2010-09-22T00:01:28 Has anyone found the problem with PlayGame-1.2.jar, that causes game clients not work as intended yet? 2010-09-22T00:01:46 (Where the bot times out, despite providing ostensibly correct outputs/not crashing on inputs.) 2010-09-22T00:01:47 *** jmcarthur has joined #aichallenge 2010-09-22T00:07:50 *** hornairs has quit IRC (Remote host closed the connection) 2010-09-22T00:08:02 *** FlagCapper has joined #aichallenge 2010-09-22T00:23:10 *** icky has joined #aichallenge 2010-09-22T00:25:18 I love my new bot even though he fails. Imma call him the hammer 2010-09-22T00:30:26 "The Hammer" 2010-09-22T00:30:27 Arekku made some silly moves in that game 2010-09-22T00:36:38 here's an example of a game that almost certainly would have changed the result if it went a little longer http://ai-contest.com/visualizer.php?game_id=4630315 2010-09-22T00:38:53 yep, i wish it had a decreasing band of ship difference allowed 2010-09-22T00:39:05 once it goes outside the band the game is declared and the band gets smaller as a game progresses 2010-09-22T00:39:33 disagree 2010-09-22T00:39:43 although i would never reduce it to 1 ship difference decides, i'd say if you cant get over 1.1 of their ships or whatever, it's a draw 2010-09-22T00:39:50 Snowplan: why? 2010-09-22T00:40:18 even just declaring a tie if the same side doesn't have both groth rate and number of ships would be better than the current method 2010-09-22T00:40:27 vulnerability : mass 2010-09-22T00:41:18 it's more to see bots battle than to have both trying to maximise ships at turn 200 2010-09-22T00:41:22 more fun* 2010-09-22T00:41:41 *** Naith has joined #aichallenge 2010-09-22T00:43:06 I think turn 400 is the cooler spectrum 2010-09-22T00:43:26 Wooho! Top Haskell bot! 2010-09-22T00:43:28 i like the band idea 2010-09-22T00:43:34 dufflebunk: nice 2010-09-22T00:43:42 i would have to top the leaderboard for best c++ bot :) 2010-09-22T00:43:55 do it 2010-09-22T00:43:55 currently 15th 2010-09-22T00:44:17 for another couple hours anyway... I think I've just been lucky with the last few games. 2010-09-22T00:44:27 i bea 3 just before 2010-09-22T00:44:41 but had 3 reasonably bad losses, so i don't expect it to hit top as yet 2010-09-22T00:46:42 *** mega1 has joined #aichallenge 2010-09-22T00:47:28 lol 2010-09-22T00:49:14 *** jesionaj has quit IRC (Ping timeout: 252 seconds) 2010-09-22T00:53:55 *** jesionaj has joined #aichallenge 2010-09-22T00:56:37 Top 10 players: kcm1700(3483), McLeopold(3476), dmj111(3436), adumlah(3356), EBraun(3309), montanalow(3294), Tracer(3265), Loks1982(3244), paulus(3224), Arekku(3220) 2010-09-22T00:57:53 *** mega1 has quit IRC (Ping timeout: 276 seconds) 2010-09-22T01:10:48 *** Snowplan has left #aichallenge 2010-09-22T01:15:32 *** Appleman1234 has quit IRC (Ping timeout: 272 seconds) 2010-09-22T01:17:46 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-09-22T01:19:09 there are too many c words in mathematics 2010-09-22T01:19:22 continous, closed, complete, compact, Cauchy, convergent etc. etc. 2010-09-22T01:22:38 left trim ;) 2010-09-22T01:22:58 ontinous, losed, omplete, ompact, auchy, onvergent 2010-09-22T01:23:04 i got to ouchy 2010-09-22T01:23:05 "there are too many o words in mathematics" 2010-09-22T01:23:06 and laughed 2010-09-22T01:23:15 etc. 2010-09-22T01:23:53 *** Utkarsh has joined #aichallenge 2010-09-22T01:23:58 tinous losed plete act uchy vergant 2010-09-22T01:24:09 problem solved. 2010-09-22T01:24:18 ha, i 2010-09-22T01:24:26 'm not sure my analysis lecturer would like me very much 2010-09-22T01:25:52 *** Palmik has joined #aichallenge 2010-09-22T01:25:53 reminds me of: a comathematician a comachine for turning theorems into ffee :P 2010-09-22T01:26:18 *cotheorems 2010-09-22T01:26:48 yep, i'm not a proper mathematician 2010-09-22T01:26:51 i don't drink coffee 2010-09-22T01:26:55 hate it 2010-09-22T01:27:00 tea? :) 2010-09-22T01:27:04 nup 2010-09-22T01:27:06 I don't drink coffee either 2010-09-22T01:27:07 waters good thanks 2010-09-22T01:27:10 but I don't hate it 2010-09-22T01:27:12 water is good. 2010-09-22T01:27:16 i like coffee 2010-09-22T01:27:34 like you like your algorithms? 2010-09-22T01:27:45 like i like cofree 2010-09-22T01:28:38 recursive... lol 2010-09-22T01:30:05 that was fun... i just ran a multithreaded 100 map test and had a bug cause none of the bot processes to exit 2010-09-22T01:30:20 1 minute load average was 57 2010-09-22T01:30:46 haha 2010-09-22T01:30:47 haha 2010-09-22T01:31:00 haha 2010-09-22T01:34:05 *** nick has joined #aichallenge 2010-09-22T01:34:09 hi me 2010-09-22T01:34:31 *** nick is now known as Guest16766 2010-09-22T01:35:31 sigh: i've almost got you on the main rankings ;) 2010-09-22T01:35:51 we need an au 1st and 2nd finish 2010-09-22T01:36:04 waiting for main ranking is so painful 2010-09-22T01:36:10 well, you already beat my bot on tcp, so I'm not surprised :P 2010-09-22T01:36:12 tell me about it 2010-09-22T01:36:25 but hopefully they'll have more servers soon 2010-09-22T01:36:26 drewrobb: where are you on the main ranking? 2010-09-22T01:36:30 there is also major leapfrogging going on 2010-09-22T01:36:31 they've been talking about it on the mailing list 2010-09-22T01:36:39 i just uploaded today, only 5 games 2010-09-22T01:36:53 have they banned starter bots? 2010-09-22T01:37:00 i've had 15, uploaded 3 days or so ago 2010-09-22T01:37:04 might have been 2 and a half 2010-09-22T01:37:10 don't think so 2010-09-22T01:37:26 drewrobb: I don't think they will ban starters 2010-09-22T01:37:34 sigh: they did last time 2010-09-22T01:37:45 oh ok 2010-09-22T01:37:52 not really a ban, just disabled those bots 2010-09-22T01:38:14 haha, imagine if they banned accounts for uploading starters 2010-09-22T01:38:24 I know what you meant :P 2010-09-22T01:38:37 yeah, especially since they tell people to do that first aha 2010-09-22T01:38:46 well, I've heard talk of limiting them to a certain amount of games and stuff, but not outright stop them from playing 2010-09-22T01:39:04 yeah, it would be silly to tell them to upload a bot, and then not see any games 2010-09-22T01:39:07 yeah i think that'd be best, or first upload from each account gets max 5 gams? 2010-09-22T01:39:11 they could bias resources to higher ranked bots 2010-09-22T01:39:27 drewrobb: they would discourage poorer competitors from competing 2010-09-22T01:40:24 arekku dropped back to 15th 2010-09-22T01:40:26 hmmm, don't they already give games based on ELO uncertainty? 2010-09-22T01:40:31 thanks to sigh too :D 2010-09-22T01:40:47 so in some way they already bias resources to the top and very bottom 2010-09-22T01:40:56 antimatroid1: haha 2010-09-22T01:41:06 drewrobb: no they don't base it on uncertainty, just on mean estimated rank 2010-09-22T01:41:32 well, they decide to schedule a bot for a game based on time since it last played a game 2010-09-22T01:41:41 and which bot to match with based on rank 2010-09-22T01:41:46 i think it'd be nice if bots got 3 or 4 games right away when uploading rather than 1 2010-09-22T01:41:58 but server is struggling so hard atm 2010-09-22T01:42:04 ahh 2010-09-22T01:42:07 awesome, it wasn't a dodgy win either, my bot played well in that game 2010-09-22T01:42:14 i think the artificial bias for new bots does more harm than good, myself 2010-09-22T01:42:38 i think mines got to the point that i've thrown so many things in my original intentions no longer work in all cases 2010-09-22T01:42:39 sure, you get that first game, but then everybody else is sitting around with 3 games waiting for their chances 2010-09-22T01:42:44 i'm completely content with dhartmei's server, i just wish more people would consistently play there 2010-09-22T01:43:00 yeah same, but it's nice for bragging rights on official rankings 2010-09-22T01:43:29 I play on tcp until I figure out why my bot is crap and then take it down :) 2010-09-22T01:43:38 repeat after I improve it 2010-09-22T01:44:15 yeah i do stuff like that, although tend to just throw a copy of .exe in a tcp folder and leave it running for at least 20 games 2010-09-22T01:44:20 whilst upating errors as it plays 2010-09-22T01:44:30 then when it gets too messy, i go through and clean everything up 2010-09-22T01:44:44 i'm terrible at developing clean code, but oh well 2010-09-22T01:46:01 yeah, my usual night is make a bunch of changes, find out one of them introduces a bug, and then undo and lose half of the good changes finding the bug 2010-09-22T01:46:25 version control is your friend :) 2010-09-22T01:46:44 using it, i'm just not diligent enough 2010-09-22T01:47:03 yeah, that can be a problem 2010-09-22T01:47:12 my version control consists of periodically zipping my bot and throwing that into a folder aha 2010-09-22T01:47:24 for me, I feel unclean if stuff isn't checked it 2010-09-22T01:47:36 antimatroid1: seriously get a proper vcs 2010-09-22T01:47:51 i use codeblocks to write stuff/compile etc. 2010-09-22T01:47:59 seems to do the job 2010-09-22T01:48:08 my test versions are always filled with debug statements, and then i don't want to check in that messy code 2010-09-22T01:48:30 just do an interactive checkin 2010-09-22T01:48:43 say which lines you do and don't want to check in, i mean 2010-09-22T01:49:10 i just have a function ostream& operator<<(ostream &o, const PlanetWars &pw); which i use to output the game each move to a file, then i can look in there if stuff is going wrong 2010-09-22T01:49:12 hmm, i wonder if there is a way to regex ignore lines? 2010-09-22T01:49:24 possibly 2010-09-22T01:49:29 yeah, i could just not check in anything with << 2010-09-22T01:49:39 ha 2010-09-22T01:49:59 grep -v 2010-09-22T01:56:32 deepblue seems to have a new bot 2010-09-22T01:56:37 Top 10 players: kcm1700(3498), McLeopold(3476), dmj111(3438), adumlah(3369), EBraun(3321), montanalow(3307), Tracer(3248), goldman(3235), Loks1982(3230), paulus(3229) 2010-09-22T01:58:37 i kind of wish that method 2 had been the winner of the "great debate on three-way battles" now. it seemed the most complex at the time, but now i think it would have been a lot simpler for my bot 2010-09-22T01:58:55 (the one where the players fight each other before fighting neutral) 2010-09-22T01:59:45 i like the way it's done now 2010-09-22T01:59:49 it works 2010-09-22T02:00:11 it would have meant that i could have merged all fleets that arrive at a planet at the same time rather than having to wait until the end 2010-09-22T02:00:27 that is, during transit 2010-09-22T02:00:28 i do that now 2010-09-22T02:00:33 but store full ship amounts 2010-09-22T02:00:44 like player 1 has this and player 2 has that 2010-09-22T02:00:50 right, that's not whati mean 2010-09-22T02:00:52 *what i 2010-09-22T02:01:00 i mean the battle could have been partly resolved in advance 2010-09-22T02:01:21 yeah, just store ship amounts then generate expected future states 2010-09-22T02:01:24 jmcarthur: I agree 2010-09-22T02:01:36 it still takes next to no time 2010-09-22T02:01:49 jmcarthur: that's the reason why I said that handling the simulation was messy 2010-09-22T02:01:58 well, in my code in particular, it would have been a godsend 2010-09-22T02:02:08 to be able to merge opposing fleets 2010-09-22T02:02:10 fair enough, it really doesn't bother me 2010-09-22T02:02:25 i would love to know min distance between planets though 2010-09-22T02:02:25 sigh: the battle itself is very simple at least 2010-09-22T02:02:42 as a once of event, yes 2010-09-22T02:03:19 the main reason i disliked method 2 was that it would have been a huge discouragement to expansion and aggression 2010-09-22T02:03:35 why? 2010-09-22T02:03:44 it's already ingrained in winning 2010-09-22T02:03:44 because neutral would have been much stronger 2010-09-22T02:03:53 so might have made it better possibly 2010-09-22T02:04:33 no way it's getting changed now anyway 2010-09-22T02:04:49 yeah i'm not proposing that we do whatsoever 2010-09-22T02:05:02 i don't regret the decision, i just would have voted differently in restrospect 2010-09-22T02:06:35 woo, my upload finally got enough games to crack the top 100 2010-09-22T02:07:43 *** Guest16766 has quit IRC (Quit: Page closed) 2010-09-22T02:07:57 i'm waiting for my next game that will hopefully place me top 10 2010-09-22T02:08:02 i'm well overdue for it 2010-09-22T02:08:23 not if it is against me :p 2010-09-22T02:08:45 is this 1.3? 2010-09-22T02:08:50 yeah 2010-09-22T02:08:52 would be luck of map 2010-09-22T02:09:05 true 2010-09-22T02:09:07 that's my current one, been to busy to play around atm 2010-09-22T02:09:22 *** jaspervdj has joined #aichallenge 2010-09-22T02:09:43 although we spent a bit of time today trying to work out proper algorithms for picking moves 2010-09-22T02:09:44 ah, assumed yours was older 2010-09-22T02:10:29 antimatroid1: you are right below me atm :P 2010-09-22T02:11:34 hmmm... only 3 aussies on the front page :S 2010-09-22T02:11:49 yeah, we aussies did very badly last contest 2010-09-22T02:11:56 we would have got ~10 but timed out 2010-09-22T02:12:07 heh 2010-09-22T02:12:36 i don't know what i'll do if that happens again after a couple of months work rather than a couple of weeks 2010-09-22T02:12:52 perhaps go on a murdering rampage under the guise of rage bot 2010-09-22T02:13:04 you didn't have timeout protection? 2010-09-22T02:13:08 we did 2010-09-22T02:13:24 i still don't know if we fucked up or the server fucked up, didn't even get to see the game that caused it 2010-09-22T02:13:37 up until it happened i assumed that was just a loss of game, first time it happened on main server :| 2010-09-22T02:13:57 but wasn't lots of people bitching about it, so have to assume it was our fault 2010-09-22T02:14:14 i still think it's an unecessarily strict rule 2010-09-22T02:14:14 hmmm... I should implement some timeout protection :S 2010-09-22T02:14:23 yeah that is tragic 2010-09-22T02:14:24 the main server had all sorts of issues relating to timeouts 2010-09-22T02:14:36 i feel like a lot of people were screwed over by it 2010-09-22T02:14:47 a bot that always times out should really have on average shorter games than other bots 2010-09-22T02:14:49 was it wall clock then as well? 2010-09-22T02:14:52 and you could set a max number of timeouts 2010-09-22T02:14:52 yes 2010-09-22T02:15:26 jmcarthur: was that "yes" for me? 2010-09-22T02:15:29 yes 2010-09-22T02:15:32 :) 2010-09-22T02:15:43 i am being absolutely maticulous about my timer this time though, there is no way i can fail to check in time 2010-09-22T02:16:03 antimatroid1: i was sure of that too, but my bot was still DQed a *ton* during testing 2010-09-22T02:16:18 antimatroid1: i ended up submitting it with a self-imposed 0.5s time limit, iirc 2010-09-22T02:16:27 and was still crossing my fingers about it 2010-09-22T02:16:39 yeah, i don't like that, i have it set at 0.9 max atm 2010-09-22T02:16:41 because in testing i had even had problems with a self-imposed 0.1s time limit :\ 2010-09-22T02:16:44 i remember there was the issue of cpu time versus user time 2010-09-22T02:16:56 and it's still an issue, IMO 2010-09-22T02:17:17 you guys were using the correct one and still had problems? 