2011-12-20T00:00:00 and yet rising 2011-12-20T00:00:03 *** Antimony has joined #aichallenge 2011-12-20T00:00:16 5 days till christmas 2011-12-20T00:00:17 ugh 2011-12-20T00:00:30 My bot just suicided a bunch of ants in a battle http://aichallenge.org/visualizer.php?game=284265&user=3845 2011-12-20T00:00:34 wow, that bot was very lucky in the last game 2011-12-20T00:03:39 oh wow that's a crap map 2011-12-20T00:03:59 lol, my bot is good on those 2011-12-20T00:04:04 so I'm looking through the rankings for low ranking bots that have won most of their games and I found this one ranked at 1705 with 3 wins and a 2nd place: http://aichallenge.org/profile.php?user=1784 2011-12-20T00:04:19 thestinger: maps where there's a chokepoint between the two bases at nearly minimum distance? 2011-12-20T00:04:23 because it cares so much about gathering food 2011-12-20T00:04:28 yeah, there were some of those on tcp 2011-12-20T00:04:46 remember there was one where I only won because I got an ant out? 2011-12-20T00:05:01 I suck at mazes like that though 2011-12-20T00:05:07 it's hard to say if my bot would be good at that 2011-12-20T00:05:24 it might just clog the corridor 2011-12-20T00:05:34 and here's one with 3 firsts and a second ranked 2154: http://aichallenge.org/profile.php?user=13537 2011-12-20T00:06:16 thestinger: but that whole map is literally a race for that 1st ant 2011-12-20T00:06:28 BenJackson: yeah, that's the part my bot is good at 2011-12-20T00:06:34 mine probably is too 2011-12-20T00:06:42 but it really only tests that one thing 2011-12-20T00:06:56 the first foragers will never go backwards or waste time forming lines to block the enemy 2011-12-20T00:07:10 "waste time" 2011-12-20T00:07:15 can you even get by? 2011-12-20T00:07:19 without cooperation? 2011-12-20T00:07:27 it's not wide enough 2011-12-20T00:07:31 don't underestimate a coward :P 2011-12-20T00:07:41 well, I dunno 2011-12-20T00:08:00 *** rebelxt has quit IRC (Ping timeout: 258 seconds) 2011-12-20T00:08:06 well, both ants can arrive at the exit and get out safely 2011-12-20T00:08:31 oh nvm 2011-12-20T00:08:35 it's closer than I thought 2011-12-20T00:08:52 *** avdg has quit IRC (Quit: Leaving.) 2011-12-20T00:09:00 I don't think the finals maps should have been left totally to rand 2011-12-20T00:09:03 has ChrisH posed a writeup of his bot yet? 2011-12-20T00:09:14 *** Warren has joined #aichallenge 2011-12-20T00:09:31 BenJackson: yeah, that's a really weird map 2011-12-20T00:09:38 the other ones like that usually have a bigger exit 2011-12-20T00:09:40 *** Warren is now known as Guest74558 2011-12-20T00:10:50 here is a bot at 2725 that has never lost (tied for 1st once) http://aichallenge.org/profile.php?user=12561 2011-12-20T00:11:26 haha wow 2011-12-20T00:11:54 greentea enumerates moves exhaustively in battle subgroups when theyre small enough 2011-12-20T00:12:27 a1k0n: for each enemy and himself? 2011-12-20T00:12:48 it looks like just his own? unsure, but no more than 7 ants 2011-12-20T00:13:08 which is reasonable 2011-12-20T00:13:21 mine will enumerate moves exhaustively if given enough time 2011-12-20T00:13:31 for itself. it only considers 5 moves for the enemy 2011-12-20T00:13:38 both enemy and his own 2011-12-20T00:13:55 *** Paradoxial has quit IRC (Quit: :) 2011-12-20T00:14:50 <_flag> Anyone know of any good way I can create animated gifs displaying my algorithmns without writing all the code myself? 2011-12-20T00:14:56 a1k0n: I do that as well, I can handle battles up to around 5 v 6, but bigger if the group is tighly packed or lots of water nearby 2011-12-20T00:15:09 <_flag> I asked this question earlier but went to sleep so I'm not sure if anyone responded 2011-12-20T00:15:25 i used perl gd for mine 2011-12-20T00:15:35 i forgot how i made the actual gif tho 2011-12-20T00:15:43 a1k0n: but I do it for both sides 2011-12-20T00:15:48 _flag: if you use the visualizer overlay stuff, you can also output to video 2011-12-20T00:15:51 he does too 2011-12-20T00:16:03 a1k0n: oh, misunderstood above comment then 2011-12-20T00:16:14 i reneged after rereading 2011-12-20T00:16:20 his code 2011-12-20T00:16:29 <_flag> McLeopold: visualizer overlay stuff? You mean just the visualizer? 2011-12-20T00:16:34 I'm curious about how bots using the memetix-style combat will do 2011-12-20T00:16:41 I bet a lot of people implemented that 2011-12-20T00:16:46 _flag: yes, have you seen my replays on paste? 2011-12-20T00:16:51 yeah, that was the downside of posting it 2011-12-20T00:16:55 I came up with a formula that would quickly estimate how many positions a battle could produce 2011-12-20T00:17:01 <_flag> McLeopold: No 2011-12-20T00:17:06 I think the variety of other combat is more interesting 2011-12-20T00:17:09 oh nice 2011-12-20T00:17:40 a1k0n: I plan to explain it in my writeup 2011-12-20T00:17:43 yeah if you count redundant positions only once you get a lot more mileage 2011-12-20T00:17:52 a1k0n: yes 2011-12-20T00:18:04 shit i should have added a cache for that 2011-12-20T00:18:11 did anyone else take a constraint satisfaction approach? 2011-12-20T00:18:15 <_flag> McLeopold: Can you link me to it/them? 2011-12-20T00:18:20 _flag: looking... 2011-12-20T00:18:44 a1k0n: and I also reshuffle the order I test enemy positions to move the best ones so far to the top of the list to enhance A-B pruning's efficiency 2011-12-20T00:19:04 _flag: https://github.com/j-h-a/aichallenge/blob/vis_overlay/VIS_OVERLAY.md 2011-12-20T00:19:21 BenJackson: I included the memetix-style stuff in my bot, and it helped a bit 2011-12-20T00:19:33 greentea orders moves too 2011-12-20T00:19:47 it let me stop generating more than one move for the opponent without ruining my combat 2011-12-20T00:19:50 *** inutard has joined #aichallenge 2011-12-20T00:19:54 so it made it a bit more aggressive 2011-12-20T00:20:12 <_flag> McLeopold: Very cool! Thanks 2011-12-20T00:20:30 what's .md? 2011-12-20T00:20:36 markdown 2011-12-20T00:20:40 @rankings 2011-12-20T00:20:41 a1k0n: Top 10 players: strcat(74.0), carlos.guia(69.5), delineate(69.5), roumenpopov(68.9), T_and_B(68.7), bombom(68.6), crispy(68.4), carpcrap(68.2), oldman(67.8), sivsoft(67.7) 2011-12-20T00:20:46 BenJackson: markdown, github uses it for markup 2011-12-20T00:20:47 wtg strcat 2011-12-20T00:21:00 strcat has a chance of holding #1 for a while 2011-12-20T00:21:02 unlike other pretenders 2011-12-20T00:21:14 i still have 4 games 2011-12-20T00:21:15 until I face one of the people better than me and get last :P 2011-12-20T00:21:34 <_flag> 4 games here as well 2011-12-20T00:22:02 *** Areks has joined #aichallenge 2011-12-20T00:22:12 _flag: http://paste.aichallenge.org/Ek04K/ 2011-12-20T00:22:37 turn on visualizations for red and you can see my food gathering logic, plus some combat logic 2011-12-20T00:22:42 a lot of bots still have 4, and I've seen one that has only 3 2011-12-20T00:23:58 Or shine time will still come. I hope :) 2011-12-20T00:24:03 *Our 2011-12-20T00:24:06 *** inutard has quit IRC (Ping timeout: 258 seconds) 2011-12-20T00:24:29 <_flag> McLeopold: That looks like it should be perfect for my write up, and possibly some Youtube videos as well :) 2011-12-20T00:24:53 _flag: talk to amstan about videos 2011-12-20T00:25:12 he and mleise worked something out 2011-12-20T00:25:16 <_flag> amstan: Is he looking for something to use as a promotion or something? 2011-12-20T00:25:16 from the visualizer 2011-12-20T00:25:21 <_flag> Oh ok 2011-12-20T00:25:28 <_flag> You mean the auto-video thing 2011-12-20T00:25:31 yeah 2011-12-20T00:25:45 used it for the annoucement 2011-12-20T00:25:52 <_flag> Yeah, that would probably be better than a screen capture, I'll ask him 2011-12-20T00:25:57 well 2011-12-20T00:26:02 idk, depends on aarossig 2011-12-20T00:26:11 he recorded the launchvideo 2011-12-20T00:26:50 <_flag> Do you know how he did it? 2011-12-20T00:27:06 which part? 2011-12-20T00:27:16 the talking part was easy, he had a mic and just talked 2011-12-20T00:27:23 <_flag> Recording the video 2011-12-20T00:27:29 i prepared the visualizer video for him 2011-12-20T00:27:30 <_flag> Was it just a screen capture? 2011-12-20T00:27:40 no, it's something that the java visualizer does 2011-12-20T00:27:48 *** sashaSochka has joined #aichallenge 2011-12-20T00:27:52 it can output all the frames as images 2011-12-20T00:27:59 then you use ffmpeg to make the movie 2011-12-20T00:28:13 <_flag> Ah, but the java visualizer can't do the debugging thing that McLeopold was talking about, right? 2011-12-20T00:28:31 ok, this profile looks funny: http://aichallenge.org/profile.php?user=2507 2011-12-20T00:28:44 _flag: i think it can, you just need to compile it 2011-12-20T00:28:58 _flag: the guy that added the debuggings forked the visualizer, it should still have all the video features 2011-12-20T00:29:18 ChrisH_: how so? 2011-12-20T00:29:24 supposedly he has only played 4 matches, but the fifth in the list looks to me like it must hav been during finals, and prior to that the sixth game looks like he was ranked #101 before finals started 2011-12-20T00:29:29 <_flag> amstan: Okay, I'll look into it 2011-12-20T00:29:31 *** jstrong has quit IRC (Quit: Leaving.) 2011-12-20T00:30:05 and now he's currently ranked #3012 2011-12-20T00:30:26 _flag: here's my last command that i have related to this: java -jar ../visualizer.jar --width=1280 --height=740 --decorated=false --video=3,png ../13821 2011-12-20T00:30:53 amstan: yeah, looks like he didn't get credit for one of his games 2011-12-20T00:31:04 _flag: and then i did: ffmpeg -i %08d.png -vcodec libx264 -vpre lossless_max movie.mp4 2011-12-20T00:31:23 ChrisH_: the third from the top? 2011-12-20T00:31:24 ChrisH_: tell janzert about it 2011-12-20T00:31:26 amstan: the 3rd one down looks like it got dropped 2011-12-20T00:31:42 janzert: this profile looks like it didn't get credit for a game: http://aichallenge.org/profile.php?user=2507 2011-12-20T00:32:06 i wonder how many more of these there are floating around 2011-12-20T00:32:24 i think they have a query to catch them 2011-12-20T00:32:32 *** sashaSochka has quit IRC (Client Quit) 2011-12-20T00:33:12 it would be a shame for a ~100 bot to miss the cutoff because of a bad first game and then not getting credit for their best win 2011-12-20T00:33:41 you could say the same about a game that was lost and not credited 2011-12-20T00:33:46 yes 2011-12-20T00:34:04 it's not about people not getting points for winning, it's about introducing noise in the rankings 2011-12-20T00:34:09 and the irony for this particular bot is that the first game it lost was to one of the bots it beat in it's last game before finals 2011-12-20T00:34:53 amstan: I think it's about both 2011-12-20T00:34:55 that bot could have updated 2011-12-20T00:34:58 the starter bots could just be dropped by comparing files 2011-12-20T00:35:07 thestinger: compare them to what? 2011-12-20T00:35:29 amstan: you mean the one he lost to in the first game? 2011-12-20T00:35:39 ChrisH_: maybe, idk 2011-12-20T00:35:52 amstan: well, I know that bot's owner personally and it wasn't updated 2011-12-20T00:36:27 what i'm saying is that you could have had the case where you lost a game and it didn't count as well 2011-12-20T00:36:39 but I would say that with the limited amount of games most bots are getting right now, not getting points for a win hurts a lot 2011-12-20T00:36:49 amstan: yeah, I agree with that 2011-12-20T00:36:50 so not getting counted for the win happens only half the time if it does 2011-12-20T00:37:20 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-12-20T00:37:23 amstan: well, you could do an sha1sum for each bot and drop bots below a cutoff rank if they have a duplicate upload 2011-12-20T00:37:29 amstan: I'm not sure I'm following you. 2011-12-20T00:37:41 then people who coded something unique (not a starter bot) can still play games, even if they aren't ranked high 2011-12-20T00:37:56 I bet almost all of the bots were just the starter zips reuploaded 2011-12-20T00:38:12 thestinger: they'll get killed after the cutoffs anyway 2011-12-20T00:38:14 thestinger: define "almost all" 2011-12-20T00:38:27 ChrisH_: below the cutoff I mean 2011-12-20T00:38:28 thestinger: the ammount of work to do that is not worth it 2011-12-20T00:38:51 ChrisH_: yeah, unfortunately there were still a few games that didn't get credited completely properly 2011-12-20T00:39:30 can you eliminate any submissions with md5 in map `md5` starterpacks 2011-12-20T00:39:33 janzert: is the cleanup effort still ongoing? 2011-12-20T00:40:21 not cleanup as such, but stopping it from occurring any more 2011-12-20T00:40:21 *** Scryer has joined #aichallenge 2011-12-20T00:40:31 thestinger: true, you could drop ANY bots where the sha matches more than one user 2011-12-20T00:40:40 since it would be illegal if it were a real user anyway 2011-12-20T00:40:45 the latest round of tries on that is looking promising at the moment, but so did the last two :/ 2011-12-20T00:41:00 janzert: is it the case that a dropped game is dropped for all bots in the game? 2011-12-20T00:41:02 so I'm trying to keep my hopes down 2011-12-20T00:41:17 it varied 2011-12-20T00:41:20 BenJackson: well, there could be an exception to that (but I doubt it) 2011-12-20T00:41:32 many of them we were able to fix up btw 2011-12-20T00:41:41 huh, I don't see how the current play rate/queue pos translates into my next game being out 2155 minutes 2011-12-20T00:41:45 by my math more like 160 2011-12-20T00:42:11 BenJackson: yeah, those numbers have seemed out of whack ever since finals started 2011-12-20T00:42:12 I need to hit one of those multi-game streaks soon 2011-12-20T00:42:25 *** jack_ has quit IRC (Quit: Page closed) 2011-12-20T00:42:27 ChrisH_: they were fine for my last run through the queue 2011-12-20T00:43:02 BenJackson: the seem better as you get closer to the front of the queue, but when you are in the bottom half they seem very inflated 2011-12-20T00:43:25 BenJackson: but that seems like a minor quible at this point 2011-12-20T00:43:35 http://aichallenge.org/profile.php?user=6158 <- starter bot with sigma barely over 2 2011-12-20T00:43:41 mine just changed from 7k to 1.6k mins after a refresh 2011-12-20T00:43:50 it's been maybe an hour :) 2011-12-20T00:43:56 37 skill seems high for a starter bot, the mus are going to be packed pretty tight on top if bots like that anchor it 2011-12-20T00:44:01 *** amstan has quit IRC (Remote host closed the connection) 2011-12-20T00:44:19 *** amstan has joined #aichallenge 2011-12-20T00:44:19 *** ChanServ sets mode: +o amstan 2011-12-20T00:44:57 BenJackson: that's about what they were running before finals 2011-12-20T00:45:05 the starter bots remind me of sci-fi laser weaponry 2011-12-20T00:45:20 where you see the slow beam of light moving :P 2011-12-20T00:45:32 this map was practically designed to put noise in the starter bot games: http://aichallenge.org/visualizer.php?game=282891&user=6158 2011-12-20T00:46:07 poor antiboty could have had free points with starter behavior 2011-12-20T00:46:49 microsoft made no profit from anyone's zunes, lol 2011-12-20T00:47:33 anyway the profile I linked above is an example of a pretty good bot (~100 before finals) that seems to have some kind of combat skill when you watch it's matches, yet is currently ranked ~3000 due to the pairings it got 2011-12-20T00:49:22 which one? 2011-12-20T00:49:35 http://aichallenge.org/profile.php?user=2507 2011-12-20T00:49:42 yeah, that one 2011-12-20T00:49:59 82 mu before the skill reset 2011-12-20T00:51:11 ah, he happened to get a 1st game with non-zombie opponents 2011-12-20T00:51:12 and it's first match in finals it got in a big game with two other decent bots, then it came in second by being late to hills in another game, won a game decisively that it didn't get credit for and won two more small games against poor competition for little mu benefit 2011-12-20T00:51:19 until xytop timed out anyway 2011-12-20T00:51:50 ChrisH_: and that "no credit" is important retroactively because it would have gotten stronger matchups 2011-12-20T00:51:57 exactly 2011-12-20T00:52:20 let's hope he doesn't know how to use irc 2011-12-20T00:52:47 there's a nice webclient link on the homepage lol 2011-12-20T00:53:15 so it is very possible to have a good bot be very low in the ranks still, and I think we need a lot more games to sort this out, but getting credit for all the games we play is important too, so I hope they get that fixed soon 2011-12-20T00:53:42 ChrisH_: there's probably some argument for how this is worse for mid tier bots 2011-12-20T00:53:51 more mid tier opponents you might have randomly met early 2011-12-20T00:53:55 and you're less dominant 2011-12-20T00:54:01 absolutely 2011-12-20T00:54:16 whereas a top 10 player with a "tough" early matchup only has a 10/8000 chance of it being a loss anyway 2011-12-20T00:55:07 by that measure I was going to poitn at nhaehnle who I beat in my 4th game, but his rank is decent 2011-12-20T00:56:09 I wonder if it is possible to write a query that would find the currently lowest ranking bot that was ever in the top 200 prior to finals 2011-12-20T00:56:45 (we really need a one word name for the period of time between beta and finals) 2011-12-20T00:57:56 the result of that query would be very interesting I think and informative about possible cutoff ranges 2011-12-20T00:57:57 group stage 2011-12-20T00:58:12 xxxx not one word 2011-12-20T00:58:35 nhaehnle has gotten some good games since we played 2011-12-20T00:58:38 and he's 20th 2011-12-20T00:58:38 hey, there are more workers going now 2011-12-20T00:59:11 what does the % fail rate actually mean? 2011-12-20T00:59:14 http://aichallenge.org/visualizer.php?game=285546&user=12357 2011-12-20T00:59:20 good, because it's been 11 hours since I had a game and I've still got 1673 people ahead of me 2011-12-20T00:59:21 he has nice defensive saves in that one 2011-12-20T00:59:41 *** JorgeB has quit IRC (Quit: Computer has gone to sleep.) 2011-12-20T01:01:00 cfg: unix-y environment 2011-12-20T01:01:10 ^ lol. 2011-12-20T01:02:24 *** rofer has quit IRC (Ping timeout: 268 seconds) 2011-12-20T01:04:36 aichallenge: janzert epsilon * rf14f5ad / website/api_game_result.php : Add big lock around the whole game insert - http://git.io/DGMNPA 2011-12-20T01:04:44 contestbot: later tell janzert when you have some time could you answer this? http://forums.aichallenge.org/viewtopic.php?f=27&t=2183#p13545 2011-12-20T01:04:44 amstan: I think that worked... 2011-12-20T01:11:34 *** rebelxt has joined #aichallenge 2011-12-20T01:12:53 *** UncleVasya has quit IRC (Ping timeout: 252 seconds) 2011-12-20T01:13:24 *** rajanaresh has joined #aichallenge 2011-12-20T01:18:55 *** rofer has joined #aichallenge 2011-12-20T01:19:28 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-12-20T01:20:40 janzert: here is another profile that may have a problem: http://aichallenge.org/profile.php?user=2010 2011-12-20T01:21:23 janzert: the ranking show it having 3 games, but the game history looks like 4 games in finals, but I don't see anything weird with the skill or mu sequence 2011-12-20T01:21:58 janzert: also it was ranking around #300 prior to finals and is currently at #3591, a data point for cutoffs 2011-12-20T01:22:00 yeah, some of the profiles we fixed up will have the game count off by one 2011-12-20T01:22:41 *** jim_ has joined #aichallenge 2011-12-20T01:23:01 I suppose I could probably go back and fix that as well but it's pretty much cosmetic 2011-12-20T01:23:23 *** Varan has joined #aichallenge 2011-12-20T01:23:45 attempt #2 at building Rust 2011-12-20T01:23:53 *** _ANTi_ <_ANTi_!63e751ed@gateway/web/freenode/ip.99.231.81.237> has joined #aichallenge 2011-12-20T01:24:13 *** Scryer has quit IRC (Quit: Page closed) 2011-12-20T01:24:55 *** jim_ has quit IRC (Client Quit) 2011-12-20T01:25:38 *** rajanaresh has left #aichallenge 2011-12-20T01:29:02 poor EvgeniyS has had four games in finals and lost to bots that were in the top 50, 20, 400, and 400 prior to finals in each of those games, that's a tough schedule and left him with a rank ~3600 now vs. ~300 prior to finals 2011-12-20T01:29:21 (second one there should be 200 not 20 2011-12-20T01:30:19 ChrisH_: are you finding these by looking at who the top players played early? 2011-12-20T01:30:24 *** lordofall has joined #aichallenge 2011-12-20T01:30:34 man so many timeouts 2011-12-20T01:30:45 troll amounts of server load 2011-12-20T01:31:07 no, I'm looking for bots with more than one submission and only 4 games then looking at game history and scrolling to the bottom to see their old rank 2011-12-20T01:31:08 *** bhasker has joined #aichallenge 2011-12-20T01:31:42 I wish I could do some queries on the DB. 2011-12-20T01:32:00 *** f4hy has joined #aichallenge 2011-12-20T01:32:01 *** smjms has quit IRC (Quit: Konversation terminated!) 2011-12-20T01:32:17 but I don't have access and don't know what the schema looks like anyway 2011-12-20T01:33:17 So clearly the match maker doesnt choose those with less matches first 2011-12-20T01:33:27 janzert also wishes he could do queries on the db ;-) 2011-12-20T01:35:01 that's not too much of a joke at times since the start of finals, the site has certainly never had this much of a load. I suppose it's good since it means lots of people are interested in the outcome. 2011-12-20T01:35:17 janzert: and reddit 2011-12-20T01:35:26 *** Scryer has joined #aichallenge 2011-12-20T01:35:45 next time no reddit post till finals end :P 2011-12-20T01:36:01 sure 2011-12-20T01:36:02 lol 2011-12-20T01:36:16 Do you really think reddit users who are not competing are checking th site often? 2011-12-20T01:36:31 f4hy: no, the match maker is trying to "sort" the bots, so it would prefer to have a bunch that play more matches to act as landmarks in the rankings I think 2011-12-20T01:37:11 ChrisH_: oh... interesting. So I didn't get picked as a landmark, so I am stuck at 4 matches? 2011-12-20T01:37:22 and I *think* the code that tries to level the number of games over a 24 hour time period didn't kick in for the first day 2011-12-20T01:37:28 ChrisH_: that's a pretty good summary of what the trueskill portion is trying to do 2011-12-20T01:38:03 if I give the matchmaker a dollar, can it choose me as a landmark ;-) 2011-12-20T01:38:13 *** Twirlip has joined #aichallenge 2011-12-20T01:38:29 janzert: I've written a system to let people rank songs by comparing them to each other and indicating which they like better, so I kinda know how these things have to "think" internally 2011-12-20T01:39:10 so you're drawing some comparison between starter bots and nickleback 2011-12-20T01:39:30 f4hy: in addition, if you are a really good bot or a really bad bot you are less likely to get picked as a landmark because winning all your matches or losing them all puts you on the ends of the scale and that's not helpful for sorting other bots 2011-12-20T01:39:44 you will notice that all the bots that get a lot of games have won some and lost some 2011-12-20T01:39:46 ChrisH_: but I am neither of those! 2011-12-20T01:39:48 *** rebelxt has quit IRC (Quit: Page closed) 2011-12-20T01:40:11 *** Twirlip has quit IRC (Client Quit) 2011-12-20T01:40:12 also all the mu in the middle are very bunched up still 2011-12-20T01:40:38 and then not everyone will be a landmark, and landmarks will then get emphasized over other bots because they are already a stronger comparison, so it will snowball 2011-12-20T01:40:59 think of quicksort and how it picks a single pivot on each pass, trueskill just has more pivots 2011-12-20T01:41:42 That is fair. I just hope I get more matches 2011-12-20T01:42:20 you know, just like the data in a list wishes it gets to be the pivot 2011-12-20T01:42:36 wow, Rust seems to have the worst build system ever 2011-12-20T01:42:37 me too, i've been waiting almost 12 hours and still have 1200 before me in the queue 2011-12-20T01:42:46 they dynamically create a bunch of .py files 2011-12-20T01:43:28 *** Palmik has joined #aichallenge 2011-12-20T01:43:43 <_flag> Any java users in here know why I might be missing the "netscape.javascript" package and can't compile the visualizer? 2011-12-20T01:43:56 BenJackson: I take it you don't like nicleback. ;) 2011-12-20T01:44:59 aichallenge: janzert epsilon * rfce1bee / manager/matchup_buffer.py : Increase pairing buffer size - http://git.io/0FdDQg 2011-12-20T01:45:30 any way to know our mu and sigma? 2011-12-20T01:46:09 LouisMartin: they're the title for the skill span (so you can see with mouseover) 2011-12-20T01:47:57 *** antimatroid-sydn has joined #aichallenge 2011-12-20T01:48:23 why so few games? 2011-12-20T01:48:23 nice 2011-12-20T01:48:24 http://aichallenge.org/profile.php?user=98 2011-12-20T01:49:03 I don't know but I wish you had -one- less game ;) 2011-12-20T01:49:15 :P 2011-12-20T01:49:56 the visualizer lags for me fairly offten, is that because I am in FF, does it run nicer in chrome? 2011-12-20T01:49:56 antimatroid-sydn: there are two reasons: 1. there have been some problems with the DB that the admins have had to deal with and the workers have not been running at full speed the whole time since finals started 2011-12-20T01:50:24 okay, well i'll be back later 2011-12-20T01:50:38 antimatroid-sydn: and 2. the matching system is still trying to establish some baseline bot skills, so matches are not being evenly distributed yet 2011-12-20T01:51:28 ChrisH_: are you certain of that, or just good guess 2011-12-20T01:51:37 Like starter bots that get 10 or 12 in a row... :P 2011-12-20T01:51:40 ChrisH_: *or* I mean that everyone should download the, like a starter kit 2011-12-20T01:52:16 *** Manwe56 has quit IRC (Quit: Page closed) 2011-12-20T01:52:39 I should read the source because I still can't imagine how my "next game in.." could be related to the adjacent stats 2011-12-20T01:52:44 f4hy: I had a very strong suspicion base on what I know from working on a similar matching and ranking system in the past, and janzert confirmed my suspicion a little ways up the chat here 2011-12-20T01:53:40 ChrisH_: I guess you could look at whether some of the high game count bots are ones that came in mid-pack in the beginning 2011-12-20T01:53:44 *** bmh has joined #aichallenge 2011-12-20T01:53:55 since win-win-win is not sorting and neither is lose-lose-lose 2011-12-20T01:55:08 at which point does the mu becomes something to consider realistically? after how many games and at which sigma? 2011-12-20T01:56:23 You can look at what sigma you had before the finals 2011-12-20T01:56:28 are those currently playing a lot of games sooner to be able to estimate their final rank? 2011-12-20T01:56:28 something like that i guess 2011-12-20T01:56:39 skill is mu-sigma*3, it's a conservative estimate of your skill (rarely overestimates) 2011-12-20T01:57:12 problem is that I never tried this version on the site and it's way better than the last I tried here which ranked ~300 after a while 2011-12-20T01:57:15 *** _ANTi_ <_ANTi_!63e751ed@gateway/web/freenode/ip.99.231.81.237> has quit IRC (Ping timeout: 258 seconds) 2011-12-20T01:58:14 before the finals xathis had a sig of 1.53 2011-12-20T01:58:36 stinger grats on rank 1 for now:) 2011-12-20T01:59:04 thestinger: have I mentioned how much I <3 haskell? 2011-12-20T01:59:23 bmh: but didn't it betray you in the end, thwarting your submission/ 2011-12-20T01:59:31 bmh: haskell is awesome 2011-12-20T01:59:35 BenJackson: No. Ulrich Drepper is the blame. 2011-12-20T01:59:42 (for everything. ever.) 2011-12-20T01:59:47 I <3 haskell. I think I am too stupid to grock it 2011-12-20T01:59:53 BenJackson: yeah, glibc can't statically link everything 2011-12-20T01:59:56 (anymore) 2011-12-20T02:00:05 I always hate IO in haskell 2011-12-20T02:00:09 f4hy: I've been playing with Haskell for three, maybe four years now, and I'm only starting to grok it. 2011-12-20T02:00:13 and ghc 7 needed a newer version of glibc than the server had 2011-12-20T02:00:46 it's true 2011-12-20T02:00:50 thestinger: Oh, #1? I wish you a very conservative estimate then ;) 2011-12-20T02:00:57 for technical reasons you basically have to dynamic link on linux now 2011-12-20T02:01:11 I read http://learnyouahaskell.com/ thought it was awesome. Coded a few tiny projects in it. but have not done anything with it in about 2 years 2011-12-20T02:01:16 BenJackson: if you use glibc :) 2011-12-20T02:01:27 if you want to make system calls 2011-12-20T02:01:42 f4hy: with the competition over, I decided to write a roguelike :) 2011-12-20T02:01:58 the kernel exposes a vdso which lets you link in platform specific syscall code 2011-12-20T02:02:07 after 25 games, do you guys think I may hang around that rank? http://aichallenge.org/profile.php?user=7025 2011-12-20T02:02:11 that's also how gettimeofday can now be done without a syscall 2011-12-20T02:02:14 *** praveen has quit IRC (Ping timeout: 258 seconds) 2011-12-20T02:02:18 which is a big win if you call it in an inner loop 2011-12-20T02:02:24 like I bet a bunch of bots do 2011-12-20T02:02:36 BenJackson: bmh's bot almost worked though 2011-12-20T02:02:38 BenJackson: my bot never queried the clock. 2011-12-20T02:02:50 just the stupid locale stuff is impossible to statically link against 2011-12-20T02:03:04 "If I can't do this in 500ms, there's no sense in wasting cycles checking the clock." 2011-12-20T02:03:24 nice, I built Rust 2011-12-20T02:03:33 just had to sed all the 'python' to 'python3' 2011-12-20T02:03:42 I mean to python2 2011-12-20T02:03:44 :P 2011-12-20T02:04:19 really annoying because they dynamically create .py files... 2011-12-20T02:05:34 no opinion? :( I'm just trying to not get excited and hoped someone could burst my bubble 2011-12-20T02:05:51 Louis how was your rank pre final? 2011-12-20T02:05:54 I put in a clock check to see whether I needed a clock check, and was aghast to see how little time I was using. Given another week or two I'd have used all that time on combat. 2011-12-20T02:06:06 unless you changed your bot dramatically i would expect you to be close to where you were prefinal 2011-12-20T02:06:21 well, I don't expect my bot to get in the top 10 at all 2011-12-20T02:06:40 I'm 80% sure it will be in the top 50, and I _hope_ it will be in the top 25 2011-12-20T02:07:15 I'd say 99% sure you're in the top 50 2011-12-20T02:07:45 based on counting off the top 10 I suspect you're outside the top 10 2011-12-20T02:07:49 BenJackson: my last upload before the finals discouraged me a bit 2011-12-20T02:07:58 I notice that no one has played a game on my server in a couple of hours. I'm probably going to take it down. 2011-12-20T02:08:06 the problem is that you had a lot of versions and no one is that familiar with 25-50 2011-12-20T02:08:13 so hard to say where 25th is these days 2011-12-20T02:08:22 yeah 2011-12-20T02:08:32 bhasker: the last version I uploaded was way better than the last one I got scored here which was going to stable at ~300. the lastest one was doing in the 400 on fluxid in the last days. 2011-12-20T02:11:33 LouisMartin: then you shouled expect about the same or lower 2011-12-20T02:14:01 in the 400 on fluxid is pretty meaningless 2011-12-20T02:14:09 not good or bad, I mean 2011-12-20T02:14:17 Louis - is the pathfinding what you were expecting in http://aichallenge.org/visualizer.php?game=282965&user=7025 where you spend a couple of hundred turns looking yearningly across the water at yellow's hive? 2011-12-20T02:14:25 there was too much junk in the db 2011-12-20T02:14:42 yeah, the burst of games on bmh's server near the end was great 2011-12-20T02:14:52 except if I improved my bot a lot after my last bot rank here. on fluxid, my top300 bot where hardly scoring any points in game, them newer version began to do that (yea!) and the last one got to rank ~400. fluxid ranks are way tougher to earn that on the contest server, so I'm hoping for a top150 position... 2011-12-20T02:15:17 BenJackson: there are a lot of bots that I've never seen on tcp, so I just wonder how many fall into the 10 to 50 range 2011-12-20T02:15:24 right 2011-12-20T02:15:33 I think the field is pretty crowded out to about 20 2011-12-20T02:15:43 to absorb the top-10 and near-top-10 we know of 2011-12-20T02:16:11 bias from the maps will probably permute things too 2011-12-20T02:16:12 Scryer: don't tell me about it :P I changed 5 lines of code 5 minutes before uploading to the contest server.... I fixed a huge problem, but cause another one. 2011-12-20T02:16:41 LouisMartin: this is why I was stressed about having to resubmit on saturday 2011-12-20T02:16:49 I didn't want to get within 24h of the deadline 2011-12-20T02:17:10 i had a better version on saturday but sadly ran into a compiler issue with 4.5 tr1 not being the same as 4.1.2:-\ 2011-12-20T02:17:18 thestinger: as I said before -- shared vm, 1/40th of a physical box :-D 2011-12-20T02:17:22 and i was busy the whole day to spend time figuring it out 2011-12-20T02:17:25 Scryer: it's not about the pathfinding. It's about not ignoring ants that are out of reach in battle... stupid bug that probably will cost me at least 100 positions if not more. I'm just in denial right now. 2011-12-20T02:17:53 http://aichallenge.org/profile.php?user=13197 mine is dropping by the minute! 2011-12-20T02:17:55 :( 2011-12-20T02:17:56 LouisMartin: But you learned a lot, right? 2011-12-20T02:18:02 will probably sit at 1000 or so 2011-12-20T02:18:54 BenJackson: Yea. I've being coding since 20 years or so. You would believe I wouldn't do such a judgement error right? Wrong. 2011-12-20T02:20:03 That said, it was worth it.. I think .. I think I gain more in the "bug exchange" that I lost, so... 2011-12-20T02:20:15 Give it a few more years. I've been coding since 1965, and decided to let it rest with 30 hours to go despite having some quick and easy ideas that "couldn't possibly go wrong." 2011-12-20T02:20:50 Scryer: Thanks for the hope. :) 2011-12-20T02:20:57 even though I didn't make any fatal mistakes (that I know of!) I still lost a lot of rigor in the final week 2011-12-20T02:21:04 not tuning as carefully 2011-12-20T02:21:13 or checking cases I didn't think I broke 2011-12-20T02:21:16 BenJackson: I did basically nothing in the past 2 weeks 2011-12-20T02:21:18 burnout basically 2011-12-20T02:21:19 I just merged some stuff 2011-12-20T02:21:23 I coded maybe 15 lines 2011-12-20T02:21:30 all in the last few hours 2011-12-20T02:21:47 I slept only 2 x 3 hours in the last 72 hours, so I think my judgement was pretty off at that time. 2011-12-20T02:22:06 well, I coded a bunch of other stuff but I didn't use any of it 2011-12-20T02:22:22 new defense, new combat, territory control stuff :P 2011-12-20T02:22:28 it all sucked 2011-12-20T02:22:48 It helped a lot that I had house guests and grandchildren running all over the house, so I *knew* I was brain-impaired. 2011-12-20T02:22:48 *** jack_ has joined #aichallenge 2011-12-20T02:22:51 *** AlliedEnvy has quit IRC (Read error: Connection reset by peer) 2011-12-20T02:22:58 amstan: I took a crack at explaining the game count issue in one of the threads since janzert confirmed my thoughts on the topic, so we'll see if people buy it or not and whether it calms them down. 2011-12-20T02:23:13 *** jack_ has quit IRC (Client Quit) 2011-12-20T02:23:30 ChrisH_: NOPE. I demand more games now 2011-12-20T02:23:57 f4hy: you will get them, for some definition of "now" 2011-12-20T02:24:20 I wish I did a lot more evolutionary experiments on tcpserver early on. Like you did with Strcat. What I've gain in 2 days of doing that was more than anything I did in weeks. 2011-12-20T02:24:55 ChrisH - I don't see how a starter bot can work well as a quicksort-like pivot. 2011-12-20T02:25:14 *** peerai has joined #aichallenge 2011-12-20T02:25:43 *** bearoff_w has joined #aichallenge 2011-12-20T02:25:47 Scryer: the matchmaking engine doesn't know it's a starter bot 2011-12-20T02:25:50 *** lordofall has quit IRC (Quit: Page closed) 2011-12-20T02:26:10 But its record should make it pretty clear. 2011-12-20T02:26:21 Scryer: and if 40% of bots are starter bots, maybe more, the odds are high that some get shoved into that job 2011-12-20T02:26:23 Scryer: there are maps where starter bots can capture hills 2011-12-20T02:26:32 some with rotational symmetry that means only some do/can 2011-12-20T02:26:37 yeah, you can remove them by patterns on certain maps 2011-12-20T02:26:55 the system is going to try and sort all the starter bots too, so some pivots need to be down in the lower ranges as well 2011-12-20T02:27:07 that is until the cutoffs get turned on 2011-12-20T02:27:08 I pointed out one (of very many) on a forum thread that had a dozen games in a few minutes where none of the bots in the game was capturing anything. 2011-12-20T02:27:36 I went from place 5000 to 1400 in my last game 2011-12-20T02:27:54 and that makes it even harder on the engine because all those bots's mu's will converge to the same value, but it's still trying to figure out which ones are better 2011-12-20T02:28:31 and there are some bots that are worse than starters, so a starter pivot sorts those out as well 2011-12-20T02:28:54 I should ignore the "next game should be withing 17 hours 2011-12-20T02:28:59 the engine isn't just trying to find the top bots, it wants to find a total ordering 2011-12-20T02:29:41 All the more reason to have an efficient way to sort out the starter bots, since they'll end up with the same value in a total ordering and poison the process while they're doing it. 2011-12-20T02:29:41 *** fasteddie_ has joined #aichallenge 2011-12-20T02:29:59 if (user_bot.Rank < mean_starter_bot.Rank) SendMail(user_bot.Email, "LOL!"); 2011-12-20T02:30:07 *** bmh has left #aichallenge 2011-12-20T02:30:09 Scryer: that may be true, but I don't think TrueSkill was designed with that in mind. 2011-12-20T02:30:11 The majority of the starter bot games are all co-equal #1. 2011-12-20T02:30:47 Here's one more point though, a starter bot pivot is a good measuring stick for deciding if a bot is better than starters or not. 2011-12-20T02:30:52 some starter bots get eliminated in a game earler as others. Yester I saw after 3 games a all starterbot game with all players ranked around 1k 2011-12-20T02:30:52 we should have an IRC REPL-bot so we could have coding fun together 2011-12-20T02:31:10 *** Corgano has joined #aichallenge 2011-12-20T02:32:31 Guess I'll hit the sack in the happy expectation that all pivots will be picked and the games will even out, xathis will be on top, and birds will be chirping when I wake up. 2011-12-20T02:32:55 I don't think it will happen that fast. :( 2011-12-20T02:33:00 maybe GreenTea will steal 1st place 2011-12-20T02:33:03 Scryer: I wont have a game by then 2011-12-20T02:33:19 I'll try to sleep in, then. nn, all. 2011-12-20T02:33:50 *** Scryer has quit IRC (Remote host closed the connection) 2011-12-20T02:36:20 *** Corgano has quit IRC (Quit: Page closed) 2011-12-20T02:38:31 *** Westicles has quit IRC (Quit: Page closed) 2011-12-20T02:41:21 great, so i had 5 games and i'm 742 2011-12-20T02:42:37 *** Rebeli has joined #aichallenge 2011-12-20T02:46:48 ouch 2011-12-20T02:51:07 Fluxid: don't worry, I'll still make you a trophy 2011-12-20T02:51:21 you want it to say 742nd or something like "top 10%" 2011-12-20T02:52:02 <_flag> "Not a starter pack" 2011-12-20T02:52:03 ;) 2011-12-20T02:52:37 <_flag> ;) 2011-12-20T02:52:53 thestinger: in the "bots in the top 50 we don't know" category the guy right behind you now, meduza, was mid-30s before the reset 2011-12-20T02:53:55 connected to tcpants.com:2081 as Accoun12301 2011-12-20T02:53:55 INFO: game 11298 cell_maze_p02_03.map : ['Accoun12301', 'QuickBot'] 2011-12-20T02:54:17 QuickBot crashes all time and games with it not showed 2011-12-20T02:54:44 connected to tcpants.com:2081 as Accoun12301 2011-12-20T02:54:44 INFO: game 11309 cell_maze_p02_03.map : ['cseyu01', 'Accoun12301'] 2011-12-20T02:54:49 same problem 2011-12-20T02:55:30 _flag: i see you in no better situation... i hope it gets better 2011-12-20T02:56:13 <_flag> Fluxid: The server won't let me play any games :( 2011-12-20T02:56:24 http://aichallenge.org/visualizer.php?game=284515&user=2962 <- broken endgame condition 2011-12-20T02:56:25 Accoun: you could just submit your bot and you wouldn't need to whine about tcp servrs' bugs 2011-12-20T02:56:33 oh wait that's a visualizer bug 2011-12-20T02:56:45 nvm 2011-12-20T02:57:05 BenJackson: no, why 2011-12-20T02:57:23 for me it's all correct 2011-12-20T02:57:28 lone survivor 2011-12-20T02:57:51 http://aichallenge.org/visualizer.php?game=283967&user=2962 <- xophoshiy really bugged on that map 2011-12-20T02:58:15 Fluxid: final score shows 1 - 3 - +2 bonus, so itl ooks like 3-3 tie 2011-12-20T02:58:23 but really orange got -1 at the end 2011-12-20T02:58:59 i don't think he did... 2011-12-20T02:59:10 hm, he did 2011-12-20T02:59:14 on meduza's page you can see it 2011-12-20T02:59:18 right 2011-12-20T02:59:19 I forgot I already reported that same bug before 2011-12-20T02:59:32 i see it now, k 2011-12-20T03:00:12 wohoo, <400 players ahead of me 2011-12-20T03:00:15 *** Antii_ has joined #aichallenge 2011-12-20T03:00:41 6316 players are ahead. 2011-12-20T03:01:01 yeah, but you've had 5 whole games :P 2011-12-20T03:01:16 and you've had 7? 2011-12-20T03:01:18 <_flag> Some of us dream of getting to play 5 games 2011-12-20T03:01:22 me, 4 2011-12-20T03:01:47 ah, ok 2011-12-20T03:02:07 and it's been over 12 hours since my last, i'm jonesing here 2011-12-20T03:02:09 occupy aichallenge 2011-12-20T03:02:18 we're the 99% and we want more games 2011-12-20T03:02:23 lol 2011-12-20T03:02:31 well, 80% 2011-12-20T03:02:55 I just got 5 games in less than an hour 2011-12-20T03:03:18 a1k0n is up to #7, the top of the board is starting to look a little more familiar 2011-12-20T03:03:36 so one just need to wait for his turn and he'll get many games at once... similar situations happened in submission phase 2011-12-20T03:04:08 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-20T03:05:14 *** Ender_ has joined #aichallenge 2011-12-20T03:06:37 a good morning/evening to all! 2011-12-20T03:06:48 *** ccsuser has quit IRC (Quit: Page closed) 2011-12-20T03:07:41 *** epicmonkey has joined #aichallenge 2011-12-20T03:09:45 *** codetiger has joined #aichallenge 2011-12-20T03:11:09 high sigma 7.43, and lowest number of games 5... does that mean anything? 