2010-10-06T00:02:10 *** Haggis has quit IRC (Quit: Page closed) 2010-10-06T00:08:35 *** delt0r___ has joined #aichallenge 2010-10-06T00:09:44 *** delt0r_ has quit IRC (Ping timeout: 245 seconds) 2010-10-06T00:14:16 *** davidjliu has quit IRC (Quit: Page closed) 2010-10-06T00:33:59 *** amstan_ is now known as amstan 2010-10-06T00:38:24 wow... dmj is back to 2nd and it's still his bot from weeks ago 2010-10-06T00:38:59 either the rankings are really unstable, or the exact players that you are playing against matters A LOT 2010-10-06T00:40:47 sigh: can you find a game where dmj beat david? 2010-10-06T00:41:05 3 games ago :) 2010-10-06T00:41:17 http://ai-contest.com/visualizer.php?game_id=5508876 2010-10-06T00:42:02 you have your answer then.. 2010-10-06T00:47:33 the other thing is that even though that bot has been up for a long time and is approaching 600 games, as of the ratings.txt that was posted earlier only 114 of them were able to count for the current scoring because of resubmissions of other bots 2010-10-06T00:47:58 so its elo confidence interval can't get all that tight 2010-10-06T00:48:37 damn this c# compile problem 2010-10-06T00:49:24 *** other_donkey has joined #aichallenge 2010-10-06T00:49:38 *** other_donkey has left #aichallenge 2010-10-06T00:59:30 Top 10 players: ruilov(3921), davidjliu(3856), dmj111(3853), Hazard(3790), bix0r4ever(3770), Arekku(3715), DocBot(3703), Rexxar(3700), EBraun(3691), sequoh(3670) 2010-10-06T01:06:07 *** otherAntimatroid has joined #aichallenge 2010-10-06T01:06:58 *** JCS^ has quit IRC (Quit: \x00) 2010-10-06T01:22:49 *** Zeiris_ has joined #aichallenge 2010-10-06T01:23:58 *** Frontier has left #aichallenge 2010-10-06T01:26:04 *** Frontier has joined #aichallenge 2010-10-06T01:30:19 *** Appleman1234 has joined #aichallenge 2010-10-06T01:59:28 Top 10 players: ruilov(3912), davidjliu(3846), dmj111(3833), Hazard(3785), bix0r4ever(3766), Arekku(3711), DocBot(3702), Rexxar(3696), EBraun(3690), sequoh(3666) 2010-10-06T02:18:42 *** amstan has quit IRC (Remote host closed the connection) 2010-10-06T02:19:00 *** mega1 has joined #aichallenge 2010-10-06T02:22:31 *** justin_pdx has joined #aichallenge 2010-10-06T02:25:45 *** Accoun has quit IRC () 2010-10-06T02:28:41 *** Palmik has joined #aichallenge 2010-10-06T02:35:56 *** sinan has joined #aichallenge 2010-10-06T02:39:54 *** sigh has quit IRC (Remote host closed the connection) 2010-10-06T02:42:24 *** fawek has joined #aichallenge 2010-10-06T02:47:25 *** prowler has quit IRC (Quit: Page closed) 2010-10-06T02:50:36 *** antimatroid has joined #aichallenge 2010-10-06T02:50:48 *** Mathnerd314 has quit IRC (Remote host closed the connection) 2010-10-06T02:50:55 *** Accoun has joined #aichallenge 2010-10-06T02:51:05 *** antimatroid2 has quit IRC (Ping timeout: 240 seconds) 2010-10-06T02:59:29 Top 10 players: ruilov(3900), davidjliu(3829), dmj111(3819), Hazard(3776), bix0r4ever(3760), Arekku(3706), DocBot(3695), Rexxar(3694), EBraun(3676), felixcoto(3650) 2010-10-06T03:07:31 *** antimatroid1 has joined #aichallenge 2010-10-06T03:07:31 *** antimatroid has quit IRC (Read error: Connection reset by peer) 2010-10-06T03:10:00 Is it true that planets retain their id all through the game? 2010-10-06T03:10:45 yes 2010-10-06T03:18:37 *** ichti has joined #aichallenge 2010-10-06T03:21:39 *** perror has joined #aichallenge 2010-10-06T03:21:41 *** Palmik has quit IRC (Remote host closed the connection) 2010-10-06T03:22:14 *** Blkt has joined #aichallenge 2010-10-06T03:22:34 *** Palmik has joined #aichallenge 2010-10-06T03:28:13 *** Titankiller has joined #aichallenge 2010-10-06T03:33:12 contestbot: where sampledb 2010-10-06T03:33:12 Zannick: sampledb = http://csclub.uwaterloo.ca/~amstan/aichallenge/sample.sql 2010-10-06T03:33:14 good morning everyone 2010-10-06T03:38:28 *** smhanov__ has joined #aichallenge 2010-10-06T03:41:50 *** smhanov_ has quit IRC (Ping timeout: 240 seconds) 2010-10-06T03:44:28 *** Florian has joined #aichallenge 2010-10-06T03:47:52 *** jaspervdj has joined #aichallenge 2010-10-06T03:49:00 Good evening. 2010-10-06T03:49:11 morning :) 2010-10-06T03:49:44 *** sigh has joined #aichallenge 2010-10-06T03:51:29 *** hellman has joined #aichallenge 2010-10-06T03:51:56 evening :) 2010-10-06T03:54:39 hey 2010-10-06T03:57:01 *** sigh has quit IRC (Remote host closed the connection) 2010-10-06T03:59:30 *** ichti has quit IRC (Ping timeout: 265 seconds) 2010-10-06T03:59:31 Top 10 players: ruilov(3916), davidjliu(3846), dmj111(3833), Hazard(3793), bix0r4ever(3775), Arekku(3723), DocBot(3714), Rexxar(3710), EBraun(3699), adumlah(3668) 2010-10-06T04:03:13 I can't wait till those notices render my name. Mmm, delusions of grandeur. 2010-10-06T04:04:40 :) 2010-10-06T04:05:36 yeah, i can't wait until i'm pinged every hour in this channel 2010-10-06T04:05:55 lol 2010-10-06T04:11:16 *** almighty000 has joined #aichallenge 2010-10-06T04:14:09 *** wvdschel has joined #aichallenge 2010-10-06T04:15:33 yo 2010-10-06T04:25:20 *** mega1 has quit IRC (Ping timeout: 240 seconds) 2010-10-06T04:26:59 *** Naktibalda has joined #aichallenge 2010-10-06T04:35:31 *** almighty000 has quit IRC (Quit: Page closed) 2010-10-06T04:37:07 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-10-06T04:39:08 *** Titankiller has quit IRC (Excess Flood) 2010-10-06T04:40:56 *** sigh has joined #aichallenge 2010-10-06T04:44:01 *** Accoun has quit IRC () 2010-10-06T04:51:11 *** Florian has quit IRC (Disconnected by services) 2010-10-06T04:51:28 *** Florian95 has joined #aichallenge 2010-10-06T04:59:31 Top 10 players: ruilov(3927), davidjliu(3850), dmj111(3840), Hazard(3806), bix0r4ever(3739), Arekku(3726), DocBot(3714), Rexxar(3707), felixcoto(3702), EBraun(3697) 2010-10-06T05:03:50 *** ichti has joined #aichallenge 2010-10-06T05:23:04 *** MyHuggingBear has joined #aichallenge 2010-10-06T05:23:07 *** smellyhippy has quit IRC (Ping timeout: 265 seconds) 2010-10-06T05:24:52 *** HakanD has joined #aichallenge 2010-10-06T05:24:58 *** ichti has quit IRC (Ping timeout: 264 seconds) 2010-10-06T05:25:30 @rankings 2010-10-06T05:25:32 HakanD: Top 10 players: ruilov(3916), davidjliu(3839), dmj111(3829), Hazard(3795), bix0r4ever(3731), Arekku(3715), DocBot(3707), Rexxar(3695), felixcoto(3691), EBraun(3685) 2010-10-06T05:26:02 *** Meatkat has joined #aichallenge 2010-10-06T05:26:29 *** Accoun has joined #aichallenge 2010-10-06T05:26:34 Have there been any changes recently regarding the ranking system and/or matchup system? 2010-10-06T05:28:10 *** smellyhippy has joined #aichallenge 2010-10-06T05:28:10 *** smellyhippy has joined #aichallenge 2010-10-06T05:28:52 Meatkat: not that I know of. Why? 2010-10-06T05:31:03 sigh: I'm not exactly sure but it seems like my bot is taking forever to 'level-up', maybe it's just the luck of the draw but I've usually started in the 500s and this time I got paired with some guy at 2600 and have had to work my way up from there. And at less than one game an hour that takes forever 2010-10-06T05:32:35 seems like that was the just an unlucky choice for first opponent 2010-10-06T05:32:49 aka QQ 2010-10-06T05:33:50 after the first game the ranking system is supposed to match you up with people close to your level more often. 2010-10-06T05:34:38 if my first game was against rank 2600, I would probably resubmit :P probably faster... haha 2010-10-06T05:34:58 I've noticed that, it would be nice if the system matched you up against someone near your old rank first instead of someone completely random 2010-10-06T05:35:19 agree, I think that is on the todo list for the admins 2010-10-06T05:35:24 Yeah I'm thinking about resubmitting, I think it'll be another 12 hours before I get anywhere at this rate. 2010-10-06T05:36:09 I came in on them discussing how to post the range of ELO ratings, do you know if they are planning to add that to our profiles? 2010-10-06T05:38:11 range? like the confidence interval for a given user? 2010-10-06T05:40:30 Yeah 2010-10-06T05:40:39 Something like 3200 (+128/-260) 2010-10-06T05:40:50 no idea 2010-10-06T05:40:55 @url issues 2010-10-06T05:40:55 sigh: issues = http://code.google.com/p/ai-contest/issues/list 2010-10-06T05:41:03 That would be really helpful to have on your profile so that you know when your bot has about leveled out 2010-10-06T05:41:31 that is the issues tracker, might find what you are looking for in there 2010-10-06T05:44:11 Wow, they have a lot on their plates! 2010-10-06T05:44:52 yup 2010-10-06T05:46:25 How is your bot coming along? You used to be in the top10, right? 2010-10-06T05:47:05 yeah, I was in the top 10 during the previous week... currently doing a big refactor/rewrite 2010-10-06T05:47:17 mostly to facilitate exploring more advanced ideas 2010-10-06T05:47:38 the current bot that is up is just a mess of heuristics :P 2010-10-06T05:48:15 same here... 2010-10-06T05:48:21 :) 2010-10-06T05:48:29 but mine is a bit messier than yours I would imagine 2010-10-06T05:48:31 where was your bot ranked before you resubmitted? 2010-10-06T05:48:57 I've usually come in on the tail end of the first 100 2010-10-06T05:49:11 but I don't really keep them in long enough to really settle 2010-10-06T05:49:29 fair enough 2010-10-06T05:49:50 Once I get heuristics down where I want them I'm probably going to something more tree-like 2010-10-06T05:49:54 keeping them there can be helpful though... get some games to see what your bots strength and weaknesses are 2010-10-06T05:50:11 Meatkat: same, that's what I want to start looking at 2010-10-06T05:50:57 I image that's what most of the top dogs are running at the moment, though it would be interesting to know 2010-10-06T05:51:23 nah, everyone is still mostly heuristics atm, I think 2010-10-06T05:51:48 more advanced search is easy enough to talk about but implementing it effectivly is another matter 2010-10-06T05:52:50 Definately, just thinking about trying to implement it with my current mess is giving me ulcers =p 2010-10-06T05:53:14 a few things I've noticed from hanging out in the channel: 2010-10-06T05:53:20 - Everyone is having the same problems 2010-10-06T05:53:34 - People think the bots above them are much more sophisticated than they actually are 2010-10-06T05:54:15 - There are some key ideas that are obvious or commonly talked about, the better bots just implement them more effectivly 2010-10-06T05:57:33 Good to know, thanks 2010-10-06T05:57:39 so true 2010-10-06T05:59:32 Top 10 players: ruilov(3923), davidjliu(3848), dmj111(3844), Hazard(3800), bix0r4ever(3734), Arekku(3720), DocBot(3710), Rexxar(3697), felixcoto(3695), EBraun(3693) 2010-10-06T06:01:39 *** MyHuggingBear has quit IRC (Ping timeout: 265 seconds) 2010-10-06T06:05:04 *** HakanD has quit IRC (Quit: brb) 2010-10-06T06:06:37 *** sinan has quit IRC (Ping timeout: 265 seconds) 2010-10-06T06:08:23 http://ai-contest.com/game_stats.php 2010-10-06T06:08:25 Server KO ? 