2010-09-22T02:17:26 i was using SIGALRM 2010-09-22T02:17:35 wall clock time 2010-09-22T02:17:42 still problems 2010-09-22T02:17:49 i had mine set to stop itself at 0.99sec, and never had issues 2010-09-22T02:17:57 i'm using a timer based off start time and current time, is that a bad idea? 2010-09-22T02:18:03 also, even when my bot had a much higher limit imposed on it, it still behaved in a way that led me to believe it was only getting like 0.1s of cpu time 2010-09-22T02:18:45 i assume the resources on my computer are much shitter than i'll get on the server, so if it works here it'll work there 2010-09-22T02:18:58 i wouldn't assume anything like that 2010-09-22T02:19:01 the server is pretty shitty 2010-09-22T02:19:07 hmm... was there any discussion on the best way to do timings? I was going to use SIGALRM too, but jmcarthur makes me unsure 2010-09-22T02:19:19 sigh: SIGALRM was the best timer i found 2010-09-22T02:19:27 ok, thank you! 2010-09-22T02:19:46 * antimatroid1 runs to google 2010-09-22T02:19:55 it also has the advantage that *if* we manage to switch to cpu time it's easy to switch SIGALRM over to that 2010-09-22T02:20:01 yeah :) 2010-09-22T02:20:47 were you doing anything more sophisticated than just outputing "go" in the signal handler? 2010-09-22T02:21:12 in the tron contest there wasn't even a "go". you just sent your command 2010-09-22T02:21:21 oh, right 2010-09-22T02:21:28 (sorry, I wasn't part of that contest) 2010-09-22T02:21:28 i did do some fancy stuff after determining that it didn't slow anything down 2010-09-22T02:22:05 what about google? 2010-09-22T02:22:16 *** otherAntimatroid has joined #aichallenge 2010-09-22T02:22:21 i had a (green) worker thread, an IO thread, and a thread that was spawned automatically for signal handling. the worker ran basically constantly and the signal and IO threads would occasionally interrupt it 2010-09-22T02:22:30 how many matroids are there? o_O 2010-09-22T02:22:44 jmcarthur: threads aren't allowed? 2010-09-22T02:22:59 sigh: green threads are allowed 2010-09-22T02:23:40 ah, ok 2010-09-22T02:23:45 yeah, that makes sense 2010-09-22T02:24:03 anyway, it was nice because i could process even while waiting for I/O 2010-09-22T02:25:11 the version that did that had a higher win rate than the one that didn't, at least 2010-09-22T02:25:44 interesting... though I'm far from doing anything that involved 2010-09-22T02:25:45 okay, i'm going to dream about array processing tonight 2010-09-22T02:25:46 sigh: just the two of us 2010-09-22T02:25:50 *** pc has quit IRC (Remote host closed the connection) 2010-09-22T02:25:56 *** pc has joined #aichallenge 2010-09-22T02:26:25 we're so damn overdue for a game, we better not lose when we get one 2010-09-22T02:26:37 *** pc has quit IRC (Remote host closed the connection) 2010-09-22T02:27:42 jmcarthur: would it really have helped that much? after like 10 levels it was hard to really penetrate the game tree much further when it was big 2010-09-22T02:33:53 http://kaggle.com/chess?viewtype=leaderboard 2010-09-22T02:34:50 i know a trick to get very high results on benzedrine 2010-09-22T02:35:13 by making multiple connections and pairing other bots through an echo loop 2010-09-22T02:36:32 bartwe: but you have to get pairs of games at the same time 2010-09-22T02:36:50 until you have an implementation, I don't believe you 2010-09-22T02:36:51 you already get to know who you are playing 2010-09-22T02:37:04 what's the point? 2010-09-22T02:37:09 couldn't you just do a master/slaves configuration? 2010-09-22T02:37:15 or am i missing something? 2010-09-22T02:37:17 *** Olathe has quit IRC (Read error: Connection reset by peer) 2010-09-22T02:37:19 *** O1athe has joined #aichallenge 2010-09-22T02:37:43 if someone really cares that much about their score on benzedrine and can be bothered, good on them aha 2010-09-22T02:37:47 how about the "write an awesome bot" method? 2010-09-22T02:38:23 McLeopold: when are you releasing a new bot? 2010-09-22T02:38:44 I'm working on it. Not sure. 2010-09-22T02:39:01 I keep finding "bugs" that make my first one really good. 2010-09-22T02:39:22 the joys of ai writing 2010-09-22T02:39:28 I'm in the middle of a total rewrite. 2010-09-22T02:39:28 errors being smarter than you 2010-09-22T02:40:26 I really thought Arekku was going to kick me ass. I was totaly wrong. 2010-09-22T02:40:34 java is surging, 8-11 2010-09-22T02:40:46 i don't think it looks as smart as some of you think 2010-09-22T02:40:54 it made some stupid moves against you 2010-09-22T02:41:07 like sending 2 fleets to claim a fairly easily claimable 69 ship neutral planet 2010-09-22T02:41:09 Maybe it was just a map that favors my style. 2010-09-22T02:41:34 68** 2010-09-22T02:41:50 did similar things a few times 2010-09-22T02:42:06 What turn was that? 2010-09-22T02:42:20 around 66 2010-09-22T02:42:30 another one happened a bit before it to a planet just to the left 2010-09-22T02:42:44 it's the 68 near centre in bottom left quadrant 2010-09-22T02:42:52 Those were my fleets... 2010-09-22T02:42:56 oh wait, that was you :P 2010-09-22T02:43:20 cough fix that :P cough 2010-09-22T02:43:43 Yeah, I fix that, then I can't beat myself. This is what I dealing with. 2010-09-22T02:44:09 *** sinan has joined #aichallenge 2010-09-22T02:44:50 I think his lack of counterplay in the center was his downfall. 2010-09-22T02:45:14 He let me take that center right planet for 11, and then did nothing. 2010-09-22T02:45:20 can't lose position bro 2010-09-22T02:45:24 that's the end 2010-09-22T02:45:51 damn i finally got a game and lost aha 2010-09-22T02:46:07 what's your bots name 2010-09-22T02:46:19 youwillfail 2010-09-22T02:46:52 ah that loss is alright, the bot has gone to favouring attacking too early which is a known error 2010-09-22T02:46:59 and he sucks at attacking once he can win too 2010-09-22T02:46:59 aha 2010-09-22T02:47:34 Did you see that tiny little fleet of 6 sneek behind you? 2010-09-22T02:47:47 cough fix that :P cough 2010-09-22T02:48:11 i ahve no idea why that wasn't picked up 2010-09-22T02:48:14 :P 2010-09-22T02:48:31 2 ships ffs was all it needed to send 2010-09-22T02:49:32 muhahahah, anyone know knows C++ check out my awesome macro sexy namespace nesting abuse :D 2010-09-22T02:49:33 http://codepad.org/GQP5Rj7l 2010-09-22T02:49:33 the incoming 24 and the outgoing 42 from the planet it crossed didn't make any difference. You could have spared 2 2010-09-22T02:50:15 yeah, i think i messed things up when i changed to estimating nearest enemies 2010-09-22T02:50:27 for obvious reasons 2010-09-22T02:51:06 *** boegel has joined #aichallenge 2010-09-22T02:56:26 w00 2010-09-22T02:56:38 Top 10 players: kcm1700(3528), McLeopold(3498), dmj111(3414), adumlah(3372), EBraun(3323), montanalow(3308), goldman(3229), paulus(3228), Tracer(3228), Arekku(3224) 2010-09-22T02:56:43 rank 145 with my last submission last night (at 2am :)) 2010-09-22T02:56:51 coming from 650+, that's not too bad :P 2010-09-22T02:56:57 nice :) 2010-09-22T02:57:27 graphitemaster: oh dear lord, what have you done? 2010-09-22T02:58:40 and I believe there's a lot of room for improvement \o/ 2010-09-22T03:02:39 antimatroid1, I made it better :P http://codepad.org/6rz2dmD8 2010-09-22T03:02:48 woah, me and Arekku have exactly the same elo right now 2010-09-22T03:03:39 3 ties in the top 15 2010-09-22T03:03:56 yeah :) 2010-09-22T03:13:09 *** tapwater has quit IRC (Quit: tapwater) 2010-09-22T03:22:37 *** antimatroid1 has left #aichallenge 2010-09-22T03:22:38 *** boegel has quit IRC (Quit: Leaving) 2010-09-22T03:25:09 *** antimatroid has joined #aichallenge 2010-09-22T03:27:36 *** aerique has joined #aichallenge 2010-09-22T03:38:12 *** graphitemaster is now known as sleep_master 2010-09-22T03:39:20 *** sigh has quit IRC (Remote host closed the connection) 2010-09-22T03:40:05 *** joshrl has joined #aichallenge 2010-09-22T03:56:39 Top 10 players: kcm1700(3546), McLeopold(3518), dmj111(3398), adumlah(3371), EBraun(3318), montanalow(3272), goldman(3241), Willjarn(3229), paulus(3223), felixcoto(3223) 2010-09-22T04:01:12 hmmm, possible reasonable condition for final tournament running time, 2010-09-22T04:01:12 final tournament has been running x amount of time AND the "average amount of change" in positions from y games is below z%? 2010-09-22T04:01:47 at least* x amount of time 2010-09-22T04:11:35 *** jesionaj` has joined #aichallenge 2010-09-22T04:13:14 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-09-22T04:15:07 why first day of the registrations always becomes a nightmare :/ 2010-09-22T04:15:38 *** jesionaj has quit IRC (Ping timeout: 265 seconds) 2010-09-22T04:15:40 *** hellman has joined #aichallenge 2010-09-22T04:18:51 *** Utkarsh has joined #aichallenge 2010-09-22T04:18:59 *** Titankiller has joined #aichallenge 2010-09-22T04:26:45 *** Alystair has quit IRC (Ping timeout: 265 seconds) 2010-09-22T04:27:04 *** kozlovsky has joined #aichallenge 2010-09-22T04:28:03 *** jesionaj` has quit IRC (Ping timeout: 240 seconds) 2010-09-22T04:28:14 *** kjeldahl has joined #aichallenge 2010-09-22T04:30:07 *** Appleman1234 has joined #aichallenge 2010-09-22T04:34:34 *** Appleman1234 has quit IRC (Ping timeout: 252 seconds) 2010-09-22T04:36:01 *** Titankiller has quit IRC (Quit: Nettalk6 - www.ntalk.de) 2010-09-22T04:36:19 *** sigh has joined #aichallenge 2010-09-22T04:38:57 *** perror has joined #aichallenge 2010-09-22T04:39:50 *** Appleman1234 has joined #aichallenge 2010-09-22T04:42:47 *** icky has quit IRC (Ping timeout: 276 seconds) 2010-09-22T04:56:39 Top 10 players: kcm1700(3532), McLeopold(3509), dmj111(3395), adumlah(3384), EBraun(3315), montanalow(3271), vinx33(3235), paulus(3225), Loks1982(3219), goldman(3218) 2010-09-22T04:58:46 *** hamstah has quit IRC (Ping timeout: 252 seconds) 2010-09-22T05:00:24 *** icky has joined #aichallenge 2010-09-22T05:01:40 *** Theudas has joined #aichallenge 2010-09-22T05:01:50 hi 2010-09-22T05:02:00 hi 2010-09-22T05:02:27 *** joshrl has quit IRC (Ping timeout: 272 seconds) 2010-09-22T05:02:38 i have a problem running the tcp client :/ ... my bot is running well but times out on the tcp client 2010-09-22T05:02:48 any suggestions? 2010-09-22T05:03:21 *** joshrl has joined #aichallenge 2010-09-22T05:03:22 *** joshrl has joined #aichallenge 2010-09-22T05:03:53 *** O1athe has quit IRC (Ping timeout: 264 seconds) 2010-09-22T05:05:25 no suggestions anyone? 2010-09-22T05:06:35 ping the tcp server 2010-09-22T05:06:44 can be a connection problem 2010-09-22T05:13:16 *** Naktibalda has joined #aichallenge 2010-09-22T05:13:23 hi 2010-09-22T05:14:30 * Naktibalda have tried to join #ai-contest again 2010-09-22T05:15:34 *** Ceridan has joined #aichallenge 2010-09-22T05:15:44 hi 2010-09-22T05:16:03 hi 2010-09-22T05:17:51 seems like it cannot handle .java files and only .jars? 2010-09-22T05:24:27 *** amstan has quit IRC (Ping timeout: 272 seconds) 2010-09-22T05:25:55 *** amstan has joined #aichallenge 2010-09-22T05:25:55 *** ChanServ sets mode: +o amstan 2010-09-22T05:26:37 *** amstan has quit IRC (Remote host closed the connection) 2010-09-22T05:30:34 *** mega1 has joined #aichallenge 2010-09-22T05:34:11 *** hamstah has joined #aichallenge 2010-09-22T05:34:26 hey 2010-09-22T05:42:20 Theudas: I don't think the tcp server sends your program to the server. It runs the program locally. 2010-09-22T05:42:59 So I think so long as your local java can run the command you give it, then you should be good. Which likely means you'll need a jar or compiled classes 2010-09-22T05:44:58 *** McLeopold has quit IRC (Ping timeout: 252 seconds) 2010-09-22T05:49:17 *** tuzzeg has joined #aichallenge 2010-09-22T05:51:16 drewrobb: looking good 2010-09-22T05:55:26 *** retybok has joined #aichallenge 2010-09-22T05:55:47 *** smhanov has quit IRC (Ping timeout: 240 seconds) 2010-09-22T05:55:52 I've submitted about 12 hours ago to the official site, haven't lost a game yet, and am not even yet in the top 100... Sigh... 2010-09-22T05:56:07 I think it takes days 2010-09-22T05:56:12 Fortunately there is the tcp server. Thanks dhartmei_ ! 2010-09-22T05:56:40 Top 10 players: kcm1700(3520), McLeopold(3508), dmj111(3399), adumlah(3371), wagstaff(3306), EBraun(3293), vinx33(3251), montanalow(3237), goldman(3233), Loks1982(3207) 2010-09-22T05:58:33 antimatroid: as long as the wins keep coming 2010-09-22T06:03:08 *** joshrl has quit IRC (Quit: Leaving) 2010-09-22T06:03:30 *** Nuriaion has joined #aichallenge 2010-09-22T06:04:35 *** HakanD has quit IRC (Ping timeout: 240 seconds) 2010-09-22T06:20:34 *** Olathe has joined #aichallenge 2010-09-22T06:27:53 *** jimmerton has joined #aichallenge 2010-09-22T06:28:14 *** icky has quit IRC (Ping timeout: 255 seconds) 2010-09-22T06:32:15 *** jimmerton_ has joined #aichallenge 2010-09-22T06:34:30 *** jimmerton has quit IRC (Ping timeout: 264 seconds) 2010-09-22T06:34:30 *** jimmerton_ is now known as jimmerton 2010-09-22T06:40:12 *** hellman has quit IRC (Remote host closed the connection) 2010-09-22T06:40:28 *** Utkarsh has quit IRC (Ping timeout: 272 seconds) 2010-09-22T06:40:32 *** kcm1700 has quit IRC (Read error: Connection reset by peer) 2010-09-22T06:45:25 *** Utkarsh has joined #aichallenge 2010-09-22T06:56:40 Top 10 players: kcm1700(3532), McLeopold(3521), dmj111(3411), adumlah(3385), vinx33(3328), wagstaff(3306), EBraun(3304), montanalow(3255), Loks1982(3227), paulus(3226) 2010-09-22T06:57:44 *** sinan has quit IRC (Ping timeout: 272 seconds) 2010-09-22T07:04:00 *** Harder has joined #aichallenge 2010-09-22T07:04:16 HEY?! 2010-09-22T07:05:05 *** Harder has quit IRC (Client Quit) 2010-09-22T07:06:38 *** jimmerton has quit IRC (Quit: jimmerton) 2010-09-22T07:07:28 *** Zaphus has joined #aichallenge 2010-09-22T07:08:34 I suspect there's a bug in my code: I lost this match :-) http://ai-contest.com/visualizer.php?game_id=4637181 wish I knew if it was a timeout or an illegal move :-) 2010-09-22T07:16:51 *** Blkt has joined #aichallenge 2010-09-22T07:19:21 Zaphus: you should play on the TCP server, it tells you whether it's a timeout or an invalid command 2010-09-22T07:20:24 retybok: Yes, it does when I'm there to run it, but even though the real server only runs one game every two or more hours, it can do it while I'm at work and I dont need to leave a PC connected. It also has more variety in opponents :-) 2010-09-22T07:21:18 retybok: in theory I could feed the playback data into the game engine so that it prompts me for that exact turn again and I could see if it were an illegal turn 2010-09-22T07:21:41 I think that you play against more opponents in TCP then in real server :) 2010-09-22T07:22:15 Zaphus: if you want to do that there is a python script on the forums to parse the playback string 2010-09-22T07:23:21 naktibalda: true, but usually the same opponents over and over again, I like seeing these stranger opponents as well - especially since I'm in Australia so half the planet is asleep when I want to play :-) 2010-09-22T07:24:31 sigh: thanks, will take a look - just going through my code and making it more solid anyway, it died immediately after the final planet was taken so the opponetn had zero and there were no other planets left, so I'm guessing illegal turn 2010-09-22T07:25:09 sigh: great progress on your bot by the way, I'm struggling to get into the top 20 to get three aussies up there! 