2011-12-20T03:11:34 *** Kurnevsky has joined #aichallenge 2011-12-20T03:12:09 not sure what you're asking codetiger 2011-12-20T03:12:33 my number of games is mostly lower than other bots 2011-12-20T03:12:47 but I have high sigma compared to them 2011-12-20T03:12:57 i don't think so, there are a lot of bots with only 4 or 5 games 2011-12-20T03:13:01 what's your mu? 2011-12-20T03:13:23 80 2011-12-20T03:13:29 then you are doing well so far 2011-12-20T03:13:35 *** sigh has joined #aichallenge 2011-12-20T03:13:36 :) 2011-12-20T03:13:42 *** praveen has joined #aichallenge 2011-12-20T03:14:00 codetiger : yes, some have played only 5 games so far (I think there are many) while I have seen others (few) who have already played as many as 19! hopefully things would get normalized within few hours :-) 2011-12-20T03:14:08 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-20T03:14:41 codetiger: you beat locutus in your last game, that's pretty good 2011-12-20T03:14:42 Ender_ the max I saw is 23 games when mine was only 3 2011-12-20T03:15:16 did you see my response to the post in the forums about the game rate? 2011-12-20T03:15:38 http://forums.aichallenge.org/viewtopic.php?f=27&t=2183&p=13556#p13556 2011-12-20T03:15:47 ChrisH_: Locutus... Good news then. My last few changes worked... may be 2011-12-20T03:16:51 codetiger: watching your bot it looks pretty good to me, it seems very efficient 2011-12-20T03:17:10 ChrisH_: Thanks. :) Informative post... Thanks 2011-12-20T03:17:28 I am wondering if it will reach top 100 2011-12-20T03:17:35 which it never did before 2011-12-20T03:19:25 *** epicmonkey has quit IRC (Ping timeout: 255 seconds) 2011-12-20T03:19:26 Chrish : Which would also mean that a decent bot which is also a pivot will get a better chance over others (since it will play more games), if the final contest actually stops in 2 days time. If the contest time is longer and if every body gets at least arnd 100 games, that that wouldn't be an issue at all I think. 2011-12-20T03:19:38 codetiger: hard to guess ranks now, especially with everyone making improvements towrads the end 2011-12-20T03:20:17 Ender_: I agree, for it to be fair the non-pivots need to get their game count caught up at some point 2011-12-20T03:20:53 put it would be better if that is done against other bots similar to them rather than random bots, so hopefully this early phase saves us wasted gams down the road 2011-12-20T03:21:00 put = but 2011-12-20T03:21:35 Chirsh : may be the organizers can change pivot values once in every 10-12 hours like. Let the bots which have played least number of games (like codetiger) to be a pivot for few hours :-) 2011-12-20T03:21:53 hey, i've only got 4 games. ;) 2011-12-20T03:22:22 Chrish : ah, then you too! may you and codetiger both be pivots soon! 2011-12-20T03:22:36 well, I know they have a limitter in the scheduler to try and equalize the games for bots over a 24 hour window, so hopefully that starts kicking in and levels it out 2011-12-20T03:22:37 *** Antii_ has quit IRC (Quit: Page closed) 2011-12-20T03:23:02 It would be nice if you could sort by sigma 2011-12-20T03:23:03 i should get a game here in about 15 minutes, then it's off to bed 2011-12-20T03:23:04 or mu 2011-12-20T03:23:14 i'd want to sort by mu 2011-12-20T03:23:44 And also global sort ...not just one page 2011-12-20T03:23:52 yeah 2011-12-20T03:23:58 but maybe thats a bit hard of the server :P 2011-12-20T03:24:13 in the earlier phases I have seen mu is most of time about 4.3 points more than a bot's actual rating. (there must be a good explanation in trueskill algo) 2011-12-20T03:25:00 so for all bots which have played at least about 60 games, their MU can be used pretty decently to predict their actual ratings 2011-12-20T03:25:35 Yes in deed 2011-12-20T03:25:38 indeed* 2011-12-20T03:26:42 Ender_: Well, skill = mu - sigma*3 2011-12-20T03:28:49 *** Chrysaor has joined #aichallenge 2011-12-20T03:29:49 matted : hey thanks! what I said becomes pretty childish now :-) (once the sigma gets near 1.5 or so .... ) 2011-12-20T03:29:52 *** visit2012 has quit IRC (Quit: Page closed) 2011-12-20T03:30:18 matthewd : 2011-12-20T03:30:55 Yeah... where sigma will be a function of the number of games played, but also the sigma of the opponents (I think) 2011-12-20T03:32:36 matthewd: sigma is also influenced by how consistently you bot scores. if it does well in many games but finishes last often as well then sigma will never go as low as a bot that consistently ranks in line with it's mu score 2011-12-20T03:32:58 *** katebus has joined #aichallenge 2011-12-20T03:33:03 "playing in a game right now" yay! 2011-12-20T03:33:17 ChrisH_: Ah yes, that too 2011-12-20T03:33:46 Personally, I think everything should've stopped right around here: http://forums.aichallenge.org/viewtopic.php?f=27&t=2183&sid=5f4ccd2b128d8d5af9bfcfb8f4c4409e#p13440 :) 2011-12-20T03:34:49 playing a game right now! wow, I have become a pivot this morning or what. have played only 4 games in first 18 hours or so but now 6 games in the last hour! 2011-12-20T03:36:45 *** Blkt has joined #aichallenge 2011-12-20T03:40:02 well, that was a fun game: http://aichallenge.org/visualizer.php?game=287348&user=10 2011-12-20T03:40:36 nice to have some worthwhile opponets for a change 2011-12-20T03:40:55 and i like that map a lot 2011-12-20T03:42:55 *** Manwe56 has joined #aichallenge 2011-12-20T03:44:18 *** mj41 has joined #aichallenge 2011-12-20T03:46:18 :) ChrisH_ Nice game... And the matching is now decend 2011-12-20T03:46:30 decend = decent 2011-12-20T03:46:38 yes, getting better 2011-12-20T03:46:44 ChrisH_:nice win 2011-12-20T03:46:47 thanks 2011-12-20T03:47:22 am still waiting for my 6th game 2011-12-20T03:47:39 *** fasteddie_ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T03:48:01 yeah, me too, that was only my 5th 2011-12-20T03:48:08 anyway, off to bed now 2011-12-20T03:48:14 ChrisH: you did a good job on the close-in combat there. That's one bit I never got working properly and it does deliver a big advantage 2011-12-20T03:48:20 *** weaslesun has joined #aichallenge 2011-12-20T03:48:38 Kettling: thanks, i put a lot of work into getting my combat code working well 2011-12-20T03:48:58 it required a lot of performance tuning to get it fast enough 2011-12-20T03:50:14 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-20T03:50:31 may be if profiles refreshed too many times, then matchmaking gives penalty :( 2011-12-20T03:51:39 anybody knows what's the contest plan for coming 2 days.. I mean will there be stages of cutoff of players (like in the last time) if so what would be cut off points.. etc ? 2011-12-20T03:52:09 codetiger: :-) 2011-12-20T03:52:34 Ender_: you seem to be my neighbor 2011-12-20T03:52:43 Srilanka. 2011-12-20T03:53:59 *** aerique has joined #aichallenge 2011-12-20T03:54:00 codetiger: how is your recent version compared to the one before? 2011-12-20T03:54:48 The last version I had in aichallenge server was v19 and now v24. v24 beats v19 99% of the times in most maps 2011-12-20T03:55:33 codetiger : oh, I didn't noticed. nice! did u join the competition late ? 2011-12-20T03:55:40 codetiger: oh, then not much of a chance for me :) - am praveen.seluka in the official server and I was ahead of you(India rankings) 2011-12-20T03:56:08 and tcp showed some improvements. However, overall my bot is now good at close neighbor maps than cellmaze or random 2011-12-20T03:56:47 praveen: my bot always loved only patticular type of maps 2011-12-20T03:57:00 Ender_: I joined late, yes. 3 weeks back 2011-12-20T03:57:05 hmm me and xoposhiy have decided to kill each other off early in the game :-( http://aichallenge.org/visualizer.php?game=287435&user=66 2011-12-20T03:57:08 codetiger: still, you were doing good in tcp server(tcpants) 2011-12-20T03:57:19 *** Garf has joined #aichallenge 2011-12-20T03:57:25 *** valydo has joined #aichallenge 2011-12-20T03:57:38 coz, fluxid had mostly close neighbor maps a lot 2011-12-20T03:57:57 codetiger: I can see that, thats the reason you were ranking little below because some matches you end up last 2011-12-20T03:58:37 yes. I didn't have much time to optimize for all maps 2011-12-20T03:59:10 praveen: Yours does a great job by the way. And I never saw you until 10 days back 2011-12-20T03:59:44 I assumption was either praveen or zyberkiddy might top our country chart 2011-12-20T03:59:46 Ya I started four weeks back but never really got it going. Spent 14+ days in exploration and I removed everything in the end 2011-12-20T04:00:06 I think you will end up in the top followed by me or zyberkiddy 2011-12-20T04:00:17 *** mj41 has quit IRC (Ping timeout: 252 seconds) 2011-12-20T04:00:38 I hope they put up a new game as soon as possible... I discovered this onetoo late :( 2011-12-20T04:00:39 wrote exploration(defense only) in the last week or so... 2011-12-20T04:01:18 praveen: I put 2 weeks in exploration and food gathering. And finally added Memetix method for combat in final week. Then optimized my strategies. 2011-12-20T04:01:23 I thought exploration was everything, and was working on crazy ideas which never really worked 2011-12-20T04:01:37 *** onion42 has joined #aichallenge 2011-12-20T04:02:09 I used collaborative diffusion kind of algo, but ended up moslty like A* 2011-12-20T04:02:35 But i tuned diffusion maps to work for one ant for one food... 2011-12-20T04:02:48 codetiger: ok ok I was not sure how to go about with these diffusion techniques 2011-12-20T04:02:58 have any blog post or something? 2011-12-20T04:03:30 I posted my source here https://github.com/codetiger/Ants-for-AI-challenge 2011-12-20T04:03:39 Blog post will be in few days 2011-12-20T04:03:49 ok thanks 2011-12-20T04:04:10 still only 5 games for my bot :( 2011-12-20T04:04:27 onion42: never mind. We are here with you 2011-12-20T04:04:33 :D 2011-12-20T04:04:57 the trick is to go out for a while and you will find that it has played 6 more games 2011-12-20T04:05:19 praveen: :) meanwhile, where r u from? 2011-12-20T04:06:21 UK, manchester :) 2011-12-20T04:06:31 and you? 2011-12-20T04:07:50 *** kire has joined #aichallenge 2011-12-20T04:09:18 India, Chennai 2011-12-20T04:11:45 oh dear, looks like my combat code has a nasty bug :( 2011-12-20T04:11:48 http://aichallenge.org/visualizer.php?game=286873&user=1724 2011-12-20T04:14:31 *** Varan has quit IRC (Ping timeout: 276 seconds) 2011-12-20T04:14:36 *** mj41 has joined #aichallenge 2011-12-20T04:16:46 *** jstemmer has quit IRC (Quit: leaving) 2011-12-20T04:18:07 *** sir_macelon has joined #aichallenge 2011-12-20T04:18:58 hi 2011-12-20T04:19:01 *** smiley1983 has joined #aichallenge 2011-12-20T04:25:02 *** nha has joined #aichallenge 2011-12-20T04:27:18 *** sparkly_princess has quit IRC (Quit: Page closed) 2011-12-20T04:27:31 *** katebus has quit IRC (Quit: Page closed) 2011-12-20T04:29:17 *** bearoff_w has left #aichallenge 2011-12-20T04:32:59 *** raulis has joined #aichallenge 2011-12-20T04:35:17 my bot has a bug that when defending the hill it goes away to collect the nearby food, leaving the hill to the enemy. wtf! lost too many hills already with this bug. 2011-12-20T04:36:20 *** weaslesun has quit IRC (Ping timeout: 258 seconds) 2011-12-20T04:36:22 if there is only one ant left to defend your hill you will lose anyway 2011-12-20T04:36:47 better get some food and hope are able to defend with 2 ants 2011-12-20T04:39:24 you might not lose if the enemy is super timid 2011-12-20T04:39:28 it has happened 2011-12-20T04:39:47 TL_: back when I had static defense the rule for having static defenders was based on ant count 2011-12-20T04:40:12 so usually it accelerated any loss because when the peripheral ants were picked off suddenly the defenders were released to gather again and boom 2011-12-20T04:40:20 now I don't have static defenders at all 2011-12-20T04:40:57 *** UncleVasya has joined #aichallenge 2011-12-20T04:41:26 I don't know why people started to use static defense anyway ;) 2011-12-20T04:42:28 BenJackson: I have static defenders that forms an X with 16 number of ants but when enemy gets close within 16 Manhattan distance they move to anticipate it. problem is when there is only 1 ant defending due and there is one enemy. my ant will leave the hill to collect the food instead of keeping the enemy at bay. 2011-12-20T04:43:04 *** locksmith3 has joined #aichallenge 2011-12-20T04:43:55 *** xyxan has joined #aichallenge 2011-12-20T04:45:20 BenJackson: the bug I've seen is that the distance to collect the food is greater than the distance for the enemy to raze the hill. so the hill is razed before another ant could spawn in the hill. 2011-12-20T04:45:37 static defense is nice protection against ninja ants 2011-12-20T04:46:17 but with good defense, it's an edge case that shouldn't matter enough to be worth the cost 2011-12-20T04:46:45 *** replore has quit IRC (Remote host closed the connection) 2011-12-20T04:47:09 I tested a lot in hill defense and finally concluded making my hill engaged rather wasting ants in defense 2011-12-20T04:47:15 sir_macelon : hi 2011-12-20T04:47:33 *** Rebeli has quit IRC (Quit: Page closed) 2011-12-20T04:47:34 *** UncleVasya has quit IRC (Ping timeout: 248 seconds) 2011-12-20T04:47:35 if one ant nears my hill I better sacrifice my ant to avoid my hill to be visible 2011-12-20T04:47:45 Ender_: hi, how is it going? 2011-12-20T04:47:56 sir_macelon : seen you after a long time.... :-) 2011-12-20T04:48:08 i have 10% of my ants in defense that goes for any ant within 16 distance of the hill 2011-12-20T04:48:09 *** kire has quit IRC (Quit: Leaving) 2011-12-20T04:48:38 *** mj41 has quit IRC (Ping timeout: 240 seconds) 2011-12-20T04:48:38 I suppose you could say all of my ants are defenders 2011-12-20T04:48:40 sir_macelon : not so good. I couldn't finish my latest bot on time (due to a bug), so had to submit a older version. 2011-12-20T04:48:41 Ender_: well, i was out for last month, busy in rl 2011-12-20T04:48:45 they just prefer to defend as far from home as possible 2011-12-20T04:48:50 so they always push ants away from my hills 2011-12-20T04:50:11 yes, that't pretty much what I do too, and I have a special rule that if I have x number of ants, the hill should always be visible 2011-12-20T04:50:12 Ender_: that's a pity 2011-12-20T04:50:26 sir_macelon : hope u manged to improve ur bot and submit a better version. BTW, ur older version was actually improving in skill in last few days b4 submission closed. I guess it was better playing against stronger opponents than against weaker ones. 2011-12-20T04:50:52 Ender_: i managed to fix some bugs over the weekend and also a couple of hours before the deadline 2011-12-20T04:51:36 Ender_: it was a crappy experiment, totally screwed exploration, cause i tried to adjust more to mazes 2011-12-20T04:51:48 *** codetiger has quit IRC (Remote host closed the connection) 2011-12-20T04:52:18 BenJackson: I should have done it like the way you do. Seems to be a better strategy. 2011-12-20T04:52:43 *** ikaros has joined #aichallenge 2011-12-20T04:52:49 Ender_: now it should do pretty well, at least i think i made all the major changes i needed 2011-12-20T04:53:32 good morning everyonbe 2011-12-20T04:53:44 I have 2 games in a row just now, and at top 25 after 7 games. Let's stop the finals! :) 2011-12-20T04:53:58 sir_macelon : ah, that's always a happy feeling compared to having a better bot at hand and looking at ur second best one performing in the finals :-) 2011-12-20T04:54:23 Blkt: good morning Blkt 2011-12-20T04:54:57 Ender_: well till the point i observe some odd behaviour 2011-12-20T04:55:47 Ender_: I haven't been able to test against best bots, so i am curious how it will fight against xathis, flag and greentea 2011-12-20T04:56:15 *** raulis has quit IRC (Quit: Page closed) 2011-12-20T04:57:18 Ender_: I am hoping to overcome them in combat, because i never get my exploration code to work as i expected 2011-12-20T04:58:48 sir_macelon: should be fun to see you vs strcat 2011-12-20T04:58:54 his theory is pretty much the reverse 2011-12-20T04:59:13 *** kire has joined #aichallenge 2011-12-20T04:59:13 explore/eat effectively, do combat ok, smother you 2011-12-20T04:59:33 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-12-20T05:00:07 wow no game since yesterday.. still only 5 :& 2011-12-20T05:00:22 *** Jak_o_Shadows has joined #aichallenge 2011-12-20T05:00:47 BenJackson: I don't remember I had any problems in handling his bot 2011-12-20T05:01:20 BenJackson: was he somewhere at the top? 2011-12-20T05:01:58 i think he was higher 2011-12-20T05:02:09 top 20 for sure 2011-12-20T05:03:09 *** antimatroid-sydn has quit IRC (Quit: Page closed) 2011-12-20T05:04:55 http://aichallenge.org/visualizer.php?game=288016&user=2107 pguillory lost here. should have razed my hill and won but the other bot discovered my hill first 2011-12-20T05:04:55 BenJackson : I too followed that strategy early on and delaying doing the combat code. It was a mistake, combat code was actually more fun (at least easier to test) to do and it require more time. how is you hill finding algorithm (SYMMETRY BREAKER) did u put that code in the final bot ? 2011-12-20T05:05:43 *** kost_ has joined #aichallenge 2011-12-20T05:06:10 Ender_: I use no symmetry detection 2011-12-20T05:06:19 when I did I found enemy hills too early 2011-12-20T05:06:26 better to find/attack them naturally 2011-12-20T05:07:07 *** Antisyd has joined #aichallenge 2011-12-20T05:07:59 i've checked my older games and he lost with my old version: http://aichallenge.org/visualizer.php?game=244927 2011-12-20T05:08:28 so I don't see there any good food gathering nor exploration 2011-12-20T05:08:46 he had so many versions in play I'm not sure what to expect 2011-12-20T05:08:47 Next game should be within 325 minutes. 2011-12-20T05:08:49 ;< 2011-12-20T05:08:52 I know he made a last minute sub 2011-12-20T05:09:00 Fluxid: better than 3250 like it was saying to many people earlier 2011-12-20T05:09:25 it was saying 2.5k when i was at home 2011-12-20T05:09:29 2 hours ago 2011-12-20T05:09:36 whatever you do, don't go home 2011-12-20T05:09:49 :) 2011-12-20T05:10:35 *** Antisyd has quit IRC (Client Quit) 2011-12-20T05:11:52 BenJackson: well anyway I'll hope I get to clash with him to see whether he improved :) 2011-12-20T05:12:30 @rankings 2011-12-20T05:12:37 BenJackson: Top 10 players: strcat(74.0), delineate(72.4), meduza(72.0), nhaehnle(70.7), oldman(70.4), prophfi(70.1), MomoBot(69.8), boring(69.7), carlos.guia(69.5), a1k0n(69.5) 2011-12-20T05:12:43 someone will have to dethrone him ;-) 2011-12-20T05:13:30 I only had like 5 games so far... 2011-12-20T05:16:27 *** Anti has joined #aichallenge 2011-12-20T05:16:45 Any news on problems yet? 2011-12-20T05:18:09 *** mj41 has joined #aichallenge 2011-12-20T05:18:28 *** ccc has joined #aichallenge 2011-12-20T05:19:41 sir_macelon: http://forums.aichallenge.org/viewtopic.php?f=21&t=2184 what do you think? 2011-12-20T05:22:10 *** UncleVasya has joined #aichallenge 2011-12-20T05:22:13 Fluxid: if you wrote a simple engine (possibly even of the form that plugs into the contest infrastructure) I'm sure you could get a few people to make test bots to see what it's like 2011-12-20T05:22:29 sir_macelon: I remember your prev version had one game against mine. probably the only game our bots have met in 1-1. this is when u had those issues in explore. 2011-12-20T05:22:33 seems like straight up a-b pruned min-max with a good heuristic and optimizations is going to win 2011-12-20T05:25:32 Benjackson: mine bot uses a straight up a-b pruned minimax with a terrible heuristic :( 2011-12-20T05:25:48 BenJackson: wasn't it the case in tron? 2011-12-20T05:25:49 onion42: I'm referring to Fluxid's idea for the next game 2011-12-20T05:26:03 so you are :) 2011-12-20T05:26:29 that game looks solvable 2011-12-20T05:27:55 well, next game i played in scholl is more similar to go 2011-12-20T05:28:13 go might be quite good 2011-12-20T05:28:17 although very hard 2011-12-20T05:28:41 lots of moves to consider 2011-12-20T05:28:57 it is much simpler than go but still game tree would be huge 2011-12-20T05:29:15 well if the game tree is big enough, then it should be ok :) 2011-12-20T05:29:55 *** pairofdice has joined #aichallenge 2011-12-20T05:30:11 http://www.instructables.com/id/Paper-and-pencil-game-Dots/ here it is! 2011-12-20T05:30:50 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-12-20T05:30:51 Fluxid: this would be loads of fun! 2011-12-20T05:30:58 sir_macelon: which one? 2011-12-20T05:31:10 soccer? 2011-12-20T05:31:46 Fluxid: yup, the one from forums 2011-12-20T05:32:02 sir_macelon: <3 2011-12-20T05:32:18 Fluxid: i've played that in school :) 2011-12-20T05:33:17 sir_macelon: that's why i asked you, i know it is popular in poland ;) 2011-12-20T05:33:42 soccer? really? 2011-12-20T05:33:48 :( 2011-12-20T05:34:17 Fluxid: :] but i don't know if the problem itself is hard enough 2011-12-20T05:34:58 i think it is slightly harder or as hard as tron. anyway, sir_macelon, di you play this or similar: http://www.instructables.com/id/Paper-and-pencil-game-Dots/ ? 2011-12-20T05:35:16 *** bearoff_w has joined #aichallenge 2011-12-20T05:35:17 Fluxid: no, never heard of it 2011-12-20T05:36:18 Fluxid: well, there's lack of maps in that idea 2011-12-20T05:36:21 http://code.google.com/p/pointsgame/ 2011-12-20T05:36:21 please review my idea about manual starter bots removing http://forums.aichallenge.org/viewtopic.php?p=13574#p13574. Is it too crazy? 2011-12-20T05:36:29 maybe only size of the pitch 2011-12-20T05:36:31 My ai for this game :) 2011-12-20T05:37:23 bearoff_w: organizers won't accept it 2011-12-20T05:37:29 You could have obstacles 2011-12-20T05:38:31 Kurnevsky: oh, so you know it! 2011-12-20T05:38:33 bearoff_w: part of te whole idea behind this contest is easy entrance for beginners, who would like to see their bot performing in the rankings, even if it is only starter 2011-12-20T05:38:44 is that ai complete? how hard it is? 2011-12-20T05:39:05 Go is too well studied FYI 2011-12-20T05:39:15 *** denes has joined #aichallenge 2011-12-20T05:39:25 No, not yet finished. But it works. 2011-12-20T05:39:29 We want something not extensively studied and exciting 2011-12-20T05:39:37 xathis' bot has played only 5 games. Mine has played 23 2011-12-20T05:39:39 :< 2011-12-20T05:39:39 sir_macelon: I understand this point, but starters really had time to see theit starter bots playing 2011-12-20T05:40:43 bearoff_w: well, having said that, they should never enter the finals, for this i would agree 2011-12-20T05:40:47 sir_macelon: and! now we can say that current owners of such bots are really not interested in competition now. 2011-12-20T05:40:50 If it's a starter bot, it's not really their bot 2011-12-20T05:41:05 I'm using my cousins iPad too, so please excuse it's corrections 2011-12-20T05:41:56 bearoff_w: yes, i agree, and so, they will be cut off anyway later 2011-12-20T05:42:38 it's essential that it can be much nore later ) 2011-12-20T05:43:17 eh, i'll write a post about it anyway 2011-12-20T05:44:20 *** smiley1983 has quit IRC (Ping timeout: 248 seconds) 2011-12-20T05:44:54 (should go away for 10 minutes, sorry if not responding) 2011-12-20T05:45:10 *** smiley1983 has joined #aichallenge 2011-12-20T05:48:21 bearoff_w: well they don't have any code to determine if a bot is a starter one. someone would have to write it 2011-12-20T05:48:39 *** xyxan has quit IRC (Quit: Page closed) 2011-12-20T05:50:34 *** onion42 has quit IRC (Quit: Page closed) 2011-12-20T05:51:28 *** Khaosza has joined #aichallenge 2011-12-20T05:51:51 *** Surya has joined #aichallenge 2011-12-20T05:52:58 *** Surya_ has joined #aichallenge 2011-12-20T05:53:20 Hi everyone 2011-12-20T05:53:52 hi 2011-12-20T05:55:26 hi, do you guys know anything about the reason that some bots played only 5-6 games while others played 5 times more? i was at the 25-30 position before the finals but right now i don't have any chance to be there if i only play 4-5 games a day. and even xathis have played only 5 games until now 2011-12-20T05:55:30 *** gustavokambara has joined #aichallenge 2011-12-20T05:57:19 denes: it should settle over time, don't worry 2011-12-20T05:57:21 denes: seems to be an artifact or the pairing system. afaik everyone will get 10 games before first cutoff at about top 2000 2011-12-20T05:57:52 *** g0llum has joined #aichallenge 2011-12-20T05:58:28 thanks! 2011-12-20T06:01:29 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-20T06:02:08 Shouldn't bots have about the same number of games at cutoff? 2011-12-20T06:02:53 I'd hope so if they cutoff by skill instead of mu 2011-12-20T06:03:35 *** bearoff_w has left #aichallenge 2011-12-20T06:03:37 *** bearoff_w has joined #aichallenge 2011-12-20T06:04:48 *** smiley1983 has quit IRC (Ping timeout: 244 seconds) 2011-12-20T06:05:40 *** smiley1983 has joined #aichallenge 2011-12-20T06:07:12 *** AlliedEnvy has joined #aichallenge 2011-12-20T06:11:42 *** flowenol has joined #aichallenge 2011-12-20T06:12:13 *** raulis has joined #aichallenge 2011-12-20T06:14:36 It would be interesting to see the rankings at the end with different orderings of games 2011-12-20T06:15:20 I'm not sure how much I like that bots with more games now had on average easier opponents for rankings 2011-12-20T06:15:45 *** Ender_ has quit IRC (Quit: Page closed) 2011-12-20T06:16:10 *** valydo has quit IRC (Ping timeout: 244 seconds) 2011-12-20T06:17:43 just having more games will already put you ahead due to lower sigma 2011-12-20T06:18:14 *** flowenol has quit IRC (Ping timeout: 244 seconds) 2011-12-20T06:18:42 Anti: But those bots will then be used to determine where everyone else sits; they are the fenceposts that determine how good you must be to get a skill of X 2011-12-20T06:19:55 So which bots get lots of games now may affect the overall skill distribution, but shouldn't cause the relative values (i.e., the ranks) to differ vs some other possible set of choices 2011-12-20T06:20:10 i hate this map: http://aichallenge.org/map.php?map=cell_maze/cell_maze_p05_29.map too many caves with dead ends, it breaks my exploration code 2011-12-20T06:20:14 (Is my very vague understanding of how the maths works out) 2011-12-20T06:21:30 *** valydo has joined #aichallenge 2011-12-20T06:23:05 *** gustavokambara has quit IRC (Quit: Page closed) 2011-12-20T06:23:12 *** Areks has joined #aichallenge 2011-12-20T06:23:12 *** Zaphus has joined #aichallenge 2011-12-20T06:24:49 *** tomb___ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T06:26:36 *** flux_w42 has joined #aichallenge 2011-12-20T06:29:59 Dang, I've played one game in the last 16 hours... and I lost it :-( I really should have tested my bots in 1-on-1 maps but I thought the contest would be about the more populated maps, never assume! 2011-12-20T06:32:18 wwwwww i want a game 2011-12-20T06:35:01 Fluxid: how many have you had ? 2011-12-20T06:35:14 5 2011-12-20T06:35:45 5 with a loss in the 5th, same as me :-) 2011-12-20T06:35:47 I had 5 too. 2011-12-20T06:36:15 How many bots will get game before you guys? 2011-12-20T06:36:44 Zaphus: that game made me sad 2011-12-20T06:36:46 3757 2011-12-20T06:36:57 6044 2011-12-20T06:37:01 2962 2011-12-20T06:37:09 225 minutes 2011-12-20T06:37:31 600 :) 2011-12-20T06:37:33 contest is meant to end in ~50 hours... I've had one game in the last 16hrs so that doesnt bode well for me recovering before the cutoffs are turned on 2011-12-20T06:38:05 ...but you're rank 690 2011-12-20T06:38:35 Zaphus: what, 50 hours? 2011-12-20T06:38:50 where did you get that information from 2011-12-20T06:39:21 Fluxid: oh, add maybe 4-5 more, got timezones confused. Front page of site says Dec 22nd, 6pm I think. No announcement in the forums though 2011-12-20T06:39:38 "Finals will end Thursday December 22nd at 7:00PM EST." 2011-12-20T06:39:54 they should turn that countdown clock back on! 2011-12-20T06:40:08 BenJackson: oh, you've got a laser cutter! 2011-12-20T06:40:30 yeah, I saw that - jealous ! 2011-12-20T06:40:49 *** flowenol has joined #aichallenge 2011-12-20T06:40:50 * Fluxid searching for something 2011-12-20T06:41:26 *** valydo has quit IRC (Ping timeout: 252 seconds) 2011-12-20T06:42:23 http://www.asahi-net.or.jp/~rh5k-isn/Puzzle/-/burr-6f.xml damn, it doesn't work 2011-12-20T06:43:50 *** bugnuts has left #aichallenge ("Leaving") 2011-12-20T06:43:59 *** bugnuts has joined #aichallenge 2011-12-20T06:44:08 ah, it deends on browser 2011-12-20T06:46:00 *** TL_ has quit IRC (Quit: Page closed) 2011-12-20T06:47:14 Hah, all my bitching last night, I finally get to my 5th game, and i'm playing sub 40 tier peopel... arse handed to me like I knew it would be 2011-12-20T06:47:26 BenJackson: for a while i was thinking about ordering laser-cut pieces for those two puzzles: http://www.asahi-net.or.jp/~rh5k-isn/Puzzle/6BoardBurr/Torture/ http://www.asahi-net.or.jp/~rh5k-isn/Puzzle/6BoardBurr/Torture/Extreme/ what do i need to know when ordering, and what material would be the best? 2011-12-20T06:47:38 *** flowenol has quit IRC (Ping timeout: 240 seconds) 2011-12-20T06:49:55 *** flowenol has joined #aichallenge 2011-12-20T06:50:22 damn 2011-12-20T06:51:58 freaking timeouts 2011-12-20T06:52:51 I found that my timeouts were caused by writing debug info to disk.... 2011-12-20T06:53:25 well im not :D 2011-12-20T06:53:44 its kinda annoying 2011-12-20T06:53:56 i cant reproduce it 2011-12-20T06:53:57 at all 2011-12-20T06:54:11 I had a weird timeout to. Here on my pc it never timed out, not even with extreme map size. Old PC 2011-12-20T06:54:36 During the finals already 1/5 of my games timed out :-( 2011-12-20T06:55:08 the version on the competition 2011-12-20T06:55:09 I tested even with turntime=400 and it worked like a charm 2011-12-20T06:55:21 only had one timeout before the competition 2011-12-20T06:55:27 now it already has 3 2011-12-20T06:55:39 every loop that can cause it 2011-12-20T06:55:43 has a safeguard 2011-12-20T06:55:46 and still 2011-12-20T06:55:58 i had that problem locally when running 8 bots 2011-12-20T06:56:10 they started taking each other cpu 2011-12-20T06:56:20 Stupid mistakes on ai class final, argh 2011-12-20T06:56:37 now the 50 hours to end sounds like joke. And will be till bots will have equal or similar games played at least 2011-12-20T06:57:20 50 hours will make sense only with cut-off starting now 2011-12-20T06:57:26 all i want is to get my rank back 2011-12-20T06:57:32 and i will be happy 2011-12-20T06:57:45 but the timeouts are not helping at all 2011-12-20T06:58:25 it would only be fair if they wait till bots will have equal or similar games played at least 2011-12-20T06:58:37 and sigma 2011-12-20T06:58:56 ending on the 22nd seems so short 2011-12-20T06:59:18 the player who gets picked is getting nearly 6 games and end up in a total 11+ games. after everybody get their 11+ cutoff should start. atleast >4000 2011-12-20T07:00:02 is there going to be a cutoff of low rank bots? 2011-12-20T07:00:10 yes there will be 2011-12-20T07:00:22 when and what is the rule 2011-12-20T07:00:27 low will be >5000 around to begin with 2011-12-20T07:00:39 its not decided yet but its there 2011-12-20T07:00:46 dont you mean less than? 2011-12-20T07:01:02 >5000 in ranking will be cut off 2011-12-20T07:01:08 >5000 will not play any more games 2011-12-20T07:01:37 it will much better if organizers do not make secrets of how they plan to do cutoffs 2011-12-20T07:01:46 ah 2011-12-20T07:01:47 nvm 2011-12-20T07:01:55 i was misinterpreting 2011-12-20T07:02:18 not enough caffeine 2011-12-20T07:02:33 *** katebus has joined #aichallenge 2011-12-20T07:02:54 bearoff_w: it would be nice indeed 2011-12-20T07:04:38 *** hanswurst has joined #aichallenge 2011-12-20T07:06:48 *** Zeta-Hoernchen has joined #aichallenge 2011-12-20T07:07:13 sup Zeta-Hoernchen how are your games 2011-12-20T07:07:52 *** Varan has joined #aichallenge 2011-12-20T07:07:52 little 2011-12-20T07:07:53 *** ademar_ has quit IRC (Remote host closed the connection) 2011-12-20T07:08:17 *** Zeta-Hoernchen is now known as Speedy_Consoles 2011-12-20T07:09:09 *** FranPeruzzi has joined #aichallenge 2011-12-20T07:10:01 Speedy_Consoles: Hi. Isn't your nick based on the name of a mouse from cartoon? 2011-12-20T07:10:25 It was something like SpeedyGonsalez 2011-12-20T07:10:49 seems like it is 2011-12-20T07:11:11 *** hanswurst has quit IRC (Quit: Page closed) 2011-12-20T07:12:11 approx. when will be the next ai challege? 2011-12-20T07:13:00 the question is WHAT it will be, i think 2011-12-20T07:13:05 time comes after 2011-12-20T07:13:52 i dont know but im afraid to know 2011-12-20T07:13:58 Speedy Gonzales is the fastest mouse in all Mexico, Speedy Consoles the fastest mouse at the computer 2011-12-20T07:13:58 i got too addicted to this one 2011-12-20T07:14:02 Yes Fluxid, how interesting the problem matters more 2011-12-20T07:14:10 praveen: In March I came here with this question. guys said in 2-3 weeks. :) 2011-12-20T07:14:24 Mouse in the Console 2011-12-20T07:14:28 Really, ants supposed to be in Spring, then in Summer... 2011-12-20T07:14:33 So nobody knows. 2011-12-20T07:15:02 the wordplay is not perfect :P 2011-12-20T07:15:11 UncleVasya: ok no expectations then 2011-12-20T07:15:16 praveen: there are some ideas on forums 2011-12-20T07:15:26 peerai: yah, writing to disk is pretty costly, however you can be disqualified for writing files 2011-12-20T07:15:27 also, who proposed ants first? 2011-12-20T07:15:32 As was said later, this is the way how non-profit projects work :) 2011-12-20T07:15:34 Fluxid: I saw few of them, like soccer and bomberman 2011-12-20T07:15:46 praveen: what do you think of soccer? ;) 2011-12-20T07:15:48 *** olexs has joined #aichallenge 2011-12-20T07:15:57 What kind of soccer 2011-12-20T07:16:12 Fluxid: soccer as in Football Manager? 2011-12-20T07:16:23 soccer as explained in post 2011-12-20T07:16:31 pairofdice: paper soccer 2011-12-20T07:16:47 *** Chrysaor has quit IRC (Quit: Page closed) 2011-12-20T07:16:51 Let me read once. 2011-12-20T07:17:15 link? 2011-12-20T07:17:20 *** ccc has quit IRC (Quit: Computer has gone to sleep.) 2011-12-20T07:17:49 http://forums.aichallenge.org/viewforum.php?f=21 2011-12-20T07:17:56 thanks 2011-12-20T07:19:58 Looks more like Tron to be honest 2011-12-20T07:21:30 *** grwip has joined #aichallenge 2011-12-20T07:24:49 käsekästchen 2011-12-20T07:26:03 watch out here comes xathis 340 -> 61 2011-12-20T07:26:43 ehh 2011-12-20T07:26:53 he passed me 2011-12-20T07:27:01 but he has one more game than me 2011-12-20T07:28:40 Next game should be within 182 minutes. 2011-12-20T07:28:42 ;_; 2011-12-20T07:29:42 243 minutes for me 2011-12-20T07:30:27 my rank after 5 games is higher than Xathis' was after 5 - that should count for something :-) 2011-12-20T07:30:51 'watch out' means 'run away from top 100 as fast as you can - xathis is there!' 2011-12-20T07:30:59 except that I can't read... and he was higher than me :-) 2011-12-20T07:31:20 Rank doesn't say much at I think 2011-12-20T07:31:58 Rank is only meaningful within a few thousand at the moment ;) 2011-12-20T07:32:11 :P 2011-12-20T07:32:29 seems wrong that the initial games ran about 1 every 2 hours or so (for each player) and now the matchmaking is turned on it's one every 10-12 hours for most people 2011-12-20T07:32:40 So all of the first 2000 should consider to have the same rank at the moment :P 2011-12-20T07:32:51 considered 2011-12-20T07:33:09 there are probably still stater bots in the top 2k 2011-12-20T07:34:34 You could make a cool picture of all the contestants sigma in one axis and mu in the other :P 2011-12-20T07:34:40 or something like that 2011-12-20T07:35:09 *** codetiger has joined #aichallenge 2011-12-20T07:35:40 xathis played his 6th game and mine didnt so my 6th game is not with xathis... happy 2011-12-20T07:36:59 yay! playing my 6th game right now :-) Thumbs up! 2011-12-20T07:37:11 sort the first page by games from low to high :P 2011-12-20T07:38:31 *** flux_w42 has quit IRC (Quit: Page closed) 2011-12-20T07:38:50 Speedy_Consoles: I just saw you on top 2011-12-20T07:38:58 yeeee 2011-12-20T07:39:21 48 mins for my 6th game 2011-12-20T07:39:52 The current game rate is about 12.2 games per minute. 2011-12-20T07:39:53 why? 2011-12-20T07:40:13 Why not? 2011-12-20T07:40:13 y u no bigger? 2011-12-20T07:40:32 because i want games! :( 2011-12-20T07:41:16 After one or two games with TrueSkill matching the ranks are a lot more accurate than ~2000 2011-12-20T07:41:25 *** TL_ has joined #aichallenge 2011-12-20T07:43:15 *** Speedy_Consoles has quit IRC (Quit: KVIrc 4.0.1 Insomnia http://www.kvirc.net/) 2011-12-20T07:47:02 pairofdice: deresh ranked 1700-1800 at close of open competition, still at 7804 at the moment. Admittedly still waiting on a second trueskill paired game 2011-12-20T07:49:01 *** mviel__ has joined #aichallenge 2011-12-20T07:49:33 *** ant_randomnumber has joined #aichallenge 2011-12-20T07:50:04 dang! 3rd straight loss in 10 games. all last places. all cell mazes with caves. my bot likes open spaces a lot. exploration might have been optimized for randomwalk and mazes. 2011-12-20T07:51:31 hey, does anyone know of anything similar to aichallenge? i kinda got hooked and now it's hard to wait for the next one... :D 2011-12-20T07:52:26 *** Khaosza has quit IRC (Quit: Page closed) 2011-12-20T07:52:27 *** mviel_ has quit IRC (Ping timeout: 240 seconds) 2011-12-20T07:53:09 ant_randomnumber: try google for ARIMAA (I'm not sure it is running now, but maybe) 2011-12-20T07:53:21 ant_randomnumber: so you can do the previous tasks 2011-12-20T07:53:28 but without competition it will not be so fun 2011-12-20T07:53:58 Also look at the robocode (or its clone Tank AI) but they are a lot different from aichallenge. 2011-12-20T07:54:13 thanks... but yeah, it's no fun without the competiton, otherwise i would just continue working on my ant code 2011-12-20T07:54:16 and there is a startcraft AI ;-) 2011-12-20T07:54:26 Computer championship registration is currently open, but a bot for that usually takes more than a month to write 2011-12-20T07:54:33 for arimaa that is 2011-12-20T07:54:43 Also I saw some competition at gomoku that seems to go every year. 2011-12-20T07:54:47 *** u_ has joined #aichallenge 2011-12-20T07:54:51 ok, thanks everyone 2011-12-20T07:55:23 ant_randomnumber: https://www.battlecode.org/info/ 2011-12-20T07:56:30 battlecode seems interesting 2011-12-20T07:57:47 *** goffrie_ has joined #aichallenge 2011-12-20T07:57:47 *** goffrie has quit IRC (Disconnected by services) 2011-12-20T07:57:48 *** goffrie_ is now known as goffrie 2011-12-20T07:58:21 TL_: oh, interesting, but there is not much info as far i can see 2011-12-20T07:59:15 yeah... which languages can you use? 2011-12-20T07:59:34 i mean, i'd prefer to program in c++ 2011-12-20T07:59:51 and it looks to be mostly for MIT studenst, or eventually groups of students from other universities... 2011-12-20T08:00:00 "We might allow non-MIT teams to compete in the January tournament this year. We will announce something soon." 2011-12-20T08:00:49 yeah, but they're mentioning an open tournament? 2011-12-20T08:01:02 *** raulis has quit IRC (Ping timeout: 258 seconds) 2011-12-20T08:01:14 *** codetiger has left #aichallenge 2011-12-20T08:02:58 *** pm_ has joined #aichallenge 2011-12-20T08:03:27 btw, when can we expect a new aichallenge challenge? 2011-12-20T08:03:42 should be fun -> http://www.marioai.org/ 2011-12-20T08:04:27 yup, thanks 2011-12-20T08:05:32 we have a new leader 2011-12-20T08:05:42 next ai challenge? previously it was twice a year... looking at planet wars, then tron, and now ants 2011-12-20T08:06:16 Fluxid: yes, there is not much info yet. there will be an open tournament in March based on the website. 2011-12-20T08:07:14 Next game should be within 4 minutes... 2011-12-20T08:07:23 *** dodecahedron_ has joined #aichallenge 2011-12-20T08:07:35 sir_macelon: lucky you! 2011-12-20T08:07:53 128 here, a nice round number 2011-12-20T08:08:01 (too) 2011-12-20T08:08:56 @katebus: yeah, but 2010 challenges were held in fall and winter... and, since ants are a "fall" challenge... but i guess we can't expect the next challenge to be in winter already? :( 2011-12-20T08:08:57 ant_randomnumber: User error, it's not my fault. 2011-12-20T08:12:46 *** meduza has joined #aichallenge 2011-12-20T08:13:08 And here he comes 2011-12-20T08:14:15 So everybody can say congrats to the new leader :) 2011-12-20T08:14:27 ;) 2011-12-20T08:15:00 Would be kind of cool to have a dump of the ranking table :P 2011-12-20T08:15:29 Varan: there is json 2011-12-20T08:15:33 also 2011-12-20T08:15:35 @rankings 2011-12-20T08:15:36 Fluxid: Top 10 players: meduza(75.5), a1k0n(73.5), analyst74(73.3), _Atsky_(73.1), strcat(72.9), JG.WAS(72.8), zarkon(72.6), nhaehnle(72.5), oldman(72.4), delineate(72.4) 2011-12-20T08:15:38 TL_ : the videos on the Battlecode site are interesting, commentary makes a huge difference when watching a match! 2011-12-20T08:15:38 Hmm 2011-12-20T08:15:47 where is a json api? 2011-12-20T08:16:23 Varan: i don't remember, take a look into php sources 2011-12-20T08:17:08 *** meduza has quit IRC (Ping timeout: 258 seconds) 2011-12-20T08:18:14 should be this: http://aichallenge.org/rankings_json.php 2011-12-20T08:18:20 *** yoden has quit IRC (Read error: Connection reset by peer) 2011-12-20T08:18:21 but doesn't work 2011-12-20T08:18:24 maybe disabled 2011-12-20T08:19:35 *** Zaphus has quit IRC (Quit: Page closed) 2011-12-20T08:23:37 *** yoden has joined #aichallenge 2011-12-20T08:24:56 *** bob__ has joined #aichallenge 2011-12-20T08:24:58 *** ant_randomnumber has quit IRC (Quit: Page closed) 2011-12-20T08:25:16 *** xyxan has joined #aichallenge 2011-12-20T08:25:59 Hmm the website doesn't seem to work anymore 2011-12-20T08:26:32 you broke it 2011-12-20T08:26:43 No I didn't :P 2011-12-20T08:26:51 I think that's literally true :/ 2011-12-20T08:27:08 lol? 2011-12-20T08:27:10 Ow then you did something bad in the code :P 2011-12-20T08:27:23 I have a feeling hitting the above url is quite an expensive query at this point 2011-12-20T08:27:26 trying to get json derailed whole server? 2011-12-20T08:27:33 ouch 2011-12-20T08:27:44 Only thing i did was call http://aichallenge.org/rankings_json.php .. which was a mistake becuase it should be ... http://aichallenge.org/ranking_json.php 2011-12-20T08:27:54 *** Raulis has joined #aichallenge 2011-12-20T08:27:58 since it returns json for the whole rankings 2011-12-20T08:28:03 ok, the second one 2011-12-20T08:28:32 and then whole channels clicks it 2011-12-20T08:28:46 Hmm ... ah :P 2011-12-20T08:28:52 that might explain it :P 2011-12-20T08:29:13 But the url i put there should not work ... it had an typo 2011-12-20T08:29:15 what happened to website? 2011-12-20T08:29:24 xyxan: see above 2011-12-20T08:29:53 janzert, https://github.com/aichallenge/aichallenge/blob/epsilon/website/ranking_json.php ... it seems to have pages ... 2011-12-20T08:29:55 lol whos json? 2011-12-20T08:30:06 so why would it ask for the whole db 2011-12-20T08:30:16 it does if you ask for a page 2011-12-20T08:30:36 yeah and else it gives 0 right? 2011-12-20T08:30:40 e.g. http://aichallenge.org/ranking_json.php?page=1 2011-12-20T08:30:46 0 means everything 2011-12-20T08:30:49 ow 2011-12-20T08:30:51 auw... 2011-12-20T08:30:58 at least it looks like to me 2011-12-20T08:30:58 sorry ... 2011-12-20T08:32:13 if ($page === 0) { 2011-12-20T08:32:13 $limit = ""; 2011-12-20T08:32:13 } 2011-12-20T08:32:16 hmm 2011-12-20T08:32:23 *** Raulis has quit IRC (Client Quit) 2011-12-20T08:32:40 same code is in the html version 2011-12-20T08:33:47 So there you also have a vulnerability 2011-12-20T08:33:57 *** mcstar has joined #aichallenge 2011-12-20T08:35:54 *** gustavokambara has joined #aichallenge 2011-12-20T08:36:03 *** sashaSochka has joined #aichallenge 2011-12-20T08:36:20 sir_macelon: congra.... uh :( 2011-12-20T08:36:22 *** xyxan has quit IRC (Quit: Page closed) 2011-12-20T08:36:25 hmm, so maybe it *was* just coincidental timing 2011-12-20T08:36:39 *** codetiger has joined #aichallenge 2011-12-20T08:36:56 so when i call a page with page=0 2011-12-20T08:37:01 it will derail server 2011-12-20T08:37:05 But still ... i would change that in you code 2011-12-20T08:37:05 * Fluxid tries 2011-12-20T08:37:07 Fluxid: i will have hard times at mazes unfortunately... 