2010-10-06T06:17:56 sigh: give me two weeks and i'ma change that 2010-10-06T06:18:06 antimatroid1: :) 2010-10-06T06:18:20 antimatroid1: hopefully I can say the same 2010-10-06T06:20:52 !predict 2010-10-06T06:20:52 I predict that edcba will win! 2010-10-06T06:21:09 !rainbow !predict 2010-10-06T06:21:09 antimatroid1: !predict 2010-10-06T06:30:46 !rainbow [predict] 2010-10-06T06:30:46 I predict that Naktibalda will win! 2010-10-06T06:31:00 thanks :) 2010-10-06T06:36:12 !predict 2010-10-06T06:36:12 I predict that Lithosphere will win! 2010-10-06T06:36:20 ?! 2010-10-06T06:37:02 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-10-06T06:37:56 !predict 2010-10-06T06:37:57 I predict that tty1 will win! 2010-10-06T06:38:29 *** sinan has joined #aichallenge 2010-10-06T06:43:25 *** Utkarsh has joined #aichallenge 2010-10-06T06:59:32 Top 10 players: ruilov(3920), davidjliu(3845), dmj111(3841), Hazard(3798), bix0r4ever(3747), Arekku(3717), DocBot(3704), Rexxar(3695), felixcoto(3693), EBraun(3690) 2010-10-06T07:01:20 *** hellman has quit IRC (Remote host closed the connection) 2010-10-06T07:03:31 *** Palmik has quit IRC (Remote host closed the connection) 2010-10-06T07:05:21 *** ak2k99 has joined #aichallenge 2010-10-06T07:07:06 *** ak2k99 has quit IRC (Client Quit) 2010-10-06T07:08:06 *** sinan has quit IRC (Ping timeout: 240 seconds) 2010-10-06T07:16:25 sigh: how's it going? 2010-10-06T07:17:03 dmj111: I'm doing well... I'm trying to convince myself that rewriting my bot is worth it :D 2010-10-06T07:17:06 how are you? 2010-10-06T07:17:49 ok. I spent three days or more rewritting mine... and now I am trying to convince myself to erase the new one. its baaad. 2010-10-06T07:18:08 ahhh... that's what I'm afraid of :( 2010-10-06T07:18:28 what did you try? or is that secret? 2010-10-06T07:19:04 my brain is almost too worn out to remember, but.... 2010-10-06T07:19:53 basically, I was trying to make sure I keep enough stuff around at each planet, accounting for the fact that they can defend each other. 2010-10-06T07:20:09 ah yes... that is difficult! 2010-10-06T07:20:17 *** mega1 has joined #aichallenge 2010-10-06T07:20:22 I tried to make my heuristics aware of that, but failed misarably 2010-10-06T07:20:52 It gave me some entertaining results, and beats my current version 50% of the time. But it makes really weird decissions. 2010-10-06T07:21:03 Glad I am not the only one who thinks it is difficult! 2010-10-06T07:21:06 :) 2010-10-06T07:21:32 I've tried modelling it as a graph and applying various closeness measures... but I couldn't get it to give me nice results 2010-10-06T07:22:59 *more accuratly: centrality measures 2010-10-06T07:23:21 dmj111: at least your current bot is still very competitive at the top! 2010-10-06T07:23:42 yeah... I am content, if possibly misled by that. 2010-10-06T07:23:52 :) 2010-10-06T07:24:12 I declared that bot an evolutionary dead end a week ago... but I laid in bed last night trying to think of how to beat ragebot with it :) 2010-10-06T07:24:40 -- for everyone watching... I have no special first move logic.... I am hoping my heuristics figure it out on their own. 2010-10-06T07:25:18 ah ok, I wasn't sure if you had done that intentionally to be more aggressive 2010-10-06T07:25:45 which I think is a valid tactic, no one is going to act like ragebot around your level :P 2010-10-06T07:26:59 ha. I wish it was intentional... It just works ok on many maps. I am pretty sure it is one of my two critical errors though. The other being too cautious to take neutrals. 2010-10-06T07:27:53 my bot loves neutrals... it often grabs them in dubious situations 2010-10-06T07:27:54 The 'RageBot' error occurs in other situations later in the game. 2010-10-06T07:28:12 fair enough 2010-10-06T07:28:46 Yeah, any fix I have done for the neutral thing just goes way too far in the other direction. I was hoping this new bot would only take stuff that it realized it could defend. In fact, it does that ok. It just does other stuff way wrong. 2010-10-06T07:28:50 http://72.44.46.68/getplayer?player=dmj_exp 2010-10-06T07:29:01 There are only a few games there... maybe I will put it back online for fun. 2010-10-06T07:30:21 ah yeah, that bot loves neutrals :) 2010-10-06T07:30:43 the hardest part for me is creating an evaluation function that can effectively compare neutrals to enemy planets 2010-10-06T07:31:26 Ok, here is possibly my biggest secret.... 2010-10-06T07:31:38 in the current bot, I don't compare enemy to neutrals. 2010-10-06T07:31:48 I alternately attack one and then the other. 2010-10-06T07:32:05 yeah, that's an interesting strategy 2010-10-06T07:32:09 ok maybe not the biggest secret. 2010-10-06T07:32:40 I couldn't figure out how to compare them, so I gave up. 2010-10-06T07:33:05 I score all planets that I want to attack, and just start attacking the best scores 2010-10-06T07:33:50 If my planet will be owned by the enemy in the future, I count sending reinforcements as an attack as well 2010-10-06T07:34:06 Yep, sounds similar. 2010-10-06T07:34:19 so all three of those situations are handled by just scoring and sorting :) 2010-10-06T07:34:39 That is what my new bot was supposed to do, but my scoring is all wonked up. 2010-10-06T07:35:56 my scoring is essentially the same since day 1... I'm suprised that it's worked so well, but other than trivial changes I haven't been able to make it much better 2010-10-06T07:36:25 *** Palmik has joined #aichallenge 2010-10-06T07:36:39 That's cool. I had hopes of improving mine, but it hasn't worked out so far. 2010-10-06T07:37:41 I keep thinking someone will get something working and I'll wake up and #1 will be 500 points ahead :D 2010-10-06T07:37:55 I mean it is so *obvious* that we can do so much better 2010-10-06T07:38:21 I agree. 2010-10-06T07:38:31 i put the new bot back on the tcp server. Let it live a little before I put it out of its misery. That, and the games can be entertaining, but I think it is optimized to beat my official bot. 2010-10-06T07:39:05 Some of the games are a real beat-down too. Maybe I will post a couple on Naktibalda's pastebin. 2010-10-06T07:39:20 Naktibalda's pastebin? 2010-10-06T07:39:33 pastebin 2010-10-06T07:39:46 http://visualizer.naktibalda.lt/ 2010-10-06T07:40:05 oooooh... nice 2010-10-06T07:40:29 what is the idea? store interesting games? 2010-10-06T07:44:09 my take was just to share interesting games. It seems that there is no list of all of the games, so it is only good as long as you have the link 2010-10-06T07:44:23 Which is still extremely useful. 2010-10-06T07:44:36 _still_ is probably not the right word there. 2010-10-06T07:45:53 cool... useful for tcp games which don't stay up very long 2010-10-06T07:48:19 http://visualizer.naktibalda.lt/94 the first easy example I can find where the new bot trounces the old bot. 2010-10-06T07:49:01 yeah, very nice 2010-10-06T07:49:12 is dmg1.8 the bot on the main contest? 2010-10-06T07:49:46 yeah. 2010-10-06T07:50:00 http://visualizer.naktibalda.lt/95 and here, the new bot pretends its ragebot. 2010-10-06T07:50:36 that's not ragebot, it waited a turn :D 2010-10-06T07:51:11 true :) 2010-10-06T07:51:19 but it seems you are having the same problem that I'm facing... you are creating a bot that is good against your previous version by exploiting their weaknesses 2010-10-06T07:51:32 but not necessarily better overall 2010-10-06T07:52:07 well, the problem most people are facing, I suppose 2010-10-06T07:53:18 I am happy, since these went about 50/50 against each other, I can have a new bot play both of them. That may be delusional, but beating two different bots should be better than beating one. 2010-10-06T07:53:33 Of course, that assumes I have _any_ new ideas to try. 2010-10-06T07:53:39 *** Meatkat has quit IRC (Ping timeout: 265 seconds) 2010-10-06T07:53:49 Last example, this one the new one loses, but it looks pretty. http://visualizer.naktibalda.lt/96 2010-10-06T07:57:15 that game is mesmerising, seems almost like it was choreographed :) 2010-10-06T07:58:06 but ouch, move 46 attacking that 5-ship enemy from the other side of the map 2010-10-06T07:58:49 if that was my bot, I'd want to know exactly why it thought that was a good idea 2010-10-06T07:59:33 Top 10 players: ruilov(3913), davidjliu(3829), dmj111(3823), Hazard(3778), bix0r4ever(3739), Rexxar(3703), Arekku(3695), felixcoto(3688), DocBot(3687), EBraun(3673) 2010-10-06T07:59:40 sorry, not 5-ship... 101-ship o_O 2010-10-06T08:07:26 *** yasith__ has quit IRC (Ping timeout: 276 seconds) 2010-10-06T08:09:05 *** delt0r_ has joined #aichallenge 2010-10-06T08:10:37 *** delt0r___ has quit IRC (Ping timeout: 252 seconds) 2010-10-06T08:12:01 *** Utkarsh has quit IRC (Ping timeout: 264 seconds) 2010-10-06T08:12:42 sigh: right, I should probably not give up on it yet, and figure out why it did that. Maybe a few days away and I will be interested in fixing it. Also, I am starting to think it does better on the maps that the attack forms front to back instead of around the perimeter. 2010-10-06T08:13:15 Maybe I could build a meta-bot that chooses which bot to use based on the map. bwahahaha... I am sure that would fail in other spectacular ways. 2010-10-06T08:13:52 dmj111: is that still souly heuristic based? 2010-10-06T08:14:11 i'm out of action for at least a week 2010-10-06T08:15:05 *** boegel has joined #aichallenge 2010-10-06T08:15:18 antimatroid1: me too. which part are you refering to for the heuristics? 2010-10-06T08:17:19 *** Utkarsh has joined #aichallenge 2010-10-06T08:18:44 antimatroid1 is referring to anything that isn't based on more generic optimisation strategies 2010-10-06T08:19:20 *** yasith__ has joined #aichallenge 2010-10-06T08:19:23 for example, he is contrasting with things like minimax search 2010-10-06T08:19:49 *** alejandro_ has joined #aichallenge 2010-10-06T08:20:07 Ahh. Still no game tree search. 2010-10-06T08:20:20 hola. Alguien habla español? 2010-10-06T08:21:01 No hable. O, un poco. 2010-10-06T08:21:06 xD 2010-10-06T08:21:28 I'm Problems with TCP server 2010-10-06T08:21:50 Ok. 2010-10-06T08:22:21 My spanish is probably too weak to help with that :( 2010-10-06T08:22:48 yeah, what sigh said 2010-10-06T08:23:11 would neureal networks be any good? or are they meant to be crappy? 2010-10-06T08:25:33 *** keynes has joined #aichallenge 2010-10-06T08:25:59 *** otherAntimatroid has quit IRC (Quit: Leaving.) 2010-10-06T08:27:10 antimatroid1: not sure. Doesn't seem so to me, but I am hardly an authority. 2010-10-06T08:27:26 *** _Utkarsh <_Utkarsh!~admin@117.201.80.38> has joined #aichallenge 2010-10-06T08:27:44 Hi! SomeOne use Java Client? 2010-10-06T08:28:07 (For the tcp server) 2010-10-06T08:28:13 :D 2010-10-06T08:28:26 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-10-06T08:35:02 Hi! SomeOne use Java Client for the tcp server? 2010-10-06T08:42:20 *** krokokrusa has joined #aichallenge 2010-10-06T08:46:10 *** Appleman1234 has quit IRC (Quit: Leaving) 2010-10-06T08:49:17 Hi! SomeOne use Java Client for the tcp server? 