2010-09-22T07:25:29 :) 2010-09-22T07:25:40 I'm about to release a new bot 2010-09-22T07:25:52 Zaphus: come on, you're letting the team down ;) 2010-09-22T07:26:02 sigh: i'm working on a new one now 2010-09-22T07:26:14 I've got my new one out on tcp atm 2010-09-22T07:26:28 fixes up a huge bug in my simulation code 2010-09-22T07:26:37 nice, i'm probably a few hours away 2010-09-22T07:26:46 restructuring lots and fixing up my decisions 2010-09-22T07:26:52 antimatroid: I'm playing around with a new idea, it's got a few holes in it. I really need to build a bot like the ones at the top so I can do some offline testing 2010-09-22T07:27:31 maybe we should all upload new bots at midnight tonight (Aussie east coast time) :-) 2010-09-22T07:27:44 i'll see how i'm going 2010-09-22T07:27:51 * antimatroid gets back to work 2010-09-22T07:28:05 i might test a few things incrementally before then anyway 2010-09-22T07:28:11 but i'll try to hit midnight for you :P 2010-09-22T07:28:16 it's 9 21pm for me right now 2010-09-22T07:31:02 this new strategy will either flop really badly or be a decent improvement 2010-09-22T07:31:31 same time here, Melbourne. you're in Tassie, right ? Sigh, where are you ? 2010-09-22T07:31:41 Melbourne 2010-09-22T07:31:59 Are there any admins here who could tell me why my bot keeps losing randomly in the middle of games of the main server but does fine on the tcp server? 2010-09-22T07:32:23 sigh: I keep confusing your name with one of my friends, slagh :-) 2010-09-22T07:32:30 o_O 2010-09-22T07:32:47 *** Ceridan has quit IRC (Ping timeout: 252 seconds) 2010-09-22T07:33:34 who owns 8-53Bot, I keep getting matched up with it 2010-09-22T07:33:36 *** boegel has joined #aichallenge 2010-09-22T07:33:43 That would be me sigh 2010-09-22T07:33:49 ah :) 2010-09-22T07:33:51 :) 2010-09-22T07:34:29 I seem to lose to you fairly consistently 2010-09-22T07:35:09 this seems relevant: http://games.slashdot.org/story/10/09/22/014255/Elo-Chess-Rating-System-Topped-By-Proposed-Replacements?from=rss 2010-09-22T07:35:14 you got me twice :P 2010-09-22T07:35:24 True :P 2010-09-22T07:36:13 I have five loops of the tcp server running all the time so I play a lot of games, which is probably why you play me so much. 2010-09-22T07:36:29 ahhh 2010-09-22T07:36:39 are they all the same bot? 2010-09-22T07:36:47 At the moment, yes. 2010-09-22T07:36:58 When I test changes I make a few of them a different bot. 2010-09-22T07:37:02 And it's easy to compare the two. 2010-09-22T07:37:13 yeah 2010-09-22T07:38:22 Unfortunately my bot just quits randomly on the main server at around 50 moves, and I have no idea why. 2010-09-22T07:38:29 ah :( 2010-09-22T07:38:37 FlagCapper: I've got to thank you, that first loss was quite instructive 2010-09-22T07:38:48 no problem :) 2010-09-22T07:39:07 my planet selection was quite crap compared to yours... currently thinking about how to mitigate that :) 2010-09-22T07:39:28 I think the only problem I have right now is my bot doesn't defend my planets properly. 2010-09-22T07:40:17 If someone only sends num_ships + 1 it just ignores it and loses the planet. 2010-09-22T07:41:00 FlagCapper: yeah... for example this game should have been in the bag for you: http://72.44.46.68/canvas?game_id=98350 2010-09-22T07:41:41 Yeah, and then you got your home planet back. 2010-09-22T07:41:57 Anyway, it should be a simple fix. 2010-09-22T07:42:02 yeah :) 2010-09-22T07:42:27 meanwhile, I need to turn the aggression back down a notch... all out on the first turn is not so good :P 2010-09-22T07:42:45 ;) 2010-09-22T07:43:03 giving my tweaked version a run on the server. 2010-09-22T07:43:07 tcp that is 2010-09-22T07:44:34 you guys are too good :) 2010-09-22T07:44:37 Just jumped in front of sigh on the Tcp server - 4 games 4 wins... it's downhill from here 2010-09-22T07:45:14 i also finally started beating sigh-2 2010-09-22T07:45:29 getting pushed out of the top 25 on benzedrine too ;p 2010-09-22T07:45:31 bartwe: finally? you never lost 2010-09-22T07:45:45 sigh: your still higher ranked 2010-09-22T07:45:58 5th on TCP, I might kill the bot now :-) 2010-09-22T07:46:41 *** phreeza has joined #aichallenge 2010-09-22T07:46:43 *** smhanov has joined #aichallenge 2010-09-22T07:46:44 doh, lost against drewrobb 2010-09-22T07:50:23 FlagCapper: are you australian as well? 2010-09-22T07:50:56 *** Naktibalda has quit IRC (Ping timeout: 252 seconds) 2010-09-22T07:51:19 *** Appleman1234 has quit IRC (Quit: Leaving) 2010-09-22T07:52:04 *** Sean_McEligot has joined #aichallenge 2010-09-22T07:53:29 yeah, getting inside the top 25 at benzedrine is getting difficult! I used to be 9th with the bot called retybok-entry1 ... 2010-09-22T07:53:52 It's 65th at the moment :-/ 2010-09-22T07:53:59 *** Naktibalda has joined #aichallenge 2010-09-22T07:54:05 .8 was in the top 5 for a while 2010-09-22T07:54:17 its getting very competative at the top 2010-09-22T07:54:23 I was 2nd about 10 minutes ago :-) 2010-09-22T07:54:37 9th now, it never lasts 2010-09-22T07:54:39 I still think a human with enough time would defeat most if not all the bots, though 2010-09-22T07:54:48 which means there is margin for improvement 2010-09-22T07:54:49 Zaphus: try again with 10+ games with atleast 4 different competitors 2010-09-22T07:55:09 retybok: removing the time limit would also improve some of the bots 2010-09-22T07:55:20 anyways, that is my tactic 2010-09-22T07:55:27 watching replays, deciding on bad moves 2010-09-22T07:55:31 bartwe: no, I liked 4 games - gave me a better result :-) 2010-09-22T07:55:45 going back into that state in the bot and tweaking the cost function 2010-09-22T07:56:00 Sure, but if you take chess for example, any computer with a 1 second thinking time can destroy 99% of humans. Granted, it took decades to arrive at this point :) 2010-09-22T07:56:07 bartwe: finally scraped a win from you: http://72.44.46.68/canvas?game_id=98637 2010-09-22T07:56:11 that was tough 2010-09-22T07:56:41 Top 10 players: kcm1700(3558), McLeopold(3552), adumlah(3417), dmj111(3410), vinx33(3367), EBraun(3335), montanalow(3306), paulus(3287), wagstaff(3259), goldman(3247) 2010-09-22T07:57:17 *** Ceridan has joined #aichallenge 2010-09-22T07:59:21 sigh: oooh, close 2010-09-22T07:59:54 yeah, that was nerve-wracking to watch 2010-09-22T08:00:36 i guess mine should have invested in the 88 neutral 2010-09-22T08:01:07 no immediatly obvious stupid moves 2010-09-22T08:01:55 yeah, that 88-neutral is a tricky one, my bot didn't want it for a while either 2010-09-22T08:02:09 it seems to have won the game for you 2010-09-22T08:02:24 i think it is due to my bot evaluating the fitness of fleets, not whole game states 2010-09-22T08:02:40 that's what I do also 2010-09-22T08:02:59 I don't look at the game state as a whole... I can't think of how to do it 2010-09-22T08:03:06 ha 2010-09-22T08:03:09 I mean a good way of doing it 2010-09-22T08:03:29 i have some things that work so far 2010-09-22T08:03:59 i was thinking of building something that i could train/evaluate against known outcomes 2010-09-22T08:04:27 yeah, possibly... but I don't want to invest the effort until I have a better understanding of what works 2010-09-22T08:05:02 so far for me, better prediction == better score 2010-09-22T08:05:15 yeah, that makes sense 2010-09-22T08:05:28 so i'm down tuning the skewed score functions and improving prediction 2010-09-22T08:05:46 hence the .12 version taking so long to beat .8 2010-09-22T08:06:06 ah 2010-09-22T08:07:05 *** Naith has quit IRC (Quit: Naith) 2010-09-22T08:08:21 ouch, getting a whipping from sigh-2 2010-09-22T08:09:18 *** Lithosphere has quit IRC (Ping timeout: 272 seconds) 2010-09-22T08:09:41 *** sinan has joined #aichallenge 2010-09-22T08:09:46 yay... climbing up the ladder 2010-09-22T08:10:24 let's see if I can't get drewrobb 2010-09-22T08:12:07 is there any way to see more than the lest 15 games on the official server? 2010-09-22T08:12:31 there is a show more games button at the bottom of your list 2010-09-22T08:12:47 "view more" 2010-09-22T08:13:03 sigh: ok, so any way to see more than the last 20 games then? :P 2010-09-22T08:13:32 it shows way more than 20 for me, are you sure your latest submission has played that many games? 2010-09-22T08:13:37 *played more games 2010-09-22T08:13:46 it won't show you games from older submissions 2010-09-22T08:14:31 *** Appleman1234 has joined #aichallenge 2010-09-22T08:14:35 *** Blkt has quit IRC (Ping timeout: 252 seconds) 2010-09-22T08:15:25 oh god drewrobb1.4 right away with a very untested bot with a few changes 2010-09-22T08:16:01 you beat him? 2010-09-22T08:17:16 sigh: you may be right, thanks 2010-09-22T08:17:27 sigh: seems like my latest bot played only 20 games 2010-09-22T08:18:22 boegel: yeah, games are not very frequent 2010-09-22T08:18:47 beating drewrobb1.4 roid1> what's the point? 2010-09-22T08:18:48 [08:30] couldn't you just do a master/slaves configuration? 2010-09-22T08:18:48 [08:30] or am i missing something? 2010-09-22T08:18:50 ehmz 2010-09-22T08:19:12 ugh 2010-09-22T08:19:17 *** Lithosphere has joined #aichallenge 2010-09-22T08:19:20 copypaste with virtualmachines suck 2010-09-22T08:19:20 O_o 2010-09-22T08:19:24 hehe 2010-09-22T08:19:37 http://72.44.46.68/canvas?game_id=98938 2010-09-22T08:19:40 beat you antimatroid :) 2010-09-22T08:20:10 I hate it when a time-out occurs when I'm winning, urgh 2010-09-22T08:20:40 do you guys see time-outs sometimes on the other server, where your bot is running locally? 2010-09-22T08:20:56 on tcp, yeah... it sucks but oh well 2010-09-22T08:21:00 bartwe: nice :) 2010-09-22T08:21:13 bartwe: you seem to be good at getting into these long drawn out battles :P 2010-09-22T08:21:37 i guess it means that it is getting stronger 2010-09-22T08:21:52 but is missing something in decicive strikes 2010-09-22T08:21:58 nope lost, it's just testing a few things, everything else is like retarded 2010-09-22T08:22:01 *** Ceridan_ has joined #aichallenge 2010-09-22T08:22:32 coordinated attacks is my next secret weapon 2010-09-22T08:22:36 *** Ceridan has quit IRC (Ping timeout: 272 seconds) 2010-09-22T08:22:50 :) 2010-09-22T08:24:08 *** ermau has joined #aichallenge 2010-09-22T08:25:48 sigh: k, thx 2010-09-22T08:25:51 oh dear, so many things are broke 2010-09-22T08:26:16 bartwe giving me some trouble 2010-09-22T08:27:10 ouch, just got annihilated by drewrobb1.4 >< 2010-09-22T08:27:16 completely 2010-09-22T08:28:30 *** Lafi has quit IRC (Ping timeout: 276 seconds) 2010-09-22T08:28:35 ouch 2010-09-22T08:29:29 looks like I'll get stuck at #4 2010-09-22T08:30:49 *** Nuriaion has quit IRC (Quit: Nuriaion) 2010-09-22T08:31:10 *** ermau is now known as ermau|work 2010-09-22T08:32:18 lol, boegel has excessively long bot names :D 2010-09-22T08:32:45 git commits, I assume 2010-09-22T08:35:27 ok, I'm getting consistently owned by drewrobb1.4 2010-09-22T08:36:36 *** Sean_McEligot_ has joined #aichallenge 2010-09-22T08:37:00 wow bartwe is quite good at comebacks 2010-09-22T08:38:00 .12 or .8 ? 2010-09-22T08:38:09 sigh: no really git commits, I'm not that organized ;) 2010-09-22T08:38:30 *** Mathnerd314 has joined #aichallenge 2010-09-22T08:38:43 bartwe: both 2010-09-22T08:39:00 sigh: I keep track of previous versions of my bots, to test against, and to avoid losing something that worked well 2010-09-22T08:39:07 sigh: maybe I should be using git, but meh :) 2010-09-22T08:39:35 boegel: well, any vcs really :P 2010-09-22T08:39:53 *** Sean_McEligot has quit IRC (Ping timeout: 252 seconds) 2010-09-22T08:39:59 *** Sean_McEligot_ is now known as Sean_McEligot 2010-09-22T08:41:16 sigh: yeah, well, maybe I will in the future 2010-09-22T08:41:35 sigh: because my pool of bots is getting pretty big already, and most of them are utter crap :P 2010-09-22T08:42:11 boegel: same :) 2010-09-22T08:42:27 drewrobb: strange, they are very different codewise 2010-09-22T08:42:36 so, who's up for uploading new bots in ~90 minutes ? just to watch them crawl the ladder together over the next couple of days (12midnight Melbourne time) 2010-09-22T08:42:47 the latest ones are doing pretty good though, and I haven't even fully implemented my strategy yet :) 2010-09-22T08:43:39 Zaphus: yeah, I might as well 2010-09-22T08:44:04 drewrobb: grouping or coordinated attacks would make you very scary 2010-09-22T08:44:21 bartwe: i have very naive support for that 2010-09-22T08:44:32 i also think your bot is strong at spreading out quickly 2010-09-22T08:44:46 more late game neutrals that it can capture at will 2010-09-22T08:45:29 hmm, tcp server seems unresponsive? 2010-09-22T08:45:57 boegel: wfm 2010-09-22T08:46:54 retybok: hmm, might be the connection here in a camping site in the South of France :P 2010-09-22T08:47:44 boegel: enchanté, comment-allez-vous? 2010-09-22T08:47:48 :) 2010-09-22T08:48:27 don't you just love this game? http://72.44.46.68/canvas?game_id=99313 2010-09-22T08:48:33 retybok: :) 2010-09-22T08:48:42 except for when I time out, that is 2010-09-22T08:48:51 *** Sean_McEligot has left #aichallenge 2010-09-22T08:49:14 *** Appleman1234 has quit IRC (Ping timeout: 252 seconds) 2010-09-22T08:49:38 boegel: very nice :) 2010-09-22T08:49:53 you could have won by spreading out early, though 2010-09-22T08:50:18 retybok: yeah, but the nearby planets are so expensive: ) 2010-09-22T08:50:35 I'll have to think about on how to fix that one :) 2010-09-22T08:50:57 retybok: also, if I would've spread out early, I would've lost the home planet 2010-09-22T08:51:10 retybok: which might cause trouble 2010-09-22T08:51:17 hard map imho 2010-09-22T08:52:00 oh wow 2010-09-22T08:52:01 i'ma get this right soon 2010-09-22T08:52:34 if you don't retaliate you could spare 20 units to take the 7 neutral 2010-09-22T08:52:50 infact i think you have a predictor bug in that you attack 2010-09-22T08:53:27 the distance gives you a 20 unit margin 2010-09-22T08:53:40 bartwe: if I give up just one more ship, I lose the home planet, which means I'm doomed imho 2010-09-22T08:53:58 nah 2010-09-22T08:54:22 if you don't retaliate, the distance is means a time off set of 4 steps 2010-09-22T08:54:27 bartwe: you don't think so? 2010-09-22T08:54:51 which at the growthrate means you have 20 units more then the enemy at time of arrival of his fleet 2010-09-22T08:54:57 then at the time he sent his fleet 2010-09-22T08:55:00 bartwe: if I don't retaliate, the other guy has more budget... 2010-09-22T08:55:06 so ? 2010-09-22T08:55:15 you have a single, onetime budget of 20 units 2010-09-22T08:55:24 bartwe: so he can kill my home planet if I spend my budget somewhere else 2010-09-22T08:55:34 untrue 2010-09-22T08:55:38 hmm 2010-09-22T08:55:46 he cant take your homeworld if you send 20 units 2010-09-22T08:55:54 but only if you dont retaliate 2010-09-22T08:56:01 yeah, just attacking the opponent start position is suboptimal 2010-09-22T08:56:09 i think 2010-09-22T08:56:40 anyone has a better bot than the default to release? :) 2010-09-22T08:56:41 drewrobb: well, how do you know it's his start position? 2010-09-22T08:56:41 Top 10 players: McLeopold(3534), kcm1700(3519), adumlah(3416), dmj111(3391), vinx33(3355), EBraun(3318), montanalow(3292), paulus(3272), wagstaff(3249), goldman(3241) 2010-09-22T08:56:46 need a better one to test against 2010-09-22T08:56:49 it is only suboptimal if he sends more then his initial budget minus home2home traveltime times growth 2010-09-22T08:56:57 atleast 2010-09-22T08:57:09 boegel: huh? 2010-09-22T08:57:10 assuming his capture is not of equal or greater immediate value 2010-09-22T08:57:38 drewrobb: well, ok, you could deduce that, but your bot doesn't know which turn it's playing, right? 