2011-12-20T08:37:08 in any case 2011-12-20T08:37:10 nope, I already tried it ;) 2011-12-20T08:37:13 ;) 2011-12-20T08:37:17 you get a really big page though 2011-12-20T08:37:23 i didn't try actually :) 2011-12-20T08:37:41 sir_macelon: why? because of discovery? 2011-12-20T08:38:06 ah, you got blocked very early 2011-12-20T08:38:24 you couldn't do anything 2011-12-20T08:38:43 and i fell below 1000 2011-12-20T08:38:46 :( 2011-12-20T08:38:46 still waiting for my 6th game 2011-12-20T08:38:56 18 mins to go... 2011-12-20T08:38:59 by doing nothing 2011-12-20T08:39:06 180 mins (again) 2011-12-20T08:39:34 why is the server count low today? 2011-12-20T08:39:46 yesterday it was much better 2011-12-20T08:39:47 lol, everyone is sitting on pins 2011-12-20T08:40:28 mcstar: Just to make sure, we dont have a game with xathis 2011-12-20T08:40:50 he won everything so far 2011-12-20T08:41:03 did he realease his code so far? 2011-12-20T08:41:10 my 5th game was 6 hours ago and i have to wait ~5 more for my 6th... :( 2011-12-20T08:41:18 szia denes! 2011-12-20T08:41:23 szia 2011-12-20T08:41:58 *** bearoff_w has quit IRC (Read error: Connection reset by peer) 2011-12-20T08:42:06 I'm sure I'm not the first one to notice that but.. why is the next game estimation doing "number of players ahead / current game rate" instead of " number of players ahead / The current player rate" ??? 2011-12-20T08:42:19 my 5th game was 10hrs ago 2011-12-20T08:42:27 *** bearoff_w has joined #aichallenge 2011-12-20T08:44:57 *** katebus has quit IRC (Quit: Page closed) 2011-12-20T08:45:11 Surya_: because most of the time most players will not be coming from ahead of you, it was doing a more complex calculation up until a few hours ago but it also had an error. This basically as accurate and much easier to calculate. 2011-12-20T08:45:14 Fluxid: I failed to optimize my code for mazes and they are very tricky anyway 2011-12-20T08:46:54 sir_macelon: you were the first one with real battle code 2011-12-20T08:47:03 how much have you changed it since? 2011-12-20T08:47:07 janzert : I got the part where players are not coming from ahead of us, but I don't get if the problem is now solved, and if the time displayed is legit? 2011-12-20T08:47:11 is it based on the same principle? 2011-12-20T08:48:15 *** Antimony has joined #aichallenge 2011-12-20T08:48:25 Surya_: basically yes, it's a bit conservative when lots of players are ahead of you and doesn't account for game rate variations so may or may not be completely accurate but it's a fairly good estimate 2011-12-20T08:49:07 janzert : Thanks for your answer & your attention :) 2011-12-20T08:49:46 daaaam, the lower i fall the longer time i need to wait... 2011-12-20T08:49:52 mcstar: it is almost untouched since the beginning 2011-12-20T08:50:00 sorry if it's been answered already... what determines the number of games a player has had up until now? Some are still on 5, others 13 2011-12-20T08:50:11 sir_macelon: whoa thats something 2011-12-20T08:51:03 mcstar: i've only find out one bug on sunday with risk evaluation for 3+ enemy ant groups and adjusted it slightly to be more aggressive 2011-12-20T08:51:09 denes: did you see my private message? 2011-12-20T08:52:02 sir_macelon: i added agressive behavior on the final day 2011-12-20T08:52:42 i clustered battle groups, then estimated a response function, when the numbers are matched, it would become agressive 2011-12-20T08:54:26 mcstar: Have you compiled Smileys submission? 2011-12-20T08:55:04 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-20T08:55:11 mcstar: with my bug it didn't want to exchange 3+ ant columns, when i intended to, and it was a bug in battle outcome evalution function, i was glad i found it 2011-12-20T08:55:15 UncleVasya: no, i compiled yours 2011-12-20T08:55:30 nice 2011-12-20T08:55:42 Do you have any results? 2011-12-20T08:56:04 I mean in testing my bot vs yours 2011-12-20T08:56:20 *** celes has joined #aichallenge 2011-12-20T08:56:54 i run it once against 5 of mine, you know, that i talked about, it held its ground 2011-12-20T08:57:19 it is better than mine, no surprise there :) 2011-12-20T08:58:57 *** sigh has quit IRC (Remote host closed the connection) 2011-12-20T08:59:01 *** gustavokambara has quit IRC (Quit: Page closed) 2011-12-20T08:59:25 *** kincody has joined #aichallenge 2011-12-20T09:00:31 *** kincody has quit IRC (Client Quit) 2011-12-20T09:01:44 mcstar: i thought it was antimatroids bot 2011-12-20T09:02:20 >Minutes per game per server 2011-12-20T09:02:22 >30.0 2011-12-20T09:02:35 uh, 3600 turns using max turntime? 2011-12-20T09:02:45 mcstar: "i run it once against 5 of mine" - I think it's bad test case as other 5 bots battle one with another and it can help your bot very much 2011-12-20T09:02:45 *** pm_ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T09:02:49 UncleVasya: yep, same setting, and i run it with yours 2011-12-20T09:03:06 bearoff_w: i totally agree 2011-12-20T09:03:14 *and vice versa in your case 2011-12-20T09:03:17 i didnt try to prove anything 2011-12-20T09:03:26 just wanted to see his bot play 2011-12-20T09:03:28 fluxid that's a simple 30 / number of games in past 30 minutes 2011-12-20T09:03:51 so first game after restarting or last game when going down will show it as 30 2011-12-20T09:04:05 aha 2011-12-20T09:04:32 and with the site going down I had to go restart several of them 2011-12-20T09:04:47 *** smiley1983 has quit IRC (Ping timeout: 252 seconds) 2011-12-20T09:04:53 so within 30 minutes there was only one game 2011-12-20T09:04:55 ;( 2011-12-20T09:05:07 ok 2011-12-20T09:05:45 *** smiley1983 has joined #aichallenge 2011-12-20T09:07:05 *** codetiger has quit IRC (Remote host closed the connection) 2011-12-20T09:13:46 *** bearoff_w has left #aichallenge 2011-12-20T09:13:53 *** bearoff_w has joined #aichallenge 2011-12-20T09:14:06 Oh, deresh won a game, still 7k++ 2011-12-20T09:15:24 is that 8k or 7001 ? 2011-12-20T09:15:53 7333 2011-12-20T09:16:22 *** GreenTea has joined #aichallenge 2011-12-20T09:16:22 *** praveen__ has joined #aichallenge 2011-12-20T09:16:33 hihi 2011-12-20T09:16:53 latest games are disappeared from my profile http://aichallenge.org/profile.php?user=398 2011-12-20T09:17:07 only i have such issue? 2011-12-20T09:17:15 xathis too 2011-12-20T09:17:22 hm 2011-12-20T09:17:28 GreenTea: possibly, i didnt notice anything on mine 2011-12-20T09:18:14 GreenTea: you were disqualified didnt you know? you violated code46 section9 2011-12-20T09:18:28 *** besh has joined #aichallenge 2011-12-20T09:19:11 Oh ma, not the section 9 2011-12-20T09:19:17 ma, ma 2011-12-20T09:19:45 sure) 2011-12-20T09:19:59 *** FranPeruzzi has quit IRC (Quit: Page closed) 2011-12-20T09:20:11 happy to get some games in again 2011-12-20T09:20:13 oh wow, 82 2011-12-20T09:20:16 good morning indeed 2011-12-20T09:21:04 this must be what sports enthusiasts feel like 2011-12-20T09:21:34 GreenTea: seriously, it is weird 2011-12-20T09:22:10 id be shitting my pants 2011-12-20T09:22:24 same, I think they posted something about missing games 2011-12-20T09:22:49 or this cutoff in action 2011-12-20T09:23:20 a have a rank 518 to low to fight with top bots XD 2011-12-20T09:23:44 GreenTea: the page cant assemble your profile from the DB engine? i dunno 2011-12-20T09:24:15 try checking other's profiles, maybe you come across the same thing 2011-12-20T09:24:17 cutoff? 2011-12-20T09:24:39 tmseiler: low ranked players will be excluded from further games 2011-12-20T09:24:44 There's not cutoff yet that I can tell 2011-12-20T09:24:49 (i only heard) 2011-12-20T09:24:53 is that a rolling thing? or a one-time thing? 2011-12-20T09:25:01 i think one time, later on 2011-12-20T09:25:06 interesting 2011-12-20T09:26:08 *** celes has quit IRC (Ping timeout: 258 seconds) 2011-12-20T09:26:12 *** praveen__ has quit IRC (Quit: Page closed) 2011-12-20T09:26:21 I wish we could have more infos about that! 2011-12-20T09:26:32 Date/Time especially 2011-12-20T09:26:42 rank 2011-12-20T09:26:47 I'm so impatient >< 2011-12-20T09:27:02 drink some herb tea 2011-12-20T09:27:07 :) 2011-12-20T09:27:22 green one? 2011-12-20T09:28:03 nah, that has coffein, it makes you nervous 2011-12-20T09:28:15 *** smiley1983 has quit IRC (Ping timeout: 252 seconds) 2011-12-20T09:29:19 *** smiley1983 has joined #aichallenge 2011-12-20T09:29:20 no, GreenTea is not nervous though he is 500+ rank and can't play with top bots ) 2011-12-20T09:29:48 i hope he will, he deserves it 2011-12-20T09:30:00 ukrain did well 2011-12-20T09:30:02 e 2011-12-20T09:30:03 I sure he will 2011-12-20T09:30:16 thanks, I'm ukrainian too ) 2011-12-20T09:30:19 *** boreus has joined #aichallenge 2011-12-20T09:30:20 i know 2011-12-20T09:30:43 French are awesome too ;) 2011-12-20T09:30:55 you have answers on all that I say ) 2011-12-20T09:30:57 *** Antimony has quit IRC (Ping timeout: 244 seconds) 2011-12-20T09:32:12 yep i have 2011-12-20T09:32:22 :) 2011-12-20T09:32:25 is it annoying? 2011-12-20T09:32:27 sorry 2011-12-20T09:32:30 no, no 2011-12-20T09:32:46 GreenTea: when you are logged in, do you see any clues? 2011-12-20T09:32:47 it's rather pleasant 2011-12-20T09:33:13 yes 2011-12-20T09:33:15 i couldnt find any other profiles with the same error, from a random sampling of mine 2011-12-20T09:33:37 mcstar, me too, thats why I asked.. 2011-12-20T09:33:48 but xathis is empty 2011-12-20T09:34:07 what do you mean? 2011-12-20T09:34:15 same issue.. 2011-12-20T09:34:19 xathis loads for me 2011-12-20T09:34:19 was empty 2011-12-20T09:34:26 now it's fine 2011-12-20T09:34:31 oh 2011-12-20T09:34:34 oh 2011-12-20T09:34:40 then its something temporary 2011-12-20T09:34:41 mcstar: just was impressed and glad that you know and remember my nationality 2011-12-20T09:34:42 im Ok now ^^ 2011-12-20T09:35:01 GreenTea: indeed :D 2011-12-20T09:35:19 bearoff_w: i just checked your profile a couple of minutes ago 2011-12-20T09:35:34 i remembered we played on tcp in the last minutes 2011-12-20T09:35:52 yep 2011-12-20T09:36:35 my profile is depressive like GreenTea, as before submission end i was in top 250 and now have no games 2011-12-20T09:37:43 will be better in a couple of days 2011-12-20T09:37:55 xathis defends his hills well 2011-12-20T09:38:03 sure, but I can't sleep ) 2011-12-20T09:38:15 bearoff_w: i can sleep better than before 2011-12-20T09:38:20 since i dont have to code 2011-12-20T09:38:25 and think about coding 2011-12-20T09:38:49 you cant control what happens from now on 2011-12-20T09:38:53 it is calming 2011-12-20T09:39:08 mcstar: no, it's nail biting! 2011-12-20T09:39:10 j/k 2011-12-20T09:39:33 are you Mr. Joking? 2011-12-20T09:39:37 :) I had no problem with bad dreams last 40 hours before deadline as I was coding 2011-12-20T09:39:43 pretty sure if they froze scores now, it's above where I will rank 2011-12-20T09:39:48 so they should just end it now 2011-12-20T09:39:50 :) 2011-12-20T09:39:53 Haters gonna hate, but #1 is USA, USA, USA! :) 2011-12-20T09:39:56 *** Anti has quit IRC (Ping timeout: 258 seconds) 2011-12-20T09:39:58 many would be disapointed 2011-12-20T09:40:12 im pretty sure its gonna be germany 2011-12-20T09:40:31 I was rank 15 overnight, dropped down to 43 now 2011-12-20T09:40:38 41 2011-12-20T09:41:16 *** cuda has joined #aichallenge 2011-12-20T09:42:32 damn, ok i admit its annoying that games are so slow 2011-12-20T09:43:03 im gonna make an herb tea 2011-12-20T09:43:29 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-20T09:45:26 *** AndrewBC_ has joined #aichallenge 2011-12-20T09:45:40 someone should have written some widget or desktop notification 2011-12-20T09:46:00 when i played a game, it would pop up, and link to it directly 2011-12-20T09:46:12 I set an alarm clock 2011-12-20T09:46:18 how? 2011-12-20T09:46:21 or did, when the times were more reliable 2011-12-20T09:46:25 ah 2011-12-20T09:46:41 i was thinking of greping the profile page 2011-12-20T09:47:11 Is site down again? 2011-12-20T09:47:21 nah 2011-12-20T09:47:22 nvm 2011-12-20T09:47:34 it'd be easy to screen-scape, but not very polite to the organizers 2011-12-20T09:47:55 whats that? "screen-scape"? 2011-12-20T09:48:07 loading the page with a script/program and parsing the data out 2011-12-20T09:48:19 I have a script that sends an email 2011-12-20T09:48:21 why is it bad? 2011-12-20T09:48:32 i mean it doesnt load images and such only text 2011-12-20T09:48:34 json a easier 2011-12-20T09:48:35 their servers are under enough load from actual humans mashing reload 2011-12-20T09:48:37 so it wouldbe minimal bandwidth 2011-12-20T09:48:56 tmseiler: I do a 5 min poll, which is slower than I would refresh at 2011-12-20T09:48:56 but a script would check in every minute or so 2011-12-20T09:49:19 i think im gonna do it 2011-12-20T09:49:40 how can i log in with wget? 2011-12-20T09:49:42 :D 2011-12-20T09:49:56 yeah, I guess I'm just saying that personal scripts are probably fine, but a site to hammer it and keep stats wouldn't be very nice 2011-12-20T09:50:06 *** pairofdice has quit IRC (*.net *.split) 2011-12-20T09:50:06 *** Kurnevsky has quit IRC (*.net *.split) 2011-12-20T09:50:06 *** murr4y has quit IRC (*.net *.split) 2011-12-20T09:50:06 *** Kingpin13 has quit IRC (*.net *.split) 2011-12-20T09:50:06 *** AndrewBC has quit IRC (*.net *.split) 2011-12-20T09:50:06 *** matthewd has quit IRC (*.net *.split) 2011-12-20T09:50:06 *** aerique has quit IRC (*.net *.split) 2011-12-20T09:50:06 *** Blkt has quit IRC (*.net *.split) 2011-12-20T09:50:07 *** raemde has quit IRC (*.net *.split) 2011-12-20T09:50:07 *** Bluedgis_ has quit IRC (*.net *.split) 2011-12-20T09:50:11 *** cuda is now known as pairofdice 2011-12-20T09:50:12 (but I wish there were an atom feed or something to subscribe to) 2011-12-20T09:50:43 im just gonna compare my skill to the previous, when it changes it will let me know 2011-12-20T09:50:49 but what should be the notification? 2011-12-20T09:50:50 *** KP13 has joined #aichallenge 2011-12-20T09:50:50 *** Kurnevsky1 has joined #aichallenge 2011-12-20T09:50:50 *** aerique has joined #aichallenge 2011-12-20T09:50:51 *** Blkt has joined #aichallenge 2011-12-20T09:50:51 *** raemde has joined #aichallenge 2011-12-20T09:50:51 *** Bluedgis_ has joined #aichallenge 2011-12-20T09:50:52 mcstar: that's what I do 2011-12-20T09:50:59 yeah 2011-12-20T09:51:03 thre is dialog 2011-12-20T09:51:06 or something 2011-12-20T09:51:13 the pops up a message box from a scrit 2011-12-20T09:51:15 pt 2011-12-20T09:51:25 or zenity? 2011-12-20T09:51:42 *** grgr has joined #aichallenge 2011-12-20T09:52:10 mcstar: http://pastebin.com/TtKfVq1R 2011-12-20T09:52:18 sends email through gmail 2011-12-20T09:52:19 dialog is ncurses? 2011-12-20T09:52:33 ncurses has no dialogs 2011-12-20T09:52:41 its a terminal graphics library 2011-12-20T09:52:45 lol 2011-12-20T09:52:50 dialog is a little app 2011-12-20T09:52:55 that can make message box 2011-12-20T09:53:03 but it did mine in ncurses 2011-12-20T09:53:08 i though it would be X 2011-12-20T09:53:33 your email doesn't give you a notification upon receiving one? 2011-12-20T09:54:16 i try not using email 2011-12-20T09:54:22 *** TheLinker has joined #aichallenge 2011-12-20T09:54:24 and there are better alternatives in this case 2011-12-20T09:54:39 rwest: thanks for the script but im not gonna use it :) 2011-12-20T09:54:45 *** pvarga has quit IRC (Quit: pvarga) 2011-12-20T09:55:24 ah ok, zenity does X dialogs 2011-12-20T09:55:48 mcstar: I wanted notifications on my phone ^^ 2011-12-20T09:56:18 ah, my phone is turned off constantly so thats not an option 2011-12-20T09:56:26 *** praveen__ has joined #aichallenge 2011-12-20T09:56:27 *** tomb___ has joined #aichallenge 2011-12-20T09:56:28 but thats sweet 2011-12-20T09:57:09 flowenol did the checking part, I just changed it from sms to email 2011-12-20T09:57:16 *** murr4y has joined #aichallenge 2011-12-20T09:57:45 *** besh has quit IRC (Quit: Page closed) 2011-12-20T09:58:05 whoa 2011-12-20T09:58:08 *** Murashka has joined #aichallenge 2011-12-20T09:58:13 did you see the source of the profile? 2011-12-20T09:58:28 its tremendous has all the country and organization entrie 2011-12-20T09:58:30 s 2011-12-20T09:59:17 *** KP13 is now known as Kingpin13 2011-12-20T09:59:31 *** Kingpin13 has joined #aichallenge 2011-12-20T09:59:34 mcstar: yea, it's so you can change yours in the dropdown box after hitting edit 2011-12-20T09:59:47 that stuff should probably be loaded via ajax after hitting edit 2011-12-20T10:00:06 but i should not be logged in 2011-12-20T10:00:12 i downloaded the page with wget 2011-12-20T10:00:17 so it cant be logged in 2011-12-20T10:00:17 heh 2011-12-20T10:00:19 so why the info? 2011-12-20T10:00:33 it must be in every page 2011-12-20T10:00:39 logged or not 2011-12-20T10:00:46 that's a lot of extra bandwidth 2011-12-20T10:00:50 yep 2011-12-20T10:01:43 >27kb 2011-12-20T10:01:48 *** matthewd has joined #aichallenge 2011-12-20T10:02:29 is it possbile to partially request a page? 2011-12-20T10:02:48 don't think so 2011-12-20T10:03:04 you can ask for it via the Range: header 2011-12-20T10:03:31 hmm, if I log out I don't get the organization or country list 2011-12-20T10:03:41 *** tob__k has joined #aichallenge 2011-12-20T10:03:52 games are so fast now, 30 minutes? 2011-12-20T10:04:12 no 2011-12-20T10:04:44 janzert: how come? i do 2011-12-20T10:04:54 my last game was 8 hours ago 2011-12-20T10:04:59 sure you're not logged in? 2011-12-20T10:05:01 yep 2011-12-20T10:05:08 i was logged in in opera 2011-12-20T10:05:14 and im downloading via wget 2011-12-20T10:05:21 and i logged out in opera 2011-12-20T10:05:38 and the two systems cant possibly be connected 2011-12-20T10:05:40 and never logged in via wget? 2011-12-20T10:05:43 no 2011-12-20T10:05:45 never 2011-12-20T10:05:49 i van try curl 2011-12-20T10:05:52 no idea 2011-12-20T10:05:53 c 2011-12-20T10:06:18 *** smjms has joined #aichallenge 2011-12-20T10:06:31 the same with curl 2011-12-20T10:06:31 still discussing the rank-change-auto-notifications? 2011-12-20T10:07:09 *** Murashka has quit IRC (Ping timeout: 258 seconds) 2011-12-20T10:07:42 ah my bad 2011-12-20T10:08:06 janzert: i never got the country/organization list with that 2011-12-20T10:08:17 ok :) 2011-12-20T10:08:28 but the file is ~30k which is large 2011-12-20T10:08:34 I wondered, what was the reasons all ranks -- even those for past versions of the bots -- were reset? I found this information very useful and a huge improvement compared to the last contest. Is this information still available somewhere? 2011-12-20T10:08:34 i somehow assumed its there 2011-12-20T10:08:45 but in fact i only saw that with opera source viewver 2011-12-20T10:09:01 there's a lot of information 2011-12-20T10:09:23 yeah, im gonna look up what a1k0n suggested 2011-12-20T10:09:38 i dont know yet what it means though 2011-12-20T10:09:42 Range: header, 2011-12-20T10:09:44 ? 2011-12-20T10:10:12 I doubt it'll work for a dynamic php page 2011-12-20T10:10:15 but you can try it 2011-12-20T10:13:50 *** tob__k has left #aichallenge 2011-12-20T10:14:01 *** katebus has joined #aichallenge 2011-12-20T10:14:15 *** Varan has quit IRC (Quit: Leaving) 2011-12-20T10:14:29 I was a little bummed about losing the historical ranks too 2011-12-20T10:15:07 mcstar: what is your login on official server? 2011-12-20T10:16:46 bearoff_w: AgentSmith 2011-12-20T10:17:01 *** boreus has quit IRC (Quit: Page closed) 2011-12-20T10:18:24 :) thanks 2011-12-20T10:19:05 *** Murashka has joined #aichallenge 2011-12-20T10:21:48 *** avdg has joined #aichallenge 2011-12-20T10:22:15 yay, my bot is doing better than the top 1500 2011-12-20T10:22:39 Next game should be within 30 minutes. 2011-12-20T10:22:47 F5F5F5F5F5F5F5F5F5F5F5F5F5F5F5 2011-12-20T10:23:04 *** gustavokambara has joined #aichallenge 2011-12-20T10:23:29 maybe there's a hidden "patience" multiplier 2011-12-20T10:23:31 *** tomb___ has quit IRC (Quit: Page closed) 2011-12-20T10:23:32 xD 2011-12-20T10:23:34 and each F5 reduces it by .1 2011-12-20T10:23:52 Fluxid: You know if there is any interest in people continuing to use servers like yours to continue to pit their bots against each other after the contest? 2011-12-20T10:24:13 I know I'm definitely going to keep playing with my friends 2011-12-20T10:24:27 this would be a great tool at computer camp, too 2011-12-20T10:24:32 i'm pretty surre that every second i hit F5, the minute counter get's lower by 1/60 2011-12-20T10:24:38 so hitting F5 helps 2011-12-20T10:24:47 lol 2011-12-20T10:24:54 Fluxid: my experience is that it goes up by 2 minutes :) 2011-12-20T10:24:59 what is pseudoscience at work 2011-12-20T10:25:01 Elderwolf: IN AVERAGE 2011-12-20T10:25:04 w->t 2011-12-20T10:25:40 Elderwolf: there is interest, but i myself am not interested in continuing to host it anymore 2011-12-20T10:25:50 Man I regret not uploading my dynamic role code before contest began. I was too chicken to put non-thoroughly tested codde out there 2011-12-20T10:26:54 Has anyone used a subset of ants to guard their hills? I haven't seen any yet. 2011-12-20T10:27:00 you know what would be awesome? a bot which during submission phase would be like 1000 places behind 2011-12-20T10:27:20 but during finals it would suddenly kick xathis's and other top bots ass 2011-12-20T10:27:59 Elderwolf: i dispatch 7 ants to wander around hill 2011-12-20T10:28:16 Fluxid: well, some people's could if there was a sudden influx of maze maps. Depending on your path finding algorithm some just do better there 2011-12-20T10:28:18 sleeper bot 2011-12-20T10:28:39 it's name would be procrastinationbot 2011-12-20T10:28:45 Fluxid: or someone that always timed out, but could reliably kill 65% of hills before time out 2011-12-20T10:29:01 its 2011-12-20T10:29:12 22 minutes 2011-12-20T10:29:28 maybe i'll get a game before i finish workday 2011-12-20T10:29:31 any news on the low rank cutoff? 2011-12-20T10:29:33 official news 2011-12-20T10:30:02 the faster it happens the faster the games will go for the others 2011-12-20T10:30:04 *** petethepig has joined #aichallenge 2011-12-20T10:30:14 Elderwolf, look at the blue guy: http://aichallenge.org/visualizer.php?game=288059&user=53 2011-12-20T10:30:50 UncleVasya: cool, thanks. HOwever i didn't know you could move diagonally? 2011-12-20T10:30:57 or is he just switching places with guards 2011-12-20T10:31:05 *** akarukov has joined #aichallenge 2011-12-20T10:31:12 no, moving diag is not possible 2011-12-20T10:31:27 flowenol: I won't do it until I'm sure the bots that are better than starter bots have had a chance to pass them up in the rankings 2011-12-20T10:31:40 UncleVasya: id' say that guard position is not idea though, since you could kill each side without dying 2011-12-20T10:31:44 janzert: oh ok 2011-12-20T10:31:50 *** Hexren has joined #aichallenge 2011-12-20T10:31:55 *** Hexren has left #aichallenge 2011-12-20T10:31:56 im stressing with it 2011-12-20T10:32:01 i want to see my bot fight more 2011-12-20T10:32:08 Next time add a SHA-1 hash or something to the starter bots ;) 2011-12-20T10:32:31 I realize now that my bot is fucked up on close multi hills >< It just kamikazes on every enemy ants so they don't see my hill >< 2011-12-20T10:32:33 a good example right now is http://aichallenge.org/profile.php?user=896 2011-12-20T10:32:41 So if there's no modification at all you can spot it easily 2011-12-20T10:33:05 rated <2000 at close of submissions >7300 currently 2011-12-20T10:33:06 http://aichallenge.org/visualizer.php?game=277755&user=53 lol, turn 244 around 43,147 2011-12-20T10:33:30 and is clearly better than a starter bot 2011-12-20T10:33:37 I think the best idea was to deny going to the final round bots that are worse then GreedyBot AND was deactivated (or deactivated for a long period of time). 2011-12-20T10:34:06 janzert: is the cutoff rank decided? 2011-12-20T10:34:10 or not yet 2011-12-20T10:34:25 what does that question mean? ;) 2011-12-20T10:34:31 the cutoff rank is dynamic 2011-12-20T10:34:40 ah ok 2011-12-20T10:34:46 janzert, may be you can spent some time to deactivate starter bots manually? 2011-12-20T10:34:49 it will first move up past the starter bots, then slowly go up from there 2011-12-20T10:35:12 bearoff_w: no, no bots will be deactivated 2011-12-20T10:35:21 hm, i wonder how could we adapt conway's game of life into aichallenge game 2011-12-20T10:35:22 just get cutoff eventually 2011-12-20T10:35:38 like, bots building gliders and other structures just to beat the enemy "somehow" 2011-12-20T10:35:40 *** Schoon has joined #aichallenge 2011-12-20T10:35:48 Finally 3 games in 20mins 2011-12-20T10:35:50 Fluxid: there was talk of that for this contest 2011-12-20T10:35:56 *** AlteredState has joined #aichallenge 2011-12-20T10:36:12 a1k0n: im might be doing something wrong, but i dont think ranges work with aichallenge 2011-12-20T10:36:22 echo -e "GET Range: bytes=0-10 /profile.php?user=12 HTTP/1.1\n\n" | nc aichallenge.org 80 |less 2011-12-20T10:36:33 12 minutes, brb popcorn 2011-12-20T10:36:33 i get the whole page no matter what is the upper bound 2011-12-20T10:36:51 basically doing battle by starting formations on your side that would travel to the other side of the board and attack your opponent's structures 2011-12-20T10:37:01 *** xathis has joined #aichallenge 2011-12-20T10:37:18 Make a spaceship formation, gain lasers 2011-12-20T10:37:20 but it would be horribly non-trivial to build anything decent 2011-12-20T10:37:25 mcstar: That's certainly something wrong... but I wouldn't expect it to work anyway, as janzert said 2011-12-20T10:37:58 *** jstrong has joined #aichallenge 2011-12-20T10:38:05 good morning all 2011-12-20T10:38:08 Are my opponents choosen by my skill or my mu? 2011-12-20T10:38:09 *** jstrong is now known as roflmao 2011-12-20T10:38:27 UncleVasya: mu is most important of the two 2011-12-20T10:38:29 matthewd: could you explain the first part? 2011-12-20T10:38:40 thanks 2011-12-20T10:39:32 actual skill isn't used at all but sigma does factor in 2011-12-20T10:39:55 *** ChrisH_ has joined #aichallenge 2011-12-20T10:39:57 for pairing that is 2011-12-20T10:40:22 *** gustavokambara has quit IRC (Quit: Page closed) 2011-12-20T10:40:43 mcstar: Your syntax for specifying an HTTP header is incorrect 2011-12-20T10:40:53 janzert: yeah, that would be hard. but i wonder, have some areas on the maps designated to build stuff, and after the stuff leaves this area you cannot do anything more with it.. hmmhmmm 2011-12-20T10:41:15 Fluxid: yep, that's the idea 2011-12-20T10:41:26 matthewd: ok, its certainly possible, what is the correct way? 2011-12-20T10:41:32 mcstar: But support for ranges is purely optional, and normally only supported against static on-disk files (or things very similar), where the server can actually benefit by not sending it all 2011-12-20T10:41:57 mcstar: When requesting a PHP script, the server has to run the entire script anyway, so it won't bother with the range 2011-12-20T10:42:05 ok i get it 2011-12-20T10:42:12 but what is the correct sytax? 2011-12-20T10:42:14 n 2011-12-20T10:42:16 or being able only to build gliders and anython else 2011-12-20T10:42:53 hmm, limiting the types of structures built instead of the area to build seems less interesting 2011-12-20T10:43:02 mcstar: wget --header='Range: bytes=0-10' ... 2011-12-20T10:43:20 also besides limiting area there was a limit on number of cells changed per turn 2011-12-20T10:43:20 http://www.youtube.com/watch?v=pSWN6Qj98Iw — weirdest 3 minutes of your day 2011-12-20T10:43:30 mcstar: On the wire: GET /foo HTTP/1.1\nRange: bytes=0-10\nHost: aichallenge.org\n\n 2011-12-20T10:43:55 *** Murashka has quit IRC (Quit: Page closed) 2011-12-20T10:44:00 matthewd: ok, so its comes after a newline 2011-12-20T10:44:02 thx 2011-12-20T10:44:11 what about a game with something a little more persistent than moving units around? 2011-12-20T10:44:14 like a base to construct 2011-12-20T10:44:26 They weren't paid to cry because nobody gets pain in NK 2011-12-20T10:44:34 paid* 2011-12-20T10:44:50 yeah, but gliders go diagonally, map would be wide enough not for gliders not to be able to reach enemys area. so you need to send gliders in such way, that colliding them with themselves/enemy gliders would cause such reaction that would cause enemy to loose (and also sending gliders so it would cancel the reaction going close to own area) 2011-12-20T10:45:22 and have 3x3 area which hitting it would cause player to loose, so reaction would need to reach it 2011-12-20T10:45:56 Fluxid: weren't you like top 40 before contest? 2011-12-20T10:46:02 i was 2011-12-20T10:46:03 definitely not a beginner friendly game ;) 2011-12-20T10:46:11 Fluxid: what happend? 2011-12-20T10:46:19 Elderwolf: finals happend 2011-12-20T10:46:33 buuut 2011-12-20T10:46:35 In Game: Playing in a game right now. 2011-12-20T10:47:04 *** besh has joined #aichallenge 2011-12-20T10:47:32 Is the problem with the match maker fixed ? 2011-12-20T10:48:00 besh: afaik there is no problem? 2011-12-20T10:48:09 There was yesterday 2011-12-20T10:48:11 the problem with game inserts is fixed 2011-12-20T10:48:26 I just read about it in the forums 2011-12-20T10:48:32 the match maker itself wasn't a problem yesterday 2011-12-20T10:48:56 *** mviel__ is now known as mvie 2011-12-20T10:48:56 *** mvie is now known as mviel__ 2011-12-20T10:49:05 besh: what is the problem you read about? 2011-12-20T10:49:11 553rd 2011-12-20T10:49:22 Next game should be within 688 minutes. 2011-12-20T10:49:26 OH COME ON 2011-12-20T10:49:28 although it does have a known race condition, that is still present, that can cause similar matches to be played close together 2011-12-20T10:49:29 that some bots are getting a lot more games than ohters. 2011-12-20T10:49:38 at this pace it'd take few days to get 10 games 2011-12-20T10:49:51 Many bots at 12 games, many other bots at 5 games 2011-12-20T10:49:51 Fluxid: I got 7 games last night 2011-12-20T10:49:52 My engine is stuck at 5 i think 2011-12-20T10:49:58 5 within 20 minutes 2011-12-20T10:50:01 besh: read my response in this thread: http://forums.aichallenge.org/viewtopic.php?f=27&t=2183&p=13582#p13582 2011-12-20T10:50:29 *** liberforce has joined #aichallenge 2011-12-20T10:50:30 *** Anilm3 has joined #aichallenge 2011-12-20T10:51:00 http://aichallenge.org/visualizer.php?game=292116&user=757 2011-12-20T10:51:14 *** flux_w42 has joined #aichallenge 2011-12-20T10:51:42 ChrisH_: i'm still not convinced, but okay 2011-12-20T10:52:04 great battle Fluxid 2011-12-20T10:52:13 boring ;) 2011-12-20T10:52:24 I guess I should be happy my bot is one of the pivots 2011-12-20T10:52:24 but feels good man 2011-12-20T10:52:36 eat dust non-pivots! 2011-12-20T10:53:03 26th place and crusing to the top! 2011-12-20T10:53:05 ChrisH_: Ok that makes sense. So it is trying to stabilize rating of some bots so that they would be used as landmarks. 2011-12-20T10:53:17 besh: yes 2011-12-20T10:53:47 but the part I don't know is how it will then compensate for the other bots later 2011-12-20T10:53:51 it still isnt conscious 2011-12-20T10:53:56 Atleast my rank is roughly what it was before finals 2011-12-20T10:54:04 sir_macelon: gz! 2011-12-20T10:54:26 and I'm just reporting behavior, i didn't write the code nor do I know what the plans are about cutoffs or anything 2011-12-20T10:54:28 chrish_: yes some of them would have to play more games than others at any time 2011-12-20T10:55:05 Fluxid: Thanks although I've lost some games already... 2011-12-20T10:55:17 but so many people were confused about the game counts it seemed important to help people understand 2011-12-20T10:56:14 old top bots are already pretty close to top 2011-12-20T10:56:18 another thing to be wary of is that many of the bots currently in the top 100 are there mostly because their sigmas are low, not because they are dominating all their matches 2011-12-20T10:56:22 and i'm loooow 2011-12-20T10:56:26 *** Ashoka has joined #aichallenge 2011-12-20T10:56:27 sadpanda 2011-12-20T10:56:31 *** akarukov has quit IRC (Quit: Page closed) 2011-12-20T10:56:49 Is trueskill also the match maker ? 2011-12-20T10:57:10 besh: I believe there is a match making component to the TrueSkill system, yes. 2011-12-20T10:58:47 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-20T10:59:05 So it seems it is trying to get ratings as quickly as possible without caring about fairness to all opponents 2011-12-20T11:00:05 yeah something is a bit funky with the way the ranking works 2011-12-20T11:00:05 but 2011-12-20T11:00:09 I'm sure it will stabilize 2011-12-20T11:00:28 like xathis wasn't even in the top 100 last night 2011-12-20T11:00:58 sir_macelon, i see many looses 2011-12-20T11:01:00 right now he's just 83 2011-12-20T11:01:20 well xathis will be there sooner or later but the rest may have problems come thursday 2011-12-20T11:01:57 sir_macelon, why in this game http://aichallenge.org/visualizer.php?game=290715&user=284 you blocade your hill? 2011-12-20T11:02:44 *** Hexren has joined #aichallenge 2011-12-20T11:02:47 *** Hexren has left #aichallenge 2011-12-20T11:02:58 GreenTea: unforunately i had problems with exploration on mazes and didn't solve it out 2011-12-20T11:04:41 *** aerique has quit IRC (Quit: ...) 2011-12-20T11:04:42 sadly 2011-12-20T11:05:08 sir_macelon: wish I had more time 2011-12-20T11:07:17 *** TheLinker has quit IRC (Quit: Bye) 2011-12-20T11:07:21 *** mj41 has quit IRC (Ping timeout: 268 seconds) 2011-12-20T11:08:13 wtf I got a timeout 2011-12-20T11:08:44 I avg 300ms with 300 ants 2011-12-20T11:08:50 yet I timeout on turn 49 2011-12-20T11:08:51 ... 2011-12-20T11:09:08 with breakpoints in place 2011-12-20T11:12:42 *** imp5imp5 has joined #aichallenge 2011-12-20T11:13:26 *** mj41 has joined #aichallenge 2011-12-20T11:13:51 *** flowenol has quit IRC (Ping timeout: 258 seconds) 2011-12-20T11:14:14 *** JorgeB has joined #aichallenge 2011-12-20T11:15:28 *** Cucuruzo has joined #aichallenge 2011-12-20T11:24:44 *** tobym has joined #aichallenge 2011-12-20T11:28:06 *** mleise has joined #aichallenge 2011-12-20T11:28:13 Next Game: 53 players are ahead., The current player rate is about 52.8 players per minute. ... Next game should be within 4 minutes. good math on site.. 2011-12-20T11:29:13 GreenTea: Shows the same to me. I hope not to be paired with you 2011-12-20T11:29:24 ^^ 2011-12-20T11:29:49 GreenTea: it's an average between games per minute and player per minute 2011-12-20T11:30:00 site won't even load for me 2011-12-20T11:32:18 *** foRei has joined #aichallenge 2011-12-20T11:34:43 http://tcpants.com/player/Accoun12301 2011-12-20T11:37:06 Finally,interesting matchups. http://aichallenge.org/visualizer.php?game=292526&user=66 2011-12-20T11:37:49 Accoun:that look cool! 2011-12-20T11:41:14 *** dvladim has joined #aichallenge 2011-12-20T11:41:30 ups it down 2011-12-20T11:41:33 Ashoka, do you worked on bot last days? 2011-12-20T11:43:14 GreenTea: yes, couldn't submit my best version. it had a bug I couldn't detect on time. ended up having to submit the second best. hope this one will not embarrass me. :-) how did u manage in last week ? 2011-12-20T11:44:21 GreenTea> http://tcpants.com/player/Accoun12301 2011-12-20T11:44:31 GreenTea> ti hotel poigrat 2011-12-20T11:44:39 *** Kurnevsky1 has quit IRC (Quit: Instantbird 1.0) 2011-12-20T11:44:46 Ashoka, last week i took vakation on work. enable rage mode, and code! code! code! 2011-12-20T11:44:55 GreenTea> 3 kopii rabotaut 2011-12-20T11:45:05 *** Kurnevsky has joined #aichallenge 2011-12-20T11:45:15 Accoun moi? 2011-12-20T11:45:27 GreenTea> ti hotel poigrat 2011-12-20T11:45:29 GreenTea> http://tcpants.com/player/Accoun12301 2011-12-20T11:45:38 i've just made a small chrome extension with popup notification if your played game count changes. grab it from here: https://github.com/adenes/ChromedAnts 2011-12-20T11:45:51 GreenTea> 2 moi kopii 2011-12-20T11:45:53 GreenTea : Looking forward to see a great fight between you, xathis, flag, pguilory etc saw you getting improved a lot lately 2011-12-20T11:46:41 GreenTea, BTW, I owe you a big thank. have u red my post ? :-) 2011-12-20T11:46:42 GreenTea> 3 moi kopii 2011-12-20T11:47:16 *** FranPeruzzi has joined #aichallenge 2011-12-20T11:47:39 Ashoka, i just read your inspiring post about planet wars. how you coding when compentition was end, and thought than man like you could not just, govi up ) 2011-12-20T11:47:42 * give up 2011-12-20T11:47:55 GreenTea> dont want? 2011-12-20T11:48:13 *** bhasker has quit IRC (Ping timeout: 276 seconds) 2011-12-20T11:48:52 *** gnu264 has joined #aichallenge 2011-12-20T11:49:08 Accoun. maby in couple hours.. 2011-12-20T11:49:18 *** Murashka has joined #aichallenge 2011-12-20T11:49:39 GreenTea> poznee? ili neskilko chasov? 2011-12-20T11:49:56 Accoun, but really don't see any sense now) 2011-12-20T11:50:28 Accon. poznee moget bit 2011-12-20T11:50:29 GreenTea> any sense? 2011-12-20T11:50:35 yes 2011-12-20T11:50:47 *** olexs has quit IRC (Quit: Leaving.) 2011-12-20T11:50:51 *** raulis has joined #aichallenge 2011-12-20T11:50:56 GreenTea> chuvsctv net? 2011-12-20T11:51:05 Accoun, a kakoy smisl esli final uze idet? 2011-12-20T11:51:19 GreenTea> ti ze hotel 2011-12-20T11:51:47 GreenTea : Tx. That work wouldn't be possible if not for shared bots by you and few others. BTW, I couldn't beat your planet wars bot even after 2 months. :-) it was a one cool bot! 2011-12-20T11:51:48 Accoun, chto hotel? 2011-12-20T11:51:50 *** Westicles has joined #aichallenge 2011-12-20T11:51:51 *** flux_w42 has quit IRC (Quit: Page closed) 2011-12-20T11:51:53 *** raulis has quit IRC (Client Quit) 2011-12-20T11:51:54 *** praveen__ has quit IRC (Quit: Page closed) 2011-12-20T11:52:02 GreenTea> co mnoj na canadskom ne poigraesh, a nr uspel adaptirovan pod ih tormasnii servera 2011-12-20T11:52:04 http://pastebin.com/2FQgiw2x 2011-12-20T11:52:08 guys try out this script 2011-12-20T11:52:12 GreenTea> co mnoj na canadskom ne poigraesh, a ne uspel adaptirovan pod ih tormasnii servera 2011-12-20T11:52:17 it notifies you if your skill changes 2011-12-20T11:52:26 you need notify-send 2011-12-20T11:52:28 *** protocolocon has joined #aichallenge 2011-12-20T11:52:32 GreenTea> co mnoj na Kanadskom ne poigraesh, a ne uspel adaptirovan pod ih tormasnii servera 2011-12-20T11:52:34 *** denes has quit IRC (Ping timeout: 258 seconds) 2011-12-20T11:52:35 and change the user=n line for your user number 2011-12-20T11:52:43 Ashoka. thanks) 2011-12-20T11:53:05 *** conor_f has joined #aichallenge 2011-12-20T11:53:21 *** Raulis has joined #aichallenge 2011-12-20T11:53:34 damn 2011-12-20T11:53:37 ? 2011-12-20T11:53:39 and use a bigger sleep than 5 2011-12-20T11:53:42 Accoun, esli hoches mogesh skachat jar http://sourceforge.net/projects/ants2011/ 2011-12-20T11:53:43 i meanlonger 2011-12-20T11:54:07 Accoun, i zapustit so svoey mashini, pod moim nikom 2011-12-20T11:54:34 GreenTea> u menia 3 kopii igraut 2011-12-20T11:54:45 *** gnu264 has quit IRC (Quit: Page closed) 2011-12-20T11:55:01 mcstar: you really have to check every 5 seconds? :P 2011-12-20T11:55:23 GreenTea : did u do min-max in planet wars? Judging by your timeout issues in Ants in the early beta stage, I guessed you were trying min-max for battles. correct ? :-) 2011-12-20T11:55:23 Accoun, nu ostav 2 i 1 mou 2011-12-20T11:55:32 janzert: i just corrected to make it longer 2011-12-20T11:55:35 it was for testing 2011-12-20T11:55:48 i use 300 now 2011-12-20T11:55:53 the server thanks you ;) 2011-12-20T11:55:54 *** tom____ has joined #aichallenge 2011-12-20T11:56:28 GreenTea> tak bolshe shans chto vtretasa 2011-12-20T11:57:10 rwest: so thats mine, if you want to use it, it sends a desktop notification instead of an email 2011-12-20T11:57:27 Ashoka, no minimax in planet wars and in ants) , in ants just check every posible combinations of moves.. 2011-12-20T11:57:42 Ashoka, but this works fast only for <7 ants 2011-12-20T11:57:49 matrix game 2011-12-20T11:57:54 sounds like minimax 2011-12-20T11:58:04 Simplex algoritm 2011-12-20T11:58:06 *** raemde_ has joined #aichallenge 2011-12-20T11:58:07 Ashoka, if more - to slow 2011-12-20T11:58:30 like poker 2011-12-20T11:58:31 Ashoka, n ^ 5 combinations.. 2011-12-20T11:58:36 for every your position you think that enemy will do best move for him? 2011-12-20T11:58:45 GreenTea: i tried that too, but i couldnt write a correct evaluator 2011-12-20T11:58:53 it was good for defense though :) 2011-12-20T11:59:03 hmm In battles, I just calculated for few selected moves from enemy. 2011-12-20T11:59:09 i did a full battle resolution on every leaf 2011-12-20T11:59:19 and scored the result acording to some metric 2011-12-20T11:59:25 So you choose Min for you from every your move and then pick Max score among these Mins. 2011-12-20T11:59:55 *** sir_macelon has quit IRC (Quit: Page closed) 2011-12-20T12:00:15 mcstar, i spend many time to find good evaluation.. 2011-12-20T12:00:22 like, a full attack and full defence, I just ignored other moves. 2011-12-20T12:00:23 much 2011-12-20T12:00:32 i guess so, i found it hard 2011-12-20T12:00:38 i mean to find one 2011-12-20T12:00:42 *** raemde has quit IRC (Read error: Operation timed out) 2011-12-20T12:00:47 so i went with a very simple battle tactics 2011-12-20T12:01:02 *** raemde has joined #aichallenge 2011-12-20T12:01:45 GreenTea> http://tcpants.com/search?name=green clear search 2011-12-20T12:01:50 calculation of scores after fight http://paste.bradleygill.com/index.php?paste_id=346878 2011-12-20T12:01:54 UncleVasya: did you try my script? 2011-12-20T12:02:10 mcstar: I guess it is for linux 2011-12-20T12:02:13 for 3 tipes of battle strategy 2011-12-20T12:02:18 *types 2011-12-20T12:02:20 yes 2011-12-20T12:02:22 ah damn 2011-12-20T12:02:29 i forgot youre on windows 2011-12-20T12:02:51 *** raemde_ has quit IRC (Ping timeout: 244 seconds) 2011-12-20T12:02:54 GreenTea: are you on linux? 2011-12-20T12:02:57 * UncleVasya is dummy :) 2011-12-20T12:02:58 Aggressive - is good for 1x1 2011-12-20T12:03:04 Defensive for FFA 2011-12-20T12:03:06 http://tcpants.com/player/Accoun12301 let play 2011-12-20T12:03:18 and Exchange when bot is almost win 2011-12-20T12:03:39 mcstar, on windows 2011-12-20T12:04:31 *** denes has joined #aichallenge 2011-12-20T12:04:33 *** katebus has quit IRC (Quit: Page closed) 2011-12-20T12:05:10 UncleVasya: hi, how are you doing ? :-) 2011-12-20T12:05:17 *** tom____ is now known as tomb___ 2011-12-20T12:05:57 the speed at which this is unfolding is stressful 2011-12-20T12:06:21 Glacial. 2011-12-20T12:06:28 UncleVasya> So you choose Min for you from every your move and then pick Max score among these Mins. very close 2011-12-20T12:06:35 pairofdice: linux? 2011-12-20T12:06:39 Nope 2011-12-20T12:06:41 http://tcpants.com/player/Accoun12301 let play comrades 2011-12-20T12:06:55 ok 2011-12-20T12:07:27 Accoun: thanks for putting a server up. will join :-) BTW, what's your bot name in the official site ? 