2010-10-06T08:51:39 me 2010-10-06T08:53:28 *** amstan has joined #aichallenge 2010-10-06T08:53:28 *** ChanServ sets mode: +o amstan 2010-10-06T08:55:46 *** area has quit IRC (Ping timeout: 240 seconds) 2010-10-06T08:59:33 Top 10 players: ruilov(3928), davidjliu(3835), dmj111(3825), Hazard(3785), bix0r4ever(3738), felixcoto(3721), Rexxar(3706), Arekku(3699), DocBot(3693), EBraun(3679) 2010-10-06T09:00:05 antimatroid1: meh neural networks 2010-10-06T09:00:12 to me it seems like they could work 2010-10-06T09:00:56 in that if there exsists some function that takes everything as arguments and returns > 0 for that condition 2010-10-06T09:01:12 that it should be possible to have weights for a neural network to curve fit that function 2010-10-06T09:01:42 *** boegel has quit IRC (Quit: This computer has gone to sleep) 2010-10-06T09:02:01 sprinkle with handwavium for the best results 2010-10-06T09:04:23 i still think i'll end up with some kind of approximate good move tree with minimax and some attempt at guessing characteristics of the opponent bot maybe 2010-10-06T09:04:58 then some testing to try and optimise parameters, and some classy way of picking optimal moves once a chosen feasible move has been selected 2010-10-06T09:05:03 not really sure what else to do 2010-10-06T09:05:26 i think i should look for solutions using dynamic programming 2010-10-06T09:05:39 hmmm, i tried that, you're gonna need some dark magic 2010-10-06T09:05:45 the ordering of moves, especially in time and coordinated strikes are allways overrunning my searchspace 2010-10-06T09:05:54 i was running 15 seconds a move with just two planets once ships got above 100 per planet 2010-10-06T09:05:57 somethign dp should do well 2010-10-06T09:05:59 *** delt0r has joined #aichallenge 2010-10-06T09:06:01 but that was a very naive implementation 2010-10-06T09:06:19 antimatroid1: maybe work with some log power of units instead of precise unit counts ? 2010-10-06T09:07:33 maybe, i think one could possibly just reduce the number of fleets on cares to consider 2010-10-06T09:07:42 but not sure you'd ever get it low enough 2010-10-06T09:07:56 especially once you have like 10 planets and a few thousand ships 2010-10-06T09:08:49 i think linear programming could possible be good 2010-10-06T09:09:00 or maybe some kind of transportation problem hookup 2010-10-06T09:09:16 possibly* I am terrible at that mistake on a keyboard of late 2010-10-06T09:10:17 linear programmings good because it doesn't really get any worse with the number of ships you have, just planets 2010-10-06T09:11:09 even running anything decent with that I think would be worse than a decent game tree, i want to combine them, which means what I do run needs to be super fast :| 2010-10-06T09:11:23 *** alejandro_ has quit IRC (Quit: Page closed) 2010-10-06T09:12:43 did that alejandro_ person pm anyone else? 2010-10-06T09:22:10 antimatroid1: but the problem is not linear 2010-10-06T09:24:26 mega1: depends how you treat the problem 2010-10-06T09:24:39 it is linear for a lot of things 2010-10-06T09:24:39 really? 2010-10-06T09:24:52 there are linear subproblems, sure 2010-10-06T09:25:04 you can set up feasibility of taking planets as a linear program fairly easily 2010-10-06T09:25:29 picking an optimal strategy for a feasible solution just requires you defining some kind of function that is linear in the size of fleets 2010-10-06T09:26:26 *** Umsturz has joined #aichallenge 2010-10-06T09:27:14 *** mceier has joined #aichallenge 2010-10-06T09:28:26 *** Umsturz has left #aichallenge 2010-10-06T09:28:35 *** amstan has quit IRC (Ping timeout: 264 seconds) 2010-10-06T09:28:59 oh, i just had a good idea wrt my linear programs if i do that :D 2010-10-06T09:29:24 I'm having a hard time imagining a useful linear function to optimize. 2010-10-06T09:31:17 hmmm, you're making me question the use of phase II now :P 2010-10-06T09:31:23 but i still think phase I is useful 2010-10-06T09:31:26 or simplex that is 2010-10-06T09:31:28 of* 2010-10-06T09:34:58 *** cfaftw has joined #aichallenge 2010-10-06T09:37:10 any word on why bots are getting played a lot less? (as per this thread: http://www.ai-contest.com/forum/viewtopic.php?f=18&t=877) 2010-10-06T09:37:41 http://ai-contest.com/game_stats.php 2010-10-06T09:38:54 what was it a few days ago? 2010-10-06T09:39:26 come on, server 0, pick up the pace! 2010-10-06T09:40:02 You can open a serv if you want 2010-10-06T09:40:09 =-O 2010-10-06T09:40:11 i can? 2010-10-06T09:40:41 http://72.44.46.68/ 2010-10-06T09:40:49 fight every time 2010-10-06T09:40:53 ohhh... the TCP thing 2010-10-06T09:40:56 yeah 2010-10-06T09:40:57 *** hellman has joined #aichallenge 2010-10-06T09:41:03 click -> fight 2010-10-06T09:41:07 :) 2010-10-06T09:41:42 problem with the TCP server is the other people have to be "online" 2010-10-06T09:41:55 no pb :) 2010-10-06T09:42:22 the official server is great because it plays all the bots automatically, even with people afk or with their computers off 2010-10-06T09:44:53 30 users simultany on TCP :) 2010-10-06T09:44:56 also... where are servers 1 - 12? 2010-10-06T09:44:57 Is enough 2010-10-06T09:45:15 and, on TCP there are hard bots, it better for training 2010-10-06T09:46:31 the only hard bots on there are dmj1.8 and CCCF-mof3-asbp8-caf12 2010-10-06T09:46:38 *** sigh has quit IRC (Remote host closed the connection) 2010-10-06T09:46:39 *** rebelxt_ has joined #aichallenge 2010-10-06T09:47:23 i'm still trying to figure out what "CCCF-mof3-asbp8-caf12" even means, haha 2010-10-06T09:48:42 any word from organizers on server slowdown? i.e. one game per 3-4 hours at the moment 2010-10-06T09:48:56 rebelxt, i was just asking about it 2010-10-06T09:49:09 Florian95 gave me this link, http://ai-contest.com/game_stats.php 2010-10-06T09:49:19 you can see that there are only two servers, and one is crawling 2010-10-06T09:50:12 rebelxt_, does somebody on freenode already have rebelxt? 2010-10-06T09:50:22 doh server #0 is too slow 2010-10-06T09:50:44 i just want to know where servers 1 through 12 went ;) 2010-10-06T09:50:55 (assuming they ever existed)\ 2010-10-06T09:51:18 cfaftw, i'm logged on from work computer as rebelxt. this one is home laptop ;) 2010-10-06T09:51:26 gotcha 2010-10-06T09:52:50 there used to be more servers for sure.. i recall seeing 40+ games per minute on the stats page 2010-10-06T09:53:46 i resubmitted last night around 9 (i think) and my new bot just now played its 6th game 2010-10-06T09:53:51 9est 2010-10-06T09:55:29 i re-submitted at 10pm est last night - 4 games only 2010-10-06T09:55:43 on another note... my bot does things that make me think "I don't remember programming it to do that..." 2010-10-06T09:56:09 lol 2010-10-06T09:56:24 :D 2010-10-06T09:57:09 sometimes i feel like it thinks for itself... "i know you made me to do this thing, but i have a better idea!" 2010-10-06T09:57:51 after submitting my last bot, i found a cool bug which was fixed weeks before - it really likes to capture 0-growth planets, such as 119 planet in the middle of map1. lol 2010-10-06T09:58:00 hahaha 2010-10-06T09:58:14 shhhh... don't blow your strategy 2010-10-06T09:58:27 0 growth planets ftw :> 2010-10-06T09:58:39 :o 2010-10-06T09:58:43 yeah! center planet is the most important one, so whatever it takes! 2010-10-06T09:58:48 0 growth planets that cost 119 ships for the win 2010-10-06T09:58:53 :) 2010-10-06T09:59:26 all the "smart" bots that do look aheads would never expect your bot to make such a move! 2010-10-06T09:59:29 remember divide by zero @ source 2010-10-06T09:59:34 Top 10 players: ruilov(3956), davidjliu(3880), dmj111(3857), Hazard(3802), bix0r4ever(3793), felixcoto(3728), Arekku(3725), DocBot(3718), EBraun(3711), Rexxar(3706) 2010-10-06T10:00:14 they probably ignore that planet first place, so its safe planet to group your ships and attack 2010-10-06T10:00:20 opponent will not never see that coming 2010-10-06T10:00:34 not never? or not ever? 2010-10-06T10:00:52 -not :) 2010-10-06T10:01:20 :D 2010-10-06T10:01:48 yay i just realized where the word "never" comes from -_- 2010-10-06T10:03:19 *** amstan has joined #aichallenge 2010-10-06T10:03:19 *** ChanServ sets mode: +o amstan 2010-10-06T10:03:32 *** Meatkat has joined #aichallenge 2010-10-06T10:05:10 I keep geting "IOError: [Errno 32] Broken pipe" and "EOFError: EOF when reading a line" when playing on the tcp server. I'm assuming because I have a bad connection... ? 2010-10-06T10:05:37 *** ichti has joined #aichallenge 2010-10-06T10:05:52 Meatkat, you using python? 2010-10-06T10:06:59 Meatkat, surround the raw_input() call with "try: ... except EOFError, e: break" 2010-10-06T10:07:11 funny i re-write a big chunk of the algorithm last night. 100% sure it's going to dominate my previous bot (by design!). and then it fails the tests miserably ;-) 2010-10-06T10:07:28 Meatkat, or "return", or whatever else you need to do to make the script terminate 2010-10-06T10:07:46 rebelxt_, yeah it happens... i've done the same 2010-10-06T10:08:13 Meatkat, did you try running a java bot against tcp server? just to rule out connection issue 2010-10-06T10:09:54 cfaftw: hehe glad i'm not alone. "highly sophisticated" algorithm loses to a simple heuristic approach ;-) 2010-10-06T10:09:58 *** Meatkat has quit IRC (Ping timeout: 265 seconds) 2010-10-06T10:11:37 *** amstan has quit IRC (Ping timeout: 252 seconds) 2010-10-06T10:11:46 see... and when i play on the TCP server, i get paired up with RageBot -_- 2010-10-06T10:13:49 sleep $(($RANDOM % 60) between TCP games ftw 2010-10-06T10:14:31 i don't run a script... i just run the command every once in a while 2010-10-06T10:15:06 last night someone was running a buggy bot against TCP.. which was just losing by invalid input and immediately re-connecting... 2010-10-06T10:15:19 nice 2010-10-06T10:16:03 free wins for everyone! 2010-10-06T10:16:40 ;-) 2010-10-06T10:18:09 rebelxt, run your bot on TCP, maybe we can play each other 2010-10-06T10:20:17 *** amstan has joined #aichallenge 2010-10-06T10:20:18 *** ChanServ sets mode: +o amstan 2010-10-06T10:20:47 *** amstan has quit IRC (Remote host closed the connection) 2010-10-06T10:22:52 cfaftw: sure, just started it - rebelxt100 2010-10-06T10:23:15 now the question is... how do we get paired together 2010-10-06T10:23:24 my bot is JetBot.2 2010-10-06T10:24:03 we could try clicking enter on 1-2-3 ;-) 2010-10-06T10:24:11 ok let me finish this game 2010-10-06T10:24:16 it's worth a shot, haha 2010-10-06T10:24:48 anyone else having trouble with uploading new submissions? 2010-10-06T10:24:52 oh man i'm gonna lose this.. 2010-10-06T10:25:39 ok 2010-10-06T10:25:41 let's do it 2010-10-06T10:25:49 um 2010-10-06T10:25:59 ready to press enter? 2010-10-06T10:26:03 yep 2010-10-06T10:26:04 i'll just say 'now' and then hit it 2010-10-06T10:26:07 k 2010-10-06T10:26:11 now 2010-10-06T10:26:21 nice 2010-10-06T10:26:24 *** sinan has joined #aichallenge 2010-10-06T10:26:24 it worked 2010-10-06T10:26:44 if you looked at the TCP server, lovebot seems to be very similar to my bot 2010-10-06T10:26:44 can i survive 100 turns? ;) 2010-10-06T10:26:49 you might win 2010-10-06T10:27:03 i mean 2010-10-06T10:27:08 i just lost to "randomtest2" 2010-10-06T10:27:22 it's going all the way! 2010-10-06T10:27:28 and i just lost to lovebot ;) 2010-10-06T10:27:38 200 it is 2010-10-06T10:28:08 gg 2010-10-06T10:28:26 waiting on replay on the site 2010-10-06T10:28:30 ditto 2010-10-06T10:28:47 *** mightybyte has joined #aichallenge 2010-10-06T10:29:56 that was close! 