2010-09-22T08:58:01 why not? 2010-09-22T08:58:26 true, i'm just saying in the initial position of that game that it is suboptimal to all in attack the opponent 2010-09-22T08:58:32 hamstah: because it's invoked for every turn seperately 2010-09-22T08:58:38 boegel ? 2010-09-22T08:58:42 it's a pipe 2010-09-22T08:59:01 the process waits for input 2010-09-22T08:59:09 but still the same process 2010-09-22T08:59:30 hamstah: hmm, well, I never looked at the bare bones of the starter packages 2010-09-22T08:59:42 hamstah: so, there's a way to pass info across turns? 2010-09-22T08:59:42 hehe 2010-09-22T08:59:47 of course. 2010-09-22T08:59:48 :) 2010-09-22T09:00:04 it's kept in memory 2010-09-22T09:00:11 k 2010-09-22T09:00:17 I should look into that probably 2010-09-22T09:00:18 :D 2010-09-22T09:00:21 drewrobb: i think it is retty close to optimal for shortrange 2010-09-22T09:00:23 probably. 2010-09-22T09:00:35 pretty effective for short range yeah 2010-09-22T09:00:42 look at ragebot when it's close to you 2010-09-22T09:00:45 because it maximizes the impact of using the budget for a second planet 2010-09-22T09:00:53 although, I haven't felt a need for it yet 2010-09-22T09:01:13 or capturing the opponents home in the first few steps which is pretty much a guaranteed win 2010-09-22T09:01:27 i have a strategy that dominates, immediately capture the 7 2010-09-22T09:01:38 drewrobb: agreed on the 7 2010-09-22T09:02:11 but not expending the units on attacking the opponents home means that the 7 is less of a tiebreaker then it can be 2010-09-22T09:02:14 so if both capture the 7, maybe it is optimal to stale mate again 2010-09-22T09:03:03 drewrobb: i think the game of two optimal bots is a stalemate, so i guess the right stalemate is optimal :) 2010-09-22T09:03:28 I wouldn't exactly call my bot optimal right now :P 2010-09-22T09:03:39 bartwe: not necessarily, the optimal might be a mixed equilibrium 2010-09-22T09:03:52 ut keeping the number of units down at the stalkemate means more influence of the remote planets 2010-09-22T09:04:29 hmm that is interesting 2010-09-22T09:04:42 so sparse territory favors low ship numbers 2010-09-22T09:05:38 i think it is more about deflating the joker of the stalemated forces 2010-09-22T09:05:57 increasing control at zero EV 2010-09-22T09:10:06 *** iouri_ has joined #aichallenge 2010-09-22T09:10:11 this game makes me think that i need to plan ahead with necessary defense ships 2010-09-22T09:10:17 http://72.44.46.68/canvas?game_id=99687 2010-09-22T09:10:32 i don't need almost 200 ships on that planet you are attacking 2010-09-22T09:11:33 drewrobb: I don't see why you're attacking the hugely expensive smallish planet either... 2010-09-22T09:11:45 you are blue 2010-09-22T09:12:06 or did you mean another planet? 2010-09-22T09:12:10 drewrobb: why are you softening neutrals instead of outright taking them ? 2010-09-22T09:12:19 drewrobb: ah, no, right, my mistake :) 2010-09-22T09:12:34 i know that you can't get the neutrals 2010-09-22T09:12:35 bartwe: to make it easier for you to win? :P 2010-09-22T09:12:55 softening is only suboptimal if you can't finish the job 2010-09-22T09:12:56 drewrobb: still, why send lower forces ? 2010-09-22T09:13:17 sending a larger force later means longer flexibility, less predictability 2010-09-22T09:13:25 meh, currently too hard to implement 2010-09-22T09:13:31 and if your prediction changes you have no investment 2010-09-22T09:13:48 i should hold the closer planet until needed 2010-09-22T09:14:46 that still could backfire if a higher priority task comes about 2010-09-22T09:15:10 probably better overall though 2010-09-22T09:18:39 cool, tasks with priotity 2010-09-22T09:18:43 sounds good 2010-09-22T09:20:37 in fact, that sounds very good, i should implement that and rule the server :) 2010-09-22T09:20:40 *** aerique has quit IRC (Quit: ...) 2010-09-22T09:22:04 this place is great to get new ideas :) 2010-09-22T09:24:10 *** sinan has quit IRC (Quit: WeeChat 0.2.6.3) 2010-09-22T09:25:38 *** sigh has quit IRC (Remote host closed the connection) 2010-09-22T09:26:02 *** Blkt has joined #aichallenge 2010-09-22T09:27:31 *** sigh has joined #aichallenge 2010-09-22T09:28:42 *** otherAntimatroid has left #aichallenge 2010-09-22T09:29:26 *** niweber has joined #aichallenge 2010-09-22T09:31:56 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-09-22T09:33:22 seems like the official server is again not handing out many games? 2010-09-22T09:36:36 *** Blkt has quit IRC (Remote host closed the connection) 2010-09-22T09:46:38 *** sleep_master is now known as graphitemaster 2010-09-22T09:48:05 *** danielvf has quit IRC (Ping timeout: 252 seconds) 2010-09-22T09:48:39 *** kcn1700 has joined #aichallenge 2010-09-22T09:56:41 Top 10 players: kcm1700(3547), McLeopold(3504), adumlah(3432), dmj111(3400), vinx33(3330), EBraun(3322), paulus(3309), montanalow(3298), wagstaff(3283), Tracer(3222) 2010-09-22T09:58:29 *** Blkt has joined #aichallenge 2010-09-22T10:00:13 antimatroid: do you think you'll be uploading a new bot soon ? 2010-09-22T10:02:11 it's playing around on tcp now, but still buggy as fuck 2010-09-22T10:02:23 what about you? 2010-09-22T10:03:50 *** danielvf has joined #aichallenge 2010-09-22T10:03:55 he keeps waiting on the first turn :| 2010-09-22T10:04:15 boegel: http://ai-contest.com/game_stats.php 2010-09-22T10:04:19 yeah, I think I might - it's only trivially better than the last one, but since it takes several days to get a good number of games on the official server I figure I should upload now so I have enough by the weekend 2010-09-22T10:04:30 wish the official server could give out games more regularly 2010-09-22T10:05:03 saying that, I've had 4 games in the last 45 minutes (official server) 2010-09-22T10:05:28 Zaphus: Within the next week, we should have at least another server online playing games. 2010-09-22T10:06:10 oh, i'm not uploading this to the main site, it's far too bad 2010-09-22T10:06:11 danielvf: that will be awesome... I think the imbalance between those that get lucky and get lots of games, and those that dont, kind of sucks. I've been on both ends. 2010-09-22T10:07:21 antimatroid: ok, thats fair enough. I wonder if I should leave mine alone to see where it settles on the main site, it's still not at the point where it's settled (but I think it's close, after 3 days) 2010-09-22T10:08:05 Zaphus: Yes, more games will rock. 2010-09-22T10:08:10 danielvf: <3 2010-09-22T10:09:04 I have just had games at these times: 2010-09-22T10:09:05 Sep 22 08:51:03 AM Sep 22 08:31:41 AM Sep 22 08:10:08 AM Sep 22 08:05:27 AM 2010-09-22T10:09:14 :-) 2010-09-22T10:10:29 Argh! (Yes I read the forums) I am loosing this Dual Pentium Laptop Friday, and My Early Mac Mini on 10.5 won't run Java 1.6 (Core, not Core2 ?) 2010-09-22T10:19:56 *** mightybyte has joined #aichallenge 2010-09-22T10:20:41 wow, just got another game on the official server - thats 5 in the last hour 2010-09-22T10:23:11 damn you :P 2010-09-22T10:23:32 i've had 15 in the last few days 2010-09-22T10:23:45 sigh: got you for once 2010-09-22T10:23:49 it's getting less buggy then :) 2010-09-22T10:28:32 ok, uploaded... threw away my top-40 spot for a few days :-) 2010-09-22T10:28:35 *** Cosmo` has joined #aichallenge 2010-09-22T10:29:44 :) 2010-09-22T10:31:51 cya later, need to sleep before work! 2010-09-22T10:32:40 *** Cosmo` has quit IRC (Remote host closed the connection) 2010-09-22T10:33:55 *** Zaphus has quit IRC (Quit: Page closed) 2010-09-22T10:33:58 *** Cosmo` has joined #aichallenge 2010-09-22T10:36:55 weird, gaining ranks on benzedrine due to bots getting removed from the ranking 2010-09-22T10:38:58 bartwe: the official ranking? 2010-09-22T10:39:08 bartwe: are they taking out "cheaters"? 2010-09-22T10:39:30 jmcarthur: would it really have helped that much? after like 10 levels it was hard to really penetrate the game tree much further when it was big <-- it would certainly have helped me, but i'm not doing game tree stuff 2010-09-22T10:42:03 boegel: no benz 2010-09-22T10:43:51 with tron i was talking about? 2010-09-22T10:45:11 oh, i don't remember the context then 2010-09-22T10:45:19 you mean the multithreading thing? 2010-09-22T10:45:32 i wasn't sure how much benefit i would get from it either, but it was measurably better 2010-09-22T10:45:38 *** hellman has joined #aichallenge 2010-09-22T10:45:57 i didn't count the number iterations my algorithm got to run per turn though 2010-09-22T10:48:01 do you know how Dred_Furst is? 2010-09-22T10:48:09 *who 2010-09-22T10:53:22 bartwe: I think bots that haven't played in 24 hours are removed 2010-09-22T10:54:20 retybok: i think that games that are older then 24 hours aren't taken into account for the elo calculation, so there is no evidence for a score, i think 2010-09-22T10:54:32 hmz, shouldn't that make the ranking unstable at the end... 2010-09-22T10:54:41 notsure, i should ask dhartmei about that 2010-09-22T10:55:20 if you look at the "Last" column in the rankings, it's never over 24 hours 2010-09-22T10:55:35 other than that, I don't know anything :) 2010-09-22T10:56:42 Top 10 players: kcm1700(3558), McLeopold(3516), adumlah(3413), dmj111(3412), wagstaff(3348), EBraun(3321), montanalow(3309), vinx33(3308), paulus(3305), Tracer(3237) 2010-09-22T11:02:43 any idea how point are calculated? 2010-09-22T11:04:01 I lost 200 in the rankings by winning twice 2010-09-22T11:06:05 hamstah: you shoudln't be winning then, should you? 2010-09-22T11:07:58 I had 1 game with my new bot, ranked 531, then for a while I had "ranking N/A", then 2 victories and now rank is 784. 2010-09-22T11:08:13 (first game was a victory as well) 2010-09-22T11:10:49 what's the local time for the official server? 2010-09-22T11:10:55 PDT? 2010-09-22T11:11:03 *** Theudas has quit IRC (Quit: Page closed) 2010-09-22T11:11:13 hmm, no 2010-09-22T11:12:05 CDT? 2010-09-22T11:15:46 *** sati has joined #aichallenge 2010-09-22T11:16:13 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2010-09-22T11:20:18 *** Utkarsh has joined #aichallenge 2010-09-22T11:22:00 hamstah: the points are a bayesian estimate of your Elo (google it) score. your score can go down for reasons other than mere wins or losses. for example, if you win against an opponent your score will usually go up, but if past opponents drop in score your score will also drop because then yours wins aren't worth as much and your losses spell an even worse fate 2010-09-22T11:22:58 oh ok. 2010-09-22T11:23:08 but 2010-09-22T11:23:20 hamstah: that is, if you lose to a bot with an estimated score of 2000 then your score is probably (but not necessarily) under 2000, but then if that bot's score drops then your score is probably, but still not necessarily, lower than initially estimated 2010-09-22T11:23:20 nevermind. 2010-09-22T11:23:36 I see 2010-09-22T11:23:55 complicated, but it makes sense 2010-09-22T11:24:03 oh i guess it's key to point out that your score is being recalculated even if you don't play games 2010-09-22T11:24:09 but since the bots are randomly chosen, isnt there a risk to drop points when playing with a player below you? 2010-09-22T11:24:15 nah 2010-09-22T11:24:20 first, they aren't entirely random 2010-09-22T11:24:38 second, your confidence interval tightens as you play more games, which makes your mean score less volatile 2010-09-22T11:25:01 ok 2010-09-22T11:25:03 third, you will only lose that many points if you are losing games to low ranking bots 2010-09-22T11:25:03 dhartmei_: Are you there? 2010-09-22T11:25:08 just have to wait for more games then 2010-09-22T11:25:11 in which case you deserve to lose those points anyway 2010-09-22T11:25:38 winning games against low ranking bots doesn't help you that much, though 2010-09-22T11:25:47 so in fact my first 531 was overestimated regarding the strength of the bots I won against 2010-09-22T11:25:52 yeah 2010-09-22T11:25:55 the scheduler tries to match bots with similar ranks, but also still adds in some variety 2010-09-22T11:26:01 good 2010-09-22T11:26:02 that is probably true 2010-09-22T11:26:32 i wish the rankings would show confidence intervals 2010-09-22T11:26:38 i think it would clear up a lot of questions about this 2010-09-22T11:26:42 yes 2010-09-22T11:27:07 would be good to have an history of ranking/points as well. 2010-09-22T11:27:24 *** kozlovsky has quit IRC (Quit: Page closed) 2010-09-22T11:27:30 we had that in the tron contest 2010-09-22T11:27:33 not sure what happened to that 2010-09-22T11:27:42 we had it in the form of a graph, that is 2010-09-22T11:28:19 and kcm made this a while ago (just a few bots over a small-ish duration) http://kcm1700.snucse.org/new/rankgraph.png 2010-09-22T11:29:07 actually that is different from when i saw it 2010-09-22T11:29:10 i guess he updated it 2010-09-22T11:29:17 *** sati has quit IRC (Ping timeout: 272 seconds) 2010-09-22T11:30:42 ok 2010-09-22T11:30:49 thanks for the explanation 2010-09-22T11:31:01 ooh very cool 2010-09-22T11:31:17 maybe someone want to do that for enzedrine ? 2010-09-22T11:31:43 i want freaking game tarballs for offline analysis 2010-09-22T11:32:17 can't you get the dump from the visualizer| 2010-09-22T11:32:21 ? 2010-09-22T11:32:25 what do you need more? 2010-09-22T11:32:26 for a single game 2010-09-22T11:32:31 and sure, you can write a scraper 2010-09-22T11:32:35 Can i include a precompiled .so file that i dlopen and run? 2010-09-22T11:32:40 but we could benefit from a gigantic tarball 2010-09-22T11:32:42 anoek, not sure :) 2010-09-22T11:32:47 would be nicer on the server and nicer for us 2010-09-22T11:32:50 the visualizer fetches the game from a php page which u can pull 2010-09-22T11:32:50 jmcarthur, true 2010-09-22T11:33:06 pgpaskar_: yeah, but scraping sucks for everyone 2010-09-22T11:33:12 dhartmei_ promised to provide it for benzedrine, but I don't know the end of this history 2010-09-22T11:33:25 there are also eventual plans for the official server to offer the same 2010-09-22T11:33:37 but i don't know how far that has gone yet 2010-09-22T11:33:43 for next contest probably 2010-09-22T11:33:49 no i think this contest 2010-09-22T11:34:17 i suggested that it be a torrent to further save bandwidth 2010-09-22T11:34:17 what's benzedrine? 2010-09-22T11:34:27 hamstah: unofficial server for testing 2010-09-22T11:34:27 the one you can dump replays? 