2011-12-20T12:07:37 so tcpants is still up? 2011-12-20T12:07:45 im not post on oficial 2011-12-20T12:07:56 to slow insatbile game servers 2011-12-20T12:08:08 no tame to make especial version for it 2011-12-20T12:08:14 no time to make especial version for it 2011-12-20T12:08:28 But what's the point... 2011-12-20T12:08:29 UncleVasya, i calc min scores for 1 my move - search amoun all enemy moves, and then for N my moves with same min take those move which is has max sum of points for all posible enemy moves.. 2011-12-20T12:08:41 *** Raulis has quit IRC (Quit: Page closed) 2011-12-20T12:09:13 *** UncleVasya has quit IRC (Ping timeout: 252 seconds) 2011-12-20T12:09:31 it matrix game 2011-12-20T12:09:32 UncleVasya, so in 2 words take more careful move, and then more agressive from several same careful moves. 2011-12-20T12:09:40 Simplex algoritm for solve it 2011-12-20T12:09:47 worck well 2011-12-20T12:09:49 *** liberforce has left #aichallenge 2011-12-20T12:10:05 oh, he is out 2011-12-20T12:10:08 bottle neck is matrix calculation 2011-12-20T12:10:12 Accoun: you use simplex to solve the payoff matrix? 2011-12-20T12:10:13 GreenTea : that's exactly what I meant by min-max. :-) 2011-12-20T12:10:18 yes 2011-12-20T12:10:22 interesting. 2011-12-20T12:10:29 Ashoka, )) 2011-12-20T12:10:30 bottle neck is matrix calculation 2011-12-20T12:10:39 simplex worck pretty fast 2011-12-20T12:10:43 but aren't there n^5 columns and m^5 rows for n,m ants on each side? 2011-12-20T12:11:12 6 vs 6 practical top for simplex 2011-12-20T12:11:15 bottle neck is matrix calculation 2011-12-20T12:11:16 i see 2011-12-20T12:11:26 so you split the ants somehow 2011-12-20T12:11:27 ti many tame and memory if more 6 vs 6 2011-12-20T12:11:38 *** nha has quit IRC (Ping timeout: 240 seconds) 2011-12-20T12:11:40 yes split 2011-12-20T12:11:47 clastering 2011-12-20T12:12:00 GreenTea : also, sounds like you have done some pruning. (alpha beta cut offs) 2011-12-20T12:12:22 Ashoka, i tried 2011-12-20T12:12:48 Ashoka, when search amoun all moves some times ants make different moves but end position is same 2011-12-20T12:12:56 pity, I didn't have time to test any of those. the version I was trying to fix in last w/e had some of those... 2011-12-20T12:13:11 Ashoka, i tried to make some hash of position 2011-12-20T12:13:24 is that what Complexity is? 2011-12-20T12:13:35 i didn't see alpha/beta cutoff in your code 2011-12-20T12:13:35 a1k0n> to me? 2011-12-20T12:13:39 no, GreenTea 2011-12-20T12:13:45 sorry 2011-12-20T12:14:12 Ashoka, and skip position if position with same cache was executed before 2011-12-20T12:14:33 thats my clustering code: https://github.com/liquid-phynix/aicbot/blob/master/cluster.cpp#L30 2011-12-20T12:14:43 GreenTea: I hear you. to reduce that cost, I tried to calculate using the front line (set of squares ants can move to make a aggresive move) then tried to produce different permutations in that set of squares instead of calculating all possible moves froma 2011-12-20T12:14:51 *** kire has quit IRC (Quit: Leaving) 2011-12-20T12:14:53 Ashoka, this is reduce count of position for, ~ 0.7 2011-12-20T12:15:14 Ashoka, so need to evaluate not 1000 but 700 positions 2011-12-20T12:15:58 Ashoka, but, then i refuse this idea because can't implement good poisition cache 2011-12-20T12:16:04 GreenTea, did u mange that within 500ms? :-) what happens when large scale battles erupts ? 2011-12-20T12:16:16 Ashoka, * position hashcode 2011-12-20T12:16:39 sad news for programmers :/ http://www.stuff.co.nz/travel/australia/6163633/Qantas-terror-blamed-on-computer 2011-12-20T12:16:42 Ashoka, i mean sometimes different positions have same hashcode 2011-12-20T12:17:05 Ashoka, this works only for <7 ants! 2011-12-20T12:17:32 Ashoka, for big battle i have another big script code ^^ 2011-12-20T12:17:34 +2 ants if write on C 2011-12-20T12:17:45 GreenTea: ok, that sounds sensible. 2011-12-20T12:17:53 have it algoritm as 2nd 2011-12-20T12:18:22 Accoun: I was going to say the same. C folks (or even c++) mayu have had a little edge due to performance. :-) 2011-12-20T12:18:39 no die algoritm worc for 8-9 ands and any amount of enemy 2011-12-20T12:18:48 no die algoritm worc for 8-9 ants and any amount of enemy 2011-12-20T12:19:12 my bot is much less complicated and does the same thing in all situations 2011-12-20T12:19:28 often it randomly suicides but overall it works pretty well 2011-12-20T12:19:36 a1k0n> simplex? 2011-12-20T12:19:45 no. i do random sampling. 2011-12-20T12:19:49 The computer code, called an algorithm, 2011-12-20T12:20:11 a1k0n> how many ants group calculate? 2011-12-20T12:20:16 a1k0n> how many ants in group calculate? 2011-12-20T12:20:34 Accoun: i don't group them. all ants are considered together. 2011-12-20T12:20:47 a1k0n: you have made a big improvement in comparatively shorter time. :-) 2011-12-20T12:20:51 i define a lattice, so that each ant has a neighbor to the north and to the west 2011-12-20T12:21:06 and the optimal move for that ant depends on what its neighbors are doing 2011-12-20T12:21:16 a1k0n: so its sub-optimal 2011-12-20T12:21:25 i mean you create an ordering 2011-12-20T12:21:28 (which means it's better at combat in some directions than others) 2011-12-20T12:21:35 it's nowhere close to optimal 2011-12-20T12:21:46 but still it rocks :D 2011-12-20T12:22:00 a1k0n: you wanna try my noptification? 2011-12-20T12:22:01 a1k0n> let play 2011-12-20T12:22:14 my code is confusing as hell because of the way i incrementally update the evaluation score 2011-12-20T12:22:18 (i hope at least you are on linux) 2011-12-20T12:22:32 Accoun: http://tcpants.com/replay.17313 2011-12-20T12:22:33 a1k0n : interesting.. how did you manage when the army is bigger. I grouped combative ants into different mutually exclusive combat groups, that help reduce the combination count. 2011-12-20T12:22:36 stole your base :) 2011-12-20T12:22:46 have 1 game 2011-12-20T12:22:58 i know, i'm waiting for more games 2011-12-20T12:23:18 Ashoka: as i said, it makes no distinction 2011-12-20T12:23:26 more ants -> fewer samples in the time limit 2011-12-20T12:24:04 it 1-1 server 2011-12-20T12:24:05 i think a lot of the bot strength comes from the food collection and routing heuristics more than the exact combat method 2011-12-20T12:24:11 ned make agresive variant 2011-12-20T12:24:29 oh wow tcpants went to 1v1 maps only? 2011-12-20T12:24:39 Accoun: why the hell are you so active now? you had your chance to make a bot and upload it! 2011-12-20T12:25:02 he has a beef with the contest organizers and is boycotting it or something 2011-12-20T12:25:17 he is only boycotting himself 2011-12-20T12:25:28 i can imagine he could have made a good bot 2011-12-20T12:25:50 i have good bot 2011-12-20T12:26:03 byt u don have good game servers 2011-12-20T12:26:22 Accoun, can you give jar? 2011-12-20T12:26:28 *** dici has joined #aichallenge 2011-12-20T12:26:37 Accoun, or exe.. 2011-12-20T12:26:42 GreenTea> can 2011-12-20T12:27:07 GreenTea> let play 2011-12-20T12:27:09 the problem is, that now we cant be sure that he is using his own code 2011-12-20T12:27:14 http://tcpants.com/player/Accoun12301 2011-12-20T12:27:33 sorry ,need to go 2011-12-20T12:27:36 *** GreenTea has quit IRC (Quit: Page closed) 2011-12-20T12:28:06 *** valydo has joined #aichallenge 2011-12-20T12:28:41 I had 5 precomputed maps of enemy moves as precondition that all enemies will move in one direction or stay in place, after that split space on 5*5 squares, do all possible moves for my ants(max 5) and get minimax estimate. Works not bad. 2011-12-20T12:29:18 anyone interested in my battle? 2011-12-20T12:29:30 it isnt especially bad, but it extremely fast 2011-12-20T12:29:31 *** onion42 has joined #aichallenge 2011-12-20T12:30:25 http://tcpants.com/replay.17376 1 more, now i win 2011-12-20T12:30:31 yep 2011-12-20T12:30:36 your combat is definitely better 2011-12-20T12:31:17 it tuned for 3 enemes now 2011-12-20T12:31:19 a1k0n : judging by your success at a shorter time, I guess your strategy may be correct after all :-) still not so clear how to justify random sampling as the enemy would select a better move for him rather than random. but I may have missed few points. :-) 2011-12-20T12:31:54 *** amstan has joined #aichallenge 2011-12-20T12:31:54 *** ChanServ sets mode: +o amstan 2011-12-20T12:32:09 thanks but I compensate it by terrible ant distribution :) 2011-12-20T12:32:13 Ashoka: i honestly wasn't expecting it to do that well. i just assumed the state space was too big to explore exhaustively, so i explore it with directed randomness 2011-12-20T12:32:38 of course you can explore exhaustively if you split your ants up 2011-12-20T12:33:23 now that im here i wished to have visited this channel before :D 2011-12-20T12:34:05 battle wasnt discussed in this depth before the deadline 2011-12-20T12:34:08 *** dr0b3rts has joined #aichallenge 2011-12-20T12:34:20 *** Migi32 has joined #aichallenge 2011-12-20T12:34:20 BTW, anybody knows who Q is ? looks like he plays pretty well. :-) 2011-12-20T12:34:37 Q is FlagCapper 2011-12-20T12:34:38 Ashoka: FlagCapper 2011-12-20T12:34:38 dici: welcome! 2011-12-20T12:34:57 thanks :) will be here for the next challenge 100% sure :) 2011-12-20T12:35:06 i suspect my bot is pretty bad at 1v1 on tcpants 2011-12-20T12:36:03 Accoun: whoa, you were robbed 2011-12-20T12:36:15 "ants not razing hills" 2011-12-20T12:36:50 a1k0n : aha! that brilliant 16 year old. :-) now I just saw olexander (sasha.sochka) also is 16. felling like we are already very old! 2011-12-20T12:36:50 ) 2011-12-20T12:37:28 oh really? 2011-12-20T12:38:05 a1k0n: from the last replay it almost seemed accoun had no battle 2011-12-20T12:38:25 17432 2011-12-20T12:38:28 i always lose on that map 2011-12-20T12:38:34 *** Surya_ has quit IRC (Quit: Page closed) 2011-12-20T12:38:54 a1k0n: do you do any type of map processing? 2011-12-20T12:38:55 you ate him up 2011-12-20T12:39:05 looking for choke points or dead ends? 2011-12-20T12:39:24 hi McLeopold thanks for the great contest 2011-12-20T12:39:28 *** gustavokambara has joined #aichallenge 2011-12-20T12:39:42 mcstar: no, he ate me up 2011-12-20T12:39:46 McLeopold: nope 2011-12-20T12:40:10 a1k0n: lol, im sorry he did, im colorblind 2011-12-20T12:40:21 as in stupid 2011-12-20T12:41:22 a1k0n: you couldnt dispatch your ants to the battle fronts in maze maps? 2011-12-20T12:41:57 i don't really dispatch ants. they follow a heuristic function, which gets spread too thin on a maze 2011-12-20T12:42:26 i see 2011-12-20T12:42:27 Heh, can't tell if i played a game or if server got slower 2011-12-20T12:42:35 *** mj41 has quit IRC (Ping timeout: 252 seconds) 2011-12-20T12:42:44 at the very end i added a priority queue for my clusters 2011-12-20T12:42:58 so the clusters are sorted by average distance from my hive 2011-12-20T12:43:01 *** fnord____ has joined #aichallenge 2011-12-20T12:43:03 any of my hives 2011-12-20T12:43:10 *** denes has quit IRC (Ping timeout: 258 seconds) 2011-12-20T12:43:10 *** peerai has quit IRC (Ping timeout: 258 seconds) 2011-12-20T12:43:24 and i must send ants to those in order, of a given number according to a response function 2011-12-20T12:43:53 the problem is, all these things are last minute, and i didnt have the time to optimize the constants 2011-12-20T12:44:42 Elderwolf: are you on linux? 2011-12-20T12:45:10 mcstar: depends on what you mean by "You" 2011-12-20T12:45:21 (damn replays keep changing yours and account's colors) 2011-12-20T12:45:30 Elderwolf: do you use linux? 2011-12-20T12:45:39 I use linux 2011-12-20T12:45:50 I'm currently on a mac ssh'd into a linux chat session 2011-12-20T12:45:51 you want to try my skill-checker script? 2011-12-20T12:46:00 thats not good 2011-12-20T12:46:07 it needs notify-send 2011-12-20T12:46:11 mcstar: not really, i'm busy pretending to work at my job 2011-12-20T12:46:15 k 2011-12-20T12:46:27 http://tcpants.com/replay.17505 let play 2011-12-20T12:46:32 seems im going to be the only user 2011-12-20T12:46:52 *** dvladim has quit IRC (Ping timeout: 255 seconds) 2011-12-20T12:46:58 Accoun: hey, did you submit a bot? I was wondering where you were 2011-12-20T12:47:12 dont 2011-12-20T12:48:05 I see that some bots get way more games than others. I was wondering: is this going to fix itself as the contest goes on? Since we do the ranking based on mu-3*sigma, it's kind of important. 2011-12-20T12:48:34 I mean, xathis is currently #104, but he would be in the top 10 if it was ranked only on mu 2011-12-20T12:48:48 Migi32: he has only played 6 games 2011-12-20T12:48:53 and won all of them 2011-12-20T12:49:14 f4hy: I know. But my question is: why has he only had 6 games and others have had double of that? 2011-12-20T12:49:42 *** kost_ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T12:49:54 Migi32: he's not cool enough 2011-12-20T12:50:01 *** Blkt has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2011-12-20T12:50:02 Migi32: When you're that good, you don't need to play as many games. 2011-12-20T12:50:08 last night when I went to bed, I had 4, just woke up, now I have 12 2011-12-20T12:50:10 who is cseyu1? 2011-12-20T12:50:13 on tc pants? 2011-12-20T12:50:15 while my brother has 11 last night and now has 12 2011-12-20T12:50:16 Property of the trueskill matching that ranks a part of the base faster than others to set anchor points 2011-12-20T12:50:20 i get second long games with him 2011-12-20T12:50:26 :@ 2011-12-20T12:50:38 I still can't get this script to send my phone a text 2011-12-20T12:50:44 It should even out 2011-12-20T12:50:47 annoying :/ 2011-12-20T12:50:48 Migi32: do you know Darhuuk? 2011-12-20T12:51:00 oh wait, different univ 2011-12-20T12:51:01 anyone here good with curl? 2011-12-20T12:51:32 #curl 2011-12-20T12:51:33 Garf: all I know is that he has a better bot than me 2011-12-20T12:51:42 Garf: but I do know qbraet and mdelbar 2011-12-20T12:52:18 conor_f: you need to add your pc a GSM antenna 2011-12-20T12:52:38 *** xar0l_ has joined #aichallenge 2011-12-20T12:53:06 mcstar: :P 2011-12-20T12:53:25 ty Anilm3 2011-12-20T12:56:21 meh, I end up fixing my own reported issue *refers to pull request https://github.com/aichallenge/aichallenge/pull/390 * 2011-12-20T12:57:19 avdg: do you know how to compile the jar? 2011-12-20T12:57:31 i haven't done that much 2011-12-20T12:57:40 aichallenge: Anthony Van de Gejuchte epsilon * r557ecda / ants/visualizer/js/Application.js : Skip onkeydown event when an input element got focus (fixes #360) - http://git.io/vjBXhQ 2011-12-20T12:57:41 aichallenge: Alexandru Stan epsilon * rbace5e8 / ants/visualizer/js/Application.js : 2011-12-20T12:57:41 aichallenge: Merge pull request #390 from avdg/fix/skipEventOnInputFocus 2011-12-20T12:57:41 aichallenge: Skip onkeydown event when an input element got focus (fixes #360) - http://git.io/xHr5Mg 2011-12-20T12:57:52 no, I didn't done that so much 2011-12-20T12:57:54 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-12-20T12:59:09 *** epicmonkey has joined #aichallenge 2011-12-20T12:59:17 *** ftjansen has joined #aichallenge 2011-12-20T12:59:35 i lived for 500 turns against accoun 2011-12-20T13:00:02 *** keith_ has joined #aichallenge 2011-12-20T13:00:19 its a strong bot 2011-12-20T13:00:42 Accoun: in the tcp, turntime is 5000 ! 2011-12-20T13:01:04 sites slowing dowwwwwn 2011-12-20T13:02:24 *** locksmith3 has quit IRC (Quit: User Quit) 2011-12-20T13:03:38 mcstar : after many 0 second games with cseyu1 I'm finally having a game with u. 2011-12-20T13:03:55 same here 2011-12-20T13:04:00 who is that guy? 2011-12-20T13:04:24 someone offline I guess. :-) 2011-12-20T13:04:27 ban please it it crashed bots 2011-12-20T13:04:41 Accoun: how? 2011-12-20T13:04:57 is tcpants your server? 2011-12-20T13:05:16 Alright, who broke the official website 2011-12-20T13:05:39 Elderwolf: ? 2011-12-20T13:05:54 16B hang 2011-12-20T13:05:58 amstan: just can't get to aichallenge.org any longer 2011-12-20T13:06:09 Elderwolf: works for me 2011-12-20T13:06:14 i get 16 bytes 2011-12-20T13:06:22 ah ok, now it loads 2011-12-20T13:06:32 avdg: do you have exotic browsers like opera or ie? 2011-12-20T13:06:40 avdg: can you test you change in there 2011-12-20T13:06:42 mcstar: finally loade dfor me too 2011-12-20T13:07:06 *** hanswurst_ has joined #aichallenge 2011-12-20T13:07:19 i have a feeling accoun's keeping his account hidden now, because he's affraid that we might change his score 2011-12-20T13:07:34 XD 2011-12-20T13:07:50 amstan: I can't test ie 2011-12-20T13:07:51 he must be cloaked as a canadian for sure 2011-12-20T13:07:57 amstan: if it's already live, it doesn't work in opera 2011-12-20T13:08:05 *** NotABug has quit IRC (Ping timeout: 258 seconds) 2011-12-20T13:08:10 rwest: doesn't work how? 2011-12-20T13:08:24 rwest: also.. hard refresh 2011-12-20T13:08:41 this was a change to make it so you can use arrow buttons in input box of game replay page right? 2011-12-20T13:08:59 rwest: well, my test was with f 2011-12-20T13:09:05 amstan: but I've based my code on jquery, so I assume it will work correctly (though I will still test it in opera now) 2011-12-20T13:09:09 if you press f and the visualizer is focused it should fullscreen 2011-12-20T13:09:19 if you select the user search and type someone with f it should work 2011-12-20T13:09:27 amstan: nvm, must have been cached 2011-12-20T13:09:31 opera is fine 2011-12-20T13:09:37 Ashoka: nice 2011-12-20T13:09:43 *** valydo has quit IRC (Ping timeout: 252 seconds) 2011-12-20T13:09:49 where were you ranked? 2011-12-20T13:09:53 ok, i'm satisfied 2011-12-20T13:09:59 amstan: not in chrome on ubuntu. If you select the user seach and type f, it still goes to full screen. 2011-12-20T13:10:07 i have chrome and it works 2011-12-20T13:10:09 refresh 2011-12-20T13:10:20 :-) 2011-12-20T13:10:26 ah, k, now it's fixed :) 2011-12-20T13:10:30 I was afraid it didn't work for a minute 2011-12-20T13:10:31 Migi32: linux? 2011-12-20T13:10:40 wanna try my notification script? 2011-12-20T13:11:10 i'm proud to say that IE is only used by 2.75% of the visitors 2011-12-20T13:11:14 even android has more 2011-12-20T13:11:31 mcstar: no thanks, I've got some work to do still, and if it kept notifying me every time my skill changes, it would distract me too much I think :) 2011-12-20T13:11:44 oh 2011-12-20T13:11:50 it changes very rarely 2011-12-20T13:11:51 amstan: developers know better 2011-12-20T13:11:54 unfortunately 2011-12-20T13:12:06 *** Belerafon_L has joined #aichallenge 2011-12-20T13:12:16 also, I'm at #15 right now and I know it'll only go down from here... 2011-12-20T13:12:23 oh yes! 2011-12-20T13:12:31 contious integration from github: https://github.com/blog/1013-janky 2011-12-20T13:13:00 :-) 2011-12-20T13:13:57 mcstar : your battles seem improved, think you were bit too cautious early in that game 2011-12-20T13:14:22 *** valydo has joined #aichallenge 2011-12-20T13:14:40 well thanks 2011-12-20T13:14:53 i dont remember what was your rank 2011-12-20T13:15:00 ~50? 2011-12-20T13:15:07 maybe better? 2011-12-20T13:15:59 Ashoka: you were eashoka right? cant find it now 2011-12-20T13:17:25 mcstar: eashoka was me at beta stage. now i'm ender. 2011-12-20T13:18:26 *** hanswurst_ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T13:18:32 damn, my machine hanged (hdd read failure) and the bot timed out. shouldn't play at tcpservers using this pc :-( 2011-12-20T13:18:40 a1k0n> by bot need tune up for 1-1 games, u win in that map set, open maps more than 2/3 2011-12-20T13:18:54 *** gustavokambara has quit IRC (Quit: Page closed) 2011-12-20T13:19:36 yeah, my bot isn't so good at mazes 2011-12-20T13:19:44 but open maps it's good at getting food 2011-12-20T13:20:30 *** Belerafon_L has left #aichallenge 2011-12-20T13:20:41 Have people started releasing code yet? 2011-12-20T13:20:54 GeorgeJ: yes, thread on forum 2011-12-20T13:20:55 "I can't believe I wasted 2 months on this. I can't believe I comprehensively wasted 2 moths on this at all!" 2011-12-20T13:21:16 I wish I wasted 2 months 2011-12-20T13:21:23 this sentence-structure is the trademark of this challenge :D 2011-12-20T13:21:37 btw that was ashoka's opening sentence on his blog 2011-12-20T13:21:42 someone wasted more 6 mounth 2011-12-20T13:23:07 *** smiley1983 has quit IRC (Ping timeout: 276 seconds) 2011-12-20T13:23:09 *** Storm__ has joined #aichallenge 2011-12-20T13:23:49 *** flowenol has joined #aichallenge 2011-12-20T13:23:55 Hmm O_o 2011-12-20T13:24:31 mcstar: thanks for visiting. that 2 months was about planet wars. I manged to waste well more than that this time :-) hmmm 2011-12-20T13:24:58 * mcstar admits he wasted much time too 2011-12-20T13:25:12 * avdg only has time to waste atm :/ 2011-12-20T13:25:14 but now i know how not to write code/develop software 2011-12-20T13:25:39 avdg: LOL 2011-12-20T13:25:45 oh, and paradoxically, im slimmer than ever 2011-12-20T13:25:49 from sitting much 2011-12-20T13:26:33 mcstar: remember you said you are finishing with academics and sending CVs around.... :-) 2011-12-20T13:26:40 I don't understand how it is wasted time 2011-12-20T13:26:54 yep, still true 2011-12-20T13:27:08 im sending my cv to ericsson soon 2011-12-20T13:28:18 pairofdice: i know i WASTED time 2011-12-20T13:28:32 because i spent too much time sometimes thinking about these problems 2011-12-20T13:28:39 and it was waste(TM) 2011-12-20T13:28:44 really 2011-12-20T13:28:49 Your brain is stronger now 2011-12-20T13:29:01 i know that you cant think that good if you spend too much time on 1 problem 2011-12-20T13:29:05 that was a mistake i made 2011-12-20T13:30:07 i dont think my brain is any better now, but i certainly not going to write a program in the way i did for this challnge, sometimes i had trouble understanding what i did, and for why 2011-12-20T13:30:30 or spending time with other frustrations 2011-12-20T13:30:38 it was hard to keep several thousand lines of spaghetti code in my mind 2011-12-20T13:31:01 * avdg learned about folding in vim while he was coding 2011-12-20T13:31:15 anyway it was fun 2011-12-20T13:31:15 i stopped using fold 2011-12-20T13:31:25 i just put things into different files 2011-12-20T13:31:32 and have them open on separate workspaces 2011-12-20T13:35:02 *** Scryer has joined #aichallenge 2011-12-20T13:35:22 i feel like a neutrino detector 2011-12-20T13:35:31 waiting for the "event" 2011-12-20T13:35:44 i.e. my red notification saying hey your skill improved 2011-12-20T13:35:45 To: the person who made a script to web text him his skill level (or anyone else): 2011-12-20T13:36:16 conor_f: it was rwest 2011-12-20T13:36:18 what did you do about AJAX? I'm guessing that thats how the text was sent, curl won't handle it though :/ 2011-12-20T13:36:22 ty mcstar 2011-12-20T13:36:25 sending email or something 2011-12-20T13:36:39 yeah, I want it to do a web text 2011-12-20T13:37:04 I managed to log in and that using curl, now all I have to do is work around ajax :/ 2011-12-20T13:37:28 flowenol has the one that texts him, but it uses a gateway in portugal 2011-12-20T13:37:46 conor_f: I don't think we're using AJAX anywhere on the site 2011-12-20T13:37:54 at least not that I can think of offhand 2011-12-20T13:38:00 *** kilae has joined #aichallenge 2011-12-20T13:38:08 *** olexs has joined #aichallenge 2011-12-20T13:38:31 for the text part janzert 2011-12-20T13:38:42 nope 2011-12-20T13:38:45 on the profile page? 2011-12-20T13:38:48 no 2011-12-20T13:38:56 for the text as in a mobile phone text 2011-12-20T13:39:06 ahh, you're sending 2011-12-20T13:39:11 yeah 2011-12-20T13:39:16 conor_f: you in the US? 2011-12-20T13:39:19 no 2011-12-20T13:39:22 europe 2011-12-20T13:39:28 thought you were trying to work around it from the aichallenge site 2011-12-20T13:39:36 I know all US carriers have emails addresses for their phone numbers that convert to text 2011-12-20T13:39:44 dunno about other countries 2011-12-20T13:39:52 so I email my phone 2011-12-20T13:39:53 a1k0n: http://tcpants.com/replay.17880 2011-12-20T13:39:56 and it comes as a text 2011-12-20T13:39:59 hmm 2011-12-20T13:40:14 after turn 250 see how i defend from your attack? 2011-12-20T13:40:18 i think it was nice 2011-12-20T13:40:23 well, my ISP doesn't let me run a mailserver :/ 2011-12-20T13:40:47 200 turns of hardcore counter-attack :) 2011-12-20T13:41:05 conor_f: you don't need a mail server, I use gmail 2011-12-20T13:41:14 how? 2011-12-20T13:41:16 http://pastebin.com/TtKfVq1R 2011-12-20T13:41:24 smtp 2011-12-20T13:42:00 yeah, I'm not allowed to have outbound SMTP afaik... 2011-12-20T13:42:32 I am pretty sure they can't block that 2011-12-20T13:42:41 ok, maybe not 2011-12-20T13:42:49 *** Redgis has joined #aichallenge 2011-12-20T13:42:57 it connects to googles smtp and uses it 2011-12-20T13:43:02 with a gmail account 2011-12-20T13:43:10 could use your isp account too 2011-12-20T13:43:14 just replace the details 2011-12-20T13:43:21 all I know is that when I tried to use php's mail() after setting up a mail server and that, I still couldn't email external addresses... 2011-12-20T13:43:59 *** Belerafon has joined #aichallenge 2011-12-20T13:44:09 :( 2011-12-20T13:44:16 that is different 2011-12-20T13:44:18 I don't think I'll be able to do all this with curl lol 2011-12-20T13:44:31 you need a mail server running on the pc that is run on with things configured right 2011-12-20T13:44:49 *** Surya has joined #aichallenge 2011-12-20T13:45:04 *** Khaosza has joined #aichallenge 2011-12-20T13:46:31 the email version was rwest 2011-12-20T13:46:36 the sms version is mine 2011-12-20T13:46:52 here i cant send sms via email 2011-12-20T13:46:58 so i setup my own gateway 2011-12-20T13:47:25 oh rwest thanks for the note on the sleep, mine was correct maybe it was a problem with the paste 2011-12-20T13:47:41 *** Redgis has quit IRC (Ping timeout: 268 seconds) 2011-12-20T13:47:49 flowenol: yea, I saw mine had crashed when i got home 2011-12-20T13:47:56 and I looked and noticed the sleep was offest 2011-12-20T13:47:58 rwest i had some other problems 2011-12-20T13:48:19 and added the difference between me and the first place in portugal 2011-12-20T13:48:25 since thats my goal 2011-12-20T13:50:04 i could kill to see my bot play a few more games 2011-12-20T13:50:32 *** Ashoka has quit IRC (Quit: Page closed) 2011-12-20T13:51:00 Is there any ideea for the next ai challenge? When do the challenges usually start? 2011-12-20T13:51:04 *** Darhuuk has quit IRC (Ping timeout: 276 seconds) 2011-12-20T13:51:32 i dont know and dont want to know 2011-12-20T13:51:41 flowenol: why not? 2011-12-20T13:51:42 because im going to get addicted again 2011-12-20T13:51:46 haha 2011-12-20T13:51:52 true 2011-12-20T13:51:58 i sleep a lot better now 2011-12-20T13:52:00 still 2011-12-20T13:52:02 I want to know before there is only a cpl weeks left 2011-12-20T13:52:15 I want a chance at top 10 next time heh 2011-12-20T13:52:19 how do you get sms messages to be sent on rank updates ? 2011-12-20T13:52:39 a script that checks every few mins 2011-12-20T13:52:48 and then uses my own gateway 2011-12-20T13:52:59 *** Darhuuk has joined #aichallenge 2011-12-20T13:53:07 i think a desktop notification is cheaper than an sms :D 2011-12-20T13:53:17 *** dodecahedron_ has quit IRC (Quit: Page closed) 2011-12-20T13:53:17 besh: what carrier do you use? 2011-12-20T13:53:17 true 2011-12-20T13:53:21 i dont pay sms for myself 2011-12-20T13:53:22 :D 2011-12-20T13:53:27 sms is free for me 2011-12-20T13:53:28 at&t 2011-12-20T13:53:28 flowenol: could your gateway send to ireland? :P 2011-12-20T13:53:34 *** icosahedron has joined #aichallenge 2011-12-20T13:53:37 conor_f yes but i would pay them 2011-12-20T13:53:45 haha, ok, i dont use phones much 2011-12-20T13:53:48 :( 2011-12-20T13:53:52 if you are willing to pay, there is clickatell 2011-12-20T13:53:58 no :P 2011-12-20T13:54:02 then nope 2011-12-20T13:54:04 besh: email to: (yournumber)@txt.att.net 2011-12-20T13:54:12 unless you can email them 2011-12-20T13:54:17 or code your own gateway 2011-12-20T13:54:17 no 2011-12-20T13:54:36 mine is a specific implementation for my operator 2011-12-20T13:54:48 rwest: where do i set the email address ? 2011-12-20T13:54:55 they have a softphone, so i coded a new client 2011-12-20T13:55:10 are you looking at the paste I put up? 2011-12-20T13:55:19 no 2011-12-20T13:55:42 http://pastebin.com/TtKfVq1R 2011-12-20T13:55:56 python script that emails upon skill update 2011-12-20T13:55:59 btw 2011-12-20T13:56:09 you may want to add a try catch 2011-12-20T13:56:19 so if something fails it doesnt die 2011-12-20T13:56:22 flowenol: true 2011-12-20T13:56:28 that was a problem on mine 2011-12-20T13:56:46 you may also want to change PT Rank 2011-12-20T13:56:55 to (your country code) Rank 2011-12-20T13:57:01 just for eye candy 2011-12-20T13:57:03 heh 2011-12-20T13:57:21 * flowenol is impatient with his bot 2011-12-20T13:57:27 besh: that script uses gmail, if you have gmail your set 2011-12-20T13:57:36 you can change it to any server though 2011-12-20T13:57:50 Ok i am trying it now. 2011-12-20T13:57:54 just change the server and disable tls if not supported 2011-12-20T13:58:08 *** FranPeruzzi has quit IRC (Quit: Page closed) 2011-12-20T13:58:16 don't forget to modify the profile link too 2011-12-20T13:58:27 or you will get rwest rank :P 2011-12-20T13:58:30 and country 2011-12-20T13:58:43 his country is US too :) 2011-12-20T13:58:57 then he is set on that 2011-12-20T13:59:29 damn i want to see my bot rise to first 2011-12-20T13:59:41 this obsession cant be healthy 2011-12-20T14:00:08 and the more games xathis plays, the lower the changes ;-) 2011-12-20T14:00:13 *chances 2011-12-20T14:00:28 he means first in his country probably 2011-12-20T14:00:31 avdg well i dont care about that first :P 2011-12-20T14:00:31 XD 2011-12-20T14:00:36 mcstar exactly 2011-12-20T14:00:46 i was there before the reset 2011-12-20T14:00:51 i will return on TCP in 45 min 2011-12-20T14:00:52 i want to go back ther 2011-12-20T14:00:55 *** Accoun has quit IRC () 2011-12-20T14:01:00 i were only 2nd in mine 2011-12-20T14:01:03 was 2011-12-20T14:01:37 im going to be a little annoyed if i dont make it 2011-12-20T14:01:43 im a bad loser 2011-12-20T14:03:50 http://aichallenge.org/visualizer.php?game=294216&user=8217 2011-12-20T14:03:59 lol watch me fail when there's no food around :P 2011-12-20T14:04:26 flowenol: I see the bug in the country rank, was missing a break 2011-12-20T14:04:53 exactly 2011-12-20T14:05:01 Ok i changed username,password,my profile page and country profile page. Anything else ? 2011-12-20T14:05:28 add a break after ptRank = content[:content.find('(')].strip() 2011-12-20T14:05:32 the next line 2011-12-20T14:05:36 username & password is for my gmail account,right ? 2011-12-20T14:05:39 or your country rank will be wrong 2011-12-20T14:05:41 besh yes 2011-12-20T14:06:01 Someone has that tuned to notice when xathis gets into a game? I know there's stalkers here! 2011-12-20T14:06:22 hah 2011-12-20T14:06:53 *** GreenTea has joined #aichallenge 2011-12-20T14:07:21 *** xathis has quit IRC (Quit: Page closed) 2011-12-20T14:08:38 is this a hatred? :) 2011-12-20T14:08:58 hey, I've got an idea for the next contest and would like some feedback before posting it on the forums: it's Haxball (http://haxball.com/), which is basically a simplified form of soccer. Each bot controls 11 circles (which move quite slow, but can kick the ball if they touch it) and the first bot to score 3 goals or something wins 2011-12-20T14:08:59 who is winning? 2011-12-20T14:09:08 it's physcis based, with continuous instead of discrete states, which I think would be really cool for an AI contest. 2011-12-20T14:09:17 A big plus is also that the physics are really simple. There are only circles and straight walls. 2011-12-20T14:09:27 We could even make it real-time, but that's probably a step too far, because of latency issues etc (especially on TCP servers). 2011-12-20T14:09:30 Is there any news about the cut of low ranks bots date? (starter packs) 2011-12-20T14:10:07 meh, flash is broken in chrome (can't type in input boxes) :/ 2011-12-20T14:10:19 avdg: not for me 2011-12-20T14:10:39 but Migi32, that was your idea for this contest already ;) 2011-12-20T14:10:52 mleise: yes, it was for this year :) 2011-12-20T14:10:53 rwest: "ImportError: No module named pyquery" 2011-12-20T14:11:03 mleise: but it obviously didn't make it :) 2011-12-20T14:11:11 I still think it could really be interesting though 2011-12-20T14:11:49 http://www.youtube.com/watch?v=z_xBdFg1RcI <-haxball 2011-12-20T14:12:00 Migi32: id like continuous space/time 2011-12-20T14:12:43 *** Anilm3 has quit IRC (Read error: Connection reset by peer) 2011-12-20T14:13:06 Some variation on that theme could be a winner 2011-12-20T14:13:26 it can perfectly be real time 2011-12-20T14:13:28 The queue comp is pretty similar aswell 2011-12-20T14:13:37 you only need an event-simulator 2011-12-20T14:13:39 After what felt like a century, I got two games in a row. But against same set of opponents ?? 2011-12-20T14:13:50 Migi32, have played this game) 2011-12-20T14:14:14 yeah, continuous things would be really interesting. There are 2 issues we'd need to think about though: rounding errors (for continuous space) and latency things (for continuous time) 2011-12-20T14:14:31 i like that idea :) 2011-12-20T14:15:07 * avdg isn't creative in choosing nicks (hints "blah") 2011-12-20T14:15:11 Migi32: no 2011-12-20T14:15:20 i mean the event are discrete 2011-12-20T14:15:26 events* 2011-12-20T14:15:35 like molecular dynamics 2011-12-20T14:15:39 besh well you are gonna need pyquery module 2011-12-20T14:15:40 you dont compute everything 2011-12-20T14:15:59 but i think this game would be perfect for some genetic thingy 2011-12-20T14:16:11 flowenol: i have python27 installed. where do i get that ? 2011-12-20T14:16:19 It is posible to make it continuos.. 2011-12-20T14:16:26 ah yeah, true. I have actually no idea what a "perfect" strategy for this game would be 2011-12-20T14:16:33 well all i did was emerge -av pyquery but then, my linux distro is gentoo 2011-12-20T14:16:34 for example 2011-12-20T14:16:59 besh http://pypi.python.org/pypi/pyquery 2011-12-20T14:17:00 there 2011-12-20T14:17:09 every 500ms bot gets on imput positions, vectors of speeds, and accelerations of all players and bolls 2011-12-20T14:17:11 Migi32: so what are playing on the video? humans? 2011-12-20T14:17:18 *** Anilm3 has joined #aichallenge 2011-12-20T14:17:20 mcstar: yes 2011-12-20T14:17:25 ah ok. 2011-12-20T14:17:41 *** jstemmer has joined #aichallenge 2011-12-20T14:17:42 and each player is 1 human. So in a 3v3 game there are actually 6 people playing 2011-12-20T14:17:42 then bot have ability to change this vectors of accelerations for it bots 2011-12-20T14:18:06 * for it players 2011-12-20T14:18:11 so far this seems to be an excellent idea 2011-12-20T14:18:20 i wanted something 3d and fps :D 2011-12-20T14:18:35 *** Accoun has joined #aichallenge 2011-12-20T14:19:06 and during replay visulized shows game smootly, without pauses of bot thinking 2011-12-20T14:20:17 So, this game can be step by stepwise but continous during visualization 2011-12-20T14:20:21 GreenTea a nice idea 2011-12-20T14:20:24 http://tcpants.com/player/Accoun12301 2011-12-20T14:20:29 Just like queue comp 2011-12-20T14:20:49 It just interpolates the missing bits (3 frames for each turn) 2011-12-20T14:21:35 But I still want explosions :< 2011-12-20T14:21:51 *** bretep has joined #aichallenge 2011-12-20T14:22:01 GreenTea: yes, that's a possibility. Do we make the computation time equal to the simulated game time? There's technically no reason to, but it would allow you to follow games at normal speed as they're being played. 2011-12-20T14:22:06 And I still want Risk like game) 2011-12-20T14:22:07 rig your monitor to blow into your face 2011-12-20T14:22:24 But that would be dangerous! 2011-12-20T14:22:29 *** conor_f has quit IRC (Ping timeout: 240 seconds) 2011-12-20T14:22:47 hmm wait actually "simulated game time" has no real meaning... 2011-12-20T14:22:50 GreenTea: risk would be awesome 2011-12-20T14:23:08 what is risk? 2011-12-20T14:23:24 @google lucky risk game 2011-12-20T14:23:25 pairofdice: http://en.wikipedia.org/wiki/Risk_(game) 2011-12-20T14:23:37 some time ago creator of http://warlight.net/ ask to make it next contest 2011-12-20T14:23:41 GreenTea: however it would be kind of trivial to make a bot to train via genetic algorithms 2011-12-20T14:23:47 isn't risk essentially a bit like planetwars though? But with restricted paths, and randomness? 2011-12-20T14:23:56 hahah 2011-12-20T14:24:00 I really wouldn't want randomness though 2011-12-20T14:24:05 they played that game in the hostel 2011-12-20T14:24:11 i mean my friends, but i hated it 2011-12-20T14:24:16 its called Riziko 2011-12-20T14:24:25 randomnes is not big issue 2011-12-20T14:24:44 just need to play more games 2011-12-20T14:24:49 mcstar: yeah you definitely killed me good there, i just overwhelmed you with numbers 2011-12-20T14:24:53 GreenTea: when playing risk, it really does feel like a big issue :) 2011-12-20T14:25:05 *** Paradoxial has joined #aichallenge 2011-12-20T14:25:06 so i guess my bot is actually really bad at combat, it just collects food efficiently 2011-12-20T14:25:07 this games will be much quicker than Ants games.. 2011-12-20T14:25:25 a1k0n: the problem is, that i though mine was good in collection too 2011-12-20T14:25:30 t 2011-12-20T14:25:44 but you always kill me :( 2011-12-20T14:25:46 i had more ants somehow 2011-12-20T14:26:00 and given that i waste so many in combat it's gotta be from eating food.. :) 2011-12-20T14:26:11 *** Corgano has joined #aichallenge 2011-12-20T14:26:19 Migi32. hehe) but good players have win rate over 80% and randomes is not worried them, i think) 2011-12-20T14:27:00 win rates should always be around 50% with any decent ranking system ;p 2011-12-20T14:27:15 (1v1) 2011-12-20T14:27:44 Dear servers, please make my bot play a few games, i could kill for a game now 2011-12-20T14:27:55 GreenTea: yeah, true. An AI for a game involving randomness would have to be very different from a game without randomness, but that's not really a bad thing. 2011-12-20T14:27:55 * flowenol starts hitting his head on the wall 2011-12-20T14:28:25 flowenol: kill some time instead 2011-12-20T14:28:29 take a walk watch a movie 2011-12-20T14:28:32 do WORK 2011-12-20T14:28:33 mcstar i cant 2011-12-20T14:28:41 i was working all day :P 2011-12-20T14:28:46 gmt time here 2011-12-20T14:28:53 also, do you guys know about Diplomacy? It's very similar to Risk, but it's focused on communication between players, and you can assist other people's attacks etc. 2011-12-20T14:28:55 Migi32, in such a game I think the most important will be learning how other player moves during game, learn its weak plases and use them, it is not so easy) 2011-12-20T14:29:00 Next challenge: Super Puzzle Fighter 2011-12-20T14:29:09 could be interesting to see how bots would play Diplomacy 2011-12-20T14:29:12 seriously, please, please deactivate all starter bots. sort submissions by md5 of their zip (or even file length) and delete anything with >5 copies 2011-12-20T14:29:25 *** ChrisH_ has joined #aichallenge 2011-12-20T14:29:33 rwest: yes! 2011-12-20T14:29:48 a1k0n couldnt agree more 2011-12-20T14:29:53 my 3 year old kid keeps beating me at that game, cuz i think doing random things is better than trying to play 2011-12-20T14:30:01 a1k0n: haha 2011-12-20T14:30:05 I offered this 2 months ago 2011-12-20T14:30:14 A quick question for anyone who might know: should I be worried that I'm stuck at only 6 games still? I was paired against two top 100 bots two games in a row and it killed my ranking a bit (I'm only shooting for top 500) 2011-12-20T14:30:27 a1k0n btw thanks for sharing your combat technique 2011-12-20T14:30:44 flowenol: sure. i also shared my code, but it may cause headaches and nausea 2011-12-20T14:30:50 rwest, omg what is it) 2011-12-20T14:30:57 a1k0n i havent read it yet, bit I will 2011-12-20T14:31:16 Corgano: I think not. Xathis also only has 6 games, and he's the one who's probably going to win this whole thing 2011-12-20T14:31:17 and dont worry, if it was not easy to write, cant be easy to read 2011-12-20T14:31:27 it was easy to write :( 2011-12-20T14:31:31 damn 2011-12-20T14:31:41 I want to understand wher I failed implementing your code 2011-12-20T14:31:48 I got a sort of works version 2011-12-20T14:31:51 but not great 2011-12-20T14:31:56 i think the heuristic function is important 2011-12-20T14:31:56 Migi32: Thanks! 2011-12-20T14:32:00 * flowenol away 2011-12-20T14:32:06 a1k0n same 2011-12-20T14:32:12 mine is too slow 2011-12-20T14:32:13 and i didn't really share mine 2011-12-20T14:32:14 brb 2011-12-20T14:33:03 GreenTea: http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=3&ved=0CD4QtwIwAg&url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DuzhKNzjz4BM&ei=2uLwTu_DN-Tf0QH94LS3Ag&usg=AFQjCNGVh74aJqRzi0BD_6CsKLqmaQqb0Q&sig2=BIxqzadDYKSsznmeIH80Cw 2011-12-20T14:33:11 *** conor_f has joined #aichallenge 2011-12-20T14:33:11 fail at youtube paste 2011-12-20T14:33:19 http://www.youtube.com/watch?v=uzhKNzjz4BM 2011-12-20T14:33:35 *** ChrisH_ has quit IRC (Client Quit) 2011-12-20T14:34:58 you don't know the bliss of the bombs 2011-12-20T14:35:07 sorry, I was away, does this look like it'll be the next game? (The soccer one) 2011-12-20T14:36:02 *** ChrisH_ has joined #aichallenge 2011-12-20T14:36:24 rwest, sorry, dont like 2011-12-20T14:37:05 @later tell janzert I was wondering if it would be a good idea to order the pairings queue by least games played with ties broken by least recently played during finals. Any thoughts on that idea? 2011-12-20T14:37:05 ChrisH_: I think that worked... 2011-12-20T14:37:59 *** petethepig has quit IRC (Quit: Page closed) 2011-12-20T14:38:01 a1k0n: "and i didn't really share mine" how come? 2011-12-20T14:38:05 I think an RTS is the most interesting game type 2011-12-20T14:38:08 you mean you didnt fully comment? 2011-12-20T14:38:26 turn based like ants is close enough 2011-12-20T14:38:30 *** iglo has joined #aichallenge 2011-12-20T14:38:53 How does the cutoff in the finals work? 2011-12-20T14:39:31 mcstar: cuz it was hard to explain 2011-12-20T14:39:38 i mean it's in the code 2011-12-20T14:39:42 but i didn't post in the forum about it 2011-12-20T14:39:53 ok, thought so 2011-12-20T14:41:34 *** breteo has joined #aichallenge 2011-12-20T14:41:48 *** Redgis has joined #aichallenge 2011-12-20T14:43:32 *** bretep has quit IRC (Ping timeout: 258 seconds) 2011-12-20T14:44:06 *** breteo is now known as bretep 2011-12-20T14:47:08 *** epicmonkey has quit IRC (Ping timeout: 240 seconds) 2011-12-20T14:47:32 *** Corgano has quit IRC (Quit: Page closed) 2011-12-20T14:49:32 hehe 2011-12-20T14:49:36 ive got my first notification 2011-12-20T14:49:40 maybe the next challenge should be devising a better ranking and matchup pairing system 2011-12-20T14:50:04 tomb___: whats the issue? 