2010-10-06T10:29:58 just watched - GG 2010-10-06T10:30:15 we exchanged planets multiple times, but your bot couldn't hold on 2010-10-06T10:30:26 yeah, very close till 100ish 2010-10-06T10:30:56 *** Nuriaion has joined #aichallenge 2010-10-06T10:33:02 here we go again 2010-10-06T10:33:06 looking at your bot's games on tcp - very nice! i can't match this.. need more tweaks to get even close 2010-10-06T10:33:19 /runs away 2010-10-06T10:33:44 that was quick ;-) 2010-10-06T10:34:06 yeah 2010-10-06T10:34:22 looking at my loss against randomtest2, i definitely need to figure out a better way to pick planets 2010-10-06T10:36:10 i am getting errors uploading 2010-10-06T10:36:12 looking at our last game, my bot did something i DID not code!... gonna replay it locally and debug ;-) 2010-10-06T10:36:20 haha 2010-10-06T10:36:40 *** Titankiller has joined #aichallenge 2010-10-06T10:37:21 "Failed to move file from temporary to permanent location." 2010-10-06T10:37:47 optimum, must be server issue. submission was working last night 2010-10-06T10:38:01 *** Meatkat has joined #aichallenge 2010-10-06T10:40:06 Sorry, what was that about Python and the tcp server, I timed out... :( 2010-10-06T10:40:16 cfaftw: i got my own tcp server too at http://ec2-72-44-46-51.compute-1.amazonaws.com/planetwars/getratings, but your bot would probably beat all of mine (especially the previous versions) easily.. 2010-10-06T10:41:00 Meatkat, put "try: ... except EOFError, e: break" around the raw_input() call in your bot 2010-10-06T10:41:15 Meatkat: what cfaftw said. also try running a java sample bot against TCP server first - to rule out connection or TCP client issue 2010-10-06T10:41:42 raw_input() raises EOFError when EOF is encountered... so your bot still works with the TCP client, it just doesn't exit nicely :) 2010-10-06T10:42:42 cfaftw: are you using python too? 2010-10-06T10:42:49 rebelxt_, yes 2010-10-06T10:43:08 cfaftw: getting close to possibility of a time-out? 2010-10-06T10:43:14 not even close 2010-10-06T10:43:20 my longest turn takes around 10ms 2010-10-06T10:43:53 cfaftw: wow.... i'm doing something wrong... 250ms longest 2010-10-06T10:44:00 wrong language 2010-10-06T10:44:22 rebelxt_, you have psyco enabled? 2010-10-06T10:44:41 it's enabled in the starter python pack, i believe 2010-10-06T10:45:57 hmm the current top 3 entries are written in java and python 2010-10-06T10:46:09 cfaftw: no, not using psyco. i just checked original starter pack - it's not ebabled 2010-10-06T10:46:16 enabled* 2010-10-06T10:46:29 that's your problem 2010-10-06T10:46:45 well i never ran my bot without psyco 2010-10-06T10:46:52 *** wvdschel has quit IRC (Quit: Leaving) 2010-10-06T10:47:00 but i'm positive it'll speed up your bot 2010-10-06T10:47:09 cfaftw: so, when you submit your bot, you don't take out psyco, and it works? 2010-10-06T10:47:20 yep 2010-10-06T10:47:42 cfaftw: nice, good hint - i'll try it now 2010-10-06T10:48:06 rebelxt_, use this: http://pastebin.com/82r0sfQ2 2010-10-06T10:48:23 not sure if psyco is enabled on the server (though i'd be surprised if it wasn't) 2010-10-06T10:48:38 that try/except clause prevents it from crashing if psyco is not available 2010-10-06T10:49:00 cfaftw: tnx 2010-10-06T10:50:59 *** rebelxt_ has quit IRC (Quit: Page closed) 2010-10-06T10:51:00 *** DanielVF has quit IRC (Read error: Connection reset by peer) 2010-10-06T10:51:07 *** DanielVF has joined #aichallenge 2010-10-06T10:51:55 Yay! http://72.44.46.68/getplayer?player=Meatkat10.4 Thanks cfaftw 2010-10-06T10:52:32 *** hexsprite has joined #aichallenge 2010-10-06T10:53:26 np 2010-10-06T10:53:53 And excuse my ignorance but I've never heard of psyco, what is it and what does it do? 2010-10-06T10:54:07 Meatkat, it's a JIT compiler for python 2010-10-06T10:54:41 speeds up your code by compiling on the fly when certain sections of your code are executed frequently 2010-10-06T10:55:03 nice! 2010-10-06T10:57:38 rebelxt_, let me know what kind of speedups you get with psyco 2010-10-06T10:57:44 i'm curious now 2010-10-06T10:59:35 Top 10 players: ruilov(3926), dmj111(3861), davidjliu(3857), bix0r4ever(3840), Hazard(3811), felixcoto(3745), Arekku(3733), DocBot(3708), EBraun(3697), Rexxar(3689) 2010-10-06T11:02:06 i'm trying to get my bot to play dmj1.8 more often 2010-10-06T11:03:36 *** dvf has joined #aichallenge 2010-10-06T11:03:36 *** DanielVF has quit IRC (Read error: Connection reset by peer) 2010-10-06T11:07:26 *** Mayank_ has joined #aichallenge 2010-10-06T11:09:06 *** Mayank_ has left #aichallenge 2010-10-06T11:10:55 *** wvdschel has joined #aichallenge 2010-10-06T11:11:42 optimum: that error message *still*? jeez 2010-10-06T11:12:31 *** thearchangel has joined #aichallenge 2010-10-06T11:12:51 hey 2010-10-06T11:13:00 submissions working? 2010-10-06T11:13:06 *** fawek has quit IRC (Ping timeout: 245 seconds) 2010-10-06T11:14:28 anyone? 2010-10-06T11:14:43 other people have complained... not sure if it's working right now 2010-10-06T11:14:48 it worked last night when i submitted 2010-10-06T11:14:53 ohk 2010-10-06T11:14:59 mine's not working 2010-10-06T11:15:07 god 2010-10-06T11:15:11 how can it be so hard? 2010-10-06T11:15:18 what? 2010-10-06T11:15:31 the reason it's not working is because they are keeping too many old bots around 2010-10-06T11:15:37 old versions and stuff 2010-10-06T11:15:40 oh 2010-10-06T11:15:43 all they have to do is not keep them 2010-10-06T11:15:49 dint kno that 2010-10-06T11:15:57 i think they know what they are doing 2010-10-06T11:16:18 there's a patch in the queue to fix it, but it's nt merged in yet 2010-10-06T11:16:40 in the meantime, jeff made a script to delete old submissions, but it only deletes a certain amount of bots per time 2010-10-06T11:16:41 well currently when you resubmit, your ranking is reset and your games list is cleared 2010-10-06T11:16:45 it's obviously not keeping up 2010-10-06T11:16:53 why that limit was put there in the first place is beyond me 2010-10-06T11:16:57 so there's no real reason to keep any old submission around 2010-10-06T11:17:02 right 2010-10-06T11:17:11 *** fawek has joined #aichallenge 2010-10-06T11:17:30 the patch in the queue just moves old bots to a nested directory structures 2010-10-06T11:17:34 *structure 2010-10-06T11:17:39 so that filesystem limit isn't hit 2010-10-06T11:18:05 i think the main reason jeff wants to keep the old ones around is "just in case" 2010-10-06T11:18:13 sure 2010-10-06T11:18:28 i'm more curious about the server slowdown 2010-10-06T11:18:43 it's going slow? 2010-10-06T11:18:56 bots are getting about 1 game every 3 hours or so 2010-10-06T11:19:02 2-3 hours 2010-10-06T11:20:56 cfaftw: looks like we lost some servers http://ai-contest.com/game_stats.php 2010-10-06T11:21:05 yeah i saw the stat page 2010-10-06T11:21:23 but, it was the first time i'd seen it, so i don't know how many there were originally 2010-10-06T11:21:46 some of them were just for testing, but i think there were four or five at once point 2010-10-06T11:21:56 either way, we did have a huge game rate at one point 2010-10-06T11:23:34 yeah we did... back in the golden days 2010-10-06T11:23:49 bots were getting like 5-6 games per hour 2010-10-06T11:23:50 amazing 2010-10-06T11:24:01 cfaftw: this wasn't long ago http://ai-contest.com/forum/viewtopic.php?f=18&t=837&p=5345#p5345 2010-10-06T11:24:13 5 days i guess 2010-10-06T11:24:29 jmcarthur, maybe they didn't pay their bills :( 2010-10-06T11:24:33 lol 2010-10-06T11:25:34 *** wvdschel has quit IRC (Quit: Leaving) 2010-10-06T11:27:31 *** shane__ has joined #aichallenge 2010-10-06T11:27:42 Hi all 2010-10-06T11:27:50 *** shane__ is now known as ShaneHudson 2010-10-06T11:27:52 yo 2010-10-06T11:28:39 cfaftw = chick-fil-a for the win 2010-10-06T11:28:44 I am just about to start my first bot, so thought it would be a good idea to connect to irc :D 2010-10-06T11:28:48 ahaha 2010-10-06T11:28:51 hi shane 2010-10-06T11:28:53 :) 2010-10-06T11:29:09 i would like to see on that same graph player amount 2010-10-06T11:29:40 Shane, i hope you're ready to devote all of your free time to this addiction 2010-10-06T11:29:55 Lol sure, my free time is basically 0 :P 2010-10-06T11:30:02 as well as non-free time such as work 2010-10-06T11:30:13 yeah... now that you've started the bot :) 2010-10-06T11:30:43 before I start properly.. is there any disadvantages to using any language for this particular case? 2010-10-06T11:30:53 *** bhasker has quit IRC (Quit: bhasker) 2010-10-06T11:30:57 I see most are java and python, anything wrong with using c# for it? 2010-10-06T11:30:58 Shane, as a general rule, no 2010-10-06T11:31:07 use whatever you are comfortable with 2010-10-06T11:31:14 Ah good 2010-10-06T11:31:28 I shall use C# then, my Java is fine, but my computer is set up for C# at the moment lol 2010-10-06T11:31:50 if you plane on doing some algorithm that is insanely expensive computationally, then you might want to use C++ or some equivilantly fast language 2010-10-06T11:32:00 there are some minor issues with C# version on the server 2010-10-06T11:32:01 but... i use python and have no problem keeping way under the 1s rule 2010-10-06T11:32:07 like versions incompatibility 2010-10-06T11:32:12 anyone know when submissions will be back up? 2010-10-06T11:32:25 I know no algorithms, I suppose I have a disadvantage in that field lol 2010-10-06T11:32:30 *** mega1 has quit IRC (Read error: Connection reset by peer) 2010-10-06T11:33:01 *** mega1 has joined #aichallenge 2010-10-06T11:33:30 Shane, that's fine... you learn heuristics as you go 2010-10-06T11:34:50 good :), I will not be trying for first place, no chance at that.. but I am looking forward to learning lots lol 2010-10-06T11:35:03 dude, go for first! 2010-10-06T11:35:23 Heh 2010-10-06T11:35:34 you'd be surprised 2010-10-06T11:36:20 well ye I would be! xD 2010-10-06T11:36:37 *** sinan has quit IRC (Quit: WeeChat 0.2.6.3) 2010-10-06T11:36:43 So are you a professional, student, hobbiest or other? lol 2010-10-06T11:36:58 well my day job is programming 2010-10-06T11:37:05 but i am a hobbiest programmer, too 2010-10-06T11:37:24 Ah fair enough 2010-10-06T11:37:28 graduated from University of Maryland last year 2010-10-06T11:37:34 Congratulations 2010-10-06T11:37:51 I hope to be going to the University of Sussex next year 2010-10-06T11:37:51 what about you? 2010-10-06T11:38:27 I am a student/website developer and hobbiest when I get time 2010-10-06T11:38:39 cool 2010-10-06T11:38:52 going to do CS at college? 2010-10-06T11:39:04 counter strike! 2010-10-06T11:39:22 *** thearchangel has quit IRC (Ping timeout: 240 seconds) 2010-10-06T11:39:29 aye, hopefully! 