2010-09-22T11:34:34 jmcarthur: dhartmei hinted to dumps 2010-09-22T11:34:48 hamstah: you connect your bot to it remotely via a tcp client and play games against other bots from your own machine 2010-09-22T11:34:57 hamstah: nice for debugging 2010-09-22T11:35:18 k 2010-09-22T11:35:21 *** mega1 has quit IRC (Ping timeout: 252 seconds) 2010-09-22T11:35:22 yeah I tried that 2010-09-22T11:35:24 bartwe: that'd be nice, but i'd still rather see them for the official server 2010-09-22T11:35:28 couldnt compile the server myself 2010-09-22T11:35:37 hamstah: you don't need the server, just the client 2010-09-22T11:35:53 yeah I know 2010-09-22T11:35:58 i wanted to host server myself :) 2010-09-22T11:36:04 ah 2010-09-22T11:36:12 to play with people I know. 2010-09-22T11:36:22 I ended up doing a netcat script to do the same 2010-09-22T11:36:45 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-09-22T11:36:56 http://72.44.46.68/canvas?game_id=101641 2010-09-22T11:36:58 damn close match 2010-09-22T11:37:04 *** Naktibalda has joined #aichallenge 2010-09-22T11:37:07 ugh why do people make webpages with a black background and dark grey text 2010-09-22T11:37:17 a capital offense against web design 2010-09-22T11:37:28 pgpaskar_: http://lab.arc90.com/experiments/readability/ 2010-09-22T11:37:37 amazing 2010-09-22T11:42:16 bot seems better again now i think 2010-09-22T11:43:01 *** ratatata has joined #aichallenge 2010-09-22T11:44:15 *** delt0r has joined #aichallenge 2010-09-22T11:51:53 *** hakand has joined #aichallenge 2010-09-22T11:53:23 *** sati has joined #aichallenge 2010-09-22T11:53:41 contestbot: rankings 2010-09-22T11:53:42 hakand: Top 10 players: kcm1700(3589), McLeopold(3539), dmj111(3461), adumlah(3437), wagstaff(3413), EBraun(3367), paulus(3342), montanalow(3297), vinx33(3288), goldman(3245) 2010-09-22T11:54:19 *** antimatroid1 has joined #aichallenge 2010-09-22T11:54:48 contestbot: rainbow H O L L A H O L L A 2010-09-22T11:54:48 pgpaskar_: H O L L A H O L L A 2010-09-22T11:54:51 lmao 2010-09-22T11:55:15 *** antimatroid has quit IRC (Ping timeout: 272 seconds) 2010-09-22T11:55:17 hmm kcm is getting ahead and sighbot is out of sight 2010-09-22T11:56:42 Top 10 players: kcm1700(3591), McLeopold(3542), dmj111(3464), adumlah(3439), wagstaff(3416), EBraun(3370), paulus(3345), montanalow(3300), vinx33(3291), pierrejoel(3249) 2010-09-22T11:58:12 contestbot: rainbow R A I N B O W 2010-09-22T11:58:12 antimatroid1: R A I N B O W 2010-09-22T12:00:08 *** Blkt has quit IRC (Ping timeout: 265 seconds) 2010-09-22T12:04:26 *** kcn1700 has quit IRC (Quit: ¡¬(^o^)£¯) 2010-09-22T12:05:13 *** Ceridan has joined #aichallenge 2010-09-22T12:05:26 *** Ceridan_ has quit IRC (Ping timeout: 240 seconds) 2010-09-22T12:10:10 *** Ceridan has quit IRC (Ping timeout: 276 seconds) 2010-09-22T12:12:48 *** herschel has joined #aichallenge 2010-09-22T12:21:00 *** sati has quit IRC (Ping timeout: 252 seconds) 2010-09-22T12:24:13 *** Kastra has joined #aichallenge 2010-09-22T12:30:07 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-22T12:30:36 *** danielvf has joined #aichallenge 2010-09-22T12:35:17 *** boegel has quit IRC (Quit: Leaving) 2010-09-22T12:36:59 *** Bobng has joined #aichallenge 2010-09-22T12:37:01 quiet, quiet 2010-09-22T12:39:46 coding :) 2010-09-22T12:43:21 *** Kastra has quit IRC (Read error: Connection reset by peer) 2010-09-22T12:45:01 *** sati has joined #aichallenge 2010-09-22T12:46:22 *** Kastra has joined #aichallenge 2010-09-22T12:47:02 *** sati is now known as sinan 2010-09-22T12:47:29 *** maiden has joined #aichallenge 2010-09-22T12:47:32 hello 2010-09-22T12:47:32 *** sinan is now known as Guest14073 2010-09-22T12:48:24 any one here that knows anything about ssh RSA login? 2010-09-22T12:49:05 you could find a better place to ask your question :) 2010-09-22T12:49:26 *** nigg_ has joined #aichallenge 2010-09-22T12:51:51 *** retybok has quit IRC (Quit: leaving) 2010-09-22T12:51:56 *** madn has joined #aichallenge 2010-09-22T12:52:30 *** Bobng has quit IRC (Ping timeout: 265 seconds) 2010-09-22T12:53:00 *** hakand has quit IRC (Ping timeout: 240 seconds) 2010-09-22T12:53:40 *** perror has quit IRC (Quit: Bye all !) 2010-09-22T12:56:12 *** Guest14073 has quit IRC (Ping timeout: 252 seconds) 2010-09-22T12:56:43 Top 10 players: kcm1700(3587), McLeopold(3542), dmj111(3464), adumlah(3446), wagstaff(3416), EBraun(3366), paulus(3344), montanalow(3298), vinx33(3284), goldman(3263) 2010-09-22T12:57:11 maiden ssh logins occur when someone attempts to telnet into a linux box but does it too loudly, what are you using, GNUquiet? 2010-09-22T12:57:35 maiden: I just posted an answer in the forums 2010-09-22T12:57:40 had the same error myself 2010-09-22T12:57:47 oki 2010-09-22T12:58:55 oops, I guess I didn't read your whole message carefully :( 2010-09-22T12:58:56 as i wrote. i have tested to run chmod 600 on the filö 2010-09-22T12:59:11 right :) 2010-09-22T13:00:09 make sure /etc/ssh/sshd_config has RSAAuthentication yes and PubkeyAuthentication yes 2010-09-22T13:00:56 also I'm not sure if it was necessary but I manually changed ownership and permissions on the jailusers ~/.ssh directory and authorized_keys file 2010-09-22T13:01:04 hahaha 2010-09-22T13:01:16 are you reading my post at all? 2010-09-22T13:01:45 slowly :) 2010-09-22T13:01:51 thanks for trying to help me :) 2010-09-22T13:02:18 the group ownership was still root for me 2010-09-22T13:02:49 on the ~/.ssh directory and user and group were root on the file 2010-09-22T13:02:49 *** Kastra has quit IRC (Quit: Nettalk6 - www.ntalk.de) 2010-09-22T13:02:56 ok and it worked? 2010-09-22T13:03:10 yeah, it works for me now 2010-09-22T13:03:36 *** Blkt has joined #aichallenge 2010-09-22T13:04:13 so i shuld chown the .ssh dirs to the correct users.. hmm lets check 2010-09-22T13:04:22 *** McLeopold has joined #aichallenge 2010-09-22T13:04:44 chown :jailuser? /home/jailuser?/.ssh 2010-09-22T13:05:05 and chown jailuser?:jailuser? /home/jailuser?/.ssh/authorized_keys 2010-09-22T13:05:27 drwx------ 2 jailuser1 root 4096 2010-09-20 17:53 .ssh 2010-09-22T13:06:17 right, the group is root. I changed that although maybe it doesn't have to be 2010-09-22T13:07:34 *** dhartmei has joined #aichallenge 2010-09-22T13:07:45 great it works! 2010-09-22T13:14:38 great 2010-09-22T13:14:51 want to post that in the forums for the next person 2010-09-22T13:22:58 *** nigg_ has quit IRC (Quit: Leaving) 2010-09-22T13:23:33 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-09-22T13:26:59 *** niweber has quit IRC (Remote host closed the connection) 2010-09-22T13:27:18 *** hellman has quit IRC (Remote host closed the connection) 2010-09-22T13:30:18 *** hamstah has quit IRC (Remote host closed the connection) 2010-09-22T13:31:40 *** Alystair has joined #aichallenge 2010-09-22T13:37:32 hmmm i think that'll do for tonight 2010-09-22T13:45:29 *** HakanD has joined #aichallenge 2010-09-22T13:49:07 *** Naktibalda has joined #aichallenge 2010-09-22T13:49:14 drewrobb: got your new bot finally 2010-09-22T13:51:47 *** Bobng has joined #aichallenge 2010-09-22T13:56:44 Top 10 players: kcm1700(3584), McLeopold(3549), adumlah(3457), dmj111(3447), EBraun(3387), paulus(3368), wagstaff(3365), drewrobb(3310), montanalow(3292), Arekku(3273) 2010-09-22T14:05:36 *** HakanD has quit IRC (Quit: Colloquy for iPhone - http://colloquy.mobi) 2010-09-22T14:06:55 *** Palmik has quit IRC (Remote host closed the connection) 2010-09-22T14:08:40 cars are the worst FWIW 2010-09-22T14:13:32 *** choas has joined #aichallenge 2010-09-22T14:13:53 hmm 2010-09-22T14:14:11 does PlayGame.jar's max_turn_time param accept it in milliseconds? 2010-09-22T14:14:15 so 1000 would be a second 2010-09-22T14:14:33 *** justin_pdx has joined #aichallenge 2010-09-22T14:14:50 lol i lost to rage bot 2010-09-22T14:15:50 antimatroid1: you exceded version PI 2010-09-22T14:16:11 man, everyone is starting to beat up on my bot 2010-09-22T14:16:17 *** Ceridan has joined #aichallenge 2010-09-22T14:16:47 doh, my code worked on my machine but broke when i uploaded it 8( 2010-09-22T14:17:03 yeah, the series diverged sorry 2010-09-22T14:17:11 and you just beat me :( 2010-09-22T14:17:17 hmm, this is really strange. pick a couple of revisions of your bot, run 1000 games among random pairs on random maps, then run bayeselo. i get wildy different ratings each time. 2010-09-22T14:18:10 my bot? 2010-09-22T14:18:14 either there's a bug in the scripts, or the random maps have a huge influence, or the revisions are too close. 2010-09-22T14:18:20 i did a lot of messing around tonight, and broke it lots of times 2010-09-22T14:18:58 i see it with my bot, basically i can't tell which revision is the best... 2010-09-22T14:19:27 run one game per bot pair + map 2010-09-22T14:19:28 i spent 8 hours on my bot yesterday only to see a 10% win improvement 2010-09-22T14:19:41 my original 3.1415 was good, but then i ruined elo score for that testing before 2010-09-22T14:19:55 expect new 4.11 to be half decent, not sure if it'll be as good as 3.1415 proper 2010-09-22T14:20:06 i compared to a full tournament, play each bot once against each other on each of 100 random maps, then reverse positions. still random ratings. 2010-09-22T14:20:33 if a re-generate the 100 maps randomly, ratings change wildly 2010-09-22T14:22:05 try it, run your last two revisions against each other, on 1000 random maps. repeat that a couple of times, compare ratings. do you get consisten t results? 2010-09-22T14:23:31 why is ELO better to use in this sort of benchmarking than W/L record? 2010-09-22T14:23:54 oh, are you testing >2 bots? 2010-09-22T14:24:11 in the first case, yes. for 2, elo is the same as w/l record, i assume. 2010-09-22T14:25:06 testing against your own bots seems like imbreeding 2010-09-22T14:26:17 bleh, bot seems to suck 2010-09-22T14:26:21 *** amstan has joined #aichallenge 2010-09-22T14:26:21 *** ChanServ sets mode: +o amstan 2010-09-22T14:26:22 but if the tournament algorithm is good, and the number of games is high enough, the ratings should be stable, no? 2010-09-22T14:26:22 oh well i shall adress it on the weekend 2010-09-22T14:26:24 night 2010-09-22T14:26:59 night antimatroid1 2010-09-22T14:27:10 night 2010-09-22T14:28:24 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-09-22T14:28:31 *** justin_pdx has joined #aichallenge 2010-09-22T14:33:46 *** Cosmo` has quit IRC (Quit: Cosmo`) 2010-09-22T14:35:34 *** maiden has quit IRC (Ping timeout: 252 seconds) 2010-09-22T14:40:06 *** Zannick has quit IRC (Ping timeout: 240 seconds) 2010-09-22T14:48:42 *** icky has joined #aichallenge 2010-09-22T14:49:57 If anyone's interested, I've improved my Java client (now accepts passwords and shows your bot's stderr): http://ai-contest.com/forum/viewtopic.php?p=3836#p3836 2010-09-22T14:50:10 *** smf68 has joined #aichallenge 2010-09-22T14:52:48 dhartmei_: how random are we talking here? 2010-09-22T14:53:02 running the map generator from the repository 2010-09-22T14:53:09 i mean the results 2010-09-22T14:53:37 are they even ballpark-consistent? 2010-09-22T14:53:41 entries jumping +- 30% in ranks 2010-09-22T14:53:53 agaisnt yourself? 2010-09-22T14:53:54 top rank changing entirely 2010-09-22T14:54:10 how many versions of your bot in a tournament? just two? 2010-09-22T14:54:16 there's nothing random in your bots is there? 2010-09-22T14:54:23 i tried 2, 5, and 7 2010-09-22T14:54:29 no, no randomness in the bot 2010-09-22T14:54:41 okay, once you have as many as 7 then the results should certainly be stabilizing a bit 2010-09-22T14:55:15 i would say that 2 or 3 isn't enough to constrain the confidence intervals at all (except perhaps for the middle bot in the 3-player case) 2010-09-22T14:56:01 dhartmei_: by "in ranks" do you mean actual numbering ranks or do you mean elo scores? 2010-09-22T14:56:45 Top 10 players: kcm1700(3552), McLeopold(3543), adumlah(3447), dmj111(3439), EBraun(3385), wagstaff(3370), paulus(3348), vinx33(3334), drewrobb(3317), montanalow(3292) 2010-09-22T14:58:36 both, in one tournament, revision A would be rank 1 with 300+ elo, revision B rank 4 with 200+ elo. next tournament, they could be reversed, or revision C would be rank 1, etc. 2010-09-22T14:58:44 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-22T14:59:00 dhartmei_: what about the differences in elo between successive players? that is the important thing about elo, after all, not absolute scores 2010-09-22T14:59:01 *** smf68 has quit IRC (Ping timeout: 240 seconds) 2010-09-22T14:59:11 weird that they actually change orders though 2010-09-22T14:59:21 is this with fairly tight confidence bounds? 2010-09-22T14:59:26 *** danielvf has joined #aichallenge 2010-09-22T14:59:33 about 10 points per rank 2010-09-22T14:59:34 *** Zannick has joined #aichallenge 2010-09-22T14:59:51 is that consistent-ish? 2010-09-22T14:59:57 i will retry several runs without re-generating maps between runs, see if at least that's static, it should be 2010-09-22T15:01:25 does bayeselo use the order of the results in some way? 2010-09-22T15:01:37 i would think it shouldn't 2010-09-22T15:02:03 could shuffle the results from a single tournament and see if bayeselo gives the same output 2010-09-22T15:02:10 good idea 2010-09-22T15:02:57 dhartmei_: are you running benzedrine on apache? if so, I'd like to see the setup 2010-09-22T15:03:07 I'd like setup a second tcp server 2010-09-22T15:03:16 *** icky has quit IRC (Ping timeout: 252 seconds) 2010-09-22T15:03:23 the apache directives that is 2010-09-22T15:03:40 it's thttpd, which is much simpler, one config file of about 5 lines... 2010-09-22T15:03:54 there's linux packages on the ec2 host 2010-09-22T15:04:00 okay 2010-09-22T15:04:19 *** smf68 has joined #aichallenge 2010-09-22T15:05:06 is you setup in the zip file? 2010-09-22T15:06:10 dhartmei_: the thttpd config file that is 2010-09-22T15:06:10 http://72.44.46.68/setup.txt 2010-09-22T15:06:14 sweet 2010-09-22T15:06:16 not yet 2010-09-22T15:06:32 server still running as root, it's not my box :) 2010-09-22T15:07:08 short term goal is a server the generates non-semetrical maps on the fly 2010-09-22T15:07:21 *** SophieB is now known as zz_SophieB 2010-09-22T15:07:21 thttpd will not server files if they are not world-readable, but cgis (running as user thttpd, not root) can read them, chroot /var/www 2010-09-22T15:07:27 and anything else crazy jmcarthur wants 2010-09-22T15:07:31 lol 2010-09-22T15:07:33 lol 2010-09-22T15:07:37 rock on 2010-09-22T15:07:43 the pursuit of science 2010-09-22T15:07:48 i mainly am interested in seeing degenerate maps 2010-09-22T15:08:00 to analyze bot behaviors more easily 2010-09-22T15:08:12 all this code will come in handy went the buggers finally find us 2010-09-22T15:08:13 but they can take many forms 2010-09-22T15:09:02 I'm going to attempt to make the growth potential from to points on the map the same, but not exact 2010-09-22T15:09:40 you know, buggers, from the ender series? 2010-09-22T15:10:14 McLeopold: yes 2010-09-22T15:10:14 the little doctor would certainly make planet wars more interesting 2010-09-22T15:10:35 Those sadly misunderstood people that would kill you 2010-09-22T15:10:35 make planets dissapear? 2010-09-22T15:10:43 *bugs 2010-09-22T15:10:46 *die die die 2010-09-22T15:11:43 danielvf: you'll need to adjust the visualizer to handle maps of distance 1000 now. :P 2010-09-22T15:12:09 Heh 2010-09-22T15:12:42 Just scale the x and y before you pass it off to the visualizer 2010-09-22T15:13:12 Or make a patch for it on git hub. :D 2010-09-22T15:13:34 maybe, first the server 2010-09-22T15:13:48 *** mega1 has joined #aichallenge 2010-09-22T15:13:52 dhartmei_: do you know what your server load is? 2010-09-22T15:15:01 *** zz_SophieB is now known as SophieB 2010-09-22T15:15:44 not high, load <0.30, 40% user time, that's probably the low select timeout, doesn't increase with the number of connections. 2010-09-22T15:16:03 the server does almost nothing (just advances turns), it's all network io bound 2010-09-22T15:16:43 20 concurrent games produce almost 512 kbit/s outgoing 2010-09-22T15:17:16 *** boiserm has joined #aichallenge 2010-09-22T15:17:20 often, a game state includes >500 fleets and bots coml 2010-09-22T15:17:36 plete games as fast as network goes (much less than one second per turn) 2010-09-22T15:18:05 I have some general questions about the starter packages 2010-09-22T15:18:19 Are there any people with access to the main server here? Who might be able to tell me why my bot keeps losing in themiddle of games? 