2011-12-20T14:50:36 not funny? joke? no? i'll go back to lurking now. 2011-12-20T14:50:45 i'm so tired of this.. 2011-12-20T14:50:55 The ranking is pretty perfect 2011-12-20T14:50:58 check this out: http://imgur.com/PQcIG upper right corner 2011-12-20T14:51:09 everyone's just suggesting how we should add more rules: "how about you dismiss games that have timeouts near eachother" 2011-12-20T14:51:13 from what i read, it works pretty well 2011-12-20T14:51:16 they don't realize how much work and testing that could cause 2011-12-20T14:51:39 mcstar: nice ;-) 2011-12-20T14:51:54 but if everyone is complaining, the next challenge would be, ok, so you think you can better. 2011-12-20T14:52:08 tomb___: no.. because everyone's complaining now 2011-12-20T14:52:10 everybody will complain, its just a rule 2011-12-20T14:52:23 and when it comes time to actually build the next engine there's going to be like one person doing it 2011-12-20T14:52:40 right. i am not complaining. 2011-12-20T14:52:40 amstan: dont forget that many are thankful too 2011-12-20T14:52:59 amstan: their pride is on the line, god forbid they do poorly because their bot sucked, more likely it is due to the rules and the engine sucking. 2011-12-20T14:53:16 :) 2011-12-20T14:53:20 well, of course 2011-12-20T14:53:26 amstan: yes i am thankful. i tried to help with a few posts on the forums. 2011-12-20T14:53:29 it's just that those posts are the majority of the forums right now 2011-12-20T14:53:40 mcstar: can I see that code plz? :P 2011-12-20T14:53:56 Part of the problem is timeouts are such a black box to people maybe just more visibility into turn times would help. 2011-12-20T14:53:57 conor_f: the notification? 2011-12-20T14:54:06 it doesnt send sms or email 2011-12-20T14:54:31 mcstar: yeah 2011-12-20T14:54:36 the notification 2011-12-20T14:54:38 24 ↑xathis 2011-12-20T14:54:39 please 2011-12-20T14:54:42 he's getting there 2011-12-20T14:55:16 And of course for some languages you don't necessarily have the fine control over the GC etc so it all seems unfair to someone who gets sort of random timeouts. 2011-12-20T14:55:16 conor_f: http://pastebin.com/yaGkjyha 2011-12-20T14:55:26 conor_f: you have to change user=12 accordingly 2011-12-20T14:55:54 As for time-outs, why was CPU-time not used? 2011-12-20T14:56:09 no one implemented it 2011-12-20T14:56:17 tyvm :) 2011-12-20T14:56:23 *** Khaosza has quit IRC (Quit: Page closed) 2011-12-20T14:56:39 I'm curious, if CPU time was used, is there a good way for a bot to know how much CPU time it used since the start of a turn? 2011-12-20T14:56:58 rdtsc 2011-12-20T14:57:04 ChrisH_: there is, but then you get threads, so you need to count for all threads at once 2011-12-20T14:57:16 ChrisH_: and is you do something like sleep(10000) 2011-12-20T14:57:22 it's not counted as cpu time 2011-12-20T14:57:52 *** Manwe56_ has joined #aichallenge 2011-12-20T14:58:14 need 1 core per bot 2011-12-20T14:58:21 mcstar: nice and simple :) 2011-12-20T14:58:22 amstan: so that sounds like the answer is no 2011-12-20T14:58:29 Does 'time.clock()' create a thread in python, or do I misunderstand? 2011-12-20T14:58:36 ChrisH_: it's possible, and ideally that's what we need to do 2011-12-20T14:58:39 Accoun: I think part of the problem now is that they're relying on the fairness of the linux scheduler for that 2011-12-20T14:58:40 just run it as a background program? 2011-12-20T14:58:43 ChrisH_: but that's code, and nobody implemented it yet 2011-12-20T14:59:29 amstan: I understand that, but I'm thinking from the bot implementer's point of view how they could write code to reliably know when their CPU time alottment was about to expire and make sure their orders are sent in time. 2011-12-20T14:59:47 i think it's fairly trivial if you only have 1 thread 2011-12-20T14:59:50 amstan: both pieces need to be doable for the whole system to work together 2011-12-20T14:59:50 you would also need to do it for windows as well as linux and for all the different languages. huge pain in the ass. 2011-12-20T15:00:09 amstan: what if your language uses more than one thread behind your back 2011-12-20T15:00:16 then no idea, lol 2011-12-20T15:00:20 go by clock time 2011-12-20T15:00:23 and you'll always be short 2011-12-20T15:00:23 amstan: so java is out 2011-12-20T15:00:31 wow, the top 100 has really shaped up overnight 2011-12-20T15:00:47 staying under the wall time will pretty much guarantee that you won't time out 2011-12-20T15:00:53 ChrisH_: it's always possible to monitor all the threads 2011-12-20T15:01:04 one possibility is that the engine provides a signal to the bot at some point 2011-12-20T15:01:14 since signal handling will probably be more accessible to more languages 2011-12-20T15:01:18 BenJackson: i'm talking about from inside the bot, not from the engine's perspective 2011-12-20T15:01:25 right 2011-12-20T15:01:36 so the engine does the monitoring so it's 100% consistent 2011-12-20T15:01:44 and then somehow tells the process 2011-12-20T15:01:45 yeah, signals would be nice then 2011-12-20T15:01:50 server could send warning via events or via input stream 2011-12-20T15:01:58 ok, I can believe that 2011-12-20T15:02:03 *** icosahedron is now known as hexahedron 2011-12-20T15:02:17 but then when do we send the signal? 2011-12-20T15:02:24 *** tobel has joined #aichallenge 2011-12-20T15:02:27 what if some people want it earlier or later? 2011-12-20T15:02:36 could be at a fixed point or it could be requested by the client at start 2011-12-20T15:02:44 amstan: you would have to send a hearbeat signal every 10 ms or something 2011-12-20T15:02:51 every 100 ms for example 2011-12-20T15:03:26 of course the bots have to be ready to receive the signal on time for it to be reliable 2011-12-20T15:03:31 even then some langauges probably won't have easy access to nonblocking IO or signals 2011-12-20T15:03:38 *** rebelxt has joined #aichallenge 2011-12-20T15:04:19 *** g0llum has joined #aichallenge 2011-12-20T15:04:34 BenJackson: you're last game had some tough competition in it 2011-12-20T15:04:41 mcstar: just so you know, that notification came up with a load of HTML ;) 2011-12-20T15:04:47 oh the other hand some of the timeouts people hate (like GC) are not simply timing errors, they're real things that are somewhat outside of the player's control 2011-12-20T15:05:37 for that I think something like allocating "5 seconds for 10 turns but no more than 1s per turn" would help 2011-12-20T15:05:42 *** Zaphus has joined #aichallenge 2011-12-20T15:05:57 honestly, for stuff like that i really think it's fair 2011-12-20T15:06:16 ChrisH_: yes, and my only real merit was being far from teapot 2011-12-20T15:06:18 but I'll take it 2011-12-20T15:06:24 in the real world it's the same, if you can't choose a language that's fairly time deterministic, you'll have issues 2011-12-20T15:06:25 If you create starter kit with threads, heartbeat and realtime, only 50-100 players want to play 2011-12-20T15:06:31 encourage people to create less garbage for the GC to cleanup . ;) 2011-12-20T15:06:32 Another way is to give portion of time for several turns. And let's bot decide what to do with it. 2011-12-20T15:06:41 ChrisH_: that too 2011-12-20T15:06:59 I ended up with using c++ anyway 2011-12-20T15:07:13 morning all 2011-12-20T15:07:40 Zaphus arrives and suddenly I look at my profile again to make sure my game mu applied 2011-12-20T15:07:41 and evening too 2011-12-20T15:07:56 ChrisH_: that's another trade off, in java it's faster to create a new array than 0 out a previous one 2011-12-20T15:07:59 I learned from a great blog article that Go is much faster if you don't create a lot of garbage, and using that idea really made my bot faster in many parts 2011-12-20T15:08:02 Zaphus: good midnight 2011-12-20T15:08:08 so I leave more garbage, but get faster run time 2011-12-20T15:08:23 rwest: which means that knowing how your language performs is key 2011-12-20T15:08:31 rwest: I used Go, and it's just the opposite 2011-12-20T15:08:46 ChrisH_: next time I will just use c/c++ and not deal with it 2011-12-20T15:09:05 BenJackson: No danger of that for me, I'd need a game to have it applied.. one game in the last 12 hours, two in the last 26 hours :-( 2011-12-20T15:09:05 rwest: no, instead you'll have to deal with cleaning up garbage manually ;) 2011-12-20T15:09:30 ChrisH_: which is fine, but I will know when I can break out of routines 2011-12-20T15:09:37 *** Garf has quit IRC (Quit: Make a new plan, Stan!) 2011-12-20T15:09:38 ChrisH_ do you have a pointer to that article? 2011-12-20T15:09:38 rwest: yep 2011-12-20T15:09:45 sure, hold on 2011-12-20T15:09:52 *** gustavokambara has joined #aichallenge 2011-12-20T15:10:18 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-20T15:10:25 rwest: I hope you practiced c++ a bit when the next competition is near ;-) 2011-12-20T15:10:44 avdg: it's my job 2011-12-20T15:10:54 avdg: mostly C 2011-12-20T15:11:30 I ran in a lot of problems with the c++ libraries and some less known language features 2011-12-20T15:11:38 jcdny: it is linked from this page (which has lots of good Go stuff): http://golang.org/doc/docs.html, but the specific article is here: http://blog.golang.org/2011/06/profiling-go-programs.html 2011-12-20T15:11:59 *** tomb___ has quit IRC (Quit: Page closed) 2011-12-20T15:12:18 Zaphus: there's a trick to it 2011-12-20T15:12:24 avdg: I have a lot of my own libraries I have written over the years in C, I really don't need C++ libraries 2011-12-20T15:12:30 I lost a game and then I got another game pretty quickly 2011-12-20T15:12:37 but then I won that it it took 3 more hours... 2011-12-20T15:13:00 *** conor_f has quit IRC (Quit: leaving) 2011-12-20T15:13:11 BenJackson: yeah, not losing is a sure way to not get games 2011-12-20T15:13:23 my last 5 games were 3rd place finishes :-/ 2011-12-20T15:13:41 BenJackson: nope, I won a game then it took 14 hours, I lost a game then it took 12 hours... ok, I guess that's quicker but it hardly counts... I really hope they even things out before the competition ends, 'cause I'm only going to end up with 10 games total at this rate, some are already past 25 2011-12-20T15:14:12 *** Westicles has quit IRC (Ping timeout: 258 seconds) 2011-12-20T15:14:14 I assume that now that I'm in 4th I'll be less interesting as a reference for the matchmaker 2011-12-20T15:14:50 ChrisH_ ah yeah I read that before. 2011-12-20T15:14:57 BenJackson: it's not your rank, but a combination of your mu and sigma 2011-12-20T15:14:58 *** ikaros has joined #aichallenge 2011-12-20T15:15:04 right 2011-12-20T15:15:20 BenJackson: if your mu is near other bots that are playing games and your sigma is low, then you are a good pairing for being pulled into games 2011-12-20T15:15:37 my mu isn't near many people 2011-12-20T15:15:52 xathis :-/ 2011-12-20T15:16:39 and me if I win another game ;) 2011-12-20T15:16:44 *** raulis has joined #aichallenge 2011-12-20T15:16:59 and the same for flagcapper 2011-12-20T15:17:05 In the top 500, There are 42 people with only 6 games, the median seems to be 13 games 2011-12-20T15:17:06 Or just xathis if you play xathis :) 2011-12-20T15:17:14 BenJackson: true 2011-12-20T15:17:42 ah teapot had some last minute subs 2011-12-20T15:17:48 but I'll probably play someone with a lower sigma than xathis 2011-12-20T15:17:53 he was stronger in that game than I remember 2011-12-20T15:18:51 ChrisH_: stupid question, but where do you guys see your mu and sigma? 2011-12-20T15:18:59 *** Antimony has joined #aichallenge 2011-12-20T15:19:03 hover the mouse over your skill 2011-12-20T15:19:03 hover your skill 2011-12-20T15:19:07 rebelxt: hover over your skill 2011-12-20T15:19:11 someone like djaarf or nhaehnle 2011-12-20T15:19:16 rebelxt: hover your mouse over your skill 2011-12-20T15:19:16 ahh, thanks all, lol 2011-12-20T15:19:18 doesnt help if you are on am iPad/android tab 2011-12-20T15:19:23 *** kilae has quit IRC (Quit: ChatZilla 0.9.87 [Firefox 8.0/20111104165243]) 2011-12-20T15:19:35 epic battle: http://aichallenge.org/visualizer.php?game=294804&user=12 2011-12-20T15:20:25 nots ure what threshold AgentSmith hit but he didn't waffle after that 2011-12-20T15:20:52 mcstar: wow, never seen a fight like that before 2011-12-20T15:21:04 thx 2011-12-20T15:21:47 BenJackson: spawn control worked very nicely in this game, it's a perfect map for it: http://aichallenge.org/visualizer.php?game=287348&user=10 2011-12-20T15:21:47 yep, great battle 2011-12-20T15:21:51 *** gustavokambara has quit IRC (Quit: Page closed) 2011-12-20T15:22:22 ChrisH_: I think I already saw that one 2011-12-20T15:22:22 *** raulis has quit IRC (Quit: Page closed) 2011-12-20T15:22:45 it was half a day ago, so that's possible if you've been checking up on my bot 2011-12-20T15:22:50 yep 2011-12-20T15:23:15 I don't know my neighbors in real life but I'm keeping track of the bots near me :) 2011-12-20T15:23:28 haha 2011-12-20T15:23:35 lol 2011-12-20T15:23:39 I'm sure a lot of us are. 2011-12-20T15:23:56 BenJackson: didnt waffle? what does that mean? :) 2011-12-20T15:24:02 Top of real life loosers 2011-12-20T15:24:12 lazarant just popped into 3rd 2011-12-20T15:24:55 mcstar: waffling over a decision: having trouble making up your mind 2011-12-20T15:25:03 ah 2011-12-20T15:25:04 slang 2011-12-20T15:25:26 or to change your opinion frequently 2011-12-20T15:25:54 I meant that once he went on the offensive and killed one rank of ants his bot didn't reevaluate and go back to the stalemate on the next turn 2011-12-20T15:26:14 speedy_consoles with a nice win: http://aichallenge.org/visualizer.php?game=295287&user=7935 2011-12-20T15:26:45 *** bluegaspode has joined #aichallenge 2011-12-20T15:27:54 *** erij has joined #aichallenge 2011-12-20T15:28:37 <_flag> So, who else is still at 6 games? 2011-12-20T15:28:37 *** erij has quit IRC (Client Quit) 2011-12-20T15:28:46 BenJackson: my bot, not his 2011-12-20T15:29:00 _flag: 1000 bots still 2011-12-20T15:29:42 whoa i got 4 notifications in 30 minutes 2011-12-20T15:29:56 me is at 6 games still :( 2011-12-20T15:30:09 <_flag> McLeopold: Do you think there will be enough time to determine accurate rankings by Friday? 2011-12-20T15:30:21 _flag: don't know 2011-12-20T15:31:09 We need a cut off so badly 2011-12-20T15:31:14 _flag, I'm at 6. 42 of the top 500 are at 6 2011-12-20T15:31:29 _flag: i'm still at 6 2011-12-20T15:31:31 we never should have allowed starter bots in the finals :( too late now 2011-12-20T15:31:37 ive got 17 games so far 2011-12-20T15:31:38 <_flag> Zaphus: Where are you getting these stats, are you counting them manually? 2011-12-20T15:32:09 _flag, yes, counted manually (you can sort each 100 by games played) 2011-12-20T15:32:17 McLeopold: what do you think of ordering the queue by least games played first, and least recently played second? 2011-12-20T15:32:25 mcstar: shut up! 2011-12-20T15:32:40 o.O 2011-12-20T15:32:44 :-) 2011-12-20T15:33:15 *** xathis has joined #aichallenge 2011-12-20T15:33:36 McLeopold: ignoring the starter bot problem, they could just let the finals run longer, end it at midnight Dec 31st or something 2011-12-20T15:33:47 ChrisH_: first order would be nice, second wouldn't work with the first 2011-12-20T15:34:32 McLeopold: time to re-run your winner projections 2011-12-20T15:34:34 McLeopold: yeah, primary sort by least games played, secondary sort (as a tie breaker) least recently played 2011-12-20T15:34:35 there might be enough data now 2011-12-20T15:34:45 ok 2011-12-20T15:34:48 can't you just stop/deactivate the starter bots manually? Select all bots with version #1 and check them manually if they just run against walls? 2011-12-20T15:34:58 *** meduza has joined #aichallenge 2011-12-20T15:35:07 bluegaspode: there are bots worse than starter 2011-12-20T15:35:19 how about just stop giving games to people at 13 games or above, for a while :-) 2011-12-20T15:35:35 or create a hashfile of the starter bot files to detect those automatically? 2011-12-20T15:35:40 greedybot has more games than me :) 2011-12-20T15:35:47 it's greedy 2011-12-20T15:35:55 I walked into that one 2011-12-20T15:35:56 What rank is greedybot now? 2011-12-20T15:36:01 1998 2011-12-20T15:36:02 Don't know how much this would save, but it's some sort of manual cutoff then 2011-12-20T15:36:07 sigma=3 2011-12-20T15:36:25 mu very close to 50 2011-12-20T15:36:30 McLeopold: I'm thinking that would level the playing field faster and then you can start cuttoffs sooner 2011-12-20T15:36:31 it would be funny if we normalized aroudn greedy=50 2011-12-20T15:37:12 greedybot timed out 2011-12-20T15:37:13 hahahaha 2011-12-20T15:37:15 Do you think deactivation of starter bots can help? This bots is faster and compute faster then 10ms 2011-12-20T15:37:25 Greedybot always times out :v 2011-12-20T15:37:27 predictions http://pastebin.com/VP5cS27V 2011-12-20T15:37:33 aichallenge.org/visualizer.php?game=289708&user=12359&turn=389 2011-12-20T15:37:49 greedybot is timing out a LOT 2011-12-20T15:38:31 *** valydo has left #aichallenge 2011-12-20T15:38:37 *** valydo has joined #aichallenge 2011-12-20T15:38:47 A lot of the bots in those projections haven't lost yet, so their mu is still determined by luck of pairings 2011-12-20T15:39:11 McLeopold: Can you include the number of games in that list, I'll bet there's a bunch on 6 games in there 2011-12-20T15:39:23 How many bots haven't lost a game so far? 2011-12-20T15:39:39 me, xathis, flag, greentea are four off the top of my head 2011-12-20T15:39:49 *** bearoff has joined #aichallenge 2011-12-20T15:40:24 ThisIsNotABug 2011-12-20T15:40:30 real men sometimes lose 2011-12-20T15:40:40 kurnevsky 2011-12-20T15:40:44 mcstar: chuck norris doesn't. 2011-12-20T15:40:45 http://pastebin.com/isVAizkY 2011-12-20T15:40:57 bix0r4ever 2011-12-20T15:40:59 rwest: my point exactly 2011-12-20T15:41:18 Yep, huge number of 6's and 7's 2011-12-20T15:41:38 mcstar: only after they get enough games to present a real challenge 2011-12-20T15:41:46 Mu is prediction of srength? 2011-12-20T15:42:09 14 of that list have 6 games - almost half the list 2011-12-20T15:42:11 top bots played the fewest games 2011-12-20T15:42:41 mcstar: more like "those with the fewest games have the largest probable skill range" 2011-12-20T15:42:47 or something like that 2011-12-20T15:42:50 GreenTea: yes 2011-12-20T15:42:55 pguillory lost http://aichallenge.org/visualizer.php?game=288016&user=589 but thats more bad luck than anything. 2011-12-20T15:42:59 but those names were top bots before 2011-12-20T15:43:22 but yeah, it could be interpreted backwards, which is probably the correct one 2011-12-20T15:43:28 his opponents capped each other on the same turn 2011-12-20T15:43:31 mcstar: yes, definitely agree there... except mine seems to be missing, must be a bug in the SQL, couldnt be because I already lost 1 of my 6 games :-) 2011-12-20T15:43:40 lol, just took a hill with 170 ants in it 2011-12-20T15:43:48 jcdny, hehe) 2011-12-20T15:43:50 capped each other, not same turn... 2011-12-20T15:44:58 McLeopold: so what do you think about changing the sort order for the seed queue? 2011-12-20T15:45:00 mcstar, what was you rank before submission end? 2011-12-20T15:45:17 bearoff: ~110 2011-12-20T15:45:36 thanks 2011-12-20T15:45:40 my bot seems to give up hills too easily sometimes damn 2011-12-20T15:45:48 why? 2011-12-20T15:45:53 do you run last version on tcpants now? 2011-12-20T15:45:55 ChrisH_: we need 2 of 3 votes for a change, and I'm the only one aroun 2011-12-20T15:45:56 trying to extrapolate? 2011-12-20T15:45:59 yes 2011-12-20T15:46:09 aha 2011-12-20T15:46:13 seems my bot should be not far from you 2011-12-20T15:46:24 McLeopold: ok, I sent the idea to janzert with a "later tell" earlier 2011-12-20T15:46:33 * avdg wonders if steve887 was fighting for the last position 2011-12-20T15:46:38 bearoff: i changed much since the last stable ranking happened 2011-12-20T15:46:52 >> my bot seems to give up hills too easily sometimes damn - I have found the bug with defending hills too 1 hour ago ( 2011-12-20T15:46:53 * flowenol back 2011-12-20T15:46:53 it SUPPOSED to be better, but we will see 2011-12-20T15:47:05 it defends only 1 hill from all in most cases ( 2011-12-20T15:47:07 *** dan__ has joined #aichallenge 2011-12-20T15:47:18 understand 2011-12-20T15:47:33 *** dan__ is now known as Guest76341 2011-12-20T15:47:57 McLeopold: if you did that change, all the 6's would get a game real fast, then the 7's (which includes the 6's that just had a game), etc... - right ? with a smattering of other numbers chosen as opponents 2011-12-20T15:48:00 i ask to compare with you as i had never let newest bot's version to play many games last days 2011-12-20T15:48:02 my bot is going to be better on multi-hill maps i suppose 2011-12-20T15:48:13 don't know how strong it became after changes 2011-12-20T15:48:13 Zaphus: that's the idea 2011-12-20T15:48:14 or mazes where they cant get to my hive soo easily 2011-12-20T15:48:22 damn last game was 6h ago 2011-12-20T15:48:48 ChrisH_ seems fair to balance the game count, I assume it would still be OK for the ranking system because opponents are still chosen 'correctly' 2011-12-20T15:49:14 Zaphus: yes, and balancing game count has a big impact on the sigma scores, which makes the skills mean more 2011-12-20T15:49:27 http://ants-score.appspot.com/?uid=AgentSmith had quite a win against mine lately 2011-12-20T15:49:31 bergmark: it seems either i kill you easily or you tie with me 2011-12-20T15:49:48 did you kill me easily? 2011-12-20T15:49:51 ChrisH_: Well, as one of the 6'ers, I'd be happy (assuming I won my games, otherwise I'll find another technical thing to blame it on :-) 2011-12-20T15:49:54 i cant find such an instance 2011-12-20T15:50:04 g0llum: :) 2011-12-20T15:50:07 cool graph 2011-12-20T15:50:35 mcstar, bet you'll get back to top100 2011-12-20T15:50:50 oh, id be infinitely glad 2011-12-20T15:50:55 g0llum, whatis this? http://ants-score.appspot.com/ 2011-12-20T15:51:09 GreenTea: a textbox? 2011-12-20T15:51:22 what graphics means 2011-12-20T15:51:23 type a bot name in there 2011-12-20T15:51:29 screen scraping user profiles 2011-12-20T15:51:42 what on X and Y axises 2011-12-20T15:51:53 and colors 2011-12-20T15:52:10 rankings + sigma 2011-12-20T15:52:11 rnak green skill blue 2011-12-20T15:52:15 looks like green is rank and blue is skill 2011-12-20T15:52:17 oh skill 2011-12-20T15:52:44 and it starts before finals 2011-12-20T15:52:54 meh, its caps sensitive 2011-12-20T15:53:00 g0llum: how does it work? 2011-12-20T15:53:06 it seems fast 2011-12-20T15:53:21 screen scraping user profiles, beatiful spup, python 2011-12-20T15:53:25 soup 2011-12-20T15:53:25 McLeopold: If you do make the change, can I suggest someone posts to the forums and explains that a change is being made to even out the game count ? 2011-12-20T15:53:26 aha, understand 2011-12-20T15:53:48 http://pastebin.com/h3LfLxg9 2011-12-20T15:53:54 I second Zaphus's suggestion 2011-12-20T15:54:00 Zaphus: yes 2011-12-20T15:54:19 Wow - 12/12 wins is impressive 2011-12-20T15:54:38 *** pomme has joined #aichallenge 2011-12-20T15:54:41 29/31 - thats just there to annoy the 6'ers isn't it :-) 2011-12-20T15:55:08 pt2091 looks like a starter bot with lots of ties for 1st 2011-12-20T15:55:08 *** pomme has quit IRC (Client Quit) 2011-12-20T15:55:08 pt2091 is a bunch of tying for 1st 2011-12-20T15:55:13 haha look at bot pt2091 2011-12-20T15:55:16 I think there is still a race condition 2011-12-20T15:55:20 lol 2011-12-20T15:55:20 *** kost_ has joined #aichallenge 2011-12-20T15:55:24 12 in a row but skill 37.6 2011-12-20T15:55:26 *** stocha has joined #aichallenge 2011-12-20T15:55:27 a bot has 55 games, and had a string of 20 in a row :( 2011-12-20T15:55:33 Zaphus: hi, how is it going ? 2011-12-20T15:56:24 stocha: Hi ! about to log off to go to work (8am Wed here) - hoping to see some more games from the top bots today 2011-12-20T15:56:24 if everyone tied for 1st, it still counts as a win 2011-12-20T15:56:48 I'm not going to curn the database looking for "true" wins :) 2011-12-20T15:56:54 Zaphus: do you think you bot will do okay ? 2011-12-20T15:57:10 McLeopold: when testing locally I tracked "number of bots you beat" - so that would be 0 in a tie, made for a good simple statistic 2011-12-20T15:57:17 so many starter bots.. 2011-12-20T15:57:21 McLeopold at this rate, dont you think the time limit is too short? 2011-12-20T15:57:30 McLeopold: how about win and mu increase from win > 1 2011-12-20T15:57:37 stocha: I should make top 100, but not much better than that - I didnt test on 2 player maps and they are now ~30% of the finals 2011-12-20T15:58:06 Zaphus: I didn't test on 2 player maps either. 2011-12-20T15:58:17 Zaphus: there was no clue about final map distribution ? 2011-12-20T15:58:23 Zaphus: other than on the TCP or main servers 2011-12-20T15:58:28 *** Belerafon has quit IRC (Remote host closed the connection) 2011-12-20T15:58:50 *** Anti has joined #aichallenge 2011-12-20T15:58:57 this is a cool game: http://aichallenge.org/visualizer.php?game=292236&user=6317 2011-12-20T15:59:04 starter bots even razing hills :) 2011-12-20T15:59:11 i would have been cursing during the whole final if i had the time to upload a bot :p 2011-12-20T16:00:00 *** ftjansen has quit IRC (Quit: Page closed) 2011-12-20T16:00:13 i think before final organizers could do special test which detect all starter bots 2011-12-20T16:00:18 and remove them 2011-12-20T16:00:21 they all seem to go north :) 2011-12-20T16:00:39 6) 2011-12-20T16:00:40 just test 100 moves on several maps 2011-12-20T16:00:46 why not doing md5sum of their files and compare? 2011-12-20T16:00:51 GreenTea: we will capture md5 hashes of uploads next time, or something 2011-12-20T16:00:52 it would remove a lot of them 2011-12-20T16:00:53 are they same as in starter bots 2011-12-20T16:01:13 stocha, you could have grinned during the whole final, by uploading your old java bot. not knowing any hills, it would have made it to 160. 2011-12-20T16:01:18 McLeopold you could add some dummy output 2011-12-20T16:01:21 * avdg wonders if starter bot could create stabilized games 2011-12-20T16:01:22 a behavior test might be more accurate 2011-12-20T16:01:22 McLeopold: I was pointing out you can eliminate any bots with identical hashes 2011-12-20T16:01:26 Maybe it was a good idea if they had a few starter bots running before the finals, and cut everyone off who ended after them. 2011-12-20T16:01:34 because even a non-starter identical must be someone with two accounts... 2011-12-20T16:01:49 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-20T16:01:55 g0llum: i did upload a random one at some point. But it appears to behave quite ineficiently :p 2011-12-20T16:02:03 k. 2011-12-20T16:02:12 BenJackson: No, you could have done the tutorial cut+paste, then you'd be identical but better than starter 2011-12-20T16:02:19 g0llum: i doubt not knowing hills nor combat would enable first 100 2011-12-20T16:02:36 BenJackson: http://aichallenge.org/quickstart.php that page was the problem :P 2011-12-20T16:02:56 it recommends just uploading a starter bot and encourages you not to make any changes before your first upload 2011-12-20T16:03:02 g0llum: i'm not even sure that one of my bot that did okay with some combat knowledge would have done well. I expect first 100th to be okay bots :) 2011-12-20T16:03:34 BenJackson: the other thing that has been mentioned before is that the starter kits haven't been the same over time and they don't have a good history for all of them, so they can't know all the hashes for all the versions of all the starter kits 2011-12-20T16:03:48 g0llum: ah 160. well. why 160 ? 2011-12-20T16:03:50 stocha, but it was very good at spreading/food hunting/fighting, and would run 'accidentally' over any hill 2011-12-20T16:03:57 The idea of a "beat this bar to compete in the finals" is growing on me - make a bot that you have to finish higher than to compete in the finals, and have it there from day 1. 2011-12-20T16:03:57 i would just ask everyone to resubmit, and telling them: pls no starter bots 2011-12-20T16:04:00 The rankings are still all over the place 2011-12-20T16:04:33 Or just something like a click-activate button 2011-12-20T16:04:35 Zaphus: I would prefer a series of submission tests that you have to win 2011-12-20T16:04:37 g0llum: i had like 50 bot in the first days of beta :p wich one are you talking about :) 2011-12-20T16:04:38 You dont need to remove ALL the starter bots, just MOST of them ... thus md5 sum for known ones should be sufficient. :) 2011-12-20T16:04:40 I kind of like the idea of someone having to activate their bot within the last week before finals in order to be part of the tournament 2011-12-20T16:04:47 Or another idea 2011-12-20T16:04:57 maybe even use that to divide contestants into levels 2011-12-20T16:05:00 bretep: you beat me to it 2011-12-20T16:05:04 About starter bots...They have a very poor logic, and so answer very quickly right? So their games should be run very quickly on the servers no? Or I am missing something? 2011-12-20T16:05:06 before final user should uprove that he wants to participate in final.. 2011-12-20T16:05:20 McLeopold: Yes, that would do it. have a second table for those who havent qualified yet, once they qualify they move to the other ladder and can compete in finals 2011-12-20T16:05:23 What's wrong with simpy cutting the lesser 20% bots using their rankings ? 2011-12-20T16:05:31 The game takes as long as the slowest bot, Manwe56_ 2011-12-20T16:05:33 many of people just upload bot and forgot about it in the same day/ 2011-12-20T16:05:34 Manwe56_: yeah, starterbots are fast 2011-12-20T16:05:43 *** Kurnevsky has left #aichallenge 2011-12-20T16:05:48 stocha: that is expected to happen progressively up the ladder as time goes on 2011-12-20T16:05:59 stocha: just like they did for planet wars 2011-12-20T16:06:00 stocha: right now, you might lose a non-starter bot who hasnt played many games and was just unlucky 2011-12-20T16:06:08 ChrisH_: isn't that enough ? 2011-12-20T16:06:14 ok so if starter are paired in the same game, it could end very quickly 2011-12-20T16:06:17 stocha: it might be 2011-12-20T16:06:33 Zaphus: which is why I suggested the change to seed selection 2011-12-20T16:06:35 stocha: Perhaps, but there's barely enough time to get enough games done 2011-12-20T16:06:37 Zaphus: hardly. Can you pick up a good bot in the worst 3500 ? 2011-12-20T16:06:50 well 600minutes more of wait 2011-12-20T16:07:00 Zaphus: the time limit is too short for sure. If they can't keep up the servers a bit more well .... 2011-12-20T16:07:10 stocha: ChrisH_ found a dozen good bots in the bottom half 2011-12-20T16:07:10 stocha: no idea, the old rankings were cleared so we can't easily find someone who was top 1000 before and bottom 3500 now 2011-12-20T16:07:11 really at this rate the time for the competition is not enough 2011-12-20T16:07:17 How quickly does the cutting go? I had a few timeouts at the start, got a somewhat weak ranking now. 2011-12-20T16:07:21 how about splitting by rank, you stabilize the upper part, after that you successively add the lower bots 1 by 1? 2011-12-20T16:07:24 g0llum: what is your bot names by the way ? 2011-12-20T16:07:27 anyway, must rush off to work - cya all later! 2011-12-20T16:07:29 I think if we could prioritize some of the bots with fewer games and get everyone up to to or 15 games minimum the cutoff point could be found fairly easily 2011-12-20T16:07:41 *** Zaphus has quit IRC (Quit: Page closed) 2011-12-20T16:07:55 thestinger: how many games did they have ? 2011-12-20T16:08:10 I think he was specifically looking at ones with 4 games 2011-12-20T16:08:23 thestinger: a lot of those that I found have now moved up with a few more games though 2011-12-20T16:08:28 thestinger: what's the fewest game count now ? 2011-12-20T16:08:32 Is anyone else iffy on how fair it is when some people get to play a bunch more games aerly on? 2011-12-20T16:08:55 ChrisH_: btw, the ranking_json.php would be great for getting data to look through stuff 2011-12-20T16:08:59 stocha, gollum 2011-12-20T16:08:59 Anti: i used to be during the early beta :p 2011-12-20T16:09:24 Because of true skill not being order independent 2011-12-20T16:09:43 Also, I think if we had an official tcp server, the 5 minute guide should talk about connecting to that, not uploading a bot right away 2011-12-20T16:10:14 McLeopold, Another idea, before final user should approve that he wants to participate in final.. Because many of people just upload bot and forgot about it in the same day. 2011-12-20T16:10:21 although, one of the goals is to have a really low bar, so anyone can participate 2011-12-20T16:10:36 GreenTea: I agree. 2011-12-20T16:10:37 g0llum: interesting, how does it work ? 2011-12-20T16:10:38 GreenTea: I'd rather not have an approval system 2011-12-20T16:10:44 GreenTea a nice idea 2011-12-20T16:10:44 I think encouraging people to upload the starter kit is good adn bad 2011-12-20T16:10:50 it probably got a few people to participate who might not have 2011-12-20T16:10:51 but you send an email notification to activate it if you want to play in the finals 2011-12-20T16:11:01 but it probably got a ton of starter kits uploaded by people who would never do anything no matter what 2011-12-20T16:11:01 McLeopold, why not? 2011-12-20T16:11:25 well, all these starter bots created a good test case for the server farm ;-) 2011-12-20T16:11:28 I don't want good bots to bow out. I really think Accoun should have posted an entry. 2011-12-20T16:11:53 avdg: yes, test failed 2011-12-20T16:11:54 Anyone want to throw me a bone here? Am I wrong or don the people who have now had like 15 games been given an easier lot of games with "higher ranked bots"s? 2011-12-20T16:12:00 Anti: i was very frustrated with the ranking system during the beta. But if Xathis indeed shows such a superioty as it deed, the winner will be clear enougth 2011-12-20T16:12:23 Anti: maybe, if you got your games early, we didn't know who was good 2011-12-20T16:12:38 GreenTea: yeah, that's a good point 2011-12-20T16:12:39 Stocha I'm not just worried about 1st place 2011-12-20T16:12:53 Anti: ah. 2011-12-20T16:13:04 i hope the first 1000 places will be stable though 2011-12-20T16:13:18 Mcleopold shouldn't the bots have been playing with similar numbers of games? 2011-12-20T16:13:30 Particularly until ranks are more stable 2011-12-20T16:13:41 Anti: yes 2011-12-20T16:13:46 we've had issues 2011-12-20T16:14:10 Anti: multiplayer is impossible to solve as far as ranking goes. 2011-12-20T16:14:13 right now there is a peak at 12-13 games and a smaller peak at 6 while we get everyone up to at least 7 games 2011-12-20T16:14:13 Anti: I think you point is valid 2011-12-20T16:14:16 not to mention that maps can interfere with ranking too 2011-12-20T16:14:32 *** olexs has quit IRC (Ping timeout: 252 seconds) 2011-12-20T16:14:36 True skill is made for multiplayer raning though :v 2011-12-20T16:15:00 *** FranPeruzzi has joined #aichallenge 2011-12-20T16:15:09 any ranking system that is dealing with the amount of uncertainty as this game has will require a large sample to get good ranks 2011-12-20T16:15:17 I understand the. Complicated parts of ranking kind off, I'm just iffy bs out. Us letting the bots have the. Extra games with 2011-12-20T16:15:17 bretep: i think i did understand how 2 player game ranking work. I hardly understand true skill 2011-12-20T16:15:23 Crazy rankings 2011-12-20T16:15:26 *** olexs has joined #aichallenge 2011-12-20T16:15:35 Sorry, on an iPad and terrible with it 2011-12-20T16:15:47 we just need a longer time 2011-12-20T16:15:54 as long as that is provided i dont worry 2011-12-20T16:16:03 McLeopold: what about financing? 2011-12-20T16:16:09 ChrisH_: yet it seems that the ranking may pick up the winner 2011-12-20T16:16:19 will you be able to keep up the server as long as the ranks dont stabilize? 2011-12-20T16:16:35 or must you cut them off at some point? 2011-12-20T16:16:35 mcstar yea the time seems short 2011-12-20T16:17:27 stocha: yes, I think the winner is easier to pick, but getting the order "correct" lower in the rankings is harder 2011-12-20T16:17:43 ChrisH_: i totally agree 2011-12-20T16:18:12 Is there a idea till what point the rankings will be (somewhat) stabilized? 2011-12-20T16:18:15 damn I can even provide my server and desktop to play games if it makes it faster 2011-12-20T16:18:18 ChrisH_: i was so frustrated with the ranking that i secretly hoped the winner will be contested :p 2011-12-20T16:18:37 Well. I'm bummed that I don't get some easier games against supposedly top bots while others get to go in. And. Fill those spots with an advantage to theirrank 2011-12-20T16:18:54 Last day it is discovered xathis uses threads or something :v 2011-12-20T16:19:07 he does 2011-12-20T16:19:08 bretep : it will be hard enough once the thing is over to answer this question retrospectively. 2011-12-20T16:19:15 but at most 1 2011-12-20T16:19:30 At least :p 2011-12-20T16:19:36 bretep: for a while I suspected the go contestants were inadvertently using multiple threads :) 2011-12-20T16:19:45 *** meduza has quit IRC (Ping timeout: 258 seconds) 2011-12-20T16:19:47 maybe that's a compliment of Go and its users :) 2011-12-20T16:19:49 Mc.leopold, will the bots with lots of games at least not be chosen for more games? 2011-12-20T16:19:49 Anti : you need at least 100 games to have a ranking. 2011-12-20T16:19:58 BenJackson dont forget java, it inherently uses some threads 2011-12-20T16:20:05 BenJackson: it's just 1 thread by default (I guess you know that though) 2011-12-20T16:20:06 * avdg heard complains about node.js using 2 threads 2011-12-20T16:20:06 *** meduza has joined #aichallenge 2011-12-20T16:20:20 BenJackson: I have an explicit call at the start of my bot that sets the maximum running threads to 1 2011-12-20T16:20:33 flowenol: you think one can somehow hack it's way through swing threads and use them for the contest ? :) 2011-12-20T16:20:33 BenJackson: but it's also the default compile mode on the server 2011-12-20T16:21:02 stocha i didnt say that, but when i attach a debugger i see a few, including a gc one 2011-12-20T16:21:11 thestinger: are you sure there is only one thread active by default ? 2011-12-20T16:21:16 BenJackson: the Go support in gcc is basically broken and it uses 1 thread per goroutine though 2011-12-20T16:21:20 stocha though it would be interesting 2011-12-20T16:21:35 flowenol: what are the other active thread ? 2011-12-20T16:21:39 thestinger: but the servers don't use gcc for go, they use the 6g compiler 2011-12-20T16:21:45 ChrisH_: yeah :) 2011-12-20T16:21:50 stocha i would have to check 2011-12-20T16:21:53 but gccgo isn't an option for that reason 2011-12-20T16:21:58 right 2011-12-20T16:22:05 6g uses N:1 goroutines:threads by default 2011-12-20T16:22:12 flowenol: it may be possible to go multihread. Yet it would hardly go unnoticed if you where top 10 :) 2011-12-20T16:22:14 but you can multiplex over more than 1 2011-12-20T16:22:28 thestinger: do you do anything else besides looking up language specifications and implementations? XD 2011-12-20T16:22:35 stocha i would try but i didnt want to compromise my country win 2011-12-20T16:22:39 mcstar: :P 2011-12-20T16:22:47 mcstar: I messed around with Go for a day or two 2011-12-20T16:23:01 typical of you 2011-12-20T16:23:08 yes, but I added runtime.GOMAXPROCS(1) at the start of my bot just to be safe 2011-12-20T16:23:23 mcstar: it's actually usually slower with more than 2 pthread 2011-12-20T16:23:27 stocha but i would try it if i wasnt interested in winning and see if i would go unnoticed 2011-12-20T16:23:30 because then it has to do locking on channels 2011-12-20T16:23:44 flowenol: now the question is .. how much having two thread give as an advantage if you gave it to someone randomly among the top 200 bot makers. 2011-12-20T16:23:45 they have a lot of optimization work to do 2011-12-20T16:23:51 unmutable concurrency ftw 2011-12-20T16:24:11 but separate threads are created when a system call is going to block, allowing other goroutines to continue running 2011-12-20T16:24:45 *** Surya has quit IRC (Quit: Page closed) 2011-12-20T16:24:50 stocha it would depend on some factors, number of cpu cores, concurrent bots running etc 2011-12-20T16:25:18 flowenol: do you think it's that easy to use the two thread efficiently rather than one ? 2011-12-20T16:25:18 what would be interesting too would use CUDA cores 2011-12-20T16:25:34 flowenol: i would have loved that though 2011-12-20T16:25:40 stocha imo depends on the algorithms being run 2011-12-20T16:25:59 my code would be pretty useless with 2 threads 2011-12-20T16:26:07 two theads means twice the cpu scheduler time (not completely true but true enough) 2011-12-20T16:26:09 wasnt developed thinking about using them 2011-12-20T16:26:14 flowenol: my bots where using bit logic. (using the 64 bits bit instructions as simd). So cuda would have been ideal for me :) 2011-12-20T16:26:27 I would add a timer that prevents me from timing out :v 2011-12-20T16:26:42 flowenol: i was pissed off by the complexity of the combat rules though. I did implement it; but it hurted my aestetic feeling for simplicity 2011-12-20T16:26:44 stocha i would like to see cuda allowed 2011-12-20T16:26:48 my bot would definitely take advantage of two cores if it had them, one for noncombat and one for combat and they are highly parallel in my code 2011-12-20T16:26:49 McLeopold: next contest at the final start it should just auto-rev everyone so that finals are a "new version" rather than having to clear out all the old info 2011-12-20T16:26:54 cuda? 2011-12-20T16:26:55 why? 2011-12-20T16:27:08 thats stream processing not multithreading 2011-12-20T16:27:17 (meanwhile I'm trying to figure out if _Atsky_ is a late challenger or if it's noise) 2011-12-20T16:27:29 Also, show the peak skill a bot has hit. :) 2011-12-20T16:27:29 did hea beat you? 2011-12-20T16:27:30 mcstar different thing, but still interesting to see what would people do with it 2011-12-20T16:27:43 flowenol: it can hardly be though. But we can still organize a cuda ant contest if we find enougth people willing to try that. I think i would be happy to contribute then. Also let's use the more elegant ant dies if on same square combat rule :p 2011-12-20T16:27:50 mcstar: he's beaten some players I recognize and he's in the top 10 right now 2011-12-20T16:27:54 flowenol: well, my algorithms wont play well on cuda :) 2011-12-20T16:28:03 stocha: we should. I have an nvidia tesla 2011-12-20T16:28:13 mcstar neither would mine 2011-12-20T16:28:29 BenJackson that card sucks, throw it in the garbage and tell me where 2011-12-20T16:28:38 :) 2011-12-20T16:28:51 i am a research engineer at nvidia and i would vote for no cuda :) 2011-12-20T16:28:54 i used fermi's, love the bandwidth :D 2011-12-20T16:29:06 BenJackson: i have a nvidia. That i never used for cuda. Although i bought it in the hope i would do some cuda stuff with it. What would the challenge look like ? Tcp only ? or a server like this one ? 2011-12-20T16:29:15 i have to confess u dont know much about cuda coding 2011-12-20T16:29:20 *i dont 2011-12-20T16:29:30 flowenol: tesla S20xx (forget the exact model) it's the 1U box with CPU and 2 teslas 2011-12-20T16:29:43 keith_: why ? 