2010-10-06T11:39:35 rabidus, nowadays it's more MC 2010-10-06T11:39:40 And hahaha a degree in counter strike would be good :P 2010-10-06T11:40:21 Uh... javac is not recognised? Surely I have installed the jdk in the last year :O! 2010-10-06T11:41:06 Meatkat, what in the world is your bot doing? it's timing out like crazy 2010-10-06T11:41:28 cfaftw: I'm in Burkina Faso, I'm my connection... :( 2010-10-06T11:41:35 (that's in West Africa) 2010-10-06T11:41:39 riiight I go to install and it says its already installed, thats strange 2010-10-06T11:41:41 oh :\ 2010-10-06T11:41:43 its* my connection 2010-10-06T11:42:23 How do you fight the bots online? I have only done it locally so far (or do you need a bot uploaded?) 2010-10-06T11:42:49 http://www.ai-contest.com/forum/viewtopic.php?f=18&t=424 2010-10-06T11:42:58 there's a TCP server you can use to spar with other bots more quickly 2010-10-06T11:43:01 and keep debugging info 2010-10-06T11:43:06 I'm looking into implementing that psyco thing now, maybe that will help though I haven't had problems timing out locally ever, it would be nice to be down to 10ms like you though 2010-10-06T11:43:15 ok, stupid me.. following a java tutorial would explain why I am trying to compile java when I am doing c#! Oh deary me lol 2010-10-06T11:43:16 or you can submit your bot to the official server, but it's playing games very slowly right now 2010-10-06T11:43:44 ah ok thanks 2010-10-06T11:44:20 Meatkat, how long do your turns take right now? 2010-10-06T11:44:49 I'm not sure, how would I find out? 2010-10-06T11:45:47 i just use time.time() 2010-10-06T11:46:11 save the time at the beginning of my turn, then log the time diff in a debug message at the end of my turn 2010-10-06T11:46:14 and just calculate it in-turn? 2010-10-06T11:46:36 right-o, give me a sec 2010-10-06T11:47:51 *** McLeopold has joined #aichallenge 2010-10-06T11:49:38 *** MarcAndrePC has joined #aichallenge 2010-10-06T11:49:59 *** irchs has joined #aichallenge 2010-10-06T11:52:22 *** fawek has quit IRC (Ping timeout: 272 seconds) 2010-10-06T11:52:58 *** BegemoT has joined #aichallenge 2010-10-06T11:53:35 *** BegemoT has quit IRC (Client Quit) 2010-10-06T11:54:56 *** Error323 has joined #aichallenge 2010-10-06T11:55:05 Hi 2010-10-06T11:55:15 syp 2010-10-06T11:55:19 i mean, sup 2010-10-06T11:55:34 cfaftw: My average turn is just over 100ms, max is around 281ms 2010-10-06T11:55:38 I assume ppl here are aware that uploading submissions is failing 2010-10-06T11:55:45 Meatkat, did you try psyco yet? 2010-10-06T11:57:01 cfaftw, yeah, I looked in the package and it imports pysco, is that all you have to do? 2010-10-06T11:57:08 Do you guys perform any type of map analysis? 2010-10-06T11:57:35 *** irchs has quit IRC (Remote host closed the connection) 2010-10-06T11:58:21 Meatkat, import psyco 2010-10-06T11:58:24 Meatkat, psyco.full() 2010-10-06T11:59:04 *** irchs has joined #aichallenge 2010-10-06T11:59:08 cfaftw, yeah's that in there, let me comment it out and see what happens... 2010-10-06T11:59:24 Error323: yes 2010-10-06T11:59:25 Meatkat, also make sure you have psyco installed 2010-10-06T11:59:35 Top 10 players: ruilov(3932), dmj111(3868), davidjliu(3863), bix0r4ever(3842), Hazard(3817), felixcoto(3746), Arekku(3740), DocBot(3709), EBraun(3700), goldman(3681) 2010-10-06T11:59:58 *** smellyhippy has quit IRC (Ping timeout: 272 seconds) 2010-10-06T12:00:05 !help 2010-10-06T12:00:05 Error323: (help [] []) -- This command gives a useful description of what does. is only necessary if the command is in more than one plugin. 2010-10-06T12:00:30 *** bhasker has joined #aichallenge 2010-10-06T12:00:33 cfaftw, can I apt-get it? 2010-10-06T12:00:44 aptitude search psyco 2010-10-06T12:00:57 should be there somewhere 2010-10-06T12:01:23 *** smellyhippy has joined #aichallenge 2010-10-06T12:01:23 *** smellyhippy has joined #aichallenge 2010-10-06T12:01:57 cfaftw, nope, not there 2010-10-06T12:02:06 cfaftw, or at least no results for psyco 2010-10-06T12:02:13 make sure not to miss-type as 'apt-get install psycho' 2010-10-06T12:02:32 unless you want to see Hitchcock once again 2010-10-06T12:02:42 :o 2010-10-06T12:03:13 i'm on fedora right now, so i can't help finding the package 2010-10-06T12:03:17 but i know there's a psyco package 2010-10-06T12:03:58 Is there an estimated time on when uploading new submissions is possible again? 2010-10-06T12:04:18 it's called python-psyco 2010-10-06T12:04:33 I can see it in aptitude, if it helps 2010-10-06T12:04:40 Error323: afaik, no 2010-10-06T12:05:17 mmm pitty 2010-10-06T12:06:56 *** Meatkat has quit IRC (Ping timeout: 265 seconds) 2010-10-06T12:07:23 rabidus: do you know how long it takes these days for a bot to reach an accurate rank? 2010-10-06T12:08:00 Error323, some servers are down right now... could take days if they don't get them up soon 2010-10-06T12:08:09 Error323, use the TCP server to test your bot in the meantime 2010-10-06T12:08:14 if servers works properly.. hour or two 2010-10-06T12:08:17 cfaftw: TCP server? 2010-10-06T12:08:42 Error323, http://www.ai-contest.com/forum/viewtopic.php?f=18&t=424 2010-10-06T12:09:22 op, that link should be added to the topic 2010-10-06T12:09:39 oh thats nice 2010-10-06T12:09:42 thanks cfaftw 2010-10-06T12:14:05 *** u1tr0n has joined #aichallenge 2010-10-06T12:14:08 hi all 2010-10-06T12:14:13 hi 2010-10-06T12:14:27 hi 2010-10-06T12:14:41 again upload didnt work :( 2010-10-06T12:15:04 it's broken again 2010-10-06T12:15:12 nobody knows when it'll be back up 2010-10-06T12:15:20 also the servers are slow and there's no word on that either 2010-10-06T12:15:27 now you're caught up :) 2010-10-06T12:16:19 why can I not figure out how to get a c# bot working lol 2010-10-06T12:17:53 because, as it turns out, you work for C# bot 2010-10-06T12:18:36 im upload to test bot with void loop late try to upload normal bot and blah... 2010-10-06T12:18:55 do you get errors like "unknown keyword at or before 'lol'" or something? 2010-10-06T12:21:40 I mean to compile it etc. for java it goes to .jar, what does c# go to? exe or project? there is little documentation 2010-10-06T12:21:58 ShaneHudson, why are you using C# if you don't know how to use it? 2010-10-06T12:22:49 it sounds like you know Java; you should probably use the Java starter kit 2010-10-06T12:22:52 I do lol, I have made loads of things in it.. I just have never used it with the engine etc 2010-10-06T12:23:19 Ye I shall do java instead, more documentation for a start lol 2010-10-06T12:23:58 and as a general rule, language specific questions should be directed to the appropriate channel 2010-10-06T12:25:01 oh? I didn't realise there were other channels, sorry! 2010-10-06T12:25:16 o_O 2010-10-06T12:26:36 haha, it's ok... FreeNode is a large IRC server with a channel for just about any programming language or compsci topic 2010-10-06T12:27:19 yeah I imagine everybody at c# channel helping like crazy after "why can I not figure out how to get a c# bot working lol" question :) 2010-10-06T12:28:32 haha, good point 2010-10-06T12:28:37 Oh general c#, thought you meant there were language specific channels for ai-contest.com 2010-10-06T12:29:08 Which is what confused me, as I had read nothing about it haha 2010-10-06T12:30:50 *** mightybyte has left #aichallenge 2010-10-06T12:31:47 *** Sylvain3665 has joined #aichallenge 2010-10-06T12:35:25 I forgot how easy it is to compile java :D 2010-10-06T12:41:04 hi, i have got a problem when i submit my bot : Failed to move file from temporary to permanent location 2010-10-06T12:41:45 submission is currently broken 2010-10-06T12:41:49 no eta on when it'll be fixed 2010-10-06T12:41:59 probably when the contest organizers get out of class :) 2010-10-06T12:42:10 ok :) thanks 2010-10-06T12:46:50 *** Bobng has joined #aichallenge 2010-10-06T12:47:37 *** justin_pdx has joined #aichallenge 2010-10-06T12:48:20 Does the visualiser usually take forever to play? 2010-10-06T12:48:45 *** Meatkat has joined #aichallenge 2010-10-06T12:49:22 *** justin_pdx has quit IRC (Client Quit) 2010-10-06T12:49:34 the java visualizer? 2010-10-06T12:51:12 *** Accoun has quit IRC (Ping timeout: 265 seconds) 2010-10-06T12:51:54 cfaftw, just got installed, it dropped my times by about 50%!!! I'm at an average of 50ms now 2010-10-06T12:52:04 *** fawek has joined #aichallenge 2010-10-06T12:52:12 Meatkat, awesome... i knew it would help 2010-10-06T12:52:36 hopefully it will prevent you from timing out on the TCP server 2010-10-06T12:53:04 One hopes, mais je panse mon connection est toujours la merde 2010-10-06T12:53:06 umm... if your previous avg time was 100ms, then your drop was 200% 2010-10-06T12:53:46 100ms to 50ms... ? 2010-10-06T12:53:49 rabidus, he meant dropped TO 50%, i think 2010-10-06T12:53:54 yeah, 2 times faster, 200% 2010-10-06T12:54:11 but if you say like TO 50%, then thats true :) 2010-10-06T12:54:24 nvrmnd, everybody understood 2010-10-06T12:54:33 la merde - I understood that! :) 2010-10-06T12:54:43 i feel like using percentages causes so much confusion 2010-10-06T12:54:43 My speed-up was 200% yes... 2010-10-06T12:54:58 or... ? whatever 2010-10-06T12:55:15 In any case 2010-10-06T12:55:18 mmm 2010-10-06T12:55:39 I am far from timing out with my bot 2010-10-06T12:55:51 *** Accoun has joined #aichallenge 2010-10-06T12:57:58 Hmm according to the simple strategy guide, planet size matters? How come the size matters? Surely it depends how many fleets are on each planet 2010-10-06T12:58:01 So I resubmitted my bot and got paired up with a 2600, I think at this rate I'll reach my old rank sometime this weekend 2010-10-06T12:58:36 ShaneHudson: The size of the planet in the visualizer matters because it tells you the planets regen rate - the bigger the planet the faster the regen rate 2010-10-06T12:59:02 you can submit to the server Meatkat ? I still get php errors 2010-10-06T12:59:10 but from the in game perspective planets don't have sizes, just positions, populations, and regen rates 2010-10-06T12:59:37 Nope, I submitted mine about 6 or so hours ago, I've got 4 games in! Score 2010-10-06T12:59:37 Top 10 players: ruilov(3924), dmj111(3863), davidjliu(3844), bix0r4ever(3798), Hazard(3797), felixcoto(3761), Arekku(3730), DocBot(3717), EBraun(3690), goldman(3688) 2010-10-06T12:59:49 Meatkat: lol 2010-10-06T12:59:50 xD 2010-10-06T13:00:14 Oh regen rate, thank you very much! 2010-10-06T13:00:48 i created a simple sh script if you wanna test against local previous versions of your bot: http://paste.pocoo.org/show/271818/ 2010-10-06T13:00:54 *** justin_pdx has joined #aichallenge 2010-10-06T13:00:56 and my connection just made me time out playing RageBot... embrassing... 2010-10-06T13:01:23 lol @ Meatkat 2010-10-06T13:01:49 you know what annoys me, my bot plays way better when it only defends 2010-10-06T13:02:08 Work on your neutral selection algorithms then 2010-10-06T13:02:21 yep 2010-10-06T13:02:34 just don't know how yet >_> 2010-10-06T13:02:38 Most bots are too agressive as is, look at all the close start maps 2010-10-06T13:03:04 yeah at close start maps my bot is epicness :D 2010-10-06T13:03:13 Hmm... well first you want to see if you CAN take a neutral planet, as in your have more troops in teh area 2010-10-06T13:03:34 lol 2010-10-06T13:03:52 then you have to figure out if you want to take it, considering where you would be pulling the resources from and if the enemy could snipe, etc... 2010-10-06T13:04:57 Then ... and profit! 