2010-09-22T15:18:23 ah, 40% user time is just bayeselo running every minute for 20+ seconds :) 98% idle without bayeselo 2010-09-22T15:19:07 on http://ai-contest.com/simple_strategy_guide.php in the "Anything to Get Ahead" section, there is no pw.Production() function 2010-09-22T15:19:18 in the java or python starter packages 2010-09-22T15:19:24 no biggie, but confusing 2010-09-22T15:19:55 Also, in the java starter package, the source for all of the example bots is the same as MyBot.java 2010-09-22T15:21:20 Is this a known issue? Already fixed? A mistake on my part? 2010-09-22T15:22:24 did you actually try a diff or md5sum of the examples against MyBot? 2010-09-22T15:22:40 i'm pretty sure most of them are not actually the same 2010-09-22T15:23:20 the executables are definitely different, but it's the sources that seem identical... let me double check the source files to make sure I'm not an idiot 2010-09-22T15:23:27 jmcarthur: do you have access to logs for the games on the official server? 2010-09-22T15:23:37 FlagCapper: i wish 2010-09-22T15:23:43 oh, they are the same in python pack too 2010-09-22T15:23:47 Are you an admin or am I mistaking you for someone? 2010-09-22T15:24:00 *** boegel has joined #aichallenge 2010-09-22T15:24:05 FlagCapper: i'm not an admin, although i have contributed a bit 2010-09-22T15:24:17 FlagCapper: j3camero? 2010-09-22T15:24:18 you may just be assuming because you saw my name somewhere 2010-09-22T15:24:27 Probably 2010-09-22T15:25:08 Yeah, maybe I was thinking of j3camero. 2010-09-22T15:25:11 FlagCapper: likely issues are timeouts, divide by zero, accessing the first element of an empty list 2010-09-22T15:25:19 i would look for those first 2010-09-22T15:25:25 boiserm: RandomBot is trully random 2010-09-22T15:25:29 I already took care of timeouts 2010-09-22T15:25:40 unfortunately there is no method currently for determining why you failed 2010-09-22T15:25:41 The biggest thing though is that the bot works just fine on the TCP server. 2010-09-22T15:26:02 there has been a patch waiting admin approval+rollout for a couple weeks now 2010-09-22T15:26:09 But it randomly loses every game on the official one. 2010-09-22T15:26:32 Ok, I'll try to contact one of the admins if I can. 2010-09-22T15:26:47 big difference between official server and tcp server may be python version 2010-09-22T15:26:55 I use C++ 2010-09-22T15:27:09 right 2010-09-22T15:27:15 jmcarthur: looks like I'm only partially an idiot. BullyBot and ExpandBot are identical 2010-09-22T15:27:19 if your machine is much more powerful than the server then it could still be timeout 2010-09-22T15:27:33 FlagCapper: I got mixed up with some above stuff. sorry 2010-09-22T15:27:42 No problem 2010-09-22T15:27:53 My machine is pretty weak, and I check for timeouts anyway. 2010-09-22T15:28:16 And I've tested the timeout checker by using an infinite loop and my bot never goes over 1 second. 2010-09-22T15:28:49 boiserm: yeah, that doesn't surprise me 2010-09-22T15:28:52 *** kjeldahl has quit IRC (Remote host closed the connection) 2010-09-22T15:28:59 boiserm: the best example bots are Dual and Rage, for the record 2010-09-22T15:29:20 jmcarthur: alright, thanks 2010-09-22T15:30:05 FlagCapper: 64 or 32 bit? 2010-09-22T15:30:21 not sure if it'll make a difference due to compiling 2010-09-22T15:30:23 32 bit 2010-09-22T15:30:42 But yeah, I don't see why it would make a difference. It's something to consider though 2010-09-22T15:31:06 All my previous bots worked however, so unless they're using a new server... 2010-09-22T15:31:16 Or I added a new feature that's incompatible. 2010-09-22T15:31:30 well, yeah, i'd recommend looking over what you added 2010-09-22T15:32:55 negative fleet size, too big fleet, etc 2010-09-22T15:33:27 If that was an issue, it wouldn't work on the TCP server. 2010-09-22T15:33:36 And I check for those things. 2010-09-22T15:33:41 the tcp server kicks you now for invalid orders 2010-09-22T15:33:50 as does the main server 2010-09-22T15:34:02 Did it ignore them before? 2010-09-22T15:34:15 the main server used to ignore them 2010-09-22T15:34:20 the tcp server used to give warnings 2010-09-22T15:34:23 *** jaspervdj has quit IRC (Quit: leaving) 2010-09-22T15:34:44 *** Naith has joined #aichallenge 2010-09-22T15:35:03 Because I've seen other people lose with and "I" for invalid next to their games, but my losses never have that. 2010-09-22T15:35:07 And they're always complete games. 2010-09-22T15:37:43 uninitialized variables/memory, i noticed that only when i ran on linux the first time, bsd zero-initializes more, it seems. 2010-09-22T15:39:59 *** RainCT has joined #aichallenge 2010-09-22T15:43:46 ...wow. Why didn't I look more closely at RageBot a long time ago? 2010-09-22T15:44:32 *** McLeopold has quit IRC (Ping timeout: 276 seconds) 2010-09-22T15:51:03 boiserm: :) 2010-09-22T15:51:50 I'm seeing lots of time-outs on the TCP server :( 2010-09-22T15:52:25 my guess was that the server is getting a little chunky and slow after all those bagels 2010-09-22T15:55:13 bagels? :) 2010-09-22T15:55:56 *** Guest14073 has joined #aichallenge 2010-09-22T15:56:17 my bot had 1 timeout 46 minutes ago, that's all 2010-09-22T15:56:43 and I was used to timeouting every second game in the old server 2010-09-22T15:56:45 Top 10 players: kcm1700(3547), McLeopold(3520), adumlah(3441), dmj111(3439), EBraun(3387), wagstaff(3370), paulus(3337), vinx33(3327), goldman(3304), montanalow(3295) 2010-09-22T15:58:17 * Naktibalda managed to get 12.333333 in available_ships :( 2010-09-22T15:58:33 *** Accoun has quit IRC () 2010-09-22T15:58:37 it isn't divided by 3 anywhere 2010-09-22T15:58:39 *** aerique has joined #aichallenge 2010-09-22T16:05:17 hi aerique 2010-09-22T16:06:10 ha:) you know bricejillis too ? :) 2010-09-22T16:07:16 If anyone's using Haskell, I made a nice new starting package for it: http://ai-contest.com/forum/viewtopic.php?p=4611#p4611 2010-09-22T16:07:55 anyone who is using doesn't need a starter pack :) 2010-09-22T16:08:27 it matters for new participants only 2010-09-22T16:08:27 not completely true 2010-09-22T16:08:57 *** area has joined #aichallenge 2010-09-22T16:09:10 often people approach things differently, and it's nice to read someone else's code sometimes 2010-09-22T16:10:06 It has some nice features over the other Haskell starter packages (like keeping state between turns and parsing speedups). 2010-09-22T16:10:32 bartwe: bricejillis 2010-09-22T16:10:41 ? 2010-09-22T16:11:28 hi aerique 2010-09-22T16:11:37 *** Ridge has joined #aichallenge 2010-09-22T16:11:50 *** Guest14073 has quit IRC (Quit: WeeChat 0.2.6.3) 2010-09-22T16:12:10 Hi 2010-09-22T16:13:47 Olathe: are you familiar with the PlanetWars starter package on github? 2010-09-22T16:13:55 *** Accoun has joined #aichallenge 2010-09-22T16:14:16 Olathe: "parsing speedups"? 2010-09-22T16:15:34 Blkt: you up and running? 2010-09-22T16:15:44 *** virtualmike has joined #aichallenge 2010-09-22T16:17:46 aerique: speakin of parsing speedups, have you tested how fast the current solution based on parse-number is? 2010-09-22T16:23:27 *** virtualmike has quit IRC (Quit: Page closed) 2010-09-22T16:25:55 mega1: no, but i can't imagine it taking much time since it only used to parse the game state from the server every second 2010-09-22T16:26:09 mega1: never hurts to check it though :) 2010-09-22T16:26:52 *** Zannick has quit IRC (Ping timeout: 252 seconds) 2010-09-22T16:27:08 boegel: Yeah, I saw it after I started my own. 2010-09-22T16:28:16 boegel: which one? - nvm 2010-09-22T16:29:17 boegel: The parsing speedups come mainly from using lazy bytestrings. 2010-09-22T16:29:22 danielvf: http://github.com/jaspervdj/planet-wars-haskell 2010-09-22T16:29:34 Olathe: hmmk, cool 2010-09-22T16:31:12 danielvf: I know the guy who made it, he's a freakin' genius with Haskell and only 19 orso :) 2010-09-22T16:33:00 i'm parsing with strict bytestrings 2010-09-22T16:33:14 mega1: less performance than expected: in a naive bench i can do about 250k parse-numbers per second (intel atom n270 @ 1.6Hhz) 2010-09-22T16:33:47 aerique: how does that compare to READ? 2010-09-22T16:34:17 mega1: how would i use that? 2010-09-22T16:34:46 aerique: as in READ-FROM-STRING? 2010-09-22T16:35:06 mega1: ah :) one moment 2010-09-22T16:35:41 mega1: takes 2.8 secs on the same test 2010-09-22T16:35:51 Olathe: that seems to be a *very* complete starter. thanks for sharing it! 2010-09-22T16:36:05 aerique: 2.8 times slower? 2010-09-22T16:36:37 jmcarthur: No problem :) 2010-09-22T16:36:48 jmcarthur: Is yours the one posted by __init__ ? 2010-09-22T16:36:52 mega1: i'm never sure how to say that but it's: parse-number:read-f-s::1.0:2.8 2010-09-22T16:37:13 mega1: so yeah :) 2010-09-22T16:37:47 *** ermau|work has quit IRC (Quit: ermau|work) 2010-09-22T16:38:07 250k parse-numbers is about 80k fleets per second 2010-09-22T16:38:10 Olathe: no, mine is this one: jake.devio.us/starter/PlanetWars.html 2010-09-22T16:38:18 Olathe: mine is more instructive than usable 2010-09-22T16:38:24 what the highest fleet count observed in the wild? 2010-09-22T16:38:30 *what's 2010-09-22T16:39:23 Olathe: i'm also not even really using much of it aside from the parsing, which i'm planning to rewrite anyway 2010-09-22T16:40:57 Olathe: i mainly did it just to go over the spec 2010-09-22T16:41:15 mega1: perhaps we could rip out most of the error checking from parse-number and use that 2010-09-22T16:41:47 aerique: or write something specialized if needed 2010-09-22T16:41:50 jmcarthur: Ahh, cool :) 2010-09-22T16:41:56 but I'm not sure if it is. 2010-09-22T16:42:48 *** wh1teside has quit IRC (Ping timeout: 272 seconds) 2010-09-22T16:43:26 mega1: "cat map*.txt | wc -l" says 2300 planets in 100 files 2010-09-22T16:43:51 yeah, it's known that planet count is 23 for now 2010-09-22T16:44:00 mega1: oh :) 2010-09-22T16:44:02 *** retybok has joined #aichallenge 2010-09-22T16:44:09 mega1: so that's 22 ships per planet per second max? 2010-09-22T16:44:16 eh, fleets 2010-09-22T16:44:30 time max distance 2010-09-22T16:44:34 *times 2010-09-22T16:44:47 *** seanmceligot has joined #aichallenge 2010-09-22T16:44:50 jmcarthur: Hmm, this might be an opportunity to learn how to write literate Haskell. 2010-09-22T16:44:56 mega1: is that known as well? 2010-09-22T16:45:07 23 * 22 * max_distance 2010-09-22T16:45:18 yes, it's about 32 I think 2010-09-22T16:45:39 mega1: about 16k then 2010-09-22T16:45:49 that's 0.2s then 2010-09-22T16:46:41 I'm sure we can optimize it if needed, but it's not too bad for now. 2010-09-22T16:46:45 average will be much lower, but i still don't like 0.2s 2010-09-22T16:46:52 yeah 2010-09-22T16:46:53 yup 2010-09-22T16:47:27 *** Ridge has quit IRC (Quit: Leaving) 2010-09-22T16:47:35 *** wh1teside has joined #aichallenge 2010-09-22T16:47:47 don't forget i tested on a netbook :) 2010-09-22T16:47:49 aerique: I plan to upload a bot for this evening 2010-09-22T16:47:58 Blkt: cool 2010-09-22T16:48:43 *** Sean_McEligot has joined #aichallenge 2010-09-22T16:49:19 mega1: I don't know about in the wild, but you start with 100 ships each, the most production per planet is 5, there are 23 planets, and the most turns is 200. So, 200 + 5*23*200 = 23200. 2010-09-22T16:49:38 *** seanmceligot has quit IRC (Ping timeout: 252 seconds) 2010-09-22T16:49:50 *** Sean_McEligot is now known as seanmceligot 2010-09-22T16:50:07 is just parsing that slow, or does that insertion into some data structure? 2010-09-22T16:50:27 parsing a couple of ints should be almost io bound... 2010-09-22T16:50:32 that was just parsing 2010-09-22T16:50:37 for floats 2010-09-22T16:51:48 Olathe: that's a good limit for the whole game 2010-09-22T16:52:34 yes, sounds much better than our impossible worst case :) 2010-09-22T16:53:13 when is the 200 turn max gonna be implemented 2010-09-22T16:53:42 *** FlagCapper has quit IRC (Quit: Page closed) 2010-09-22T16:55:11 mega1: just checked parse-number.lisp and there's a lot that can be cut 2010-09-22T16:55:11 herschel: I think it's up now. 2010-09-22T16:55:41 *** FlagCapper has joined #aichallenge 2010-09-22T16:55:47 herschel: For example: http://ai-contest.com/visualizer.php?game_id=4640131 2010-09-22T16:56:47 Top 10 players: kcm1700(3542), McLeopold(3517), adumlah(3447), dmj111(3398), wagstaff(3375), EBraun(3372), vinx33(3356), paulus(3331), goldman(3331), montanalow(3291) 2010-09-22T17:03:36 what are the graphs below each game show on tcp? total ships owned? 2010-09-22T17:04:47 *** RainCT has quit IRC (Remote host closed the connection) 2010-09-22T17:05:15 *** Ceridan has quit IRC () 2010-09-22T17:05:52 yes 2010-09-22T17:07:37 *** sigh has joined #aichallenge 2010-09-22T17:11:45 *** dhartmei has quit IRC (Quit: good night) 2010-09-22T17:15:22 *** EhMatt has joined #aichallenge 2010-09-22T17:16:10 *** Titankiller has joined #aichallenge 2010-09-22T17:18:15 *** seanmceligot has quit IRC (Ping timeout: 252 seconds) 2010-09-22T17:18:53 *** seanmceligot has joined #aichallenge 2010-09-22T17:22:45 *** iouri_ has quit IRC (Quit: Page closed) 2010-09-22T17:25:27 Olathe: literate haskell is dead simple, really 2010-09-22T17:26:18 Olathe: just use bird marks for code and everything is normal (except you unfortunately can't start lines with '#') 2010-09-22T17:26:27 *everything else is 2010-09-22T17:30:41 *** evgenius has joined #aichallenge 2010-09-22T17:31:07 *** Naith has quit IRC (Read error: Connection reset by peer) 2010-09-22T17:32:52 is everything OK with compile notifications? 2010-09-22T17:33:28 evgenius: Yeah. 2010-09-22T17:33:40 evgenius: I got a much-needed compiler error message this morning. 2010-09-22T17:33:51 jmcarthur: Ahh, OK :) 2010-09-22T17:34:25 *** Macuyiko has joined #aichallenge 2010-09-22T17:36:50 Olathe: the only unique thing about my literate haskell is that i used markdown so i could run it through pandoc to generate the html :) 2010-09-22T17:41:31 is the 200-turn limit set in stone yet? 2010-09-22T17:43:04 untill it's physically wired into server, it isn't :) 2010-09-22T17:43:04 nope 2010-09-22T17:45:29 are all maps always 'mirrored'? 2010-09-22T17:45:40 I mean, do both sides get the same opportunities? 2010-09-22T17:45:51 *** seanmceligot has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.10/20100914125854]) 2010-09-22T17:45:59 yes 2010-09-22T17:46:11 Naktibalda: that's done to ensure fairness, right? 2010-09-22T17:46:22 *** SophieB is now known as zz_SophieB 2010-09-22T17:46:53 *** seanmceligot has joined #aichallenge 2010-09-22T17:47:08 they might introduce asymmetric maps later that your bot plays twice, one from each side 2010-09-22T17:47:43 Mathnerd314: ah, right... so not a good idea to take that into account in the strategy then :) 2010-09-22T17:48:46 probably not, though a few mirrorbots have done somewhat well 2010-09-22T17:48:59 *** drewrobb has quit IRC (Read error: Connection reset by peer) 2010-09-22T17:49:32 Mathnerd314: the idea is there to anticipate the moves of the opponent, and try and beat him to it? 2010-09-22T17:49:53 *** Appleman1234 has joined #aichallenge 2010-09-22T17:49:54 no, just see what your opponent did and mirror it 2010-09-22T17:50:06 it was done allready 2010-09-22T17:50:29 yeah, I can imagine 2010-09-22T17:50:55 Mathnerd314: how would just mirroring your opponent lead to a win? you're always one step behind... 