2011-12-20T16:29:44 on newer cards you could multiplex multiple bots on one card 2011-12-20T16:29:46 not the S2050 that's 4 teslas (but no cpus, only external PCIe) 2011-12-20T16:29:47 BenJackson how much did it cost? 2011-12-20T16:29:56 I don't know! We bought it for some stupid reason at work 2011-12-20T16:29:57 keith_: ant game seems very well suited for that doesn't it ? 2011-12-20T16:30:03 but we didn't really use it so only I use it 2011-12-20T16:30:19 BenJackson: what do you do with it ? 2011-12-20T16:30:24 BenJackson: we have supermicro workstattions, with 8 cores + 4 teslas + 48gb memories 2011-12-20T16:30:32 screw around with CUDA 2011-12-20T16:30:36 I use my nvidia to do some md5 and lm crunching 2011-12-20T16:30:51 we bought it because our CTO was buddies with a guy we hired as a consultant to evaluate GPU computing for our projects 2011-12-20T16:30:52 *** Murashka has quit IRC (Quit: Page closed) 2011-12-20T16:30:54 though for lm hashes now I use rainbow tables 2011-12-20T16:30:56 I saw his final report, it was shitty 2011-12-20T16:31:03 but we bought the box for him to work on 2011-12-20T16:31:11 That's a very good idea. I should do just that. Build an ant cuda bot. Even if it take a few month to build up my most basic stuff. It could be fun. 2011-12-20T16:31:21 lol 2011-12-20T16:31:25 and time wasting 2011-12-20T16:31:26 ant cuda bot would be neat 2011-12-20T16:31:31 i wonder how fast it can go. 2011-12-20T16:31:34 any problem that can benefit much from cuda computation would probably come down to a competition in problem decomposition for parallel processing, exploitation of memory efficiencies, etc. not an AI challenge anymore. people who are very fluent in cuda, like me would have a huge advanage 2011-12-20T16:31:38 boy the GAME ENGINE would run GREAT in CUDA 2011-12-20T16:31:45 * avdg wonders if there would be an optimized solution to this game 2011-12-20T16:31:47 i built my bot principle right over the simd concept :) 2011-12-20T16:31:48 stocha: just try to make a minimax on cuda first 2011-12-20T16:32:03 mcstar: who did a minimax bot in there ? 2011-12-20T16:32:08 many 2011-12-20T16:32:12 stocha tried, failed 2011-12-20T16:32:13 did they ? 2011-12-20T16:32:15 i mean lets not call it minimax 2011-12-20T16:32:18 arrgh I have so many ideas for totally rewriting my bot 2011-12-20T16:32:20 most of the time its 1 ply 2011-12-20T16:32:33 I imagine myself in a few weeks.. "guys... guys?? anyone want to play my new bot?? it slaughters xathis!!" 2011-12-20T16:32:37 butsome did monte carlo move evaluation 2011-12-20T16:32:37 mcstar: then simd is fine. 2011-12-20T16:32:41 *** Palmik has quit IRC (Read error: Connection reset by peer) 2011-12-20T16:32:47 BenJackson: I wanted to do the auto rev this time, but ran out of time 2011-12-20T16:32:50 well, i would call it brute force best combat position finder 2011-12-20T16:32:51 mcstar: simd is still fine for that 2011-12-20T16:33:08 I need a goot excuse to learn cuda 2011-12-20T16:33:11 *a good 2011-12-20T16:33:20 cuda ants would be a nice one 2011-12-20T16:33:24 stocha: i know random number generation and monte carlo stuff is good for cuda 2011-12-20T16:33:26 * avdg was thinking at coverage as best combat position finder 2011-12-20T16:33:32 flowenol: quite obviously the best strategy is a mixed strategy (with random elements) 2011-12-20T16:33:33 but still, its a lot of work to make an ants bot on cuda 2011-12-20T16:33:54 btw my card died 2011-12-20T16:33:57 stocha note that im not using minimax 2011-12-20T16:33:59 flowenol: same here. And i do want to do something too complex to start with. So ant would be ideal i think. 2011-12-20T16:34:12 its just a 9400gt in my desktop, but it stopped working 2011-12-20T16:34:25 keep in mind that for most cuda workflows you want thousands of threads ... 2011-12-20T16:34:27 the most complex projects are the best for learning imo 2011-12-20T16:34:29 mcstar: i think you can do one okay bot quite easilly indeed. 2011-12-20T16:34:32 i liked it cause i didnt have to ssh into the institute 2011-12-20T16:34:56 mcstar: it would be especially interesting to try brute force montecarlo 2011-12-20T16:35:03 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-20T16:35:11 ok, if you have the time... 2011-12-20T16:35:13 600minutes someone kill me 2011-12-20T16:35:14 nothing stops you 2011-12-20T16:35:15 Brute force monte carlo? 2011-12-20T16:35:28 *** ChanServ sets mode: +o Zannick 2011-12-20T16:35:29 i think some had success with montecarlo for combat in the 500ms constrain 2011-12-20T16:35:32 *** Zannick changes topic to "AI Challenge (sponsored by Google): http://aichallenge.org || Channel Logs: http://contestbot.aichallenge.org || Code Repo: http://github.com/aichallenge/aichallenge || Final rankings will be out in a few days (amstan) || Final maps: http://forums.aichallenge.org/viewtopic.php?f=3&t=2170#p13344 (amstan)" 2011-12-20T16:35:37 *** Zannick sets mode: -o Zannick 2011-12-20T16:35:48 (fixed minor typo) 2011-12-20T16:35:54 bretp : like you do 10 000 turn evaluation per second, and just try random moves and keep the best one :) 2011-12-20T16:36:06 Zannick thanks for warning me, i was looking for anything new 2011-12-20T16:36:08 *** erij has joined #aichallenge 2011-12-20T16:36:34 stocha: antimatroid said he picked move randomly 2011-12-20T16:36:46 I thought you meant something different with 'brute force monte carlo' 2011-12-20T16:36:49 flowenol: heh 2011-12-20T16:36:51 i use something based on a1k0n idea 2011-12-20T16:36:54 so he did some kind of sampling, i dont know the details 2011-12-20T16:36:57 Although monte carlo is kinda brute force 2011-12-20T16:37:15 but to get them to do something different i first moved them all into the enemys and kept them static 2011-12-20T16:37:20 bretep: i meant a very close to random things, with many many evaluation. Rather than clever simulation with good efficiency 2011-12-20T16:37:28 i was enough to influence them in the right way, most of the time 2011-12-20T16:38:04 flowenol: so it's a montecarlo evaluation of a sort ? 2011-12-20T16:38:09 i wonder is it easy to implement A* with cuda 2011-12-20T16:38:09 *** Anti has quit IRC (Ping timeout: 258 seconds) 2011-12-20T16:38:18 stocha dont ask, i dont know how to answer 2011-12-20T16:38:19 flowenol: how much position do you evaluate ? 2011-12-20T16:38:51 flowenol: i think my path solving would be very easy to implement with cuda. 2011-12-20T16:38:52 well with around 5 ants only to evaluate, around 10000 (on my computer) 2011-12-20T16:38:58 i think recursive algorithms are tough to implement with cuda 2011-12-20T16:39:13 more like inefficient 2011-12-20T16:39:18 for food i had a pseudo A* 2011-12-20T16:39:28 flowenol: was 10 000 enough to get good behaviors ? 2011-12-20T16:39:36 stocha yea 2011-12-20T16:39:45 flowenol: what's your bot name ? 2011-12-20T16:39:55 but with too many ants in combat i only would to around 50 iterations 2011-12-20T16:39:57 same as my nickname 2011-12-20T16:40:02 flowenol: don't you loose an ant here or there from bad luck ? 2011-12-20T16:40:14 all the ants I lose are bad luck 2011-12-20T16:40:16 stocha one ant vs 2 is rare 2011-12-20T16:40:16 the ones I kill are skill 2011-12-20T16:40:17 mcstar: i have seen one attempt to implement alpha-beta for tic-tac-toe 3d that had some success 2011-12-20T16:40:27 BenJackson: lol :) 2011-12-20T16:40:28 well, implementing something like A* in cuda is pretty easy -- it comes down to ensuring high occupancy on the cuda cores in order to achieve high parallelism. unless you are running thousands of A* searches at once, this is difficult to do 2011-12-20T16:40:31 mostly bad luck i would say 2011-12-20T16:40:36 * avdg didn't test how many ants his bot would need to time out 2011-12-20T16:40:45 *** Anti has joined #aichallenge 2011-12-20T16:41:00 avdg neither did i, but it works well with around 400 2011-12-20T16:41:10 mine already hit 600 ;-) 2011-12-20T16:41:14 but my code has checks for timeout on critical sections 2011-12-20T16:41:26 A* is easier i think but serial algorithms like alpha-beta are tough to parallelize efficiently 2011-12-20T16:41:37 Pity this ended so http://aichallenge.org/visualizer.php?game=295989&user=98 2011-12-20T16:41:47 Soon 2011-12-20T16:41:55 stocha I made it to around 150 before the competition started 2011-12-20T16:42:41 Anti: haha nice match 2011-12-20T16:42:53 What's your current rank, flowenol ? 2011-12-20T16:43:05 flowenol: i think we should expect the competition to be quite harder than it was 24h before it begun. 2011-12-20T16:43:12 *** ltriant has joined #aichallenge 2011-12-20T16:43:15 *** Anti has quit IRC (Client Quit) 2011-12-20T16:43:19 Anti: also I watched http://aichallenge.org/visualizer.php?game=294841&user=98 of yours. Bad luck 2011-12-20T16:43:40 bretep lower than i would want 570 2011-12-20T16:43:54 i might be going ballistic 2011-12-20T16:43:57 *** mj41 has joined #aichallenge 2011-12-20T16:43:59 stocha as long as i get first in my country, im ok 2011-12-20T16:44:00 delineate seems to have exploration problems with this new maps 2011-12-20T16:44:15 im running on tcpants, i lost to accoun and paratrechina a lot 2011-12-20T16:44:20 i changed 1 think in my code 2011-12-20T16:44:34 and now i tied against account and had razed hills from paratrechina 2011-12-20T16:44:40 *** rebelxt has quit IRC (Ping timeout: 258 seconds) 2011-12-20T16:44:41 antimatroid bot seem okay. How high do you think this one will be ? 2011-12-20T16:44:45 i changed <= to < 2011-12-20T16:45:11 *** Anti has joined #aichallenge 2011-12-20T16:45:11 * avdg wonders if mcstar needed > 2011-12-20T16:45:13 mcstar always one character 2011-12-20T16:45:30 i didnt have time to properly test it 2011-12-20T16:45:54 i will be angry if i see that my bor really is better on tcp with this change 2011-12-20T16:45:55 *** a9wegf9a has joined #aichallenge 2011-12-20T16:45:56 t 2011-12-20T16:46:00 does anybody know who are A and Q on Fluxid's server? 2011-12-20T16:46:15 flagcapper? 2011-12-20T16:46:35 yes 2011-12-20T16:46:47 that would be #1 on the tcp faq 2011-12-20T16:46:50 not sure if he was B too 2011-12-20T16:46:58 no, someone else was B, to his chagrin 2011-12-20T16:47:15 flagcapper is only 16. 2011-12-20T16:47:16 B on fluxid confused me because B was one of my local test bots (last pre-combat version) 2011-12-20T16:47:25 so whenver I saw B do anything "smart" on fluxid it was jarring 2011-12-20T16:47:30 amazing. i want to be 16 too. 2011-12-20T16:47:48 I'm glad I'm not 16 now 2011-12-20T16:47:51 stocha: where is your time machine ? 2011-12-20T16:47:56 but it would have been amazing to have stuff like this around at age 16 2011-12-20T16:48:03 oh man... I just realized why my ant does so bad on multi hill maps. Oops 2011-12-20T16:48:07 *** bob__ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T16:48:09 Congrats BenJackson " 1 ↑ BenJackson" 2011-12-20T16:48:14 so flagcapper was both A and Q? 2011-12-20T16:48:20 Stocha I'm hoping it is top 30 quality 2011-12-20T16:48:26 Besh : i think i still have to build it. But then i'll be back to when i'm 10. So i can give it back to myself 2011-12-20T16:48:30 It definitely isn't top 10 2011-12-20T16:49:02 Anti: top 30 ? that's seem quite good. I though it was a simple bot. 2011-12-20T16:49:09 *** kost_ has quit IRC (Quit: Page closed) 2011-12-20T16:49:15 when i was 16 i don't know what computers are 2011-12-20T16:49:28 haha. BenJackson wins! 2011-12-20T16:49:33 when i was 16, i didn't know what internet was. 2011-12-20T16:49:37 that's pretty good for 10 games 2011-12-20T16:49:45 stocha: it didn't exist :) 2011-12-20T16:49:58 someone learned me about linux when I was 16 2011-12-20T16:50:03 on irc 2011-12-20T16:50:09 Stocha it's not too basic either 2011-12-20T16:50:09 besh: it probably did. there was like 10 users :p 2011-12-20T16:50:09 honestly when i was 16, i was actually 20 2011-12-20T16:50:18 Anti: what does it do ? 2011-12-20T16:50:21 mcstar: haha 2011-12-20T16:50:38 Anti: i think it try to handle it's frontier with priority to the closest one to the nest 2011-12-20T16:50:44 I'm 16 :P 2011-12-20T16:50:47 oh 2011-12-20T16:50:50 *** a9wegf9a is now known as conor_f 2011-12-20T16:50:52 :P 2011-12-20T16:51:08 Anti: with a litle bit of combat solving most probably. I would expect just about every bot to do just that. 2011-12-20T16:51:08 came a bit late to the conversation there lol 2011-12-20T16:51:10 conor_f: what do you mean with the html output or whatnot? 2011-12-20T16:51:16 a1k0n: it looks like you ant lazarant took each other out? 2011-12-20T16:51:35 oh yeah, it was outputting HTML in the notification, by design or not I don't know 2011-12-20T16:51:43 it shouldnt 2011-12-20T16:51:50 I changed it so it just has my rank followed by me skills 2011-12-20T16:51:50 a1k0n: screenshots taken :) 2011-12-20T16:51:59 conor_f: will you beat flagcapper bot then ? :) 2011-12-20T16:52:04 Stocha I'm on an iPad which I can barely use, so can't go into too much detail 2011-12-20T16:52:13 what was wrong? cause it is working for me 2011-12-20T16:52:14 stocha: hahah I wish :P :/ 2011-12-20T16:52:16 Anti: did you post on the forum ? 2011-12-20T16:52:25 conor_f: how is your bot doing ? 2011-12-20T16:52:28 mcstar: I changed the getskill function then 2011-12-20T16:52:31 it's now this 2011-12-20T16:52:33 VAL=`curl -s http://aichallenge.org/profile.php?user=8217 | sed -r 's/^.* //g;s/ [0-9]+.[0-9]+//g;/^[^0-9]/d;/^$/d'` 2011-12-20T16:52:36 Basically it just sends ants towards targets with various orders 2011-12-20T16:52:47 stocha: around the 700 mark lol :P fail much 2011-12-20T16:52:48 conor_f now thats a big regex 2011-12-20T16:52:50 Yeah, there's a link to it in the forums 2011-12-20T16:53:00 flowenol: wasn't me, not taking credit for it 2011-12-20T16:53:02 conor_f: do you have a gist for that (or something simular) :p 2011-12-20T16:53:08 conor_f oh ok 2011-12-20T16:53:18 but i keep my comment about the regex itself 2011-12-20T16:53:18 avdg: sorry, I don't understand you? 2011-12-20T16:53:28 he wants the code 2011-12-20T16:53:28 yeah, its a damn scary regex.... 2011-12-20T16:53:28 conor_f: no combat, a bit of pathfinding ? What other features ? 2011-12-20T16:54:03 stocha: no combat, bit of pathfinding :P 2011-12-20T16:54:06 pretty much it 2011-12-20T16:54:15 conor_f: but do you know why it didnt work? 2011-12-20T16:54:17 it razes hills when they are in sight and thats about it 2011-12-20T16:54:17 ah 2011-12-20T16:54:19 ah 2011-12-20T16:54:20 conor_f: what pathfinding did you implement ? 2011-12-20T16:54:21 ah 2011-12-20T16:54:23 yo dont have nc ?? 2011-12-20T16:54:25 you 2011-12-20T16:54:27 meh, its still sed -r 2011-12-20T16:54:29 I do I think.... 2011-12-20T16:54:37 stocha: breadth-first search 2011-12-20T16:54:40 made a cost map 2011-12-20T16:54:48 conor_f: you try to spot the closest frontier and defend it also don't you ? 2011-12-20T16:54:53 with different weights for hills and food 2011-12-20T16:55:03 stocha: no, nothing so advanced :P 2011-12-20T16:55:05 conor_f: ah, nice. 2011-12-20T16:55:24 conor_f: how did you handled distance through crooked cooridors then ? 2011-12-20T16:55:27 2 big failings I have with it, no defense which has ruined me 2011-12-20T16:55:55 and and 2: no exploring so on maps with low food count, it just gets stuck 2011-12-20T16:56:04 stocha: bfs handled it for me ;) 2011-12-20T16:56:12 conor_f: how much time did you invest in it ? 2011-12-20T16:56:25 put all the food locations into a list and set that to 0 on the cost map 2011-12-20T16:56:27 right now my mu+3sigma is 99.97 2011-12-20T16:56:32 *** kost_ has joined #aichallenge 2011-12-20T16:56:35 *** rebelxt has joined #aichallenge 2011-12-20T16:56:37 then +1 for each step away from it 2011-12-20T16:56:51 I put in too much time at the start trying to learn about it :P 2011-12-20T16:56:56 so it just crossed under 100 2011-12-20T16:57:05 and then I ripped up the starter bot and put in my own ant class 2011-12-20T16:57:07 bearoff: interesting, after the change, i stand my ground better against paratrechina and accoun, and lose to you ! 2011-12-20T16:57:10 conor_f: how do you add secondary objectives then ? 2011-12-20T16:57:15 BenJackson: wait +3sigma? What does that tell you 2011-12-20T16:57:18 and then rmd my code :( but still got it done 2011-12-20T16:57:29 stocha: negative values ;) 2011-12-20T16:57:33 better win percent query http://pastebin.com/SMXKxpLe 2011-12-20T16:57:33 oh wait, wrong sigma 2011-12-20T16:57:40 VAL=`curl -s http://aichallenge.org/profile.php?user=8217 | sed -r 's/^.* //g;s/ [0-9]+.[0-9]+//g;/^[^0-9]/d;/^$/d'` 2011-12-20T16:57:44 sorry 2011-12-20T16:57:49 conor_f: rewriting one's code is a good way to learn. And maybe a good way to build software in fact. 2011-12-20T16:57:49 no, it's still 104.5 2011-12-20T16:57:50 meant to open that link >_< 2011-12-20T16:58:09 conor_f please dont post that again, when I see a regex i try to read it 2011-12-20T16:58:10 conor_f: although you can find yourself doing that just too much at some point :) 2011-12-20T16:58:13 stocha: I didn't like it all that much ;) 2011-12-20T16:58:27 stocha: where are you ranked? 2011-12-20T16:58:31 f4hy: http://www.cs.uni.edu/~campbell/stat/z-score.html 2011-12-20T16:58:40 maybe that's not the best link on normal distrubtion 2011-12-20T16:58:50 http://en.wikipedia.org/wiki/Normal_distribution 2011-12-20T16:58:51 there! 2011-12-20T16:59:19 everyone know normal distributions, it is natures blessing 2011-12-20T16:59:25 s 2011-12-20T16:59:38 mcstar i wonder if everyone does 2011-12-20T16:59:39 I still think the difference in game counts is a decent problem that you all sem to not care about? 2011-12-20T16:59:41 conor_f: unfortunatly i haven't uploaded a good bot. Although i could have picked one that worked quite okay. (i did test one on the tcp server that seems do okay. Considering it had no combat knowledge and no hill awarness) 2011-12-20T16:59:45 stocha: we'll probably play at some time, we're ranked similarly 2011-12-20T16:59:49 conor_f: it's stocha. 2011-12-20T16:59:54 actually log-normal 2011-12-20T17:00:01 * avdg is probably the only one here that knows what normal distribution is 2011-12-20T17:00:20 avdg indeed 2011-12-20T17:00:25 * kost_ lols 2011-12-20T17:00:28 tell me 2011-12-20T17:00:30 conor_f: the problem is my bot don't have combat nor hill awarness. So against similarly ranked bot, it will mostly conquer the map, and loose because of lowest score :) 2011-12-20T17:00:36 at home 2011-12-20T17:00:45 *** sashaSochka has quit IRC (Quit: Page closed) 2011-12-20T17:00:47 conor_f: although its particularly bad at conquest this particular one. 2011-12-20T17:01:04 somehow I was confusing stocha and sashaSochka 2011-12-20T17:01:14 so was was confused about the comments about "no combat awareness" 2011-12-20T17:01:15 avdg: are you Russian? 2011-12-20T17:01:17 BenJackson: yeah, close names 2011-12-20T17:01:23 luckily sashaSochka disconnected to clear it up for me 2011-12-20T17:01:25 no, flemish 2011-12-20T17:01:36 is anyone else still working on their bot even though finals are on? 2011-12-20T17:01:50 stocha: yeah, I can see from your games ;) 2011-12-20T17:02:03 I just implemented hill awareness the night of the competition close lol 2011-12-20T17:02:07 *** bluegaspode has quit IRC (Quit: Page closed) 2011-12-20T17:02:10 BenJackson: in fact i did build bots in the early few weeks of the early beta. In those days there was flagcapper at the top with sir-macelon. And then one of my bot. And then a few more of my bot in the top bots :) 2011-12-20T17:02:10 erij: not me. but i have this urge to connect to tcp servers 2011-12-20T17:02:11 erij: I am, because my finals ended yesterday so I am finally free to work on it 2011-12-20T17:02:27 I think the most I'm going to do is profile the games my bot timed out, barely any ants no fights, yet it times out :/ 2011-12-20T17:02:31 so... BenJackson is the boss now 2011-12-20T17:02:35 *** mattswe has joined #aichallenge 2011-12-20T17:02:40 besh: do you play on tcpants? 2011-12-20T17:02:41 Anilm3: yes, for now, anything he says you have to do 2011-12-20T17:02:49 haha 2011-12-20T17:02:52 besh/f4hy: I am. using the tcpservers and competing against some other friends who have sent compiled/jar versions of their bots 2011-12-20T17:03:05 mcstar: yes. i can connect now 2011-12-20T17:03:12 BenJackson: but bot flagcapper and sir-macelon were combat aware bots. mine weren't. 2011-12-20T17:03:13 Anilm3: I will be a benevolent ruler 2011-12-20T17:03:22 Anilm3: unless your ants get too close to my hill 2011-12-20T17:03:36 http://aichallenge.org/visualizer.php?game=292888&user=6680 2011-12-20T17:03:43 i dont need a rules i can draw straight lines 2011-12-20T17:03:48 ruler* 2011-12-20T17:03:53 stocha: around 600 your ants look pretty combat aware :O 2011-12-20T17:04:04 BenJackson: don't worry my ants will be respectful 2011-12-20T17:04:28 conor_f: i think i uploaded this one because i liked how it looked when moving. Although it really seems drunk :) 2011-12-20T17:04:39 stocha: wow you really snuck around blue's hill avoiding it there 2011-12-20T17:04:46 multiple times! 2011-12-20T17:04:51 stocha: its really good 2011-12-20T17:05:00 if only you'd just equated hills with food 2011-12-20T17:05:13 lol, making a circle around the hill 2011-12-20T17:05:22 red dies by getting his hill between you and lunch 2011-12-20T17:05:23 BenJackson: most probably i would equate them with ants 2011-12-20T17:05:38 ut oh, blue has a food behind it 2011-12-20T17:05:41 this could be the end 2011-12-20T17:05:52 NOOOO 2011-12-20T17:05:54 my bot had basically no ai in it :/ yet it still does ok for some reason... 2011-12-20T17:05:54 hahaha 2011-12-20T17:05:56 at turn 850ish 2011-12-20T17:06:01 I like that map :) 2011-12-20T17:06:17 i really have no idea what this particular bot is trying to do exactly. i only know that it doesn't see hills and doesn't know the combat rule 2011-12-20T17:06:21 Food searching is smart anyway 2011-12-20T17:06:31 My bot would send several for that food 2011-12-20T17:06:44 Where he goes around the blue hill 2011-12-20T17:07:13 stocha: oh, we played each other 2011-12-20T17:07:22 http://aichallenge.org/visualizer.php?game=292206&user=58 2011-12-20T17:07:39 BenJackson: you still have 15 skill to go 2011-12-20T17:07:47 I want to play mcstar, he's one place higher then me in the pypy ranking. :) 2011-12-20T17:08:14 600 minutes 2011-12-20T17:08:24 kill me or the starter bots 2011-12-20T17:08:39 bretep: sorry? 2011-12-20T17:08:40 trouble is, my bot don't realize he does own hills :) 2011-12-20T17:08:48 who is? 2011-12-20T17:08:54 mcstar: you're not agent smith? 2011-12-20T17:08:56 and i keep saying 2 days are not enough 2011-12-20T17:09:00 yes 2011-12-20T17:09:03 AgentSmith 2011-12-20T17:09:04 my bot has played 2 games in 28 hours :( 2011-12-20T17:09:05 stocha: ya sorry for caping your hill :p 2011-12-20T17:09:06 not agent_smith 2011-12-20T17:09:33 f4hy: wich one is yours ? 2011-12-20T17:09:38 stocha: moduspwned 2011-12-20T17:09:39 Sorry, anyhow I'm PeterB 2011-12-20T17:10:08 @reverse "bretep" 2011-12-20T17:10:08 mcstar: Error: The command "reverse" is available in the Filter and Format plugins. Please specify the plugin whose command you wish to call by using its name as a command before "reverse". 2011-12-20T17:10:12 damn 2011-12-20T17:10:21 bretep: come to tcpants.com 2011-12-20T17:11:07 i need to get some rest. 2011-12-20T17:11:09 see you guys. 2011-12-20T17:11:26 smith...agent....smith 2011-12-20T17:11:34 post the cuda ant contest on the forum :p 2011-12-20T17:11:36 *** Westicles has joined #aichallenge 2011-12-20T17:11:38 :) 2011-12-20T17:11:42 i love that guy 2011-12-20T17:11:45 best bad guy ever 2011-12-20T17:12:06 bb stocha 2011-12-20T17:12:10 was nice to see you 2011-12-20T17:12:40 meh http://tcpants.com/replay.18086 <- mcstar would be disapointed 2011-12-20T17:12:51 s/disapointed/happy/ 2011-12-20T17:12:59 * avdg looked from the wrong side 2011-12-20T17:12:59 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-12-20T17:13:08 *** Jak_o_Shadows has joined #aichallenge 2011-12-20T17:13:20 Looks like TrueSkill is doing a good job of sorting the bots... but by language rather than skill! 3 C++ followed by 4 Java atm. 2011-12-20T17:13:26 wow, just got 9 games in the last hour. After being one of the naysayers on the forum, I guess I should have had faith in the code. 2011-12-20T17:13:47 okay, I'm playing on tc-pants now 2011-12-20T17:13:53 FranPeruzzi i still dont have much although i have around 14 games 2011-12-20T17:14:07 McLeopold: is it a real bot? 2011-12-20T17:14:08 Anyone else happy with their current rank? Join with me while I drive up the EC2 spot price for the next 3 days 2011-12-20T17:14:09 I have 15 total now 2011-12-20T17:14:10 I only really tried to grow fast, combat sucks 2011-12-20T17:14:14 Connected to tcpants, mcstar 2011-12-20T17:14:16 mcstar: yes, kinda 2011-12-20T17:14:23 I'm going to be computing digits of pi and I'm willing to pay $20/cpu hr 2011-12-20T17:14:27 lets see :) 2011-12-20T17:14:51 mine is better than what I expected, so I don't think I should be sad about that one 2011-12-20T17:14:52 I'll probably time out 2011-12-20T17:15:03 I need to comment out some logging... 2011-12-20T17:15:05 BenJackson: would be better to donate that money for these servers 2011-12-20T17:15:21 I liked the leaderboard better after my 4th game. :) 2011-12-20T17:15:30 anyway, Pi is best served in symbolic form 2011-12-20T17:15:58 mcstar: see you mc. I'm disapointed i couldn't have those 20 hours i need to submit at least something :) 2011-12-20T17:15:58 mcstar: no no, I'm in 1st, I don't want more games played 2011-12-20T17:16:10 *** ChrisH_ has joined #aichallenge 2011-12-20T17:16:31 BenJackson: congrats on 1st so far 2011-12-20T17:16:34 mcstar: i had to take care of my babies though. so ... maybe i'll build a cuda based bot and post it on the forum :p in a few month 2011-12-20T17:16:46 k, good luck 2011-12-20T17:16:58 ChrisH_: thanks. I was just conspiring to drive up the EC2 spot price for the remainder of the contest 2011-12-20T17:17:08 lol 2011-12-20T17:17:21 ChrisH - still 6 games, eh? You're getting punished for standing up on the forum for the matching algorithm. :) 2011-12-20T17:17:24 donations from anyone who likes their current rank accepted 2011-12-20T17:17:48 Scryer: ChrisH_ actually has a good suggestion for fixing them 2011-12-20T17:17:57 Scryer: maybe, but I think it has more to do with having a high sigma because I've won all six 2011-12-20T17:17:59 i like it too. top100 is all i wanted . so yes. 2011-12-20T17:18:01 but janzert is probably on a plane to the bahamas by now, fleeing the contest 2011-12-20T17:18:10 amstan: ping 2011-12-20T17:18:17 http://tcpants.com/replay.18087 2011-12-20T17:18:19 we can get it with a vote from amstan though 2011-12-20T17:18:20 from turn 365 2011-12-20T17:18:27 mcstar: pong 2011-12-20T17:18:32 :D 2011-12-20T17:18:32 McLeopold: pong 2011-12-20T17:18:34 I think we should put all contestants names into a big hat, then first person drawn out is the winner 2011-12-20T17:18:35 too many mcs 2011-12-20T17:18:36 oh is tcpants working again? 2011-12-20T17:18:37 lol 2011-12-20T17:18:44 amstan: I want to change the seed player to ones with the least games 2011-12-20T17:18:45 *** Nefbrethilion has joined #aichallenge 2011-12-20T17:18:50 I should see what all those games were when its web was down 2011-12-20T17:18:50 do you concur? 2011-12-20T17:18:53 ok quiet everyone while McLeopold and amstan talk 2011-12-20T17:19:00 I agree with onion42 2011-12-20T17:19:04 McLeopold: what? why? 2011-12-20T17:19:05 Increases my odds 2011-12-20T17:19:10 haha me too 2011-12-20T17:19:17 oh, bj_v13 is #1, I didn't even know 2011-12-20T17:19:18 so we can get everyone to 10 games and start the cutoffs 2011-12-20T17:19:22 i'll be happy to finish in the top 500 :) 2011-12-20T17:19:32 McLeopold how about those with > than 10 2011-12-20T17:19:33 oh everyone else had left :) 2011-12-20T17:19:51 i'm not exactly sure, what does janzert think? 2011-12-20T17:20:02 amstan: equal number of games is very important for far skills, otherwise variance in the sigma dominates too much 2011-12-20T17:20:04 onion42: will you be ok if we give you two more games ? 2011-12-20T17:20:10 *** stocha has quit IRC (Quit: Page closed) 2011-12-20T17:20:14 fair skills 2011-12-20T17:20:22 two more games is better than no more games :) 2011-12-20T17:20:23 ChrisH_: yes, but sigma is not the same as game count 2011-12-20T17:20:29 amstan: http://pastebin.com/ehdJs2n0 game counts 2011-12-20T17:20:35 amstan: I'd like to say that I'm 100% satisfied with the finals at this point :) 2011-12-20T17:20:42 lol 2011-12-20T17:20:44 ;) 2011-12-20T17:20:45 you won't hear that from too many people 2011-12-20T17:20:45 amstan: no, but *highly* correlated when there are very few games played 2011-12-20T17:20:57 that is how they shut me up. I was on 4 for long and they threw a bone at me to quieten me :) 2011-12-20T17:21:10 wow, who's got 55 games? 2011-12-20T17:21:15 I'm happy enough :) 2011-12-20T17:21:16 besh: that's what I was jokingly thinking too 2011-12-20T17:21:17 some starter bot :( 2011-12-20T17:21:18 McLeopold: that is a funny looking table 2011-12-20T17:21:46 i didn't realize that we had another peak at the start 2011-12-20T17:21:48 i bet they punish me by throwing me in with xathis for a couple of 1vs1s lol 2011-12-20T17:21:52 McLeopold: anyway, what does janzert think? 2011-12-20T17:22:01 janzert isn't around 2011-12-20T17:22:02 nice histogram 2011-12-20T17:22:07 I had to play Cathie already 2011-12-20T17:22:09 55 games? lucky bastard 2011-12-20T17:22:19 yeah, that table is very intersting and informative 2011-12-20T17:22:20 it's probably a starter 2011-12-20T17:22:28 McLeopold: ok, well.. change it, if nothing breaks it's probably fine 2011-12-20T17:22:34 when are they killing the noobs? :P 2011-12-20T17:22:46 lol 2011-12-20T17:23:01 conor_f":^ 2011-12-20T17:23:01 amstan: I'll kick in the logic as long as there is a bot with 6 games, then it will turn off 2011-12-20T17:23:08 hopefully by that time janzert is back 2011-12-20T17:23:22 yay, i might get some easy games :P 2011-12-20T17:23:48 @later tell g0llum theres a bug in your web app, it gives the incorrect user when i enter user id 2011-12-20T17:23:48 mcstar: As you wish. 2011-12-20T17:23:50 McLeopold: do that for seeds only, right? 2011-12-20T17:24:01 otherwise the 6s can only play other 6s and we still won't learn about their skills 2011-12-20T17:24:05 BenJackson: seeds only 2011-12-20T17:24:18 uhoh 2011-12-20T17:24:27 and, it might not even make a difference 2011-12-20T17:24:35 so, i've got 6 games and 1217 in the queue before me, this will jump me at least halfway down the queue 2011-12-20T17:24:39 from my current spot 2011-12-20T17:24:44 good point 2011-12-20T17:24:51 my next games coming up in just over an hour 2011-12-20T17:25:10 yeah, only seeds is what I was suggesting 2011-12-20T17:25:18 *** Westicles has quit IRC (Ping timeout: 258 seconds) 2011-12-20T17:27:29 *** dici has quit IRC (Ping timeout: 252 seconds) 2011-12-20T17:27:35 I'm in the place I finished in for planet wars 2011-12-20T17:27:42 *** dici has joined #aichallenge 2011-12-20T17:28:19 McLeopold: is it ok for me to post on the forums about the change? 2011-12-20T17:28:52 McLeopold: I'll describe as a trial until all the 6 game bots have 7 2011-12-20T17:29:39 it's a trial until ChrisH_ is in #1 2011-12-20T17:29:40 aichallenge: McLeopold epsilon * r9d293de / sql/2_generate_matchup.sql : switch seed to least played for finals - http://git.io/70NrIw 2011-12-20T17:29:45 LOL 2011-12-20T17:30:01 BenJackson: are you feeling nervous ? 2011-12-20T17:30:18 *** tobym has quit IRC (Remote host closed the connection) 2011-12-20T17:30:20 ChrisH_: yes, please do 2011-12-20T17:30:26 haha BenJackson :P 2011-12-20T17:30:49 ChrisH_: nowhere to go but down now! 2011-12-20T17:30:57 I'm pretty sure I'll make the cutoff though 2011-12-20T17:30:59 as a matter of interest, are there any prizes for this? :P 2011-12-20T17:31:05 *** TheLinker has joined #aichallenge 2011-12-20T17:31:07 conor_f afaik no 2011-12-20T17:31:09 round 1: top 8000. round 2: top 1. 2011-12-20T17:31:17 :P 2011-12-20T17:31:19 well there is one for my ego 2011-12-20T17:31:22 the prize is an elephant 2011-12-20T17:31:22 conor_f: I was offering to make something 2011-12-20T17:31:29 good point, what are the cutoff points? 2011-12-20T17:31:54 the top 10 contestants will have the chance to meet at a pub in hungary, ill pay the beer 2011-12-20T17:32:03 mcstar: ;) 2011-12-20T17:32:21 that will become very expensive, very fast :P 2011-12-20T17:32:23 i should have said top 3, people can drink much.... 2011-12-20T17:32:24 who pays the trip? :p 2011-12-20T17:32:31 oops 2011-12-20T17:32:33 mcstar: exactly :) 2011-12-20T17:32:39 14:32 < McLeopold> oops 2011-12-20T17:32:39 I created a match for mister 55 games :( 2011-12-20T17:32:47 <- me reload rankings 2011-12-20T17:32:51 mr 56 games now 2011-12-20T17:33:01 McLeopold while you are at it create a few more for me pls :D 2011-12-20T17:33:07 FYI a1k0ns performance in torn helped him get a job at goggle 2011-12-20T17:33:20 Google 2011-12-20T17:33:21 aichallenge: McLeopold epsilon * r1f0f35a / sql/2_generate_matchup.sql : switch seed to least played for finals - http://git.io/WYDrVQ 2011-12-20T17:33:22 Anti: wow, really? :) 2011-12-20T17:33:24 the goggles do nothing 2011-12-20T17:33:27 thats impressive 2011-12-20T17:33:40 * avdg doesn't know if getting a job would be good for him 2011-12-20T17:33:46 He's a pretty smart guy though 2011-12-20T17:33:48 lol 2011-12-20T17:33:55 it is good for anyone 2011-12-20T17:34:00 it makes you a man from a boy 2011-12-20T17:34:11 i hope i get one soon :D 2011-12-20T17:34:23 He's always had the neatest ideas IMO and is the only one to always pull off non trivial ones 2011-12-20T17:34:24 *** Guest76341 is now known as dapplegate 2011-12-20T17:34:27 *** Jak_o_Shadows1 has joined #aichallenge 2011-12-20T17:34:29 meh, its a mess here 2011-12-20T17:34:53 Anti: I read his writeup on tron. His grasp of graph theory would have gotten him a job at google anyway 2011-12-20T17:35:14 Yeah, that's what I'm trying to say essentially 2011-12-20T17:35:29 i would mention in passing, that game developers have a bit of an edge in these contests 2011-12-20T17:35:34 *** Scryer has quit IRC (Remote host closed the connection) 2011-12-20T17:35:37 *** tobym has joined #aichallenge 2011-12-20T17:35:38 *** dici has quit IRC (Quit: ( www.nnscript.com :: NoNameScript 4.22 :: www.esnation.com )) 2011-12-20T17:35:46 I'm really going to have a go at the next challenge, whatever it is 2011-12-20T17:35:50 *** grgr has quit IRC (Quit: Page closed) 2011-12-20T17:35:54 amstan: change made, wait for the server crash now 2011-12-20T17:35:58 Mcstar: lots of people have their own advantages 2011-12-20T17:36:00 they teach you a lot when you take them serious 2011-12-20T17:36:19 * avdg hopes he can plan things better without being disturbed 2011-12-20T17:36:23 Anti: lets say, on average 2011-12-20T17:36:26 if this goes badly, we'll stop, otherwise I'll raise it to 7 games 2011-12-20T17:36:37 I already coded at night this competition 2011-12-20T17:36:50 They do, I've been too lazy about learning more programming specific stuff, I'm going to spend more time on that this Sumer 2011-12-20T17:37:17 I am going to participate in the beta stage of the next contest whatever it is. 2011-12-20T17:37:21 Anti: whats your bot? 2011-12-20T17:37:23 *** Jak_o_Shadows has quit IRC (Ping timeout: 252 seconds) 2011-12-20T17:37:32 yeah, I have school and them free stanford courses (since they're free, I'm compelled to do them :P) so I didn't have much time for this :/ 2011-12-20T17:37:33 mcstar anti stroud 2011-12-20T17:37:39 Anti stroud 2011-12-20T17:37:46 Bah stupid ipD 2011-12-20T17:37:46 besh: can you sign up to participate in the beta? 2011-12-20T17:37:50 mcstar: I grew faster :) http://tcpants.com/replay.18091 2011-12-20T17:37:56 btw, http://www.youtube.com/watch?v=L_dttkuqJwM 2011-12-20T17:37:57 that 2011-12-20T17:38:48 conor_f: I think so. Many strong engines including xathis were developed at that time 2011-12-20T17:38:51 McLeopold: yep, id love to have that code in my bot 2011-12-20T17:39:04 but you surprised us with your timeout 2011-12-20T17:39:09 interesting... 2011-12-20T17:39:14 it looks like you maintain coverage, which slows you down 2011-12-20T17:39:44 http://aichallenge.org/profile.php?user=98 2011-12-20T17:39:48 Mcstar there 2011-12-20T17:40:06 mcstar: it's my combat code, no timeout checks in there 2011-12-20T17:40:12 I need to put one in 2011-12-20T17:40:15 McLeopold: post made 2011-12-20T17:40:25 Anti: ? 2011-12-20T17:40:27 http://forums.aichallenge.org/viewtopic.php?f=27&t=2183&p=13630#p13630 2011-12-20T17:40:32 you linked in antimatroid's profile 2011-12-20T17:40:56 I am anti stroud 2011-12-20T17:41:01 I'm in Sydney 2011-12-20T17:41:13 Bloody iPad 2011-12-20T17:41:18 are you nick? 2011-12-20T17:41:53 I am nick ham yes 2011-12-20T17:42:00 ah ok 2011-12-20T17:42:09 why are yo keep saying anti stroud? 2011-12-20T17:42:13 Sorry I'm trying to use my cousins iPad, and not doing well aha 2011-12-20T17:42:19 aha 2011-12-20T17:42:23 ok that i remember 2011-12-20T17:42:24 iPad. Auto corrects it to that 2011-12-20T17:42:29 McLeopold: my player's ahead number hasn't dropped like I would expect, will that take time to adjust? 2011-12-20T17:42:58 maybe he's just skipping seeds 2011-12-20T17:43:06 ChrisH_: no, I'm not recalculating that 2011-12-20T17:43:31 *** Anilm3 has quit IRC (Quit: leaving) 2011-12-20T17:43:32 McLeopold: oh 2011-12-20T17:43:36 so, that may piss off some people who were next 2011-12-20T17:43:45 cool it says I am playing a game right now 2011-12-20T17:43:46 McLeopold: ok, so 6 game bots will just jump the queue unexpectedly 2011-12-20T17:43:55 mcstar: timing code added, here goes game 2 on the pants server... 2011-12-20T17:44:18 ChrisH_: yes 2011-12-20T17:44:21 ok 2011-12-20T17:44:26 good to know if anyone asks 2011-12-20T17:44:32 the playing in a game right now will be accurate 2011-12-20T17:44:40 ok 2011-12-20T17:45:50 McLeopold thanks, am waiting for my 7th game too 2011-12-20T17:46:23 the problem was, that i only tested against my bot to get a good food collecting behavior, and it seemed goo enough for me, looks like it not the best around 2011-12-20T17:46:50 yeah, even paratrechina beats me on food, I thought I was pretty good 2011-12-20T17:47:13 *** bluegaspode has joined #aichallenge 2011-12-20T17:48:25 and it will matter much, if you have territory, you can be much more aggresive, obviously 2011-12-20T17:48:27 damn 2011-12-20T17:48:50 yeah, I won a game! 2011-12-20T17:49:04 *** Anti has quit IRC (Ping timeout: 258 seconds) 2011-12-20T17:49:11 *** antimatroid has joined #aichallenge 2011-12-20T17:49:23 ah ok 2011-12-20T17:49:26 thats easy 2011-12-20T17:49:29 :) 2011-12-20T17:49:57 i got ownef 2011-12-20T17:50:02 adam_ant doesnt put up much fight 2011-12-20T17:50:07 who's adam_ant? 2011-12-20T17:50:17 dunno 2011-12-20T17:51:03 antimatroid: want that bad 2011-12-20T17:51:16 you couldnt expand 2011-12-20T17:51:49 it still went badly 2011-12-20T17:51:59 all you 6 game losers, let me know when you hit 7 2011-12-20T17:52:01 but yeah, my bot does a lot better when it can set itself up a bit first 2011-12-20T17:52:19 McLeopold please tell me that there will be a cutoff soon 2011-12-20T17:52:27 McLeopold: i've had 2 new games and i'm now on 11 2011-12-20T17:52:29 flowenol: after everyone hits 10 games 2011-12-20T17:52:33 *** erij has quit IRC (Quit: Page closed) 2011-12-20T17:52:50 McLeopold: where will be the cutoff? 2011-12-20T17:52:51 Does anyone have an ant that does different behavior when the turn count gets close to 1000? 2011-12-20T17:53:04 ok now where is dat table 2011-12-20T17:53:12 f4hy: i think chrisH said he stops pathfinding at turn 1000 2011-12-20T17:53:16 or something like that 2011-12-20T17:53:24 f4hy: i have 2011-12-20T17:53:34 I think the behavior of my bot is already close to "almost undefined" 2011-12-20T17:53:36 on tuns-50 it prefers KILL 2011-12-20T17:53:43 less than 400 6 games bots 2011-12-20T17:53:45 antimatroid: turn limit is 1000 iirc 2011-12-20T17:53:46 and close to end - even DIE 2011-12-20T17:53:55 *** erij has joined #aichallenge 2011-12-20T17:54:05 I imagine you could have it so if you know the enemy cant make it to your hill in the next 50 turns, just go all out attack turn off your attack code even and jsut charge at the enemy hill 2011-12-20T17:54:29 *** Storm__ has quit IRC (Quit: Page closed) 2011-12-20T17:54:34 bearoff: nice 2011-12-20T17:54:36 http://tcpants.com/replay.18105 i tied with the big names 2011-12-20T17:54:43 f4hy:i tried to attack more when the turn limit gets close. 2011-12-20T17:55:10 I think a1k0n valued ants in relation to turns left 2011-12-20T17:55:18 so ants where more valuable at the start 2011-12-20T17:55:44 ac.coun is really stupid for not submitting 2011-12-20T17:55:47 Glad people thought of that. I only thought of it just now. 2011-12-20T17:56:08 f4hy: there were many bots who became more agressive near the end 2011-12-20T17:56:41 considering the number of games to be done and current player rate 2011-12-20T17:56:41 mcstar: don't you have an example at the moment? 2011-12-20T17:56:48 besh: i know, i just mean when the hills aren't too close at the start 2011-12-20T17:56:51 80mins to level at 10 games more or less 2011-12-20T17:56:52 accoun is the bode miller of ai 2011-12-20T17:57:04 how good is accouns bot? 2011-12-20T17:57:09 antimatroid: pretty good 2011-12-20T17:57:12 if it is his 2011-12-20T17:57:14 and i wonder if it would run on the contest servers 2011-12-20T17:57:14 *** erij has left #aichallenge 2011-12-20T17:57:22 i expect it is 2011-12-20T17:57:25 i mean he only let it play after many pople released their code 2011-12-20T17:57:27 s 2011-12-20T17:57:30 he's had strong bots for tron and planet wars 2011-12-20T17:57:34 but probably it is his 2011-12-20T17:57:57 mcstar: are the games slow with him? 2011-12-20T17:58:09 i cant say 2011-12-20T17:58:12 *** gaussin has joined #aichallenge 2011-12-20T17:58:17 am talking to you not watching the console 2011-12-20T17:59:19 *** tobel has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2011-12-20T17:59:25 I saw some slowness of accoun, but he had above 300 ants and there was much food 2011-12-20T17:59:51 McLeopold: if you say only 400 6 games bots, should my profile then also say that only 400 player are ahead? 2011-12-20T18:00:00 if I'm a 6 play bot? 2011-12-20T18:00:05 no 2011-12-20T18:00:06 this part is not fixed 2011-12-20T18:00:36 McLeopold: did you stop playing? 2011-12-20T18:00:42 no 2011-12-20T18:00:47 in a game 2011-12-20T18:01:18 down to 300 6 game bots 2011-12-20T18:01:43 300 / 50 games / minute = 6 minutes 2011-12-20T18:01:46 going fast 2011-12-20T18:01:52 ok - not a time to go to bed now :) 2011-12-20T18:02:04 but then it will be adjusted for the 7 game bots right? 