2010-10-06T13:05:20 was that vague enough not to help at all? :P 2010-10-06T13:06:18 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-10-06T13:06:22 not vague no =) 2010-10-06T13:06:50 but obvious indeed :P 2010-10-06T13:06:55 1. code unbeatable bot 2. ?????? 3. profit 2010-10-06T13:07:10 :> 2010-10-06T13:07:29 i think "upload successfully" should be part of step 2. 2010-10-06T13:08:43 LOL my bot had 1 enemy planet left, but went to ALL the neautral planets first, i guess that needs changing lol 2010-10-06T13:08:50 Zannick: LOL 2010-10-06T13:09:11 +1 on that 2010-10-06T13:09:37 Who is in charge of the tcp server? 2010-10-06T13:09:45 dhartmei 2010-10-06T13:10:29 *** amstan has joined #aichallenge 2010-10-06T13:10:29 *** ChanServ sets mode: +o amstan 2010-10-06T13:10:52 *** justin_pdx has joined #aichallenge 2010-10-06T13:10:55 I'm going to ask him if he could up the time limits a bit for responses, if someone really wants to cheat and use extra time then great, but it won't help them in the contest at all and for me maybe I'll actually be able to play some games 2010-10-06T13:12:01 Meatkat, you can play multiple games at once. 2010-10-06T13:12:16 *** justin_pdx has quit IRC (Client Quit) 2010-10-06T13:12:52 McLeopold I'm playing from Burkina Faso, my connection is very, very, very bad. I can't respond in time: http://72.44.46.68/getplayer?player=Meatkat10.4 2010-10-06T13:13:22 ouch 2010-10-06T13:14:01 are you guys also still students? =) 2010-10-06T13:14:06 yeah... :( and now with the main servers going down I'm not getting much to on in terms of feedback from games 2010-10-06T13:14:43 Meatkat, are you using python? 2010-10-06T13:14:51 Is Ademar.Gonzalez on or anyone know what nick he goes by here? 2010-10-06T13:15:41 McLeopold Yeah, cfaftw just told me about installing psyco and that dropped my times from about 100ms to 50ms, I also changed the code to handle the EOFError, you were the one that posted that on the forums weren't you? 2010-10-06T13:15:51 yeah 2010-10-06T13:16:17 Thanks, that helped us all a lot. 2010-10-06T13:16:39 meatkat, are you connected to the internet through telegraph wire or something? 2010-10-06T13:16:52 McLeopold: I remember the first week you were #1 with a star, you going to be able to take it back? 2010-10-06T13:16:53 how does it take 900ms to send your responses? 2010-10-06T13:17:12 Meatkat: yes, I will be #1 next week. 2010-10-06T13:17:32 xD 2010-10-06T13:17:34 mcleopold, why next week? why not NOW 2010-10-06T13:17:37 McLeopold Excellent 2010-10-06T13:17:46 cfaftw: bugs... 2010-10-06T13:18:09 so you've written The One bot, eh? 2010-10-06T13:18:21 did the server die yet? 2010-10-06T13:18:24 any problems with it? 2010-10-06T13:18:29 #rankings 2010-10-06T13:18:30 amstan: Top 10 players: ruilov(3917), dmj111(3858), davidjliu(3830), Hazard(3797), bix0r4ever(3791), felixcoto(3752), Arekku(3717), DocBot(3715), EBraun(3687), goldman(3676) 2010-10-06T13:18:36 it's just slow, and it won't accept submissions 2010-10-06T13:18:37 Well, it's good. It doesn't play position yet, but I'm working on that. 2010-10-06T13:18:44 oh nto that again.. 2010-10-06T13:18:49 cfaftw I'm using a cellular modem to connect to the internet, which then has to be relayed out of one of the poorest countries in the world to one of the two (three?) lines leading out of Africa to the states 2010-10-06T13:19:06 Meatkat, so... worse than a telegraph wire :) 2010-10-06T13:19:15 contestbot: repo 2010-10-06T13:19:16 amstan: repo = http://code.google.com/p/ai-contest/ 2010-10-06T13:19:18 amstan: I have a script in the move-old-submissions branch to move out all the old submissions 2010-10-06T13:19:20 cfaftw haha... well almost 2010-10-06T13:19:29 And there was me wanting to move to Japan for faster internet, you deserve it more than I :P 2010-10-06T13:19:32 Janzert: i can't even get to the bash shell 2010-10-06T13:19:43 ugh 2010-10-06T13:19:51 ShaneHudson where are you now? 2010-10-06T13:20:04 *** Titankiller has quit IRC (Read error: Connection reset by peer) 2010-10-06T13:21:22 k 2010-10-06T13:21:26 guys.. try uploading now 2010-10-06T13:21:34 Janzert: i don't understand why we need them moved.. 2010-10-06T13:21:53 amstan: still failing 2010-10-06T13:21:53 to avoid the filesystem limit on number of files in a directory 2010-10-06T13:22:06 Janzert: why isn't the delete script good enough? 2010-10-06T13:22:12 why do we need to move them instead? 2010-10-06T13:22:14 misremembered branch name btw it's old-submission-cleanup 2010-10-06T13:22:15 Meatkat: I am in England, pay for a 8mb connection and get 0.3 2010-10-06T13:22:44 eesh 2010-10-06T13:22:55 so you still have them if they are ever needed. sounded like j3camero didn't like the idea of deleting them 2010-10-06T13:23:26 ShaneHudson Fun.... :( That sucks 2010-10-06T13:23:26 it could easily be changed to delete them of course 2010-10-06T13:23:55 we have a delete script already.. 2010-10-06T13:23:57 I'm not sure how the current one works, but it will also never delete a submission that is still in use 2010-10-06T13:24:01 janzert, amstan, any word on when more servers will be back up? 2010-10-06T13:24:08 s/delete/move 2010-10-06T13:24:58 if we move them we'll have the same problem in the destination.. 2010-10-06T13:24:59 cfaftw: I think danielvf has to deal with that 2010-10-06T13:25:01 exactly the same problem 2010-10-06T13:25:13 it moves them to a different directory each time it is run 2010-10-06T13:25:24 basically old_submissions/date-time/ 2010-10-06T13:27:52 one problem it does have is once there are around 10000 active participants it will allow the active submissions directory to start hitting the limit again because it keeps the 3 latest submissions for each participant in the submissions directory 2010-10-06T13:28:08 *** Meatkat has quit IRC (Ping timeout: 265 seconds) 2010-10-06T13:28:08 the problem is with this supid delete script.. 2010-10-06T13:28:49 it deletes the old ones 2010-10-06T13:28:58 except.. it doesn't mark them deleted in any way 2010-10-06T13:29:07 so next time you run it, it deletes the same ones 2010-10-06T13:29:16 oops 2010-10-06T13:29:20 ...huh 2010-10-06T13:29:47 unless you use increasing n numbers 2010-10-06T13:29:49 mine does check that it is only moving scripts that are actually still in the submissions directory :) 2010-10-06T13:30:05 i want a link 2010-10-06T13:30:25 http://code.google.com/p/ai-contest/source/browse/branches/old-submission-cleanup/planet_wars/backend/move_old_submissions.py 2010-10-06T13:32:09 the other reason I liked moving them instead of deleting is that if it somehow did mess up and move the wrong submissions they can always just be moved back 2010-10-06T13:32:26 Janzert: idk about the performance of your script 2010-10-06T13:32:33 there's over 40000 rows in that table 2010-10-06T13:32:38 not a good idea to do a select without a limit 2010-10-06T13:34:01 does anyone know who "CCCF-mof3-asbp8-caf12" is on the tcp server? 2010-10-06T13:34:15 creative name 2010-10-06T13:34:37 Hmm, I wouldn't expect 40000 to be a problem but I've not used mysql. 2010-10-06T13:35:12 I've pulled from tables with over a million lines without a problem before on sqlite and postgres 2010-10-06T13:35:24 million rows that is 2010-10-06T13:36:06 also I suppose that was on an otherwise unloaded database as well 2010-10-06T13:37:27 *** Florian has joined #aichallenge 2010-10-06T13:37:28 yeah, idk, jeff has the final say for it 2010-10-06T13:37:34 *** Florian has quit IRC (Remote host closed the connection) 2010-10-06T13:38:04 yeah, ok. I left him a code review for it :) 2010-10-06T13:38:19 *** MarcAndrePC has quit IRC (Quit: Page closed) 2010-10-06T13:39:04 *** amstan has quit IRC (Quit: Konversation terminated!) 2010-10-06T13:39:16 *** amstan has joined #aichallenge 2010-10-06T13:39:16 *** ChanServ sets mode: +o amstan 2010-10-06T13:39:44 *** Florian95 has quit IRC (Ping timeout: 240 seconds) 2010-10-06T13:39:53 if a column was added to mark submissions that have been moved out then you could be sure no more than 32000 rows would ever be returned since otherwise the filesystem limit would have been hit already :) 2010-10-06T13:40:09 yeah, that would work 2010-10-06T13:40:14 but then we need schema changes 2010-10-06T13:40:17 this is so fucked up 2010-10-06T13:40:24 right 2010-10-06T13:40:25 lol 2010-10-06T13:41:06 oh man.. i wish we could just redesign 2010-10-06T13:42:47 are the maps in the starterpackages equal to those on the server? 2010-10-06T13:43:16 amstan: um 2010-10-06T13:43:24 the script can just test to see if the files exist 2010-10-06T13:43:39 rather than simply try to rm them 2010-10-06T13:44:04 why would you need a column? 2010-10-06T13:44:41 the added column was just so the script can skip pulling the submission from the database at all 2010-10-06T13:44:50 ah 2010-10-06T13:45:36 something like select * from submissions where submissions.old = 0; 2010-10-06T13:45:40 or whatever 2010-10-06T13:46:07 *** keynes has quit IRC (Quit: Bye) 2010-10-06T13:47:20 on a different subject any one know if ademar.gonzalez comes here and under what nick if he does? 2010-10-06T13:50:45 amstan: to answer your yesterday night question, my forum user is also jfrank... im glad that didn't cause too much trouble. I was going to suggest deleting and recreating the org add and change thread with a "post using this format" OrgName|id,id,id then getting more orgchanges.txt would be simple 2010-10-06T13:51:17 jfrank: i'm going to go ahead and lock them then 2010-10-06T13:51:31 jfrank: it would be awesome if you could compile me a list with updates from there 2010-10-06T13:51:59 amstan: from the new thread? or the one you're locking 2010-10-06T13:52:07 the one i'm locking 2010-10-06T13:52:12 new thread should be fine.. 2010-10-06T13:52:30 amstan: I took everything from your last post, assuming you did everything before that 2010-10-06T13:52:38 up till yesterday 2010-10-06T13:53:45 but there are a few more. lets see... one second 2010-10-06T13:54:22 *** flezru has joined #aichallenge 2010-10-06T13:55:12 *** Meatkat has joined #aichallenge 2010-10-06T13:56:53 i love my bot for being stupid yet successful 2010-10-06T13:57:25 cfaftw how stupid? 2010-10-06T13:57:36 cfaftw: how successfull? :P 2010-10-06T13:57:46 Meatkat, well, just that i don't do anything "fancy" 2010-10-06T13:57:59 define fancy =) 2010-10-06T13:58:10 So.... RageBot v2.0? 