2010-09-22T17:51:15 you send more ships, so sometimes you take over more planets 2010-09-22T17:51:25 Mathnerd314: hmm, right 2010-09-22T17:51:29 it's not that great a strategy :-) 2010-09-22T17:51:34 bot sends the same amount of ships 2010-09-22T17:51:39 *** zz_SophieB is now known as SophieB 2010-09-22T17:51:40 doesn't seem like something that would work always though 2010-09-22T17:51:41 err, opponent 2010-09-22T17:52:46 create such bot, it should be easy 2010-09-22T17:53:16 then create another, that can beat mirrot bot, it should be a much harder exercise :) 2010-09-22T17:54:07 *** Zaphus has joined #aichallenge 2010-09-22T17:54:50 * Naktibalda should do it as an exercise 2010-09-22T17:55:03 `/msg nickserv thisisbullshit fordearlinux@gmail.com 2010-09-22T17:55:27 fail :p 2010-09-22T17:55:32 :D 2010-09-22T17:55:50 epic fail 2010-09-22T17:56:01 great password though 2010-09-22T17:56:15 this is going on bash.org 2010-09-22T17:56:46 Top 10 players: kcm1700(3578), McLeopold(3507), adumlah(3445), wagstaff(3413), dmj111(3371), vinx33(3361), EBraun(3353), goldman(3328), paulus(3312), drewrobb(3296) 2010-09-22T17:57:45 http://bash.org/?930469 :) 2010-09-22T17:57:53 *** retybok has quit IRC (Remote host closed the connection) 2010-09-22T17:57:54 (I removed the nicks and stuff, don't worry 2010-09-22T17:58:03 at least it 2010-09-22T17:58:10 is not his gmail password 2010-09-22T17:58:43 *** sigh has quit IRC (Remote host closed the connection) 2010-09-22T17:58:57 it was 2010-09-22T18:00:08 Naktibalda: really? :D 2010-09-22T18:00:15 no 2010-09-22T18:00:28 that would have been truly epic 2010-09-22T18:01:02 u tot so ? 2010-09-22T18:01:13 *** evgenius has left #aichallenge 2010-09-22T18:01:34 Titankiller: erm, what? 2010-09-22T18:02:22 aerique; really tot its my gmail passwrd ? 2010-09-22T18:02:38 *** choas has quit IRC (Quit: leaving) 2010-09-22T18:03:02 no, but i never hurts to check 2010-09-22T18:03:50 heheh 2010-09-22T18:03:55 :-) 2010-09-22T18:03:59 how's everyone doing? 2010-09-22T18:04:31 lol 2010-09-22T18:04:38 doing great. This contest is sucking my life away though :-P 2010-09-22T18:04:45 good, other than overload :p 2010-09-22T18:05:16 amstan, still doing real life stuff, like job interview, though I have a pretty decent design / plan developed 2010-09-22T18:11:29 *** Bobng has quit IRC (Ping timeout: 272 seconds) 2010-09-22T18:11:33 *** n2kra has quit IRC (Ping timeout: 252 seconds) 2010-09-22T18:11:33 *** Zaphus has quit IRC (Quit: Page closed) 2010-09-22T18:11:38 *** aerique has quit IRC (Quit: ...) 2010-09-22T18:19:06 *** mega1 has quit IRC (Ping timeout: 240 seconds) 2010-09-22T18:19:11 *** Appleman1234 has quit IRC (Ping timeout: 240 seconds) 2010-09-22T18:23:36 *** smhanov has quit IRC (Ping timeout: 265 seconds) 2010-09-22T18:24:31 amstan: hoping to get finished with real life stuff so I can continue developing my bot :) 2010-09-22T18:25:32 boegel: what function do you use to generate a name of your bot? uuid() ? 2010-09-22T18:25:53 Naktibalda: no function :) 2010-09-22T18:26:15 Naktibalda: I just use the md5 of my source code, and a simple date stamp 2010-09-22T18:26:29 *** Macuyiko has quit IRC (Ping timeout: 255 seconds) 2010-09-22T18:26:29 Naktibalda: I keep track of old versions of my code (manually) 2010-09-22T18:26:33 ah, right 2010-09-22T18:26:47 I should be using some kind of version control though, but I'm too lazy to look into it 2010-09-22T18:26:52 same here 2010-09-22T18:26:57 :) 2010-09-22T18:27:07 just found a stupid bug in my code, urgh 2010-09-22T18:27:13 and I can't use github, because it would be public :) 2010-09-22T18:27:27 I was rounding distances to the nearest integer, instead of the first larger integer 2010-09-22T18:27:29 stupid bug 2010-09-22T18:27:34 Naktibalda: :D 2010-09-22T18:27:37 Naktibalda: right 2010-09-22T18:28:37 dude, version control is trivial if you don't go with crap like cvs or svn 2010-09-22T18:28:42 no reason not to use it 2010-09-22T18:29:05 1) d/l git, 2) go to where your source is, 3) git init 2010-09-22T18:29:15 *ahem* s/git/darcs/ 2010-09-22T18:29:21 meh 2010-09-22T18:29:38 git/darcs/mercurial are all fine 2010-09-22T18:29:51 git and mercurial beat svn at least 2010-09-22T18:30:01 oh yes they do 2010-09-22T18:30:48 but whenever i use git i always find myself missing real cherry picking 2010-09-22T18:30:55 can you get a virgin copy of ghc HEAD using darcs now? 2010-09-22T18:31:01 hell yeah 2010-09-22T18:31:03 or is it still busted 2010-09-22T18:31:05 in like 2.5 minutes 2010-09-22T18:32:12 and then you can do it in like 2.5 seconds after that because it caches it locally 2010-09-22T18:33:46 the only thing from git i really miss is stash, and an equivalent to that is in the works 2010-09-22T18:34:15 well, and there's another feature i'd love to have that would be a more darcs-like alternative to in-place branching 2010-09-22T18:34:48 which is to be able to toggle patches on and off at will 2010-09-22T18:34:58 without an auxillery repo or patchfile 2010-09-22T18:35:28 although the stash alternative in the works can do a similar thing by explicitly specifying a patchfile at least 2010-09-22T18:42:43 *** madn has quit IRC (Ping timeout: 252 seconds) 2010-09-22T18:45:52 *** jimmerton has joined #aichallenge 2010-09-22T18:56:47 Top 10 players: kcm1700(3573), McLeopold(3514), adumlah(3445), wagstaff(3423), EBraun(3389), dmj111(3381), vinx33(3367), goldman(3338), paulus(3319), drewrobb(3303) 2010-09-22T18:58:30 *** dmj111` has joined #aichallenge 2010-09-22T18:58:45 *** danielvf_ has joined #aichallenge 2010-09-22T18:58:51 *** sigh has joined #aichallenge 2010-09-22T18:59:08 *** smhanov has joined #aichallenge 2010-09-22T19:00:19 *** dmj111 has quit IRC (Ping timeout: 252 seconds) 2010-09-22T19:01:26 *** danielvf has quit IRC (Ping timeout: 240 seconds) 2010-09-22T19:01:26 *** danielvf_ is now known as danielvf 2010-09-22T19:03:07 hi there 2010-09-22T19:03:45 so has drewrobb said how his bot works? 2010-09-22T19:05:25 no secrets revealed yet afaik 2010-09-22T19:05:41 hmm. 2010-09-22T19:05:54 i got hopelessly hooked on minecraft 2010-09-22T19:06:11 and my really smart bot code apparently really sucks. 2010-09-22T19:06:20 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-22T19:06:29 *** danielvf has joined #aichallenge 2010-09-22T19:07:42 a1k0n: story of my past week 2010-09-22T19:07:55 a1k0n: minus the minecraft 2010-09-22T19:07:59 :-) 2010-09-22T19:08:26 boiserm: so.. i've been stuck the last few hours, playing minecraft 2010-09-22T19:08:31 heh 2010-09-22T19:08:34 boiserm: wtf did they put in it? 2010-09-22T19:08:41 it's like drugs 2010-09-22T19:08:42 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-22T19:09:06 a1k0n: What is Minecraft? 2010-09-22T19:09:10 why is it that I'm always pessimistic when I'm seeing my bot fight on TCP, but that it turns out OK on the official server? :) 2010-09-22T19:09:19 *** Zaphus has joined #aichallenge 2010-09-22T19:09:20 i've been meaning to make a 2d browser version of that (more like dwarf fortress) (engine is here: http://a1k0n.net/code/jscroll/b/ ) 2010-09-22T19:09:27 justin_pdx: minecraft.net 2010-09-22T19:09:40 *** danielvf has joined #aichallenge 2010-09-22T19:09:41 Yeah I'm looking at it and watched the video but I still don't get it. 2010-09-22T19:09:42 justin_pdx: this is a good introduction: http://www.youtube.com/watch?v=4bh4EexJO4I&p=E005D335B57338D1&index=1 2010-09-22T19:11:02 *** smellyhippy has quit IRC (Ping timeout: 240 seconds) 2010-09-22T19:13:09 randomly generated worlds, hehe 2010-09-22T19:14:48 floating trees 2010-09-22T19:14:54 it just gets crazier 2010-09-22T19:15:16 Kinda reminds me of A Tale in the Desert. 2010-09-22T19:16:01 whats up with the official server handing out games? I uploaded 9 hours ago and have already had 13 games :-) did somebody fix something or did I get incredibly lucky with the roll of the dice ? 2010-09-22T19:16:33 *** smellyhippy has joined #aichallenge 2010-09-22T19:16:33 *** smellyhippy has joined #aichallenge 2010-09-22T19:18:15 Zaphus: same for me 2010-09-22T19:18:28 Zaphus: it's good, right? 2010-09-22T19:19:19 it was like that for some time.. one game per 40m was the average I always had 2010-09-22T19:21:23 urgh, I hate it when you fix a bug and suddenly your bot loses more than before :( 2010-09-22T19:21:38 ratatata: This time yesterday I was averaging about 180 minutes between games, I like the change. dang, just played another one - lost to sighbot :-( 2010-09-22T19:21:54 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-22T19:22:21 boegel: funny when you make a change and start losing more, but beat some bots you'd been losing against before 2010-09-22T19:22:26 * boegel just played one to and won, w00 2010-09-22T19:22:35 dwchandler: yeah :) 2010-09-22T19:23:04 well, my bot isn't fully functional yet, but I was winning a fair amount of games before *with* the bug, urgh 2010-09-22T19:23:15 I should try and learn from that, probably :) 2010-09-22T19:23:36 Zaphus: sighbot is an old version :P 2010-09-22T19:24:02 *** smhanov has quit IRC (Ping timeout: 265 seconds) 2010-09-22T19:24:20 Zaphus: you're right, my bot played more games too.. I also like the fact that 1 week old bot has been hanging around rank 120-140 for quite some time :) 2010-09-22T19:24:29 sigh: I do have to fix the logic that lets you crush me every time, it's pretty obvious the mistakes it's making, I could at least put up a fight! 2010-09-22T19:25:06 New server rule: If (owner is in IRC chat) then playmoregames 2010-09-22T19:25:40 heh 2010-09-22T19:25:50 I love it how all bots lose a fair amount of games, keeps it interesting 2010-09-22T19:27:06 dwchandler: Yeah, I tested only against my previous bot, thought I had improved, but then uploaded it and ...not so much. 2010-09-22T19:27:50 that's why I like to have a variety of strategies to test against locally 2010-09-22T19:29:14 I run my latest creation in a large round robin before promoting to tcp (or main server) 2010-09-22T19:29:28 sigh: wouldnt want to throw me one of your old bot executables to test against locally would you ? :-) 2010-09-22T19:29:39 :) 2010-09-22T19:29:41 haha 2010-09-22T19:30:05 * sigh hands Zaphus ragebot 2010-09-22T19:30:27 sigh: I haven't invested enough time in this to have much variety in my strategies. 2010-09-22T19:30:32 most are incremental improvements 2010-09-22T19:30:42 my most recent version is the biggest change so far 2010-09-22T19:30:53 but it's too soon to know how it will do 2010-09-22T19:31:11 boiserm: you'll find that sometimes small tweaks of parameters can change the strategy dramatically 2010-09-22T19:31:50 my changes have mostly been incremental too, in terms of code 2010-09-22T19:31:58 sigh: true. I need to work on parameterizing things more, probably. 2010-09-22T19:33:09 for example changing your evaluation function should be one line of code, but results in wildly different strategies 2010-09-22T19:33:42 even if it doesn't do as well as your main bot, it's still good for testing 2010-09-22T19:34:17 indeed 2010-09-22T19:34:19 sigh: good point 2010-09-22T19:34:43 here's a question... after 200 turns, the winner is the one with the most ships on planets or in fleets? 2010-09-22T19:34:44 I had a bad bot, but it was opportunistic about taking over the "home world" 2010-09-22T19:34:49 or just the most ships on planets? 2010-09-22T19:34:52 dwchandler: ragebot? :P 2010-09-22T19:34:59 made a couple of nice wins 2010-09-22T19:35:15 sigh: no, one of my dwc-### bots 2010-09-22T19:35:19 boiserm: ships anywhere, plants or fleets 2010-09-22T19:35:26 sigh: ok, thanks 2010-09-22T19:35:31 I do not think I've ever lost to ragebot, btw 2010-09-22T19:35:41 map30 ? 2010-09-22T19:35:42 which is odd, because my bots are not good 2010-09-22T19:36:41 I still lose to ragebot on a few maps... fixing that should be my next job, because I can see that weakness affecting my tournament results 2010-09-22T19:36:45 sigh: =-O ruh-roh 2010-09-22T19:37:04 *** smhanov has joined #aichallenge 2010-09-22T19:37:08 boiserm: ? 2010-09-22T19:37:24 sigh: my ragebot-based strategy just bit the dust on that map when I ran it against my previous best... 2010-09-22T19:37:30 sigh: me too, I know how to beat it (on all but one) but havent coded it up either - I lose on about 8 of the 100 maps 2010-09-22T19:38:10 *** seanmceligot has joined #aichallenge 2010-09-22T19:38:34 another game played, I'm averaging one every 15 minutes for the last hour and a half 2010-09-22T19:38:37 sigh: yet both my current submission and previous best just got pwnd by RageBot on that same map :-( 2010-09-22T19:39:40 boiserm: yeah... there's a lot of rock-paper-scissors situations 2010-09-22T19:41:11 good night all 2010-09-22T19:41:13 *** ratatata has quit IRC (Quit: Page closed) 2010-09-22T19:50:45 *** danielvf has joined #aichallenge 2010-09-22T19:51:00 boegel: new version inherited the same weakness 2010-09-22T19:51:06 boiserm: ^^ 2010-09-22T19:52:03 eh 49 to 808 2010-09-22T19:53:35 first game was played 12 seconds after submission 2010-09-22T19:54:25 Naktibalda: you're comments are very confusing :P 2010-09-22T19:55:12 *** jesionaj has joined #aichallenge 2010-09-22T19:56:11 *** Zannick has joined #aichallenge 2010-09-22T19:56:48 Top 10 players: kcm1700(3581), McLeopold(3520), adumlah(3455), wagstaff(3436), EBraun(3405), dmj111(3381), vinx33(3365), goldman(3347), paulus(3315), Tracer(3299) 2010-09-22T19:57:11 contestbot top 20 2010-09-22T19:57:12 Zaphus: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2010-09-22T19:57:29 contestbot help 2010-09-22T19:57:29 Zaphus: (help [] []) -- This command gives a useful description of what does. is only necessary if the command is in more than one plugin. 2010-09-22T19:57:37 genius 2010-09-22T19:57:54 you have no gotten 2010-09-22T19:57:57 :/ 2010-09-22T19:58:04 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-09-22T19:59:05 contestbot help top 2010-09-22T19:59:05 boegel: Error: There is no command "top". 2010-09-22T19:59:19 contestbot list 2010-09-22T19:59:19 boegel: AIChallenge, Admin, Alias, BadWords, Channel, ChannelLogger, ChannelStats, Config, Ctcp, Dict, Dunno, Factoids, Filter, Format, Games, Google, Herald, Internet, Later, Limiter, Math, Misc, Network, News, Note, Owner, Plugin, Quote, QuoteGrabs, RSS, Reply, Scheduler, Seen, Services, ShrinkUrl, Status, String, Success, Time, Todo, Topic, URL, Unix, User, Utilities, and Web 2010-09-22T19:59:47 contestbot help AIChallenge 2010-09-22T19:59:47 boegel: Error: There is no command "aichallenge". 2010-09-22T20:00:06 whateva 2010-09-22T20:00:36 *** justin_pdx has joined #aichallenge 2010-09-22T20:00:41 contestbot 2010-09-22T20:00:49 contestbot help top 2010-09-22T20:00:49 Zaphus: Error: There is no command "top". 2010-09-22T20:01:03 how do we find out the contestbot commands to get help on them ?! 2010-09-22T20:01:04 contestbot: rankings 20 2010-09-22T20:01:05 amstan: Top 20 players: kcm1700(3581), McLeopold(3520), adumlah(3455), wagstaff(3436), EBraun(3405), dmj111(3381), vinx33(3365), goldman(3347), paulus(3315), Tracer(3299), montanalow(3290), felixcoto(3288), drewrobb(3277), Loks1982(3214), OneLife(3211), sighbot(3211), 3WorldsCollide(3181), zaphus(3180), exaide(3178), pierrejoel(3173) 2010-09-22T20:01:10 Zaphus: that? 2010-09-22T20:01:25 amstan: yep, you knew the magic incantation 2010-09-22T20:01:47 amstan: just wanted to see my name in lights 2010-09-22T20:01:53 that's the secret 2010-09-22T20:02:00 amstan: help rankings 2010-09-22T20:02:05 amstan: list 2010-09-22T20:02:09 contestbot: help rankings 2010-09-22T20:02:09 amstan: (rankings [how many]) -- Get the top $(how many) players. 