2011-12-20T18:02:24 or maybe rather go to bed than to be depressed when my bot loses 2011-12-20T18:02:28 as yet none challenge my hegemony 2011-12-20T18:02:29 *** mj41 has quit IRC (Ping timeout: 240 seconds) 2011-12-20T18:02:30 hard decision 2011-12-20T18:02:45 bluegaspode sleep 2011-12-20T18:02:47 its better 2011-12-20T18:02:59 *** Nefbrethilion has quit IRC (Quit: Page closed) 2011-12-20T18:03:18 Did pguillory release his code ? 2011-12-20T18:03:53 some of the new maps seem pretty well set up for stalemates 2011-12-20T18:04:08 there are some really bad new maps 2011-12-20T18:04:22 there's one with a 3 or 4 wide chokepoint really close to both player's spawns 2011-12-20T18:04:30 which is only going to measure who gets one ant there first 2011-12-20T18:04:40 can I get game lists by map? 2011-12-20T18:04:46 "In Game: Playing in a game right now." 2011-12-20T18:04:49 Huzzah! 2011-12-20T18:05:11 am still in queue.. 2011-12-20T18:05:33 and my rank is going back up 2011-12-20T18:05:40 without playing games 2011-12-20T18:05:57 flowenol: a sign things are stabilizing 2011-12-20T18:06:24 all i want is see the game rate going up 2011-12-20T18:06:31 like one every hour or so 2011-12-20T18:06:41 http://aichallenge.org/map.php?map=maze/maze_p02_37.map 2011-12-20T18:06:42 im way too addicted to this 2011-12-20T18:06:42 onion42: Be a good boy now :) 2011-12-20T18:06:53 :) 2011-12-20T18:06:56 McLeopold: can you get me a list of games on that map? 2011-12-20T18:07:35 BenJackson interesting map 2011-12-20T18:07:35 yeah 2011-12-20T18:07:37 it might have the shortest average duration 2011-12-20T18:07:37 That seems to be a map that's quickly decided 2011-12-20T18:07:52 uh 2011-12-20T18:07:55 171.70 +/- 169.48 2011-12-20T18:07:59 haha 2011-12-20T18:08:00 its on the page 2011-12-20T18:08:06 oh, sry 2011-12-20T18:08:12 I have "maximize" saved for visualizer 2011-12-20T18:08:20 I always forget there's stuff "under" the map 2011-12-20T18:08:40 that's a great map, I love it 2011-12-20T18:09:23 http://aichallenge.org/visualizer.php?game=290921 2011-12-20T18:09:24 hahaah 2011-12-20T18:09:25 yeah it really tests intelligence 2011-12-20T18:09:27 ok not what I epxected 2011-12-20T18:09:43 if you stop there, and fight you wont get food 2011-12-20T18:09:54 and the other a coupl eof turns later will overpower you 2011-12-20T18:09:59 you have to fight, the gap isn't wide enough 2011-12-20T18:10:03 to zerg, or not to zerg 2011-12-20T18:10:04 *** rebelxt has quit IRC (Quit: Page closed) 2011-12-20T18:10:10 you have to leave a couple of ants 2011-12-20T18:10:13 but dont fight 2011-12-20T18:10:15 I think 2 strong bots will meet there and not get out 2011-12-20T18:10:18 gather food, and then fight 2011-12-20T18:10:22 that map may be mad for my bot 2011-12-20T18:10:29 *bad 2011-12-20T18:10:38 my bod probably will fight 2011-12-20T18:10:40 I thikn my bot may be great at it 2011-12-20T18:10:50 but if i remmeber correctly there are strictly food collectors 2011-12-20T18:10:51 I can get you a list ordered by rank 2011-12-20T18:10:51 you just have to get one ant out 2011-12-20T18:10:56 due to some metrics i put in to try to force fast expand on those kinds of maps 2011-12-20T18:11:00 i dont kknow if i commented it out or not 2011-12-20T18:11:17 McLeopold: ideally by their rank NOW rather than as played 2011-12-20T18:11:25 yes, let me change that 2011-12-20T18:12:41 McLeopold: check your latest game :) 2011-12-20T18:13:03 *** roflmao has quit IRC (Quit: Leaving.) 2011-12-20T18:13:11 McLeopold: how many 6-games bots left? 2011-12-20T18:13:19 *** fnord___1 has joined #aichallenge 2011-12-20T18:13:30 BenJackson: http://aichallenge.org/visualizer.php?game=284265 2011-12-20T18:13:30 *** fnord____ has quit IRC (Ping timeout: 245 seconds) 2011-12-20T18:13:45 arg 2011-12-20T18:13:51 so close 2011-12-20T18:13:51 I think a good solution is to have my bot be the seed for every match. I think that would equalize the ranks fastest ;-) 2011-12-20T18:14:21 at the current game rate all 400 should have played by now 2011-12-20T18:14:28 McLeopold: i lost a lot of ants in battle 2011-12-20T18:14:40 188 6 gamers left 2011-12-20T18:14:49 bluegaspode: I'm still with 6 and not in game ) 2011-12-20T18:14:58 me too 2011-12-20T18:14:58 *** foRei has quit IRC (Quit: Bye) 2011-12-20T18:14:59 thanks, McLeopold 2011-12-20T18:15:01 There's quite a few 'extra' games 2011-12-20T18:15:08 cheer up! ) 2011-12-20T18:15:10 bluegaspode 4/5mins maybe 2011-12-20T18:15:15 McLeopold: btw, you could have taken blue there 2011-12-20T18:16:10 yeah, my combat doesn't think about hills 2011-12-20T18:16:56 McLeopold: my combat didn't think about hills, and when I added that, that component of my scoring function overwhelmed all the other components and made my ants extremely dumb near hills. Lesson learned. 2011-12-20T18:17:15 pity 2011-12-20T18:17:17 *** Surya has joined #aichallenge 2011-12-20T18:17:21 *** f4hy is now known as moduspwned 2011-12-20T18:17:29 hills are like food to my ants 2011-12-20T18:17:41 mcstar delicious? 2011-12-20T18:17:46 My ants try to eat hills as well 2011-12-20T18:17:51 *** conor_f has quit IRC (Ping timeout: 252 seconds) 2011-12-20T18:18:01 I think the real problem is my combat doesn't attempt to overwhelm small numbers of ants 2011-12-20T18:18:02 the only difference, that they step on it instead near it 2011-12-20T18:18:02 my ants try to raze food 2011-12-20T18:18:10 Hmm. chrisH lost in one of my games 2011-12-20T18:18:10 in my bot as well. these leads to fast hill razing on multi hill maps 2011-12-20T18:18:35 *** protocolocon has quit IRC (Ping timeout: 258 seconds) 2011-12-20T18:18:51 http://aichallenge.org/profile.php?user=568 2011-12-20T18:19:08 wasn't he on the undefeated list? 2011-12-20T18:19:45 now I won't feel bad if I lose my 7th game 2011-12-20T18:19:55 wow, playing a game 2011-12-20T18:20:03 *** matthewd has quit IRC (Ping timeout: 245 seconds) 2011-12-20T18:20:04 down to 4 undefeated: GreenTea, xathis, FlagCapper and cs_bot 2011-12-20T18:20:14 the last 2 are only at 6 games 2011-12-20T18:20:21 wow 2011-12-20T18:20:24 what is undefeated? 2011-12-20T18:20:28 I'm one of the last two at 6 games 2011-12-20T18:20:45 every game in finals finished better than everyone else 2011-12-20T18:20:46 still not playing :) 2011-12-20T18:20:50 so no ties for first 2011-12-20T18:20:57 thanks 2011-12-20T18:21:09 damn strong bots 2011-12-20T18:21:29 flag still blows my mind 2011-12-20T18:21:36 how so? 2011-12-20T18:21:42 the winner, cool ) 2011-12-20T18:21:43 with his 16 years 2011-12-20T18:21:50 but 19 isnt that much either 2011-12-20T18:21:55 ? 2011-12-20T18:22:34 what coincidence makes me the last one of the 6games players? Is it my need to still go to bed to be not too tired tomorrow? 2011-12-20T18:22:54 playiiingggg ... 2011-12-20T18:22:56 now you must wait 2011-12-20T18:23:12 sayng spells or smth like this 2011-12-20T18:23:18 im also playing now bluegaspode... maybe we are attacking each other 2011-12-20T18:23:29 :) 2011-12-20T18:23:31 *** tdubellz_ has joined #aichallenge 2011-12-20T18:23:39 mcstar: we both crash after losing our hills :P 2011-12-20T18:23:52 i dont 2011-12-20T18:24:03 i havent watched it yet 2011-12-20T18:24:27 *** fnord___1 has quit IRC (Quit: Lost terminal) 2011-12-20T18:24:38 McLeopold: i didnt even move, it was a timeout from the start 2011-12-20T18:24:54 oh 2011-12-20T18:25:03 I guess the visualizer is wrong 2011-12-20T18:25:32 *** dmj111` has joined #aichallenge 2011-12-20T18:25:54 hm 2011-12-20T18:25:55 *** dmj111 has quit IRC (Read error: Connection reset by peer) 2011-12-20T18:25:55 *** Darhuuk has quit IRC (Ping timeout: 245 seconds) 2011-12-20T18:26:07 so i can just pattern match on Maybe types? 2011-12-20T18:26:09 *** rofer has quit IRC (Ping timeout: 240 seconds) 2011-12-20T18:26:13 *** tdubellz_ has quit IRC (Client Quit) 2011-12-20T18:26:23 *** matthewd has joined #aichallenge 2011-12-20T18:26:24 *** rofer has joined #aichallenge 2011-12-20T18:26:25 function Just x = balbal ? 2011-12-20T18:26:26 *** Darhuuk has joined #aichallenge 2011-12-20T18:26:39 any haskeller here? 2011-12-20T18:26:41 mcstar: haskell? 2011-12-20T18:26:51 You can do that, but it might fail, I think. 2011-12-20T18:26:57 a1k0n: what happened? http://aichallenge.org/visualizer.php?game=296187&user=432 2011-12-20T18:27:01 *** bretep has quit IRC (Ping timeout: 258 seconds) 2011-12-20T18:27:11 dmj111`: i mean how can you use something that is Just x ? 2011-12-20T18:27:20 find return Just 4 for example 2011-12-20T18:27:25 now what do i do with it? 2011-12-20T18:27:27 s 2011-12-20T18:27:32 Come on Deresh, you can do it 2011-12-20T18:27:49 onion42: let me know when you loose .. cause I am playing too :) 2011-12-20T18:27:57 pattern matching is one way to go. My haskell is rusty... but you can pattern match to extract the data, else handle the None case. 2011-12-20T18:28:03 i lost :) 2011-12-20T18:28:07 (i am much more familiar with F#/ocaml) 2011-12-20T18:28:29 my combat code is giving me some problems on maze maps... too many standstills :( 2011-12-20T18:28:31 are bots timing out that shouldn't? 2011-12-20T18:28:32 amstan: I happened ;-) 2011-12-20T18:28:34 Nothing 2011-12-20T18:28:35 that's a serious problem 2011-12-20T18:29:32 antimatroid: whats your wpm on the ipad? 2011-12-20T18:29:32 My brothers bot times out his first two matches and he claims it should never time out, but hasnt happened since his first two matches 2011-12-20T18:29:59 34 bots left 2011-12-20T18:30:11 mcstar: i've switched to a computer for the moment 2011-12-20T18:30:23 :) 2011-12-20T18:30:25 whats the minimum number of games before the starter bots get culled? :) 2011-12-20T18:30:26 *** grwip has quit IRC (Quit: Leaving) 2011-12-20T18:30:32 10 2011-12-20T18:30:36 playing for 8 minutes. Is this a good or pad sign? 2011-12-20T18:30:39 mcstar: http://book.realworldhaskell.org/read/code-case-study-parsing-a-binary-data-format.html 2011-12-20T18:30:47 antimatroid: ill be gone for the holidays too 2011-12-20T18:30:53 i'll be lucky not to get culled myself :S 2011-12-20T18:30:56 probably beofore we know the resutls 2011-12-20T18:31:04 mcstar: there is a big ugly example, but it uses pattern maching on Maybe's 2011-12-20T18:31:04 bad sign, I lost :( 2011-12-20T18:31:06 and there aint gonna be internet 2011-12-20T18:31:06 too many bugs in my final version :( 2011-12-20T18:31:16 dmj111`: thx 2011-12-20T18:31:19 onion42 too 2011-12-20T18:31:26 too few features in mine 2011-12-20T18:31:38 come and join me in last palce bluegaspode :) 2011-12-20T18:32:04 bluegaspode, this map is for licky bots 2011-12-20T18:32:04 *** dmj111` is now known as dmj111 2011-12-20T18:32:08 *lucky 2011-12-20T18:32:20 *** olexs has quit IRC (Quit: Leaving.) 2011-12-20T18:32:22 I probably just had to make a separate map with targets 2011-12-20T18:32:24 mcstar: i'm going to be at my cousins house most of the time, so i'll have net :) 2011-12-20T18:32:24 doesn't matter ) 2011-12-20T18:32:30 my bot is so bad at maze maps now, its embarrassing lol 2011-12-20T18:32:57 what it 'licky bots' ? 2011-12-20T18:33:10 looks like you got teamed up on bluegaspode :( 2011-12-20T18:33:20 orange and green surrounded you 2011-12-20T18:33:27 *lucky - i mean it's much more luck than skill needed on this map 2011-12-20T18:33:30 wrapped you in a blanket of ants 2011-12-20T18:33:47 *** matthewd has quit IRC (Ping timeout: 245 seconds) 2011-12-20T18:33:53 McLeopold damn finish that count you are making me nervous 2011-12-20T18:34:14 i'm up to 9 games now :D 2011-12-20T18:34:15 Do we have any new infos on when the bot cut will occur? 2011-12-20T18:34:38 *** matthewd has joined #aichallenge 2011-12-20T18:34:39 when all bots will have 10 games 2011-12-20T18:34:45 and onion got trapped pretty fast as well 2011-12-20T18:34:51 My last game is like it never happend: Last place :( 2011-12-20T18:34:54 I think we agree on 'bad map' :) 2011-12-20T18:34:59 bearoff: official info? 2011-12-20T18:35:07 flowenol: what count? 2011-12-20T18:35:07 *** tomb___ has joined #aichallenge 2011-12-20T18:35:10 McLeopold said 2011-12-20T18:35:15 ok, thx 2011-12-20T18:35:19 mcstar: You can also use "isJust" to test. "fromJust" to get the value if you are SURE that you have a Just value. Or maybe (result if Nothing) (function if Just) (maybe value).. maybe :: b -> (a -> b) -> Maybe a -> b 2011-12-20T18:35:21 i like the map where i start with 10 hill and everyone else only gets 1 :) 2011-12-20T18:35:22 http://tcpants.com/replay.18132 2011-12-20T18:35:26 McLeopold the bots left count 2011-12-20T18:35:34 i could beat accoun, by razing hills faster 2011-12-20T18:35:34 10 2011-12-20T18:35:43 McLeopold: look at that link 2011-12-20T18:35:43 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa 2011-12-20T18:35:44 McLeopold: cs_bot lost too 2011-12-20T18:35:48 Is it known how many will make the cut? 2011-12-20T18:35:49 besh: try to beat this one http://ants-score.appspot.com/?uid=Steve887 :-) 2011-12-20T18:35:50 * flowenol starts running around 2011-12-20T18:36:06 Steve887 is just a slow starter 2011-12-20T18:36:14 :D 2011-12-20T18:36:16 dr0b3rts: ah ok, ill get that in time 2011-12-20T18:36:17 he will win the contest in the end 2011-12-20T18:36:26 FranPeruzzi : I'd like to know it too 2011-12-20T18:36:28 Steve887 prefers stability 2011-12-20T18:36:41 bearoff: in rank? yeah 2011-12-20T18:36:49 dr0b3rts: did haskellers release code? 2011-12-20T18:37:08 Surya: looks like we're in comparable position 2011-12-20T18:37:10 @rankings 2011-12-20T18:37:11 antimatroid: Top 10 players: BenJackson(79.0), delineate(78.5), Belerafon(78.0), _Atsky_(78.0), a1k0n(77.1), lazarant(77.0), meduza(77.0), JG.WAS(76.8), strcat(76.6), antimatroid(76.3) 2011-12-20T18:37:15 yeahhhh 2011-12-20T18:37:19 http://aichallenge.org/visualizer.php?game=297213&user=3474 2011-12-20T18:37:28 Steve887 has invented the perpetual motion machine, its a miracle but sadly it always times out 2011-12-20T18:37:34 I haven't looked recently. I haven't released mine. 2011-12-20T18:37:45 *** Zed_ has joined #aichallenge 2011-12-20T18:38:01 I've seen an OCaml bot, which is somewhat like Haskell. 2011-12-20T18:38:08 Steve887 has his own goals. Getting his ranking below outer space temperature. 2011-12-20T18:38:09 onion42: its a miracle it passed the tests :p 2011-12-20T18:38:10 and now the 7games bot will be preferred until they reach 23 games like others? 2011-12-20T18:38:19 *** Speedy_Consoles has joined #aichallenge 2011-12-20T18:38:26 sort the first page by game number :P 2011-12-20T18:38:30 antimatroid: you can join me in my quest to drive up EC2 spot prices and end the contest now 2011-12-20T18:38:30 antimatroid: you could have taken those hills with that shitload of red dots :) 2011-12-20T18:38:33 FranPeruzzi: Yeah! 2 spots behind you! 2011-12-20T18:38:45 *** amstan_ has joined #aichallenge 2011-12-20T18:38:45 *** ChanServ sets mode: +o amstan_ 2011-12-20T18:39:00 Speedy_Consoles: lol 2011-12-20T18:39:25 you just beat xathis that way 2011-12-20T18:39:32 Surya: You have higher mu and sigma though 2011-12-20T18:39:45 did xathis lose? 2011-12-20T18:39:55 my sigma went down way too fast imo 2011-12-20T18:40:17 those 3 timeouts didnt help at all 2011-12-20T18:40:17 onion42: xathis didn't lose yet 2011-12-20T18:40:30 *** bearoff has left #aichallenge 2011-12-20T18:40:38 just sort the list by played games 2011-12-20T18:40:45 beat as in rankings, i suppose 2011-12-20T18:40:49 sigma doesn't matter does it? It's the variance. The higher it is, the less the system is sure about the mu. Am I right? But yeah, I got higher mu, so I keep playing good bots (and eventually lose :)) 2011-12-20T18:40:58 its not just me on top then, its universität zu lübeck on top :D 2011-12-20T18:41:09 Speedy_Consoles: looks like they don't like our university ;) 2011-12-20T18:41:19 mcstar: could have, didn't 2011-12-20T18:41:21 Speedy_Consoles: do you know xathis in real life then? 2011-12-20T18:41:24 my bot is not at its potential 2011-12-20T18:41:26 but it sort of works 2011-12-20T18:41:31 he introduced me into ants :) 2011-12-20T18:41:33 1 bot left... 2011-12-20T18:41:36 :p 2011-12-20T18:41:46 McLeopold and then what? 7 game bots? 2011-12-20T18:41:51 1 bot left before what?* 2011-12-20T18:42:10 amstan_: I'm going to let the normal seeds curn for a little bit, then do the same for 7 game bots, sound okay? 2011-12-20T18:42:24 sure 2011-12-20T18:42:46 curn? 2011-12-20T18:42:51 churn 2011-12-20T18:43:01 process 2011-12-20T18:43:08 what worries me is when will the cut be done 2011-12-20T18:43:14 time seems so short 2011-12-20T18:43:28 No worry 2011-12-20T18:43:39 When there's <100 bots left the games accumulate fast 2011-12-20T18:43:42 your algorithms are just bad :D 2011-12-20T18:43:47 is GreedyBot still being used as the benchmark for cutting bots? 2011-12-20T18:43:49 I didn't get the 7 bots thingy :s 2011-12-20T18:44:12 we just need to send the servers into space really fast to speed up the process 2011-12-20T18:44:13 Surya bits with only 7 games 2011-12-20T18:44:36 preferable in a black hole maybe? 2011-12-20T18:44:36 good idea, good idea... and how to get them back? 2011-12-20T18:44:43 other way around 2011-12-20T18:44:44 onion42: nothing like that will happen for cutoffs 2011-12-20T18:44:50 pairofdice <100 bots worries me, mine will not make it there 2011-12-20T18:44:59 Such is competition 2011-12-20T18:45:00 if the servers go at relativistic speeds, they'll experience less time than us :P 2011-12-20T18:45:05 Zannick: from the servers perpective, we will be slow :) 2011-12-20T18:45:20 Zannick, time also slows down near mass 2011-12-20T18:45:33 we should send the servers back in time and run the competition last week :D 2011-12-20T18:45:55 Near mass effect is probably higher than orbital velocity 2011-12-20T18:45:57 lol, I wish we had portable mass boxes 2011-12-20T18:46:04 or put them in narnia, where time runs faster :P 2011-12-20T18:46:06 we would control time that way 2011-12-20T18:46:20 *** matthewd has quit IRC (*.net *.split) 2011-12-20T18:46:22 or maybe not 2011-12-20T18:46:25 I don't think narnia has electricity... 2011-12-20T18:46:31 1 net split? 2011-12-20T18:46:34 they have lamps, i think 2011-12-20T18:46:34 that's rather small 2011-12-20T18:46:41 oil lamps 2011-12-20T18:46:43 good point, but no wires? 2011-12-20T18:46:43 take some electric generators 2011-12-20T18:46:47 where will be the cutoff? 2011-12-20T18:46:48 oil powered servers? 2011-12-20T18:46:58 the proper sorting function changes with time, right now it is: (sortBy (on compare (\x->abs(x-198)))) 2011-12-20T18:47:11 we could just run a cord through the wardrobe 2011-12-20T18:47:22 McLeopold works for me 2011-12-20T18:47:36 i never understood that 2011-12-20T18:47:47 you get time dilatation by going near the speed of light 2011-12-20T18:47:52 but relative to what? 2011-12-20T18:47:57 lol 2011-12-20T18:47:59 :D 2011-12-20T18:48:04 relative to another frame of reference 2011-12-20T18:48:18 but what is that other reference 2011-12-20T18:48:20 who's to say that earth is not going at the speed of light the other way and the rocket is stationary? 2011-12-20T18:48:25 we just need more speedy consoles 2011-12-20T18:48:32 exactly amstan 2011-12-20T18:48:37 frames of reference 2011-12-20T18:48:44 bring industrialization to narnia 2011-12-20T18:48:48 amstan_: that doesnt matter 2011-12-20T18:48:51 they need it 2011-12-20T18:48:51 so.. which one experiences more severe time dilatation then? 2011-12-20T18:48:55 but accleration changes the picture 2011-12-20T18:48:58 i mean their king is a lion 2011-12-20T18:49:01 since you don't know which one goes faster 2011-12-20T18:49:05 lol 2011-12-20T18:49:19 theres some serious errors in your thought process 2011-12-20T18:49:22 sorry 2011-12-20T18:49:59 so.. think of the classic example, like in planet of the apes 2011-12-20T18:50:02 amstan_: if you want to study relativity first try to understand what simultaenity is 2011-12-20T18:50:08 you get a rocket leaving earth, and then coming back 2011-12-20T18:50:14 and from at least wikipedia, not from here 2011-12-20T18:50:19 Dr Zaius, Dr Zaius! 2011-12-20T18:50:21 thast insane 2011-12-20T18:50:24 pairofdice: so is 100 the official cutoff number? 2011-12-20T18:50:26 the rocket accelerates 2011-12-20T18:50:40 FranPeruzzi: no 2011-12-20T18:50:41 FranPeruzzi: no, not at all 2011-12-20T18:50:42 you have to integrate everything by proper time 2011-12-20T18:50:47 rocket's explode! 2011-12-20T18:51:06 *** matthewd has joined #aichallenge 2011-12-20T18:51:31 The acceleration is fairly important 2011-12-20T18:51:35 mcstar: so what frame of reference is narnia in? 2011-12-20T18:51:38 FranPeruzzi: more like a couple of thousands 2011-12-20T18:51:47 :P 2011-12-20T18:51:50 what? 2011-12-20T18:51:57 grab some solar panels, the servers and move them into narnia 2011-12-20T18:51:57 what is narnia? 2011-12-20T18:51:57 oh and if you sort by games 2011-12-20T18:52:03 there are also 3 germans on top :D 2011-12-20T18:52:05 well time travels faster in narnia 2011-12-20T18:52:29 ah, ok 2011-12-20T18:53:13 *** Speedy_Consoles has quit IRC (Quit: Page closed) 2011-12-20T18:53:47 hmmm... what happens if you shine a laser into a narnia wardrobe? does the light speed up? :P 2011-12-20T18:54:09 you cant change the speed of light 2011-12-20T18:54:15 (You can) 2011-12-20T18:54:18 no 2011-12-20T18:54:19 it's fiction 2011-12-20T18:54:20 Make it slower 2011-12-20T18:54:25 maybe in some non-linear theory 2011-12-20T18:54:31 they have talking lions, i think the normal rules of physics may not apply 2011-12-20T18:54:32 how? 2011-12-20T18:54:40 glass slows light down to about 2/3 2011-12-20T18:54:42 lol 2011-12-20T18:54:45 thats not it 2011-12-20T18:54:48 It is 2011-12-20T18:54:56 That's light slowed down :) 2011-12-20T18:54:56 that is just interaction with matter 2011-12-20T18:55:03 you cant slow it down 2011-12-20T18:55:24 wrt c, the value of the speed of light in a vacuum, that doesn't really change 2011-12-20T18:55:38 unless youre in narnia 2011-12-20T18:55:45 we dont know of any effect that would change it 2011-12-20T18:55:50 when all bets are off 2011-12-20T18:56:04 Michelson and Morley shouldve done more tests in narnia 2011-12-20T18:56:06 no, the speed of light in a vacuum would be the same in narnia relative to narnia 2011-12-20T18:56:24 Zannick: ah thats true :D 2011-12-20T18:56:31 neutrino experiment 2011-12-20T18:56:31 you're unlikely to be able to compare in narnia with outside of narnia unless your portal was in a vacuum 2011-12-20T18:56:31 *** iglo has quit IRC (Remote host closed the connection) 2011-12-20T18:56:34 you can measure lesser c if you make light travel more 2011-12-20T18:56:42 for example 2011-12-20T18:56:48 and even then your measuring instruments take place in one or the other 2011-12-20T18:56:51 i shine light into a cavity 2011-12-20T18:56:56 light bounces off the walls 2011-12-20T18:57:03 i leave it in ther efor a couple of days 2011-12-20T18:57:08 and after that i let it out 2011-12-20T18:57:16 so light travelled only a meter in a day 2011-12-20T18:57:19 or 2 2011-12-20T18:57:21 lol 2011-12-20T18:57:33 thats what happens when light interacts with matter 2011-12-20T18:57:34 maybe thats whats been happening with the "faster than light" neautrinos in CERN... the particles travelled through a narnia wardrobe on their journey 2011-12-20T18:57:43 http://aichallenge.org/visualizer.php?game=296700&user=2255 meh 2011-12-20T18:57:43 photons are absorbed and emitted constantly 2011-12-20T18:58:04 but light doesnt behave then there, like a wave 2011-12-20T18:58:04 *** bluegaspode has quit IRC (Ping timeout: 258 seconds) 2011-12-20T18:58:07 in vacuum 2011-12-20T18:58:12 so you cant slow down light 2011-12-20T18:58:17 but you can absorve itsquantas 2011-12-20T18:58:23 and later re-emit it 2011-12-20T18:58:31 absorbe* 2011-12-20T18:58:42 I don't like that map at all :P 2011-12-20T18:59:02 *** gaussin has left #aichallenge 2011-12-20T19:00:20 McLeopold: http://tcpants.com/replay.18138 you didnt collect food faster here 2011-12-20T19:00:45 i collect food efficiently, while i have a thrust from the other ants behind, and this was a relatively small map 2011-12-20T19:02:37 *** matthewd has quit IRC (Ping timeout: 245 seconds) 2011-12-20T19:02:49 *** onion42 has quit IRC (Quit: Page closed) 2011-12-20T19:04:28 *** Redgis has quit IRC (Read error: Connection reset by peer) 2011-12-20T19:05:13 *** Redgis has joined #aichallenge 2011-12-20T19:07:46 okay, starting the 7 game bots... 2011-12-20T19:07:54 682 2011-12-20T19:08:05 McLeopold :D 2011-12-20T19:08:22 *** valydo has quit IRC (Ping timeout: 248 seconds) 2011-12-20T19:08:34 interesting there were ~400 before, does that mean a lot of them got a game during the 6 game bot run? 2011-12-20T19:08:51 because there were 600+ 6 game bots that should all be at 7 now 2011-12-20T19:08:51 yeah, 7 is after 6 :P 2011-12-20T19:09:17 there were at the paste McLeopold posted 2011-12-20T19:09:22 441 7 game bots 2011-12-20T19:09:50 http://pastebin.com/f8daAb3k 2011-12-20T19:09:54 updated? 2011-12-20T19:09:59 yeah, for a total of about 1000 with either 6 or 7, now only 682 with 7, so it's like killing 2 birds with 1 stone 2011-12-20T19:10:03 cool 2011-12-20T19:10:24 did you change the logic to pick seeds from the lowest game count then? 2011-12-20T19:10:28 yes 2011-12-20T19:10:29 holy ****, 66 games 2011-12-20T19:11:05 I noticed that those with 7 games are very stong engines ! 2011-12-20T19:11:12 i.e on the first page 2011-12-20T19:11:32 so in 1 hour or so 2011-12-20T19:11:39 all bots should have 10 games 2011-12-20T19:12:02 682 bots can't all possibly fit on the first page ;) 2011-12-20T19:12:13 xathis,flagcapper,fourmidable,speedy_consoles,greantea,komaki ... conicidence ? 2011-12-20T19:12:30 i bet they would suck against starter bots 2011-12-20T19:12:37 pairofdice: ok. but still 2011-12-20T19:12:44 yeah, they're only good against "good" bots 2011-12-20T19:12:52 The bots with 7 on any given page could be pretty strong compared to the rest on that page, wouldn't they? 2011-12-20T19:12:52 so who's to say which bots are actually good?! 2011-12-20T19:13:22 trueskill shoild 2011-12-20T19:13:23 win predictions http://pastebin.com/k11f3f2U 2011-12-20T19:13:26 *should 2011-12-20T19:13:30 (yes i'm just kidding) 2011-12-20T19:13:34 They have to have higher mu value to balance out the higher sigma that (generally) goes with less games. 2011-12-20T19:13:45 So trueskill thinks they're better, but isn't sure yet 2011-12-20T19:14:15 those sigmas are huuuge 2011-12-20T19:14:19 McLeopold: at least let the games play while your win prediction wont stabilize :) 2011-12-20T19:14:47 http://pastebin.com/fUideMFN 2011-12-20T19:15:02 with rounding so the table doesn't suck 2011-12-20T19:15:57 i'm kind of amazed it stabilized that well in seven games 2011-12-20T19:16:40 TrueSkill is fairly impressive for a MS product 2011-12-20T19:16:54 it's not so much a product as a research paper 2011-12-20T19:16:56 it was a paper 2011-12-20T19:17:05 Well yea 2011-12-20T19:17:09 *** dapplegate has quit IRC (Quit: Ex-Chat) 2011-12-20T19:17:13 speaking of that 2011-12-20T19:17:16 the actual implementation being used was not microsoft's doing ;) 2011-12-20T19:17:16 christopher bishop gives good lectures on graphical models 2011-12-20T19:17:22 anyone has a link to the paper? 2011-12-20T19:17:47 http://research.microsoft.com/apps/pubs/default.aspx?id=67956 2011-12-20T19:17:47 * avdg wonders who thinks that microsoft is really that bad 2011-12-20T19:18:04 Personally I think Apple is far worse ;p 2011-12-20T19:18:11 avdg haters? 2011-12-20T19:18:11 sure they have hypnotized people and marketing that doesn't make sense, but they also have good things 2011-12-20T19:18:24 microsoft research turns out a lot of great papers (so does yahoo labs for that matter) 2011-12-20T19:18:33 at google we are paranoid 2011-12-20T19:18:48 so you dont publish? 2011-12-20T19:19:06 well sometimes. mapreduce, gfs, etc 2011-12-20T19:19:09 we can't answer that. 2011-12-20T19:19:12 =P 2011-12-20T19:19:14 haha 2011-12-20T19:19:29 oops i already said too much nooooooooo 2011-12-20T19:19:44 fired 2011-12-20T19:19:57 like in Back to the future 2011-12-20T19:20:13 at least we don't have a loyalty team like apple :P 2011-12-20T19:20:13 yes, i just got a huge ascii art fax in my car 2011-12-20T19:20:38 apple can burn 2011-12-20T19:20:42 <- apple hater 2011-12-20T19:20:47 Google does have it's fans 2011-12-20T19:20:54 <- google fan 2011-12-20T19:21:00 yes, we do have fans. 2011-12-20T19:21:00 * avdg moves away from apple, but isn't an apple hater 2011-12-20T19:21:06 gotta stay cool in the summer, you know 2011-12-20T19:21:32 Zannick AC? 2011-12-20T19:21:35 fans are bad for that 2011-12-20T19:21:44 *** replore_ has joined #aichallenge 2011-12-20T19:21:45 AC fans. :P 2011-12-20T19:21:46 you better have a shower 2011-12-20T19:22:08 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-20T19:22:12 I blame the loss in my 7th game on pkmiec losing his hills so fast. ;) 2011-12-20T19:22:13 *** TL_ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T19:22:29 of course we have AC! We'd run out of batteries really fast otherwise. 2011-12-20T19:22:49 google backup power must be huge 2011-12-20T19:22:53 tons of batteries 2011-12-20T19:23:01 i have no comment on that 2011-12-20T19:23:27 I wonder how much does Youtube use bandwidth 2011-12-20T19:24:26 i wonder more on how much storage gmail uses and how do they reduce it 2011-12-20T19:24:26 pairofdice: it is estimated that during peek viewing netflix streaming accounts for 20% of all bandwidth 2011-12-20T19:24:48 thats nothing compared to spam XD 2011-12-20T19:24:50 netflix of course is much high res than most of youtube 2011-12-20T19:24:55 but i bet they cant answer that 2011-12-20T19:25:00 Ungodly amount 2011-12-20T19:25:11 image all that pron 2011-12-20T19:25:12 no, i can't answer anything about netflix either 2011-12-20T19:25:23 for different reasons, though 2011-12-20T19:25:29 Zannick its easier to ask, what can you answer :P 2011-12-20T19:25:39 * avdg doesn't know how his dad gets to pron while he can't even use the computer properly 2011-12-20T19:25:44 flowenol: i can't answer that. :P 2011-12-20T19:25:51 Zannick damn 2011-12-20T19:25:51 all he can is starting up the computer and change background 2011-12-20T19:25:53 chrisH_: that was my engine which rushed pkmeic, sorry 2011-12-20T19:25:53 avdg: he has your mother 2011-12-20T19:26:02 yeah, she was mad 2011-12-20T19:26:10 besh: which is you? 2011-12-20T19:26:12 and I also said yuck a half year ago 2011-12-20T19:26:12 so you cant answer anything 2011-12-20T19:26:24 (my mother only discovered one week ago) 2011-12-20T19:26:35 i am dshawul fighting to get top 100 2011-12-20T19:26:45 dad hides everything (looks like he is kinda scared of something dunno) 2011-12-20T19:26:47 besh: ahh, very cool 2011-12-20T19:27:15 besh: we had a lot of battles on TCP, you and I, I used Ombedor there 2011-12-20T19:27:28 meh dunno, my problems are just that worse I don't even care about a job 2011-12-20T19:27:28 My bots tactic is basically to rush who ever is closer to me. Works sometimes. 2011-12-20T19:27:35 *** kost_ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T19:27:42 besh: you cant beat me always 2011-12-20T19:27:43 ChrisH_: I always loose :) 2011-12-20T19:28:09 besh: well, that map was also relatively low food, which I usually have a harder time with, so that didn't help 2011-12-20T19:28:20 someone needs counceling here 2011-12-20T19:28:31 avdg: come on life is a joyride 2011-12-20T19:28:44 you have to believe in this crap 2011-12-20T19:28:49 Yes very big map with low food count 2011-12-20T19:28:55 look at me what happens if you dont 2011-12-20T19:28:58 *** pguillory has joined #aichallenge 2011-12-20T19:29:10 besh: I have confidence I'll make it up in my next game. 2011-12-20T19:29:21 mcstar: I already found something, sad that in this country, we have to wait for every permission 2011-12-20T19:29:43 ChrisH_: Ofcourse. You will forget your loss sooner than you think. I have 3 losses after that. 2011-12-20T19:29:47 what is that? 2011-12-20T19:29:55 are you joining the armed forces? 2011-12-20T19:30:01 half work, half help 2011-12-20T19:30:45 I'd really like to see an extra round after the finals, with the top players, last-man-standing, no turn-limits nor rank stabilization winning conditions 2011-12-20T19:30:55 *** kost_ has joined #aichallenge 2011-12-20T19:32:00 oh well, everybody outside home says I'm the one with problems (because the parents have enough power to decide what to happen with their child and not visa versa if the parents are kind of stupid) 2011-12-20T19:32:11 meh, I should stop being negative 2011-12-20T19:33:42 someone who watches dexter probably shouldnt give you advices, so i stfu 2011-12-20T19:33:57 :p 2011-12-20T19:34:13 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-20T19:35:11 *** Jak_o_Shadows has joined #aichallenge 2011-12-20T19:36:55 *** Redgis has quit IRC (Ping timeout: 252 seconds) 2011-12-20T19:37:04 *** matthewd has joined #aichallenge 2011-12-20T19:37:09 *** Darhuuk has quit IRC (Ping timeout: 240 seconds) 2011-12-20T19:37:09 *** timos has quit IRC (Ping timeout: 240 seconds) 2011-12-20T19:38:03 *** timos has joined #aichallenge 2011-12-20T19:38:05 *** Darhuuk has joined #aichallenge 2011-12-20T19:38:10 *** Jak_o_Shadows1 has quit IRC (Ping timeout: 252 seconds) 2011-12-20T19:38:35 xathis do a writeup on his bot yet? 2011-12-20T19:38:50 arscan: i'm working on it 2011-12-20T19:39:09 cool, looking forward to reading that 2011-12-20T19:39:09 *** TL_ has joined #aichallenge 2011-12-20T19:39:34 heh, I wonder how much techniques he added in his bot 2011-12-20T19:39:47 avdg: bfs, all the way 2011-12-20T19:40:27 xathis: do you have minmax in your battle ? 2011-12-20T19:40:43 *** Manwe56_ has quit IRC (Quit: Page closed) 2011-12-20T19:40:51 *** matthewd has quit IRC (Ping timeout: 245 seconds) 2011-12-20T19:41:40 besh: i only look one turn ahead, i would not call that minimax 2011-12-20T19:42:23 xathis: but you do try to evaluate all possible moves and then score? 2011-12-20T19:42:27 That is interesting. Looking at your 2 vs 1 combats I was convinced some of the behaviour must come from search. 2011-12-20T19:42:34 xathis: for considering the turn ahead, do you just consider the maximum number of enemies that could be "threatening" any given square? I was planning on trying that really simple approach had I time to work on my bot 2011-12-20T19:43:13 *** ikaros has joined #aichallenge 2011-12-20T19:43:21 i was not just looking at the number, i really simulated the moves, but of course only for small groups of ants 2011-12-20T19:43:21 http://aichallenge.org/visualizer.php?game=297515&user=2107 it's a good thing i have lots of ants in the hive. :) 2011-12-20T19:43:43 xathis, 7 ants in group? ) 2011-12-20T19:43:50 ChrisH_: i yes i evaluated the states but i did not do all moves 2011-12-20T19:44:41 not sure what that means, I guess I'll figure it out when you publish your writeup (working on mine in another window BTW) 2011-12-20T19:44:50 GreenTea: v1 had 7 hardcoded, now it's a bit different, i don'T look at all moves and thus can do a bit more but it depends on the way they are located 2011-12-20T19:44:58 what was your heuristic for selecting which moves to look at? 2011-12-20T19:46:49 Extrarius: not a heuristic, moving away from the enemy does not need to be considered in some cases 2011-12-20T19:47:06 xathis, it is smart! 2011-12-20T19:47:32 GreenTea: hehe thanks ;) 2011-12-20T19:47:48 if ants is more then enemy ants then no need to consider moves away from enemy ants 2011-12-20T19:48:14 GreenTea: oh no, i did not mean that 2011-12-20T19:48:56 its my guess) 2011-12-20T19:49:04 aichallenge: McLeopold epsilon * r04b482d / sql/2_generate_matchup.sql : switch seed to least played for finals (include scheduled games) - http://git.io/6dr9MQ 2011-12-20T19:49:08 ..a...b.. <- in this case you have to look what happens when you move a right, but you don't need to test what heppens when a moves left 2011-12-20T19:49:43 why? 2011-12-20T19:49:46 but thats tricky, generalizing this alg 2011-12-20T19:49:48 also if ..a...b.. in this example my ant a stays there i don't have to look at b at all 2011-12-20T19:50:14 ah, that 2011-12-20T19:50:17 sure 2011-12-20T19:50:45 McLeopold so how is it going 2011-12-20T19:50:47 that is a reasonable "pruning" stragegy i suppose 2011-12-20T19:50:48 how many remain 2011-12-20T19:51:12 i considered it but i thought it would be hard to implement 2011-12-20T19:51:15 *** Zed_ has quit IRC (Quit: Page closed) 2011-12-20T19:51:31 flowenol: 249 2011-12-20T19:51:38 and i was disappointed that i could only handle ~7-8 ants 2011-12-20T19:51:42 maybe 9 2011-12-20T19:51:45 i dont know 2011-12-20T19:51:50 i know i timed out at 10 2011-12-20T19:52:03 i could have been 9 2011-12-20T19:52:21 but at that time, i didnt have the battle code that i have now 2011-12-20T19:52:31 with this i cuold have handled large clusters 2011-12-20T19:52:40 and smaller ones with enumeration 2011-12-20T19:52:48 I wonder if it'd be worth it to precompute tables of possibilities for squares of AxB and use chess-compression techniques to store and search a huge number of tables 2011-12-20T19:53:03 i also have 2 distinct different battle calculations for <=7 ants and for >7 ants. have no time to make some hybryd 2011-12-20T19:53:03 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-20T19:53:15 Extrarius: even storing some things while enumerating had a performance hit 2011-12-20T19:53:24 i would say just compute everything on the fly 2011-12-20T19:53:42 Extrarius: I split all ants into combat groups, then memorized formations for both sides 2011-12-20T19:53:57 it helped with all the small battles 2011-12-20T19:53:59 mcstar: storing has a penalty, but there is always a time/space tradeoff 2011-12-20T19:54:10 it is too big if you consider attac_radius+2 zone which had like 70 something squares 2011-12-20T19:54:28 *** Surya has quit IRC (Quit: Page closed) 2011-12-20T19:54:28 70? 2011-12-20T19:54:47 off the top of my head. I am not sure. 2011-12-20T19:54:49 besh: but you don't have to consider all the squares, you cuold do something like generate tables for up to X enemy ants and Y of your ants 2011-12-20T19:54:51 You only need to consider a small fraction of those 2011-12-20T19:54:54 that seems too much 2011-12-20T19:54:57 ah ok 2011-12-20T19:55:07 yep ok 2011-12-20T19:55:19 i remember it was bigger than a chess board. 2011-12-20T19:55:19 has the same magnitude 2011-12-20T19:55:39 Ants at certain positions relative to your ants are an impossibilty due to combat 2011-12-20T19:55:40 may good idea will be to store battle patters for different enemies and try to recognize them in future.. to predict enemy moves 2011-12-20T19:55:49 lol for the next contest lets apply chess rules! 2011-12-20T19:55:50 it is bigger than a chess board, but the number of moves per piece is far fewer 2011-12-20T19:55:52 but this is very complicated, i think 2011-12-20T19:56:03 *** kost_ is now known as teapotahedron_ko 2011-12-20T19:56:05 *** hexahedron is now known as teapotahedron_ni 2011-12-20T19:56:27 mcstar: lets not do 'real' chess, but how about a game like ants but with different moves, call it 'bees' or something and you can move like a knight in chess 2011-12-20T19:56:33 and the engine spawns the ants with a specific role 2011-12-20T19:56:47 NOOO I go play skyrim and look what happens 2011-12-20T19:56:48 @rankings 2011-12-20T19:56:49 BenJackson: Top 10 players: teapotahedron(79.3), BenJackson(79.0), delineate(78.6), Belerafon(78.0), _Atsky_(78.0), a1k0n(77.1), lazarant(77.0), meduza(77.0), Kurnevsky(76.7), strcat(76.6) 2011-12-20T19:56:50 no more ants 2011-12-20T19:57:14 is it a concidence that teapotahedron just joined? 2011-12-20T19:57:28 I want more ants. I hope some good people release source so I can work on my bot and have something to play against =-/ 2011-12-20T19:57:41 McLeopold, how many 7 games bots left? 2011-12-20T19:57:50 192 2011-12-20T19:57:55 tnx 2011-12-20T19:58:22 i wonder if at the same time 8game bots are moving into 9 2011-12-20T19:58:33 cs_bot lost 2011-12-20T19:58:41 GreenTea is still undefeated... 2011-12-20T20:00:15 not for long I think 2011-12-20T20:00:21 when are you going to prune starter bots? 2011-12-20T20:00:44 janzert is taking the helm now 2011-12-20T20:00:47 ask him :P 2011-12-20T20:00:52 haha 2011-12-20T20:01:22 * avdg wonders if focussing on starter bots makes much more sense than focussing on the last players with time outs 2011-12-20T20:01:42 one of the timeout prone players is GreedyBot 2011-12-20T20:01:45 python startup issues? 2011-12-20T20:01:52 I'm definitely timeout-prone 2011-12-20T20:02:03 pguillory: you there? You said that you wanted to upload a version which predicted enemy moves before the finals but you did not resubmit 2011-12-20T20:02:09 GreedyBot has derp issues 2011-12-20T20:02:10 avdg consider my bot 2011-12-20T20:02:20 python on the server is way slower than it is locally 2011-12-20T20:02:21 Extrarius: You would need to consider all of the squares I think. f.e for a 5 man chess endgame you still cosider the 64 squares 2011-12-20T20:02:22 had 3 timeouts in the official competition 2011-12-20T20:02:28 *** GreenTea has quit IRC (Ping timeout: 258 seconds) 2011-12-20T20:02:31 flowenol: are you actually last ;-) 2011-12-20T20:02:43 none before that with this version 2011-12-20T20:02:47 last? 2011-12-20T20:02:53 in rank 2011-12-20T20:02:58 yeah 2011-12-20T20:03:03 600ish atm :D 2011-12-20T20:03:04 besh: in chess, some pieces can move across the board. Ants can only move 1 square at a time, so you don't need to consider nearly as much state space per piece 2011-12-20T20:03:26 im aiming for top200 2011-12-20T20:03:54 lets see if i reach it 2011-12-20T20:03:56 you would, though, need to consider squares at the periphery of your 'table' radius, but only for moves that possibly bring ants into the radius 2011-12-20T20:04:05 I won't start raising the cutoff until all the better than starter bot bots have worked themselves above the starter bots in the rankings 2011-12-20T20:04:10 Etrarius: Yes if you consider only one ply search 2011-12-20T20:04:32 janzert and how will you know that 2011-12-20T20:04:45 by watching :) 2011-12-20T20:04:54 thats a lot of work 2011-12-20T20:05:06 yes 2011-12-20T20:05:06 hmm, no stats? :/ 2011-12-20T20:05:12 Also the attack zone could have waters in it 2011-12-20T20:05:14 * avdg has sad face 2011-12-20T20:06:06 janzert: what about a cutoff of bottom 100 per hour? that would be 2400 bots in a day 2011-12-20T20:06:14 besh: of course, which means you'd need a nifty storage mechanism so you could do something like prune moves that would move into water (so you wouldn't need to have water in your tables, just remove invalid moves) 2011-12-20T20:06:30 TL_: why per hour? 2011-12-20T20:06:59 why absolute amount and not a percentage 2011-12-20T20:07:12 TL_: it'll move fast past the starter bots once they are sorted out, probably around 10 games for them 2011-12-20T20:07:13 avdg: just a random idea 2011-12-20T20:07:21 *** gustavokambara has joined #aichallenge 2011-12-20T20:07:29 janzert: Is there a rank that divides starter bots from others? I'm just wondering if my bot at 6965 is anywhere near the line =-P 2011-12-20T20:07:32 just saying :p 2011-12-20T20:08:14 Extrarius: if the ranks were settled it would be quite a bit below the line 2011-12-20T20:08:53 and being that it times out every game is probably going to stay below the line :/ 2011-12-20T20:09:00 I figured as much, being around the 12th percentile and all =-) 2011-12-20T20:09:22 Total Games: 298108 . Is this number correct ? 