2010-10-06T13:58:21 Stick to the basics... :P 2010-10-06T13:58:28 my bot doesn't defend at all 2010-10-06T13:58:31 no minimax, simplexes, graphs 2010-10-06T13:58:43 no lookahead 2010-10-06T13:59:06 *** krokokrusa has quit IRC (Ping timeout: 255 seconds) 2010-10-06T13:59:07 mine just uses a weighted average with naive lookahead 2010-10-06T13:59:10 minimax is quite unfeasable with the actionspace that is available 2010-10-06T13:59:15 of course i'll have to step up my game soon 2010-10-06T13:59:18 really? everything is made with state 0 data? 2010-10-06T13:59:37 Top 10 players: ruilov(3922), dmj111(3861), davidjliu(3837), Hazard(3797), bix0r4ever(3779), felixcoto(3773), Arekku(3718), DocBot(3711), EBraun(3688), goldman(3667) 2010-10-06T13:59:48 the more time that passes, the smarter bots are becoming 2010-10-06T14:00:12 do you know of anyone using minimax or alpha beta or the like? 2010-10-06T14:00:35 i've only heard talk of it... but i'm positive some of the top 10 are using it 2010-10-06T14:00:38 Remember the good old days when you could have an Elo of 3200 and be #1? I was very dissapointed when I hit 3200 and was barely on the first page! 2010-10-06T14:01:12 mmm i wanted to use that lol 2010-10-06T14:01:21 but its insane 2010-10-06T14:01:38 how would you apply that given the massive action space. 2010-10-06T14:01:49 massive pruning :) 2010-10-06T14:01:53 *** rebelxt has joined #aichallenge 2010-10-06T14:01:59 massive massive pruning 2010-10-06T14:02:19 massive massive massive pruning? 2010-10-06T14:02:23 *** oPless has quit IRC (Quit: This computer has gone to sleep) 2010-10-06T14:02:27 well, you can narrow the problem by just thinking of taking a planet as a move 2010-10-06T14:02:57 but what about partial moves? where people don't launch fleets till the last minute 2010-10-06T14:03:01 that leaves you with |planets|! actions 2010-10-06T14:03:06 jfrank: cool: http://www.ai-contest.com/forum/viewtopic.php?f=4&t=881&p=5629#p5629 2010-10-06T14:03:11 *** smellyhippy has quit IRC (Ping timeout: 252 seconds) 2010-10-06T14:03:12 that is faculty 2010-10-06T14:03:15 limit action space to conceivably rational actions, rank actions, select top ranks? 2010-10-06T14:03:31 optimum: yeah 2010-10-06T14:03:36 but there in lies the problem ;0 2010-10-06T14:03:48 amstan: here are the rest from the locked thread http://pastebin.com/raw.php?i=B2DMuZGb 2010-10-06T14:04:03 amstan: any word on server slowdown? 2010-10-06T14:04:47 i wonder if it is possible to formulate the problem with a convex objective function 2010-10-06T14:04:52 *** anoek has quit IRC (Remote host closed the connection) 2010-10-06T14:04:53 rebelxt: no idea 2010-10-06T14:05:52 player el percutor not found 2010-10-06T14:06:05 amstan: hopefully someone can take a look. seems like 2 servers are processing games, and server #0 is only doing 9 GPM 2010-10-06T14:07:11 how many games are played simultaniously? 2010-10-06T14:07:35 *** Zeiris_ has quit IRC (Ping timeout: 252 seconds) 2010-10-06T14:08:06 solved it, there was an underscore in there 2010-10-06T14:08:29 jfrank: thanks 2010-10-06T14:11:33 amstan: welcome ttyl 2010-10-06T14:11:45 *** jfrank has quit IRC (Quit: Page closed) 2010-10-06T14:12:19 optimum: convex objective function? 2010-10-06T14:12:42 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-10-06T14:13:15 *** RainCT has joined #aichallenge 2010-10-06T14:14:15 Error323: yeah... 2010-10-06T14:14:54 I never heard of it 2010-10-06T14:17:51 http://en.wikipedia.org/wiki/Optimization_(mathematics) 2010-10-06T14:19:05 cfaftw: gonna be slow grind up to the top. stats show only single server playing games at the moment. 9gpm ;-( 2010-10-06T14:19:13 =-O 2010-10-06T14:19:20 server stats link please 2010-10-06T14:19:42 check out CCCF-mof3-asbp8-caf12 on the TCP server... that bot is good 2010-10-06T14:19:47 http://ai-contest.com/game_stats.php 2010-10-06T14:20:12 hahahaha... that is not good 2010-10-06T14:20:20 maybe my bot will get one more game... this week 2010-10-06T14:22:10 ok, just one more game 2010-10-06T14:22:15 may your wish come true 2010-10-06T14:22:27 j/k 2010-10-06T14:23:05 wow 2010-10-06T14:26:23 i wonder what's wrong with the servers 2010-10-06T14:29:55 oh no it's down to 8gpm 2010-10-06T14:30:21 the main server is always quite low and the cloud servers seem to go up and down pretty frequently 2010-10-06T14:30:31 not sure of why they do though :/ 2010-10-06T14:30:38 they're hosted by Amazon, right? 2010-10-06T14:30:46 *** felixcoto has joined #aichallenge 2010-10-06T14:30:48 amstan: any chance we could ping DanielVonFange about server issue 2010-10-06T14:30:52 yeah EC2 2010-10-06T14:31:06 that doesn't speak well for Amazon :) 2010-10-06T14:31:21 (assuming it's a hardware issue and not buggy server code) 2010-10-06T14:31:24 rebelxt: he's not online.. 2010-10-06T14:31:41 hehe, well, amazon EC2 is actually great. i think it's actually servers getting bogged down by our buggy bots ;-) 2010-10-06T14:32:01 down to 7gpm! 2010-10-06T14:32:17 she's gonna crash! 2010-10-06T14:32:26 pull up! pull up! 2010-10-06T14:32:58 amstan: got it. sent him a PM 2010-10-06T14:33:10 I will say after jumping into the bitchfest that overall they've done a pretty wonderful job with everything. There have been a couple minor hitch ups here and there but overall it's been solid. 2010-10-06T14:33:58 agreed 2010-10-06T14:34:04 yeah my expeience has been that the likely problem when faced with either hardware or your code... is your code ;) 2010-10-06T14:34:10 cfaftw: it got your message, back to 8 2010-10-06T14:35:02 First rule of the AI challenge: the more you refresh the stats page, the slower the server gets 2010-10-06T14:35:30 I think I am finally getting the hang of this! Though for some reason my bot takes over ALL planets rather than just the green and some neutral! lol 2010-10-06T14:35:35 that's why i'm playing on the TCP server 2010-10-06T14:36:09 ShaneHudson Good work mate 2010-10-06T14:36:25 I avenged my loss against RageBot! http://72.44.46.68/canvas?game_id=341624 2010-10-06T14:36:41 good jaerb 2010-10-06T14:36:56 cfaftw what bots do you have running on the TCP server? 2010-10-06T14:37:07 JetBot.2 2010-10-06T14:38:19 i think my bot just played a map that had no planets... or only two planets... or planets that were right on top of each other 2010-10-06T14:38:28 http://72.44.46.68/canvas?game_id=341869 2010-10-06T14:38:31 the game was a draw 2010-10-06T14:38:38 but no moves were performed, and the game doesn't render 2010-10-06T14:39:25 Weird 2010-10-06T14:41:03 Woah, now thats clever lol 2010-10-06T14:42:20 *** Meatkat has quit IRC (Quit: Page closed) 2010-10-06T14:43:40 btw, how is google sponsoring this challenge ? 2010-10-06T14:48:38 server#0 is at 6 gpm 2010-10-06T14:48:48 I have no idea, but it certainly adds to the "trust" of the site I think. As soon as I read Google AI Contest I signed up lol 2010-10-06T14:49:21 *** Migi32 has joined #aichallenge 2010-10-06T14:49:43 *** perror has quit IRC (Quit: Bye all !) 2010-10-06T14:54:42 drodil: probably helping with server bills, maybe more. also noticed lots of google employees are participating 2010-10-06T14:56:04 okay 2010-10-06T14:57:09 the google employees aren't doing very well 2010-10-06T14:57:53 of course how well you do is directly linked to the amount of effort you put in 2010-10-06T14:59:10 exactly. rating = function(time, brain, programming skill, luck) 2010-10-06T14:59:37 Top 10 players: ruilov(3919), dmj111(3850), davidjliu(3828), bix0r4ever(3818), Hazard(3798), felixcoto(3776), DocBot(3709), Arekku(3707), EBraun(3696), Rexxar(3696) 2010-10-06T15:04:56 *** Accoun has quit IRC () 2010-10-06T15:08:50 *** Meatkat has joined #aichallenge 2010-10-06T15:10:13 *** SANDIMUS has joined #aichallenge 2010-10-06T15:11:56 *** Meatkat has quit IRC (Client Quit) 2010-10-06T15:13:20 *** SANDIMUS has quit IRC (Client Quit) 2010-10-06T15:13:27 i haven't put a lot of effort into a runnable bot yet 2010-10-06T15:16:18 is it normal that the elo of the best bots keeps increasing? There don't seem to be too many new signups 2010-10-06T15:18:30 perhaps it is because of the correlation between ELO and resubmitting? 2010-10-06T15:18:45 no wait that should draw points *away* from the game 2010-10-06T15:18:57 Migi32: you might be right about the resubmitting 2010-10-06T15:19:11 Migi32: a resubmitted bot is consider a new bot for elo 2010-10-06T15:20:14 *** SANDIMUS has joined #aichallenge 2010-10-06T15:20:46 *** hellman has quit IRC (Remote host closed the connection) 2010-10-06T15:21:29 *** amstan has quit IRC (Quit: Konversation terminated!) 2010-10-06T15:21:39 *** amstan has joined #aichallenge 2010-10-06T15:21:42 @rankings 10 2010-10-06T15:21:43 SANDIMUS: Top 10 players: ruilov(3912), dmj111(3843), davidjliu(3820), bix0r4ever(3809), Hazard(3790), felixcoto(3769), DocBot(3701), Arekku(3699), EBraun(3683), Rexxar(3670) 2010-10-06T15:21:45 *** ChanServ sets mode: +o amstan 2010-10-06T15:22:02 *** Accoun has joined #aichallenge 2010-10-06T15:22:03 @msg 2010-10-06T15:22:04 SANDIMUS: No! 2010-10-06T15:22:14 @msg contestbot commands 2010-10-06T15:22:15 SANDIMUS: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2010-10-06T15:22:39 @msg contestbot more 2010-10-06T15:22:40 SANDIMUS: User error, it's not my fault. 2010-10-06T15:22:58 @/msg contestbot commands 2010-10-06T15:22:59 McLeopold, which might be a flaw. In the ELO algorithm no bots are supposed to enter/leave the system, which keeps the average the same, but if you remove/add points when resubmitting you break ELO's assumptions. 2010-10-06T15:22:59 SANDIMUS: User error, it's not my fault. 2010-10-06T15:23:31 @help 2010-10-06T15:23:31 Migi32: (help [] []) -- This command gives a useful description of what does. is only necessary if the command is in more than one plugin. 2010-10-06T15:23:38 *** Sylvain3665 has quit IRC (Quit: Page closed) 2010-10-06T15:23:55 @list 2010-10-06T15:23:55 Migi32: 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 2010-10-06T15:24:05 @Dunno 2010-10-06T15:24:06 Migi32: I have no idea what you mean. 2010-10-06T15:24:06 Migi32: well, this is bayeselo. Chess elo's had the same problem. 2010-10-06T15:24:10 *** Meatkat has joined #aichallenge 2010-10-06T15:24:31 Migi32: Generally the high mark is a function of the number of entrants. 2010-10-06T15:24:50 oh ok so it's supposed to keep rising 2010-10-06T15:25:10 Migi32: I wouldn't worry about it. We don't need to rank today's bots with bots in 50 years or anything. 2010-10-06T15:25:51 Migi32: it might shrink too. Watch the tcp server ratings. The top spot will rise and fall depending on the total number of bots remembered. 