2010-09-22T20:02:19 it works! 2010-09-22T20:02:28 contestbot help commands 2010-09-22T20:02:28 Zaphus: (commands takes no arguments) -- Returns a list of the commands offered by the bot. 2010-09-22T20:02:50 commands doesn't give very useful output 2010-09-22T20:02:54 i've tried it 2010-09-22T20:03:15 yeah, theres no way to find out what commands it takes that I can see 2010-09-22T20:03:25 @commands 2010-09-22T20:03:25 sigh: action, add, alert, alias, allyourspace, announce, announce add, announce list, announce remove, any, aol, apply, apropos, at, author, azn, ban add, ban list, ban remove, base, binary, bold, bye, cache, calc, call, capabilities, capability add, capability list, capability remove, capability set, capability setdefault, capability unset, capitalize, change, changename, channel, channels, (5 more messages) 2010-09-22T20:03:49 ^^ that's not so useful for us :P 2010-09-22T20:04:01 contestbot: list aichallenge 2010-09-22T20:04:01 amstan: game, match, and rankings 2010-09-22T20:04:07 ah 2010-09-22T20:04:15 match is not implemented yet 2010-09-22T20:04:21 if only contestbot help told you that :-) 2010-09-22T20:04:21 @help match 2010-09-22T20:04:21 sigh: (match player1 [player2]) -- Get the most recent match of $player1. If $player2 is specified, then the most recent match between $player1 and $player2 2010-09-22T20:04:24 oh 2010-09-22T20:05:13 *** EhMatt has quit IRC (Ping timeout: 252 seconds) 2010-09-22T20:07:58 cool 2010-09-22T20:08:25 none of this "matches more often" goodness for me yet :-/ 2010-09-22T20:09:21 boiserm: it must be broken, it's been 25 minutes since my last match :-) 2010-09-22T20:09:35 ha 2010-09-22T20:09:48 *** mightybyte has quit IRC (Quit: leaving) 2010-09-22T20:09:49 I'm headed for 2 hours pretty soon here, if I'm not mistaken 2010-09-22T20:16:01 boiserm: I'm working on automating bot server creation 2010-09-22T20:16:10 Now that I'm off from work 2010-09-22T20:17:12 danielvf: I'm not sure I understand 2010-09-22T20:17:22 danielvf: but it sounds good :) 2010-09-22T20:17:31 does the official server mark invalid input or time-outs? 2010-09-22T20:17:34 Working on another server to play games. 2010-09-22T20:17:38 http://ai-contest.com/visualizer.php?game_id=4651647 2010-09-22T20:17:49 ^ it says I lost this game, when I clearly didn't 2010-09-22T20:17:52 boegel - it doesn't save it, but it needs too 2010-09-22T20:18:47 boegel: Logs show a timeout on that game 2010-09-22T20:18:50 danielvf: I mean, can one tell what went wrong (time-out, invalid input, ...) 2010-09-22T20:19:12 at this point i would recommend someone make a new patch for it 2010-09-22T20:19:21 danielvf: ok, but there's no way to tell without the logs? 2010-09-22T20:19:23 It's probably a bug in the server - random time-outs happen too often. 2010-09-22T20:19:27 boegel: no 2010-09-22T20:19:36 the error_messages branch is probably too out of date by now 2010-09-22T20:19:40 danielvf: ah, it's the server? 2010-09-22T20:19:52 boegel: Once we have more server's playing games, then we'll get error reporting in. 2010-09-22T20:20:07 Zannick: I can still manually merge 2010-09-22T20:20:16 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-09-22T20:20:18 danielvf: ok fine, but are you saying the timeout isn't the fault of my bot? 2010-09-22T20:20:31 danielvf: i mean, sure, but it needs some modifications to tables 2010-09-22T20:20:37 Zannick: but after getting new server (s) up. 2010-09-22T20:20:41 Zannick: yeah 2010-09-22T20:20:41 yes yes 2010-09-22T20:21:11 boegel: The odds are at least pretty good that it's not your bot's fault. 2010-09-22T20:21:26 danielvf: hmmk, thx 2010-09-22T20:21:29 boegel: Unless you forgot to send a "go" 2010-09-22T20:21:43 *** justin_pdx has joined #aichallenge 2010-09-22T20:21:45 i have been too swamped in stuff to really work on anything here 2010-09-22T20:22:04 danielvf: I haven't seen time-outs before in my bots... 2010-09-22T20:22:24 boegel: rock on then. 2010-09-22T20:22:39 and rejoice 2010-09-22T20:22:47 danielvf: too bad I'm now rnaked 1500 places lower because of that :( 2010-09-22T20:22:54 my friends will laugh at me! :( ;) 2010-09-22T20:23:03 resubmit. :P 2010-09-22T20:24:21 *** tapwater has joined #aichallenge 2010-09-22T20:25:27 danielvf: heh, well, I guess I could 2010-09-22T20:25:48 danielvf: but the server will pick my adversary based on my last ranking, right? 2010-09-22T20:26:00 No, not that smart yet 2010-09-22T20:26:20 It always picks your first adversary from the middle of the pack 2010-09-22T20:26:23 danielvf: ah 2010-09-22T20:26:28 ok, I see 2010-09-22T20:26:39 danielvf: you guys have work to do, chop chop :P 2010-09-22T20:26:44 :D 2010-09-22T20:26:58 i expect my bot to make a massive jump in rating when i submit 2010-09-22T20:27:14 namely, because an old version of copybot is running as me on the official 2010-09-22T20:27:21 and a real bot is in the works 2010-09-22T20:27:43 Zannick: lol 2010-09-22T20:28:19 *** boiserm has quit IRC (Ping timeout: 252 seconds) 2010-09-22T20:29:38 that's more like it: http://ai-contest.com/visualizer.php?game_id=4651840 \o/ 2010-09-22T20:29:51 Zannick: what's your name in the main contest? 2010-09-22T20:30:07 Zannick: I hope you won't be disappointed there... :P 2010-09-22T20:30:31 sigh: bswolf 2010-09-22T20:30:39 boegel: i've seen a few people implement ideas i've had 2010-09-22T20:30:56 besides, testing, testing, testing. 2010-09-22T20:31:00 Zannick: thanks 2010-09-22T20:31:02 before submitting :P 2010-09-22T20:31:03 Zannick: doesn't mean you implemented them correctly :P 2010-09-22T20:31:14 well, of course i haven't/ 2010-09-22T20:31:22 i haven't implemented nearly anything yet 2010-09-22T20:31:24 :P 2010-09-22T20:32:03 the official games are probably not very exciting 2010-09-22T20:32:14 due to being paired with people near me 2010-09-22T20:32:33 and most of them as i'm looking now are the sample bot 2010-09-22T20:33:12 Zannick: if you keep winning them you'll move up fast, my version from last night is now 14th after about 15 games. you'll settle in a place where you win about half the games you play 2010-09-22T20:34:42 sounds good 2010-09-22T20:35:16 *** jorge__ has joined #aichallenge 2010-09-22T20:35:45 hi 2010-09-22T20:36:02 pacotabaco: hi? are you there? 2010-09-22T20:37:40 Zannick: Whats your user ID ? (can't search a list of 2000 entries to find bswolf) 2010-09-22T20:37:53 *** Olathe has quit IRC (Excess Flood) 2010-09-22T20:37:55 ai-contest.com/profile.php?user_id=3886 2010-09-22T20:38:10 yeah, you won't find it, it's 1500 or so 2010-09-22T20:38:20 the games are boring, trust me 2010-09-22T20:38:21 *** Olathe has joined #aichallenge 2010-09-22T20:38:50 and copybot doesn't do very well on the tcp server, when i run it 2010-09-22T20:38:59 because everyone there basically trounces it 2010-09-22T20:39:26 anyway 2010-09-22T20:39:27 * Zannick -> 2010-09-22T20:40:45 well.. yeah, because the tcp server has smarter bots 2010-09-22T20:40:48 on average 2010-09-22T20:41:42 Epic fight http://ai-contest.com/visualizer.php?game_id=4644228 2010-09-22T20:41:47 turn 95 on. 2010-09-22T20:42:12 * at turn 95 2010-09-22T20:43:49 danielvf: please update visualiser in server 2010-09-22T20:44:04 Naktibalda: with? 2010-09-22T20:44:37 with visualizer from http://planetwars.zerd.net/ 2010-09-22T20:45:21 Ah. 2010-09-22T20:45:38 Eventually. :( 2010-09-22T20:45:46 Naktibalda: what's different in that one? 2010-09-22T20:46:17 ship count/growth rate at least 2010-09-22T20:46:49 Naktibalda: k, good stuff 2010-09-22T20:48:09 *** Naith has joined #aichallenge 2010-09-22T20:49:14 *** drewrobb has joined #aichallenge 2010-09-22T20:49:52 amstan, do you have access to game logs on the main server? 2010-09-22T20:49:59 If you're still here. 2010-09-22T20:51:18 FlagCapper: game id? 2010-09-22T20:51:51 *** td123 has joined #aichallenge 2010-09-22T20:52:08 Any of my games: http://ai-contest.com/profile.php?user_id=4138 2010-09-22T20:52:31 I do fine on the tcp server and lose them all on the main server right in the middle 2010-09-22T20:52:41 I'm curious as to why 2010-09-22T20:54:17 timeout - could be a server problem. We are will be working on that next - after getting up a new server. 2010-09-22T20:55:08 OK, because I have something in my code to prevent timeouts. It's based on clock speed though, so maybe if you compiled it on a different server than you ran it there could be a problem? 2010-09-22T20:55:39 Same. 2010-09-22T20:56:48 Top 10 players: kcm1700(3577), McLeopold(3519), adumlah(3485), wagstaff(3448), EBraun(3403), dmj111(3378), goldman(3346), vinx33(3337), felixcoto(3307), paulus(3307) 2010-09-22T21:07:43 (random 1) 2010-09-22T21:07:49 lol sorry 2010-09-22T21:07:57 I thought was the REPL 2010-09-22T21:12:01 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-09-22T21:13:22 FlagCapper: define game logs 2010-09-22T21:13:50 FlagCapper: also.. yeah, danielvf can also help you 2010-09-22T21:14:06 *** jimmerton has quit IRC (Quit: jimmerton) 2010-09-22T21:15:14 amstan: Our current server is a 64 or a 32 bit? 2010-09-22T21:15:23 i think it's 32 2010-09-22T21:15:46 though any code/software packages should be able to run on 64 bit, since other servers might be diff 2010-09-22T21:16:26 It's looking like both amazon and rackspace are using AMD processors. 2010-09-22T21:16:42 That would mean we would have to recompile all bots. :( 2010-09-22T21:17:07 To test them out 2010-09-22T21:17:10 how much money would it be to move this to EC2? 2010-09-22T21:17:34 *** justin_pdx has joined #aichallenge 2010-09-22T21:17:35 seems like it would be a lot 2010-09-22T21:17:40 jmpespxoreax: I'm guessing that's not the cheapest way to do it. 2010-09-22T21:18:02 jmpespxoreax: Since EC2 isn't the cheapest way to get CPU power 2010-09-22T21:18:07 danielvf: weren't we running the distribution? 2010-09-22T21:18:14 danielvf: why are we considering ec2? 2010-09-22T21:18:19 amstan: Eh? 2010-09-22T21:18:27 you're assume that my question => considering ec2 2010-09-22T21:18:42 well.. i tought we were just using this server for http and distribute the bots on other servers 2010-09-22T21:18:44 *** jimmerton has joined #aichallenge 2010-09-22T21:18:54 i thought the current server was 64 bit... 2010-09-22T21:19:02 consider instead that I haven't been keeping up to date with ai contest news 2010-09-22T21:19:02 that's what i was told in the past anyway 2010-09-22T21:19:38 Right now I'm just building up a script to go from empty box, to running game server. 2010-09-22T21:20:00 After that is done, it doesn't matter where it's hosted. 2010-09-22T21:20:05 my assumption is either cloud or user contributed servers 2010-09-22T21:20:08 sorry for the mix up 2010-09-22T21:20:29 either cloud or rented dedicated server 2010-09-22T21:21:08 the latter may be easier to set up a VM-based sandbox on 2010-09-22T21:22:04 or at least a fast VM-based sandbox, that is 2010-09-22T21:22:08 danielvf: can your setup be run inside a VM? 2010-09-22T21:22:29 amstan: Just using the current code right now 2010-09-22T21:22:33 *** justin_pdx has quit IRC (Ping timeout: 264 seconds) 2010-09-22T21:22:48 amstan: Which can run in a vm. bots in the same vm 2010-09-22T21:25:42 *** boegel has quit IRC (Quit: Leaving) 2010-09-22T21:27:07 neat 2010-09-22T21:40:53 *** pacotabaco has quit IRC (Quit: Page closed) 2010-09-22T21:41:19 *** jorge__ has quit IRC (Remote host closed the connection) 2010-09-22T21:41:47 *** pacotabaco has joined #aichallenge 2010-09-22T21:46:39 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-09-22T21:46:49 *** delt0r_ has joined #aichallenge 2010-09-22T21:48:14 *** Blkt has quit IRC (Remote host closed the connection) 2010-09-22T21:48:15 *** Baphomet has quit IRC (Ping timeout: 272 seconds) 2010-09-22T21:50:07 *** delt0r has quit IRC (Ping timeout: 240 seconds) 2010-09-22T21:51:35 *** Baphomet has joined #aichallenge 2010-09-22T21:56:45 *** drewrobb has quit IRC (Ping timeout: 264 seconds) 2010-09-22T21:56:49 Top 10 players: kcm1700(3561), McLeopold(3518), adumlah(3477), wagstaff(3447), EBraun(3399), dmj111(3395), goldman(3352), vinx33(3319), felixcoto(3319), paulus(3307) 2010-09-22T22:06:56 *** SophieB is now known as zz_SophieB 2010-09-22T22:13:56 *** zz_SophieB is now known as SophieB 2010-09-22T22:18:39 *** jesionaj has quit IRC (Ping timeout: 272 seconds) 2010-09-22T22:23:28 *** Zaphus has quit IRC (Quit: Page closed) 2010-09-22T22:26:31 *** naglijo has joined #aichallenge 2010-09-22T22:26:50 the FAQ says this: ou can have as many source files as you want, as long as they're all in the same language and they're all in the same directory. 2010-09-22T22:27:54 does the mean im prohibited from having code in languages A and B being run even if my source files are all in language A? 2010-09-22T22:28:10 if so, what about DSLs made by me for the contest? 2010-09-22T22:30:23 *** Appleman1234 has joined #aichallenge 2010-09-22T22:32:40 naglijo: it will compile the files of one language with whatever compiler is standard for that language. You can also have whatever other files you want, and they'll be available when your program is running. 2010-09-22T22:32:40 DSLs? 2010-09-22T22:33:51 can you execute files? 2010-09-22T22:34:25 no (they don't have +x set), but jmcarthur is pushing to allow submitting binaries 2010-09-22T22:34:54 *** smhanov has quit IRC (Ping timeout: 255 seconds) 2010-09-22T22:36:13 see http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/compile_anything.py for more details 2010-09-22T22:38:03 you can also just ask for whatever language you're writing in 2010-09-22T22:38:18 python executining haskell and visa-versa 2010-09-22T22:38:25 sp* 2010-09-22T22:39:18 that probably won't work, because there will be multiple processes 2010-09-22T22:48:03 *** smhanov has joined #aichallenge 2010-09-22T22:53:07 *** jimmerton has quit IRC (Quit: jimmerton) 2010-09-22T22:55:10 *** Theorevery has joined #aichallenge 2010-09-22T22:56:50 Top 10 players: kcm1700(3547), McLeopold(3511), adumlah(3482), wagstaff(3438), EBraun(3392), dmj111(3358), goldman(3350), felixcoto(3326), vinx33(3311), Tracer(3298) 2010-09-22T22:58:44 *** tapwater has quit IRC (Ping timeout: 252 seconds) 2010-09-22T23:04:59 *** tapwater has joined #aichallenge 2010-09-22T23:08:49 *** seanmceligot has quit IRC (Remote host closed the connection) 2010-09-22T23:11:07 *** Appleman1234 has quit IRC (Ping timeout: 276 seconds) 2010-09-22T23:13:03 multiple processes aren't strictly necessary for haskell and python to interop 2010-09-22T23:13:41 thus the "probably" 2010-09-22T23:15:03 *** Titankiller has quit IRC (Quit: Nettalk6 - www.ntalk.de) 2010-09-22T23:26:05 *** td123 has quit IRC (Changing host) 2010-09-22T23:26:05 *** td123 has joined #aichallenge 2010-09-22T23:40:15 *** AlliedEnvy has quit IRC (Ping timeout: 240 seconds) 2010-09-22T23:50:05 *** jesionaj has joined #aichallenge 2010-09-22T23:56:52 Top 10 players: kcm1700(3551), McLeopold(3515), adumlah(3482), wagstaff(3438), EBraun(3405), dmj111(3363), goldman(3357), felixcoto(3335), Tracer(3312), zaphus(3307)