2011-12-20T20:09:25 for the next competition, there needs to be an option for pypy 2011-12-20T20:09:29 *** amstan has joined #aichallenge 2011-12-20T20:09:29 *** ChanServ sets mode: +o amstan 2011-12-20T20:09:36 pypy was an option 2011-12-20T20:09:41 well there actually is a bot below you that should be around 2000 2011-12-20T20:09:48 pypy was available 2011-12-20T20:09:52 besh: yes 2011-12-20T20:09:59 what did i have to do for pypy? 2011-12-20T20:10:05 .pypy 2011-12-20T20:10:08 *** meduza has quit IRC (Ping timeout: 258 seconds) 2011-12-20T20:10:12 call MyBot.py MyBot.pypy 2011-12-20T20:10:31 too bad I didn't know that, my bot might get an extra 50%-100% more moves before timing out =-/ 2011-12-20T20:10:38 *** meduza has joined #aichallenge 2011-12-20T20:10:58 https://github.com/aichallenge/aichallenge/blob/epsilon/worker/compiler.py#L528 I'm surprised no one took advantage of that 2011-12-20T20:11:09 janzert: Ok. I see it includes games before the finals 2011-12-20T20:11:12 mcstar: \o/ http://tcpants.com/replay.18191 2011-12-20T20:11:21 you could compile your binary locally and upload it along with an empty MyBot.ss 2011-12-20T20:11:40 McLeopold 7game done? 2011-12-20T20:11:52 I've logged off the server... 2011-12-20T20:12:03 time for dinner... 2011-12-20T20:12:06 McLeopold: sometimes up sometimes dow, in the next game the opposite happens 2011-12-20T20:12:10 thestinger: one person at least tried to but couldn't get the binary working with the server glibc 2011-12-20T20:12:16 janzert: yeah, bmh 2011-12-20T20:12:27 ahh, you know that :) 2011-12-20T20:12:35 yeah, I'm bad at open maps 2011-12-20T20:12:38 ghc 7.x needs a very recent glibc and drepper ruined static linking :P 2011-12-20T20:12:40 * avdg wonders if McLeopold's bot sucks more than his bot 2011-12-20T20:13:02 avdg: what is your bot name on tcpants? 2011-12-20T20:13:07 I think submitting binary is risky. Maybe interpreted java code.. 2011-12-20T20:13:09 *** amstan_ has quit IRC (Ping timeout: 240 seconds) 2011-12-20T20:13:13 I didn't submit one (yet) 2011-12-20T20:13:16 41 game 7 people left 2011-12-20T20:13:34 but I would definitely loose many games :p 2011-12-20T20:14:10 i hope i get games while im asleep 2011-12-20T20:14:21 and that begins right now 2011-12-20T20:14:26 bb 2011-12-20T20:14:28 mcstar: dream of ants so they would be generous. 2011-12-20T20:14:29 *** mcstar has quit IRC (Quit: WeeChat 0.3.6) 2011-12-20T20:15:00 *** teapotahedron_ni is now known as octahedron 2011-12-20T20:15:10 *** teapotahedron_ko is now known as kost_ 2011-12-20T20:15:33 aichallenge: McLeopold epsilon * r1d76a0a / sql/2_generate_matchup.sql : switch seed to least played for all time - http://git.io/oBdb0A 2011-12-20T20:15:55 @later tell mcstar Your dream has come true... santa is giving you two wins agains steve8877 :) 2011-12-20T20:15:55 besh: OK 2011-12-20T20:16:07 *** TheLinker has quit IRC (Remote host closed the connection) 2011-12-20T20:16:27 lol 2011-12-20T20:16:48 @rankings 2011-12-20T20:16:49 octahedron: Top 10 players: lazarant(79.4), teapotahedron(79.3), BenJackson(79.0), delineate(78.6), runevision(78.0), Belerafon(78.0), _Atsky_(78.0), Kurnevsky(77.4), a1k0n(77.2), JG.WAS(77.1) 2011-12-20T20:18:43 <_flag> Interesting, GreenTea and a1k0n just lost to someone named runevision 2011-12-20T20:19:03 im out 2011-12-20T20:19:08 hope my bot gets some games 2011-12-20T20:19:24 cya 2011-12-20T20:19:41 cya tomorrow 2011-12-20T20:20:31 *** Randex has joined #aichallenge 2011-12-20T20:22:14 anybody thinking about a game for the next contest? 2011-12-20T20:23:02 * avdg wants to know the concept of the next game, probably even start trying to fail-develop for the new contest 2011-12-20T20:23:41 avdg: there isn't a next game yet, so the thinking needs to be about picking one not solving it =-) 2011-12-20T20:24:07 *** flowenol has quit IRC (Ping timeout: 252 seconds) 2011-12-20T20:24:11 meh, I don't know if picking a game do any good for my brain 2011-12-20T20:25:15 I go with fluxid's soccer proposal for next year. 2011-12-20T20:25:39 besh: is that on the forums somewhere? 2011-12-20T20:25:55 *** octahedron has quit IRC (Quit: Page closed) 2011-12-20T20:26:00 *** sparkly_shows has joined #aichallenge 2011-12-20T20:26:18 http://forums.aichallenge.org/viewtopic.php?f=21&t=2184 2011-12-20T20:29:09 hrmm, I think I'm with 'Gajet' - it sounds like the state space isn't big enough 2011-12-20T20:31:02 anyone wants icpc challenge? with pushers and markers. "Coericone" 2011-12-20T20:31:58 coercion 2011-12-20T20:31:59 * avdg needs to refresh some basic trigonometric knowledge for that one 2011-12-20T20:32:15 ahh 2011-12-20T20:32:16 ) 2011-12-20T20:32:29 kost_: It looks interesting, but not as 'exciting' somehow as ants is IMO 2011-12-20T20:32:58 it would be more interesting in rl 2011-12-20T20:33:39 avdg: definitely, but robots are a whole 'nother universe from AI challenges like this 2011-12-20T20:33:54 true 2011-12-20T20:34:28 its sometimes hard to believe how wide it is spreaded 2011-12-20T20:35:50 Extrarius, may be) but i was excited when i saw the finalists of ICPC playing it 2011-12-20T20:36:59 xathis: that's right, i didn't resubmit. the situation was that i wouldn't get #1 with the improvements, i'd probably get top 10 either way, and didn't feel the reward of getting slightly *higher* in the top 10 was worth the risk of introducing a disastrous bug 2011-12-20T20:38:32 *** gustavokambara has quit IRC (Quit: Page closed) 2011-12-20T20:38:54 *** matthewd has joined #aichallenge 2011-12-20T20:40:18 i'm not delighted that some people have over twice as many games played as i do at this point 2011-12-20T20:41:02 how do you run a bot made in javascript? i tried with wscript MyBot.js but it dosen't work... 2011-12-20T20:41:42 nodejs 2011-12-20T20:42:23 thanks mate 2011-12-20T20:45:48 @rankings 2011-12-20T20:45:49 kost_: Top 10 players: xathis(79.7), lazarant(79.4), teapotahedron(79.3), BenJackson(79.0), delineate(78.6), Belerafon(78.0), _Atsky_(78.0), runevision(78.0), bix0r4ever(77.9), meduza(77.7) 2011-12-20T20:47:45 aichallenge: janzert epsilon * rfcf36ac / website/api_get_task.php : Don't generate a matchup in php but wait for one to become available - http://git.io/APjdrQ 2011-12-20T20:47:45 aichallenge: janzert epsilon * r3839d64 / sql/2_generate_matchup.sql : Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/THThhw 2011-12-20T20:48:15 *** woot_ has joined #aichallenge 2011-12-20T20:48:38 *** xathis has quit IRC () 2011-12-20T20:50:22 *** Scryer has joined #aichallenge 2011-12-20T20:50:40 *** woot_ has quit IRC (Client Quit) 2011-12-20T20:50:41 *** rebelxt has joined #aichallenge 2011-12-20T20:53:49 C:\Python27\python.exe "%~dp0playgame.py" -So --engine_seed 42 --player_seed 42 --end_wait=0.25 --verbose --log_dir game_logs --turns 1000 --map_file "%~dp0maps\maze\maze_04p_01.map" %* "python ""%~dp0sample_bots\python\HunterBot.py""" "python ""%~dp0sample_bots\python\LeftyBot.py""" "python ""%~dp0sample_bots\python\HunterBot.py""" "C:\Program Files (x86)\nodejs\node MyBot.js" | java -jar visualizer.jar 2011-12-20T20:54:01 i used nodejs for the defautl javascript bot 2011-12-20T20:54:10 what am i doing wrong? because the ants are just not moving 2011-12-20T20:55:36 just as a joke: using windows? 2011-12-20T20:55:49 @rankings 2011-12-20T20:55:50 BenJackson: Top 10 players: xathis(79.7), lazarant(79.4), teapotahedron(79.3), BenJackson(79.0), delineate(78.6), Belerafon(78.0), _Atsky_(78.0), runevision(78.0), bix0r4ever(77.9), meduza(77.7) 2011-12-20T20:55:55 the time for pretenders at #1 is over 2011-12-20T20:56:45 yeah 2011-12-20T20:57:05 i only have server linux boxes 2011-12-20T20:57:05 So who let xathis play another game? I thought we had a deal!! 2011-12-20T20:57:07 this is how it works 2011-12-20T20:57:36 don't have any desktop linux box so i want to make it work on windows 2011-12-20T20:57:46 you play, struggle for 1st place and than comes xathis 2011-12-20T20:58:01 ? 2011-12-20T20:59:37 *** pairofdice has quit IRC (Quit: When cryptography is outlawed, bayl bhgynjf jvyy unir cevinpl.) 2011-12-20T20:59:39 The next milestone is getting rid of 6,000 starter bots. 2011-12-20T21:00:37 ? 2011-12-20T21:01:22 any ideea why the bots don't even move when i test the command from a little up? the bot is the default one that tries to move any way it can 2011-12-20T21:01:28 37% of the last 200 games played on the server were "Won at turn 1". 2011-12-20T21:03:02 *** pguillory has quit IRC (Ping timeout: 258 seconds) 2011-12-20T21:03:45 wow 2011-12-20T21:04:46 Scryer: 6000 starter bots and Extrarius's bot. Mine is written from scratch but times out at turn 100 or so =-/ 2011-12-20T21:05:00 are we still waiting for everyone to get to 10+ games? 2011-12-20T21:05:32 Extrarius: I have no problem with people who write terrible bots from scratch and lose faster than a starter bot with them. No reason not to let them play until the cutoff. 2011-12-20T21:06:04 690 bots below 10 games 2011-12-20T21:06:08 Scryer: Yeah, but the only reasonable way to cutoff is by rank, score, or some other stat that'll be really low for loser bots 2011-12-20T21:06:21 McLeopold: tnx for update 2011-12-20T21:06:29 My bot can win when the hills are close enough though =-) 2011-12-20T21:06:35 McLeopold: what's your goal game count? 2011-12-20T21:06:45 a bazillion 2011-12-20T21:07:18 Extrarius: As I mentioned on the forum, you can do a cryptographic hash to see if someone has changed anything from the starter bots. If the SHA1 of an executable matches SHA1 of a starter bot, no reason to measure its performance accurately. 2011-12-20T21:07:50 I agree with the concept of pruning failbots gradually, but the starter bots can be taken out instantly. 2011-12-20T21:09:45 Scryer: I wish they'd have implemented a simple diff system, where it'd diff the source on upload and mark it as a "Start Bot" if it's less than 5% (for ex) different from the closest start bot 2011-12-20T21:11:04 I think *any* change should be allowed to compete and get pruned as normal, but opinions may differ. Even 5% source diffs would be enough to make a sample improvement, considering how much of the starter bot is simply bookkeeping. 2011-12-20T21:11:44 *** Storm_ has joined #aichallenge 2011-12-20T21:11:52 scryer: if they used a diff system, you could easily circumvent it by adding a large comment block or something if you wanted, but it'd prune out people that didn't care 2011-12-20T21:12:22 Momobot has 1st version ) 2011-12-20T21:12:24 is there any way to launch a visualizer in a linux terminal? 2011-12-20T21:12:32 I'm suggesting a hash of the executable, not a diff of the source. 2011-12-20T21:13:13 well, you would need to hash the content of all the files 2011-12-20T21:13:15 I would prefer a behavior test 2011-12-20T21:13:19 Randex: don't know) you can run without a visualizer and save replay to html 2011-12-20T21:13:45 hashes get messy and you can write some comments and have a "new" bot 2011-12-20T21:13:46 Scryer: Many of the languages don't produce executables 2011-12-20T21:14:15 for ants, it would be pretty easy to watch a bot ignore close food, go north and suicide 2011-12-20T21:14:22 McLeopold: The point isn't to filter out people that don't want filtering, just to let people play a tiny bit and not get counted if they don't want 2011-12-20T21:14:48 I'd also prefer an official tcp server as the first step, not submitting a "final" bot 2011-12-20T21:15:23 I agree that pruning on behavior is important and necessary. I'm just saying that eliminating sample bots will clear up a great deal of the dead wood. 2011-12-20T21:15:31 McLeopold: Definitely! Having a way to test immediately (and distribute the cpu requirements) would have been nice 2011-12-20T21:15:49 tcp client baked into playgame 2011-12-20T21:16:02 yeah 2011-12-20T21:18:09 *** JorgeB has quit IRC (Ping timeout: 240 seconds) 2011-12-20T21:19:59 *** mleise has quit IRC (Quit: Leaving.) 2011-12-20T21:20:49 *** visit2112 has joined #aichallenge 2011-12-20T21:23:26 *** pvarga has joined #aichallenge 2011-12-20T21:26:31 people are still playing on tcpants? 2011-12-20T21:26:44 I'm getting my butt kicked by a1kon right now 2011-12-20T21:26:53 a1k0n 2011-12-20T21:26:55 i'll probably die, but antsinyourpants will play a few games 2011-12-20T21:27:28 curl + perl: 340 of the last 1000 games were won at turn 1: 34% 2011-12-20T21:27:38 BAM! http://aichallenge.org/visualizer.php?game=298393&user=4513 2011-12-20T21:27:45 contest over 2011-12-20T21:28:16 *** Zedenstein has joined #aichallenge 2011-12-20T21:29:04 good set of opponents too 2011-12-20T21:29:14 3 of the last undefeated 2011-12-20T21:29:20 or, all of them 2011-12-20T21:29:25 now just 1 2011-12-20T21:30:46 heh 2011-12-20T21:30:49 xathis being xathis :) 2011-12-20T21:30:57 hmm, how do I test the java visualizer? 2011-12-20T21:33:06 *** Israfel has quit IRC (Read error: Connection reset by peer) 2011-12-20T21:33:29 *** Israfel has joined #aichallenge 2011-12-20T21:34:07 *** jmcarthur has quit IRC (Quit: WeeChat 0.3.4) 2011-12-20T21:34:20 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-20T21:34:36 I was wondering if the cutoffs that have been talked about are actually happening / going to happen in these finals 2011-12-20T21:35:12 * avdg wonders if the cut off is listed on http://aichallenge.org/server_stats.php 2011-12-20T21:35:47 *** pvarga has left #aichallenge 2011-12-20T21:37:03 *** jmcarthur has joined #aichallenge 2011-12-20T21:37:51 hmm on that page I see that some servers have a pretty high "fail" percentage. Like 25% for server #661. What are these fails and what causes them? 2011-12-20T21:38:11 timeouts and crashes 2011-12-20T21:38:16 bad bot code 2011-12-20T21:38:28 lol I just looked up this address I have an interview at on thursday 2011-12-20T21:38:31 it's an apartment 2011-12-20T21:38:38 I know it's a startup, but cmon 2011-12-20T21:38:51 rwest_: maybe it's an office building 2011-12-20T21:39:06 amstan: street view confirms apartment 2011-12-20T21:39:08 2 family house 2011-12-20T21:39:25 they could have a nice office in there, you'll never know 2011-12-20T21:39:45 Can you see the oubliette and the rotating knives from street view? 2011-12-20T21:39:48 amstan: we'll see, I just find it funny that it's on a residential street and in an apartment 2011-12-20T21:39:59 in romania there's a ton of dental places that are inside appartments 2011-12-20T21:40:04 they looked quite classy 2011-12-20T21:40:07 McLeopold: oh, so a bot crash also counts as a fail? In that case, it's fine :) 2011-12-20T21:40:43 parking in boston sucks anywhere bu residential areas, so that's a plus 2011-12-20T21:42:03 *** sparkly_shows has quit IRC (Quit: Page closed) 2011-12-20T21:44:42 *** visit2112 has quit IRC (Quit: Page closed) 2011-12-20T21:44:47 meh, just done a dirty fix, I should probably be fired as programmer 2011-12-20T21:45:12 everybody bugs 2011-12-20T21:45:48 and paratrechina is much better than me 2011-12-20T21:50:43 aichallenge: McLeopold epsilon * r20fe8a5 / website/sql.php : partial fix to next game calc - http://git.io/dZrUQw 2011-12-20T21:50:43 aichallenge: McLeopold epsilon * r2ea69c6 / website/api_get_task.php : Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://git.io/suarLQ 2011-12-20T21:51:31 wow, I was scared that my fix was getting commited 2011-12-20T21:52:24 *** tomb___ has quit IRC (Quit: Page closed) 2011-12-20T21:53:28 doh, players ahead counter just jumped from 1K to 6K :-( 2011-12-20T21:53:43 rebelxt: you have too many games :) 2011-12-20T21:54:28 McLeopold: i'd like 1000 more games please ;) 2011-12-20T21:54:36 it's actually longer than that 2011-12-20T21:54:55 cause everyone with 7 games is in line 7 times before you 2011-12-20T21:55:55 ouch. no cutoffs before everyone gets to 14? i don't mind at all, just want to play more ;-) wish i could donate some EC2 servers 2011-12-20T21:56:22 hey, I survived a1k0n \o/ 2011-12-20T21:56:44 :D 2011-12-20T21:57:06 hmm, I backed up in queue 2011-12-20T21:58:10 McLeopold: gratz 2011-12-20T21:58:14 *** dr0b3rts has quit IRC (Ping timeout: 258 seconds) 2011-12-20T21:58:32 the cutoff kicked in before he wiped out my last 2 ants 2011-12-20T21:58:35 meh, now I'm loosing more ranks 2011-12-20T21:58:41 ouch. I have 17 games atm, this probably means I won't get any more games for a long, long time... 2011-12-20T21:58:46 but that's fair 2011-12-20T21:58:47 oh well 2011-12-20T21:59:53 McLeopold: a few days ago i was running a local tourney between my 16 bots on EC2's biggest cluster instance. the one with 60 GB of RAM and 16 cores. games were flying.. was fun to watch 2011-12-20T22:00:31 lucky man) 2011-12-20T22:00:48 we have laptops 2011-12-20T22:02:03 * avdg wonders when we can use gpu's 2011-12-20T22:02:23 kost_: so do i. i just had 16 minor revisions with different scoring function weights.. 6 hours was enough time to get rankings to stabilize after ~1K games 2011-12-20T22:04:06 *** Ashoka has joined #aichallenge 2011-12-20T22:04:15 rebe1xt: nice) i ran just ~100 games on my laptop... and didn't gather stats to adjust parameters 2011-12-20T22:07:36 *** Storm_ has quit IRC (Quit: Page closed) 2011-12-20T22:11:17 rebelxt: did one of those actually come out as a clear winner? 2011-12-20T22:12:15 I would rather expect that there would be some who are obviously bad, and all others would be about equally good. 2011-12-20T22:15:08 Migi32: yep. there were 4 revisions that were way ahead of the rest, and some were really bad. i had hard time choosing between them so just did picked one that "felt" right 2011-12-20T22:20:36 if the next contest will also involve so many heuristics and scoring function weights as this one, I should really look into some ways of doing automated parameter tuning 2011-12-20T22:20:50 evolutionary algorithms, that kind of thing 2011-12-20T22:21:09 I wonder how much bias there is by only playing against yourself, though 2011-12-20T22:21:25 * avdg wants to ask why migi32 is awake as well 2011-12-20T22:21:49 heh. Why are you still awake? :) 2011-12-20T22:22:05 because I need my rest ;-) 2011-12-20T22:22:10 its kinda silly 2011-12-20T22:22:33 *** ChrisH_ has joined #aichallenge 2011-12-20T22:22:50 my brain overloads at day 2011-12-20T22:23:06 Migi32: too much noise from evolutonary algorithms makes them impractical 2011-12-20T22:24:33 that is my opinon from a little experience i had tuning another bot. 2011-12-20T22:24:47 avdg: I kept finding small bugs and possible improvements in my bot and i kept working until like 2 hours before the deadline. Then for 2 hours I tested if the new bot was actually any better than my old bot. It wasn't. Anyway, by that point it was 6 am and now my sleep schedule is completely messed up 2011-12-20T22:25:01 but school is over anyway so who cares :p 2011-12-20T22:25:10 heh 2011-12-20T22:25:22 I don't know if I ever make deadlines again 2011-12-20T22:25:34 the more home work I got, the more I skipped 2011-12-20T22:26:09 sad most people thought that I had problems, they only see the consequence of a problem 2011-12-20T22:26:43 besh: there are other things out there than evolutionary algorithms to do parameter tuning, right? 2011-12-20T22:27:14 besh: it was just an example. I really know very little about it 2011-12-20T22:28:59 avdg: yeah I know what you mean. My brain also does it very best to submit on the last second of the deadline every time... 2011-12-20T22:29:18 I have literally done that once, submitted on 0:00:00 :p 2011-12-20T22:29:30 it was rejected :/ 2011-12-20T22:29:34 :D 2011-12-20T22:29:51 Migi32: Sure. But my expeince with ga algorithms is bad in general I don't recommend it to anyone. Usually there are far simpler alternatives. 2011-12-20T22:31:37 will the cutoff kick in at 10 games ? 2011-12-20T22:32:53 *** dom7b5 has joined #aichallenge 2011-12-20T22:33:14 <_flag> @rankings 2011-12-20T22:33:15 _flag: Top 10 players: FlagCapper(81.7), lazarant(80.7), delineate(79.8), xathis(79.7), teapotahedron(79.3), BenJackson(79.0), runevision(78.8), meduza(78.2), Belerafon(78.0), _Atsky_(78.0) 2011-12-20T22:33:26 GreedyBot is at 2000 right now so it will save a lot 2011-12-20T22:33:31 what happened? 2011-12-20T22:33:35 congrats flag 2011-12-20T22:33:40 <_flag> Not for long though 2011-12-20T22:33:46 <_flag> xathis just hasn't gotten the games 2011-12-20T22:33:58 yeah, he still didn't lost 2011-12-20T22:34:00 <_flag> But I'll just enjoy the moment 2011-12-20T22:34:21 take a pic or it didn't happen ;-) 2011-12-20T22:34:25 take a screenshot 2011-12-20T22:34:32 <_flag> Contestbot says it happened :) 2011-12-20T22:34:33 _flag: I have no idea what you mean. 2011-12-20T22:34:40 <_flag> Shh... don't lie 2011-12-20T22:34:46 <_flag> You know you saw it 2011-12-20T22:34:47 :D 2011-12-20T22:35:24 _flag: We're finally getting some games, and you're doing well, nice job. 2011-12-20T22:35:37 _flag: I've had a couple of not great games, but finally in top 100 2011-12-20T22:36:06 <_flag> ChrisH_: Thanks, hopefully I can secure second. I've tested against GreenTea locally and I think I can beat him, but we'll have to see 2011-12-20T22:36:25 <_flag> ChrisH_: Hope you move up too :) 2011-12-20T22:36:56 *** xar0l_ has quit IRC (Quit: Page closed) 2011-12-20T22:37:25 _flag: I think I will, I've had a couple of maps with poor access to all the neighbors and imbalanced other bots hurt me chances to score, but that happens 2011-12-20T22:37:25 _flag: nice, greantea is a great opponent to have for local testing 2011-12-20T22:38:04 GreenTea released his code, so I guess anyone could do it now. 2011-12-20T22:38:08 <_flag> rebelxt: Well he released his source, so it's not like I got to test against him during the contest 2011-12-20T22:38:19 <_flag> I'm doing it after the fact just out of curiousity 2011-12-20T22:38:37 <_flag> ChrisH_: Ants is like that, games are very variable 2011-12-20T22:38:43 yep 2011-12-20T22:39:15 _flag: ahh, got it. myself and a few co-workers had a private server running for weeks. it was big help 2011-12-20T22:39:16 meh pguillory has a pretty big code base 2011-12-20T22:39:23 https://github.com/pguillory/aichallenge2011-c 2011-12-20T22:39:57 rebelxt: hey, good to see you! 2011-12-20T22:40:07 luckily that some of them are just tests 2011-12-20T22:40:11 that must be his old code, because he switched to golang midway through the competition 2011-12-20T22:40:15 dmj111: hey man, u2! 2011-12-20T22:41:00 Game 298777: I took the whole brunt of the attack from marcello834 (on two fronts) and then MypaBeu comes and snitches the unguarded hole ... that stinks 2011-12-20T22:41:02 you are up to your old EC2 tricks, I see! I need ot learn to do that 2011-12-20T22:41:46 <_flag> Woah! xathis lost a game 2011-12-20T22:42:05 <_flag> And I didn't even die in it 2011-12-20T22:42:08 dmj111: haha, well, i had to time to write smart bot this time, so decided to optimize using brute force. it did not work out that well ;) 2011-12-20T22:42:09 <_flag> http://aichallenge.org/visualizer.php?game=299185&user=31 2011-12-20T22:42:25 <_flag> Although that game dropped me to second, it was worth it ;) 2011-12-20T22:43:09 dmj111: next contest i'm just going to bake and share my AMI so everyone can use EC2 for testing 2011-12-20T22:43:43 _flag: now you got a big chance! 2011-12-20T22:43:49 _flag: nice. 2011-12-20T22:43:56 rebelxt: that would be aweseom 2011-12-20T22:44:18 MacLeopold: Your undeafeated list is now null :) 2011-12-20T22:44:34 rebelxt: since it is the holidays around here... I immediately thought of christmas cookies when I saw "bake" 2011-12-20T22:45:57 http://pastebin.com/T76ehrFv 2011-12-20T22:46:11 surprizingly meduza got only 1 points for that performace 2011-12-20T22:46:28 wow, so now nobody won all games 2011-12-20T22:46:39 dmj111: haha. still waiting to get paired with ur bot. probably not gonna end well for me 2011-12-20T22:47:04 <_flag> McLeopold: Can you add mus to that list? 2011-12-20T22:47:06 289 bots at 9 games 2011-12-20T22:48:02 McLeopold: I'm actually a bit surprised how quickly it is bringing the lower game bots up. 2011-12-20T22:48:23 McLeopold: but that's good 2011-12-20T22:48:47 @rankings 2011-12-20T22:48:49 avdg: Top 10 players: meduza(81.4), FlagCapper(81.4), lazarant(80.7), BenJackson(80.0), delineate(80.0), runevision(79.8), _Atsky_(79.6), teapotahedron(79.6), Speedy_Consoles(78.9), Belerafon(78.0) 2011-12-20T22:48:52 http://pastebin.com/4AkCKwsk 2011-12-20T22:48:57 the top bots are being paired with each other. 2011-12-20T22:49:26 <_flag> McLeopold: Thanks 2011-12-20T22:49:43 Call it a contest and make that the final rankings! 2011-12-20T22:49:48 kidding everyone 2011-12-20T22:49:56 *** rlanday has joined #aichallenge 2011-12-20T22:49:58 <_flag> dmj111: Nah, let me play one more game against a few starter packs 2011-12-20T22:50:07 <_flag> Then we can call it a finals :) 2011-12-20T22:50:19 _flag: that could increase your sigma :) 2011-12-20T22:50:43 you mean decrease it ? 2011-12-20T22:51:08 new predictions http://pastebin.com/YQWu8nTi 2011-12-20T22:51:11 _flag: ok.... I guess :) Good luck flag! 2011-12-20T22:51:21 <_flag> dmj111: Thanks 2011-12-20T22:52:58 janzert: how are we looking for doing a cutoff? 2011-12-20T22:53:13 McLeopold: how about top 100 rankings based on mu only? 2011-12-20T22:53:32 I'm probably a bad story telling on the forums :p 2011-12-20T22:54:22 http://pastebin.com/AqpdDaad 2011-12-20T22:54:38 Interesting... 2011-12-20T22:55:51 McLeopold: If it's not too much work, I'd am curious to see the games played histogram again. 2011-12-20T22:56:42 http://pastebin.com/QmYQG2QR 2011-12-20T22:57:05 thanks, that looks a lot better than before the change 2011-12-20T22:57:11 yeah 2011-12-20T22:58:18 I think we are churning through good bots, because we are down to 6.2 games per minute 2011-12-20T22:58:18 you've made the change "permanent" i see from github, so i'm going to update my forum post 2011-12-20T22:58:20 gn 2011-12-20T22:58:26 ChrisH_: yes 2011-12-20T22:58:47 *** avdg has quit IRC (Quit: Leaving.) 2011-12-20T23:00:37 *** Antimony has joined #aichallenge 2011-12-20T23:02:39 *** dvladim has joined #aichallenge 2011-12-20T23:03:48 McLeopold: I also see that the players ahead counter has been updated to at least partially reflect the new queue order. How often does it get recomputed? 2011-12-20T23:04:20 ChrisH_: should be every page hit 2011-12-20T23:04:36 the error now is that if you have 15 games and someone has 10, they are in line 5 times before you 2011-12-20T23:05:08 McLeopold: ok, it is curious that my counter doesn't go down when other games are played. 2011-12-20T23:05:09 so, technically, it should say (there are 30,000 bots ahead of you) 2011-12-20T23:05:37 ChrisH_: I also didn't do any ordering, so every bot at 15 will have the same "ahead" number 2011-12-20T23:05:41 McLeopold: I have 10 games, so I would expect someone with 9 to play and slot in behind me based on least recent play tiebreak 2011-12-20T23:05:51 McLeopold: ahhh, ok 2011-12-20T23:06:02 you won't see it go down until someone goes above you 2011-12-20T23:06:13 hmm... 2011-12-20T23:06:23 I did less than or equal 2011-12-20T23:06:35 so someone else with 10 appears ahead of you 2011-12-20T23:06:48 which makes it a worst case counter 2011-12-20T23:06:56 the whole thing is nearly impossible to predict anyway 2011-12-20T23:07:11 yeah, I figured getting a game earlier is nicer 2011-12-20T23:07:16 absolutely 2011-12-20T23:07:20 it's all perceptions anyway 2011-12-20T23:07:24 underpromise and over deliver always 2011-12-20T23:07:44 *** ltriant has quit IRC (Quit: Get MacIrssi - http://www.sysctl.co.uk/projects/macirssi/) 2011-12-20T23:07:51 90 9 game bots left 2011-12-20T23:08:17 my counter just went down by 3, so some of the 10 game bots must be getting pulled in 2011-12-20T23:08:22 which is good 2011-12-20T23:08:26 highly likely 2011-12-20T23:08:30 ChirsH, have you written up your bot design yet? 2011-12-20T23:08:36 working on it 2011-12-20T23:08:40 oh ok 2011-12-20T23:08:49 I'm curious how your spawn control works 2011-12-20T23:09:09 *** kost_ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T23:09:17 I'm going to post a high level overview first and then add edits with more details and answer questions in replies 2011-12-20T23:09:20 *** Ashoka has quit IRC (Quit: Page closed) 2011-12-20T23:09:41 ChrisH_: we also almost ran out of disk space, then janzert shutoff the workers and fixed it, if anyone asks 2011-12-20T23:09:43 Antimony: I plan to cover that in more detail since not many bots do it. 2011-12-20T23:09:49 McLeopold: ok 2011-12-20T23:09:53 so, the game rate plummetted earlier 2011-12-20T23:10:11 McLeopold: i missed that so it was probably while I was watching a movie with the Mrs. 2011-12-20T23:10:19 http://forums.aichallenge.org/viewtopic.php?f=27&t=2187 2011-12-20T23:10:26 that title is accually kinda funny 2011-12-20T23:10:50 ok, I'll reply that just to clarify 2011-12-20T23:12:16 *** bergmark has quit IRC (Ping timeout: 240 seconds) 2011-12-20T23:13:57 pguillory must be one of the last 9 game bots left 2011-12-20T23:14:56 42 left 2011-12-20T23:15:37 I'm always curious about the details. 2011-12-20T23:15:40 http://aichallenge.org/visualizer.php?game=299185&user=3473 <- I'd say several bots including mine didn't play that very well 2011-12-20T23:15:42 McLeopold: I'm curious why only the profile owner gets to see the player's ahead stats on their profile page 2011-12-20T23:15:59 ChrisH_: caching? 2011-12-20T23:16:00 ChrisH_: to prevent griping 2011-12-20T23:16:52 McLeopold: reasonable precaution. There's plenty of griping already, I wonder how much worse it would be. ;) 2011-12-20T23:16:56 Anyone can blog "I explore by looking for places I haven't seen", but it isn't very informative. 2011-12-20T23:17:16 Did you do bfs? From what point? How do you resolve ties? 2011-12-20T23:17:33 Do you do from each ant or all at once? 2011-12-20T23:18:07 McLeopold: Yeah, I'm going to try to get into the weeds of my bot in my writeup, but I don't want to bore people not interested in the details, so I'll summarize first, and people can read as far down as they want. 2011-12-20T23:18:13 *** dvladim has quit IRC (Ping timeout: 255 seconds) 2011-12-20T23:18:31 mega1 left me dazed with last years explaination 2011-12-20T23:18:38 McLeopold: and i didn't want to release a novel in two weeks, so I'm going to serialize it a bit at a time as I write it up. 2011-12-20T23:18:50 I can't read lisp very well, and I thought the explaination was lacking 2011-12-20T23:18:55 BenJackson: flagcapper happened 2011-12-20T23:19:02 *** ltriant has joined #aichallenge 2011-12-20T23:19:24 ouch 2011-12-20T23:20:11 that game is a giant upset 2011-12-20T23:20:28 ChrisH_: I'd prefer a novel slowly released over time 2011-12-20T23:20:44 yeah, that's what I'm going to do 2011-12-20T23:20:54 I understand how hard it is to explain, the tutorials were a pain to write 2011-12-20T23:20:57 as long as I have the energy and there is interest 2011-12-20T23:21:53 ChrisH_: thanks for diving on the PR grenade 2011-12-20T23:22:04 <_flag> McLeopold, amstan: Just an FYI, I'm not sure if there's anything you can do about this, but the java version of the visualizer overlay has a tendency to crash on me (I assume this is mleise's responsibility?) 2011-12-20T23:22:06 What I haven't decided is how much effort to put into making illustrations. 2011-12-20T23:22:17 ChrisH_: step 1, make a tool, step 2, we'll all use it 2011-12-20T23:22:17 _flag: yes 2011-12-20T23:22:20 <_flag> This is after heavily using the visualizing features 2011-12-20T23:22:24 _flag: but it worked pretty well for me 2011-12-20T23:22:33 *** carpcrap has joined #aichallenge 2011-12-20T23:22:40 <_flag> I'm basically changing the colour of almost every tile to show how my bot sees the game 2011-12-20T23:22:51 _flag: the added features could be an explanation for the crashes 2011-12-20T23:22:52 McLeopold: so how many bots left at what low game number are left? 2011-12-20T23:23:09 BenJackson: someone needed to do some PR, but first we needed to find out what the facts were. Once that was established it was easy to type the messages. 2011-12-20T23:23:36 <_flag> amstan: Yes, I'm pretty sure the visualizer overlay features are at fault because it works fine without them 2011-12-20T23:23:51 _flag: yell at the creator then 2011-12-20T23:24:04 <_flag> @seen mleise 2011-12-20T23:24:04 _flag: mleise was last seen in #aichallenge 9 hours, 13 minutes, and 24 seconds ago: but Migi32, that was your idea for this contest already ;) 2011-12-20T23:24:10 I guess I could start downloading bots and testing my voting theory 2011-12-20T23:24:15 *** u_ has quit IRC (Quit: u_) 2011-12-20T23:24:15 <_flag> Oh, I didn't know he was still active 2011-12-20T23:24:18 _flag: not him, the other guy, with the debugging ones 2011-12-20T23:24:27 *** keith__ has joined #aichallenge 2011-12-20T23:24:28 <_flag> Who's the other guy? 2011-12-20T23:24:31 isn't that what you meant? 2011-12-20T23:24:37 they crash when they debugging features are enabled 2011-12-20T23:25:02 <_flag> It crashes when this: https://github.com/j-h-a/aichallenge/blob/vis_overlay/VIS_OVERLAY.md is enabled 2011-12-20T23:25:12 yes 2011-12-20T23:25:13 <_flag> But I don't know who is responsible for that, which is why I'm asking 2011-12-20T23:25:15 so report it to j-h-a 2011-12-20T23:25:20 it's his repo 2011-12-20T23:25:34 BenJackson: at last report there ~40 bots with 9 games left, pguillory is one and still has nine, so they are not done yet. 2011-12-20T23:25:48 <_flag> Oh, I didn't even realize it wasn't part of aichallenge 2011-12-20T23:25:51 <_flag> Nevermind, ignore me then 2011-12-20T23:26:23 BenJackson: and I have 10 games and my profile says 703 players in front, which means 703 (or 704) 10 or 9 game bots to go 2011-12-20T23:26:49 _flag: his name is jab_bott on irc 2011-12-20T23:26:53 not sure if he's still around 2011-12-20T23:26:56 @seen jab_bott 2011-12-20T23:26:56 thestinger: jab_bott was last seen in #aichallenge 3 days, 11 hours, 8 minutes, and 40 seconds ago: heh 2011-12-20T23:26:59 _flag: maybe this: https://github.com/aichallenge/aichallenge/pull/390#issuecomment-3228770 2011-12-20T23:27:35 <_flag> amstan: Mine crashes before starting and not on an event 2011-12-20T23:27:42 BenJackson: at the rate my game counter is dropping, I'd say we are into the 10 game bots now, but not all the 9 game bots's games are finished playing yet 2011-12-20T23:27:46 ChrisH_: this is right up your alley other than the fact that you can't do anything about it: http://forums.aichallenge.org/viewtopic.php?f=21&t=2201&sid=85d34a5beccbfd55a1d95fa7cbe2a5fd 2011-12-20T23:28:06 *** u_ has joined #aichallenge 2011-12-20T23:28:19 yeah, i read that, very nice writeup, but I chose not to respond for the reason you stated 2011-12-20T23:28:43 ChrisH_: care to bias me with comments before I watch your last few games? 2011-12-20T23:29:16 I didn't do well for two reasons, bad positioning and low food. 2011-12-20T23:29:24 but that happens 2011-12-20T23:30:17 in ...029 I'd almost say that your multi-hill put the strength on the wrong hill 2011-12-20T23:30:23 not that it could know 2011-12-20T23:30:40 pguillory just lost again 2011-12-20T23:31:08 BenJackson: yeah, it looks great when it pulls those out, and not so great when it doesn't 2011-12-20T23:31:15 McLeopold: this guy is right: http://forums.aichallenge.org/viewtopic.php?f=21&t=2201&sid=85d34a5beccbfd55a1d95fa7cbe2a5fd 2011-12-20T23:31:21 McLeopold: note the mus for the top 20 people 2011-12-20T23:31:23 *** keith__ has quit IRC (Ping timeout: 258 seconds) 2011-12-20T23:31:37 McLeopold: if you were to sort by mu, you would have xathis, a1k0n, flagcapper and BenJackson near the top 2011-12-20T23:31:41 BenJackson: and it couldn't pull it out because of relatively low food and pkmiec rolled over too easy for the others 2011-12-20T23:31:41 not with other bots in between 2011-12-20T23:32:18 I'm already in 4th aren't I? 2011-12-20T23:32:20 @rankings 2011-12-20T23:32:22 BenJackson: Top 10 players: meduza(81.4), FlagCapper(81.4), lazarant(80.7), BenJackson(80.0), delineate(80.0), runevision(79.8), _Atsky_(79.6), teapotahedron(79.6), Speedy_Consoles(78.9), Belerafon(78.0) 2011-12-20T23:32:42 amstan: I agree that he makes an excellent argument in his post. 2011-12-20T23:32:42 oh I didn't realize xathis got bumped 2011-12-20T23:32:45 I didn't expect that 2011-12-20T23:32:53 BenJackson: yes, but your mu is second or something 2011-12-20T23:33:13 amstan: I also agree with him but did not make that point mainly because I expected the tournament to run until sigmas were essentially equal 2011-12-20T23:33:54 is that possible though? 2011-12-20T23:34:35 if game selection optimized for it 2011-12-20T23:34:38 *** Antimony_ has joined #aichallenge 2011-12-20T23:34:41 antimatroid's not doing too bad either 2011-12-20T23:34:53 it seemed that sigmas were converging to about 1.3 on the official server 2011-12-20T23:35:00 on tcp servers they would stay high due to unevenness in the matchups 2011-12-20T23:35:17 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-20T23:35:20 *** Antimony_ is now known as Antimony 2011-12-20T23:36:00 on the scoreboard on the website, a quick scan shows sigmas from 2 to almost 5 2011-12-20T23:36:14 http://aichallenge.org/visualizer.php?game=299033&user=10 <- I saw you sit on your hill and immediately stopped to figure out where the multi-hills were 2011-12-20T23:36:18 yes, given a roughly equal number of games and a large enough sample, the sigmas should become much smaller and have less impact on the skill, but even before finals I always looked at the mu to see how I was doing. 2011-12-20T23:36:24 your bots look awfully "thin" in that game 2011-12-20T23:37:36 BenJackson: That map seemd to have low food too, if that's what you mean. 2011-12-20T23:38:12 k 2011-12-20T23:38:14 g2g 2011-12-20T23:41:46 I'm .01 ahead of delineate and he has 22 (!) games, wow 2011-12-20T23:42:29 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-20T23:44:50 i'm back 2011-12-20T23:44:51 *** gaussin has joined #aichallenge 2011-12-20T23:44:53 how's everyone going? 2011-12-20T23:45:17 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-20T23:45:37 *** Antimony has joined #aichallenge 2011-12-20T23:45:59 *** andrei_c has joined #aichallenge 2011-12-20T23:46:10 antimatroid: waiting for games ;-) 2011-12-20T23:46:21 hi all 2011-12-20T23:46:38 can someone explain one moment in the game? I'm really confused 2011-12-20T23:46:59 andrei_c: ask your question, someone here can probably help 2011-12-20T23:47:37 http://aichallenge.org/visualizer.php?game=298652&user=11532 turn 460 coord 106,13 2011-12-20T23:48:10 why the hell my ants (white) could not capture the hill with only one protector??? 2011-12-20T23:48:18 is it a bug? 2011-12-20T23:48:24 no 2011-12-20T23:48:41 *** u_ has quit IRC (Quit: u_) 2011-12-20T23:48:49 the blue player still has 25 ants in it's hive trying to come out 2011-12-20T23:49:02 *** rebelxt has quit IRC (Quit: Page closed) 2011-12-20T23:49:27 the order of actions in the turn is 1. movement, 2. combat, 3. hill razing, 4. ant spawning, 5. food spawning 2011-12-20T23:49:47 *** meduza has quit IRC (Ping timeout: 258 seconds) 2011-12-20T23:49:50 so you kill the one defender in step 2, but before you get to move again a new ant appears on the hill 2011-12-20T23:49:55 Unless he moves off the hill you need to kill each defender. 2011-12-20T23:49:58 in step 4 2011-12-20T23:50:18 but why my ants die? they are in overwhelming majority 2011-12-20T23:50:36 and playing the game forward a bit it looks like blue is eating food so fast you can't kill them all 2011-12-20T23:50:58 if your ant steps on the hill while an enemy ant is there they both die in step 2 no matter how many friends you have nearby 2011-12-20T23:51:09 ant on ant collision is always death during combat 2011-12-20T23:51:13 andrei_c: ants aren't allwed to collide 2011-12-20T23:51:28 well, they are allowed to, but it always kills them 2011-12-20T23:51:42 *** rlanday has quit IRC (Ping timeout: 258 seconds) 2011-12-20T23:51:46 even if you step on your own ant, they both die with no enemy around 2011-12-20T23:51:48 ok. thanks for explanation! I was totally unaware of this scenario. This had to be programmed specially :( 2011-12-20T23:52:04 no problem 2011-12-20T23:52:09 the other thing you have to do specially is NOT STEP OFF your hill when it is surrounded 2011-12-20T23:52:10 sorry it didn't work out for you there 2011-12-20T23:52:15 if you want to take advantage of the hive 2011-12-20T23:52:17 I think I forgot to do that :) 2011-12-20T23:53:10 i didn't do that explicitly but my combat ants will suicide an enemy to stop it from stepping on the hill if needed, so that logic probably deals with it implicitly 2011-12-20T23:53:15 I actually came up with a specific strategy there but never implemented it 2011-12-20T23:53:49 to whit: if there are 2 enemies threatening your hill but none standing right next to it, step toward the nearer enemy and hope to suicide him 2011-12-20T23:53:53 rather than just dying in combat 2011-12-20T23:53:58 ohh, i'm playing in a game now. :) 2011-12-20T23:54:06 Looks like the failbot cutoff hasn't happened yet: 322 of the last 1000 games were won at turn 1 2011-12-20T23:54:20 It was 34% a couple of hours ago also. 2011-12-20T23:54:26 Scryer: at least those are fast 2011-12-20T23:54:33 *** andrei_c has quit IRC (Quit: Page closed) 2011-12-20T23:55:01 BenJackson: true enough. 2011-12-20T23:55:15 it probably takes longer to update the DB than play the game 2011-12-20T23:56:04 Does somebody still have that piratething top 10 list from a couple nights ago? 2011-12-20T23:56:35 I'd like to know who's missing. 2011-12-20T23:57:35 http://aichallenge.org/visualizer.php?game=299185&user=4513 2011-12-20T23:57:41 xathis had his first loss 2011-12-20T23:57:54 ChrisH_: you should have brokered a treaty between pguillory and nhaehnle 2011-12-20T23:58:00 seriously 2011-12-20T23:58:34 did you have any special code for a wall of ants 3 wide and 13 deep vs 2 defenders? 2011-12-20T23:58:40 i almost got to joholm's hill at the end, but too little to late. 2011-12-20T23:59:11 BenJackson: no 2011-12-20T23:59:16 I was watching your fight with hellow so when I glanced over and saw panela holding off a flood of purple I was pretty surprised 2011-12-20T23:59:46 I would have expected pguillory not to get so bunchy