2010-10-06T15:29:21 *** info-del has joined #aichallenge 2010-10-06T15:29:34 Are there any plans to be able to submit multiple bots? 2010-10-06T15:30:29 ShaneHudson: why? 2010-10-06T15:30:32 I hope not 2010-10-06T15:30:54 you can submit as many bots as you like on your account, only one will run 2010-10-06T15:31:07 you can't have multiple accounts participating 2010-10-06T15:31:41 we will ban them if there's multiple active accounts 2010-10-06T15:32:05 if you want to test your bots, either do them offline, or use the tcp server 2010-10-06T15:33:42 i wish more people used the TCP server 2010-10-06T15:33:53 why? 2010-10-06T15:35:12 @amstan: Well I was just thinking it would be good to be able to submit ones that work different ways, but I suppose it is not important. Another question though, will the server etc still work after the challenge ends? 2010-10-06T15:35:13 ShaneHudson: User error -- Replace user. 2010-10-06T15:35:40 ShaneHudson: ideally just work on one 2010-10-06T15:35:42 Pardon? "ShaneHudson: User error -- Replace user" whats that mean lol 2010-10-06T15:35:50 Ok thats fine 2010-10-06T15:36:04 if we were to allow multiple entries, people could code bots that loose to some people intentionally 2010-10-06T15:36:07 to boost their rank 2010-10-06T15:36:12 @everytime you start your sentence with an @ contestbot gives you an error 2010-10-06T15:36:13 Migi32: I have no idea what you mean. 2010-10-06T15:36:25 yeah.. don't start your sentences with @ 2010-10-06T15:36:29 he almost passes the Turing test :P 2010-10-06T15:37:19 amstan, to make testing my bot more accurate 2010-10-06T15:37:36 cfaftw: you can't have both.. 2010-10-06T15:37:40 lol 2010-10-06T15:38:19 :( 2010-10-06T15:39:02 cfaftw: I don't see your name on tcp? 2010-10-06T15:39:16 JetBot.2 2010-10-06T15:39:45 the name has nothing to do with anything 2010-10-06T15:39:45 hmm... As soon as I make my code runnable, I'll be back on. :) 2010-10-06T15:39:45 haha 2010-10-06T15:39:48 good 2010-10-06T15:40:43 McLeopold, just so you know, you were my inspiration 2010-10-06T15:40:49 starting out i studied many of your games 2010-10-06T15:40:52 McLeopold, with wormbot? 2010-10-06T15:41:20 I feel so good now. 2010-10-06T15:41:38 Okay, WormBot and MoldBot are running. 2010-10-06T15:41:48 ew 2010-10-06T15:41:49 The real bot will be there this afternoon 2010-10-06T15:41:53 there's a MoldBot? I got to see :D 2010-10-06T15:41:59 Oh, it's bad 2010-10-06T15:42:08 First attempt at positional play 2010-10-06T15:42:16 WormBot, MoldBot, DirtBot, Poop Bot 2010-10-06T15:42:43 Well, mold because it's suppose to spread like spores... 2010-10-06T15:43:08 Version 2 will be frog 2010-10-06T15:43:47 My inspiration was deepblue, but now where is he? 2010-10-06T15:44:27 yeah he dropped off the map 2010-10-06T15:45:16 yeah McLeopold your MoldBot sucks bad 2010-10-06T15:45:23 what was the idea? 2010-10-06T15:45:37 *** justin_pdx has joined #aichallenge 2010-10-06T15:45:41 oh, spread like spores 2010-10-06T15:46:08 I don't wanna know how poopbot plays 2010-10-06T15:46:23 ask cfaftw, it's his idea 2010-10-06T15:46:40 :P 2010-10-06T15:47:05 http://72.44.46.68/canvas?game_id=342734 2010-10-06T15:47:16 Here is a win. You can see the idea I was going for. 2010-10-06T15:47:38 It's an attempt to value planets on more than just growth rate. 2010-10-06T15:47:58 you definitely have to value planets on more than growth rate in this contest 2010-10-06T15:48:25 Actually, not too much... 2010-10-06T15:48:30 Yet... 2010-10-06T15:48:52 Do many bots currently work by distance? 2010-10-06T15:49:04 The best ones do. 2010-10-06T15:49:12 distance and growth rate 2010-10-06T15:49:22 Obviously a close planet is faster than one across the map. 2010-10-06T15:49:47 Precisely, yes I kept seeing my fleets go across the maps.. so next time I am working on it I shall change that 2010-10-06T15:50:02 But for now I have a website due in 2010-10-06T15:50:15 another thing to consider, there's a higher risk when sending your ships further to hit a planet 2010-10-06T15:51:06 btw cfaftw, what does cfa stand for? 2010-10-06T15:51:12 That the planet would have more fleets there by the time you got there? And less time to renew 2010-10-06T15:51:22 cfaftw == chick-fil-a, for the win 2010-10-06T15:51:24 cfaftw: thus my relay system on my official bot 2010-10-06T15:51:25 lol 2010-10-06T15:51:30 now i want chick-fil-a 2010-10-06T15:51:49 i always want chick-fil-a 2010-10-06T15:51:56 One good thing about sending fleets a long way is that if they are your last fleets, it gives you a last chance 2010-10-06T15:52:07 shanehudson, not really 2010-10-06T15:52:10 ShaneHudson, it really doesn't 2010-10-06T15:52:18 so perhaps having "if fleets are low, send to furthest away that has low population" 2010-10-06T15:52:21 no? 2010-10-06T15:52:25 nah 2010-10-06T15:52:25 Ah ok, I am new to this :P 2010-10-06T15:52:34 enemy can easily reinforce that planet your ships are drifting toward 2010-10-06T15:52:35 opponent has *plenty* of time to attack the planet you are headed toward 2010-10-06T15:52:41 Every planet that fleet flies across can respond. 2010-10-06T15:53:02 my bot keeps losing against that CCCF bot 2010-10-06T15:53:04 so fail 2010-10-06T15:53:04 Can you make bots that sensor where the opponents fleets are? 2010-10-06T15:53:30 *in terms of ones that are moving 2010-10-06T15:53:32 shanehudson, you can do anything you want :) 2010-10-06T15:53:44 the game is fully observable 2010-10-06T15:53:46 so yes 2010-10-06T15:54:03 Haha that would involve changing the actual game files rather than just the bot... but certinally something to think about! 2010-10-06T15:54:15 ShaneHudson, use fleet.DestinationPlanet() and fleet.TurnsRemaining() 2010-10-06T15:54:26 *** TwoHappyBiscuits has joined #aichallenge 2010-10-06T15:54:32 who is CCCF? 2010-10-06T15:54:36 no clue 2010-10-06T15:54:37 for the opponent? oo :) 2010-10-06T15:54:52 TwoHappyBiscuits: OSC? 2010-10-06T15:55:08 his bot plays VERY similar to my bot, but seems to make the better starting move 2010-10-06T15:55:10 oh mercy 2010-10-06T15:55:14 nevermind :) 2010-10-06T15:55:20 brb 2010-10-06T15:55:48 cfaftw: he poached your planet 2010-10-06T15:56:00 turn 56 http://72.44.46.68/canvas?game_id=342853 2010-10-06T15:56:05 he does that every game 2010-10-06T15:56:12 jmcarthur: hey man :) 2010-10-06T15:56:20 you went for a planet closer to him. 2010-10-06T15:56:24 back 2010-10-06T15:56:30 *** rogue780 has joined #aichallenge 2010-10-06T15:56:36 *** EfTwelve has joined #aichallenge 2010-10-06T15:57:09 * rogue780 feels ashamed for not working on ErrBot for over 2 weeks. 2010-10-06T15:57:54 yep -_- 2010-10-06T15:57:56 I blame a1k0n 2010-10-06T15:57:57 I'ma make a VortexBot which only sends ships clockwise :P 2010-10-06T15:58:04 wut 2010-10-06T15:58:06 is there a known bug that causes you to lose on the first turn because you have no starting planet? 2010-10-06T15:58:07 oh wait there's already a guy named vortex 2010-10-06T15:58:33 cfaftw: you built faster, but allowed him to pass you up. 2010-10-06T15:58:33 should make for some really cool base-trades 2010-10-06T15:58:36 Migi32: vortexbot. that's awesome 2010-10-06T15:58:52 you can't just sit there. push your advantage or grow bigger 2010-10-06T15:59:01 mcleopold, yeah i'm aware that my bot plays rather conservatively 2010-10-06T15:59:25 EfTwelve: the java visualizer does that if one of the players timed out on the first turn, iirc 2010-10-06T15:59:25 i made an extremely advanced bot which basically spams everything randomly and dies 2010-10-06T15:59:32 If I'm your inspiration, I expect 2nd place. 2010-10-06T15:59:37 Top 10 players: ruilov(3914), dmj111(3855), davidjliu(3820), bix0r4ever(3808), Hazard(3790), felixcoto(3766), DocBot(3712), Arekku(3696), EBraun(3677), sequoh(3661) 2010-10-06T15:59:46 but it does it with a very thorough monte-carlo search 2010-10-06T15:59:47 haha 2010-10-06T15:59:51 a1k0n: the bot you've been working on secretly? 2010-10-06T15:59:56 yeah it fucking sucks 2010-10-06T15:59:58 aw 2010-10-06T16:00:01 monte-carlo? 2010-10-06T16:00:05 i hope mine doesn't do that :\ 2010-10-06T16:00:17 Is it worse than MoldBot? 2010-10-06T16:00:22 yeah that's why you need to start running it and spend less time on the backend 2010-10-06T16:00:24 i tink mine should at least do better than monte carlo 2010-10-06T16:00:30 *think 2010-10-06T16:00:40 strange, because there is no delay, it says 'turn1' and 'player 1 wins' immediatly, in the viewer on startup there is no player 2 planet. 2010-10-06T16:01:34 Hmm a question 2010-10-06T16:01:38 You say that we can do anything 2010-10-06T16:01:45 Is it possible to freeze a fleet mid flight 2010-10-06T16:01:48 or is that cheating 2010-10-06T16:01:49 lol 2010-10-06T16:01:58 you can see anything happening in the game, not *do* anything 2010-10-06T16:02:03 i guess i was thinking of a more limited "anything" 2010-10-06T16:02:06 Ah right 2010-10-06T16:02:09 :p 2010-10-06T16:02:25 you can't freeze\ 2010-10-06T16:02:29 I guessed it was not possible, but always worth asking 2010-10-06T16:03:50 a1k0n: so you are simulating many games fully to completion? i'm curious how many simulations per turn you are able to do, if so 2010-10-06T16:04:28 here i go against wormbot 2010-10-06T16:06:45 won at turn 200 2010-10-06T16:07:19 *** smellyhippy has joined #aichallenge 2010-10-06T16:08:31 *** delt0r_ has quit IRC (Read error: Connection reset by peer) 2010-10-06T16:08:39 ok definitely just spotted a bug in my bot 2010-10-06T16:08:41 thanks wormbot 2010-10-06T16:08:43 *** rogue780_ has joined #aichallenge 2010-10-06T16:08:47 Are there any in the top of the high scores that do not use really advanced algorithms? Or even none at all? I want to learn some but would be interesting how advanced it can go without them 2010-10-06T16:09:12 * cfaftw raises hand 2010-10-06T16:09:13 depends on what you mean by advanced algorithms 2010-10-06T16:09:18 *** delt0r_ has joined #aichallenge 2010-10-06T16:09:20 Uh, define an algorithm ShaneHudson. 2010-10-06T16:09:47 true 2010-10-06T16:09:48 ShaneHudson: nothing more advanced than basic dynamic programming... but yeah, what's basic? ;-) 2010-10-06T16:10:15 ahaha very true point, same as what's normal! 2010-10-06T16:11:06 *** rogue780 has quit IRC (Ping timeout: 240 seconds) 2010-10-06T16:12:10 And by algorithm, I meant the stuff taught in CS classes :P I will not be doing those until next year, but until then hopefully I can use this to have a strong base for starting! 2010-10-06T16:12:17 *** flezru has quit IRC (Ping timeout: 265 seconds) 2010-10-06T16:12:42 jmcarthur: no, i'm doing something really dumb that doesn't work. ... out to whatever the universe diameter is 2010-10-06T16:12:57 rebelxt: really, you have a DP recurrence? 2010-10-06T16:14:18 *** fawek has quit IRC (Ping timeout: 240 seconds) 2010-10-06T16:14:42 *** jmcarthur has quit IRC (Ping timeout: 265 seconds) 2010-10-06T16:16:43 a1k0n: What is a DP recurrence? 2010-10-06T16:20:06 *** Palmik has quit IRC (Read error: Connection reset by peer) 2010-10-06T16:20:50 *** McLeopold has quit IRC (Read error: Connection reset by peer) 2010-10-06T16:21:23 *** TwoHappyBiscuits has left #aichallenge 2010-10-06T16:21:27 *** area has joined #aichallenge 2010-10-06T16:28:33 *** amstan has quit IRC (Ping timeout: 265 seconds) 2010-10-06T16:29:20 *** jmcarthur has joined #aichallenge 2010-10-06T16:38:00 *** u1tr0n has quit IRC () 2010-10-06T16:40:46 *** ratatatta has joined #aichallenge 2010-10-06T16:41:19 got quiet in here 2010-10-06T16:41:39 *** Arthur_ has quit IRC (Ping timeout: 250 seconds) 2010-10-06T16:41:48 everyone's pondering what a1k0n said about monte carlo 2010-10-06T16:42:40 how do you guys debug your bot? 2010-10-06T16:43:37 I get an error and I'd like to do a back-trace but I can't run any debugger on it because that would conflict with PlayGame.jar 2010-10-06T16:43:52 i use the python logging module 2010-10-06T16:44:04 but i don't do any live debuggign 2010-10-06T16:44:30 I log stuff as well but I can't start logging *all* my calls to that function 2010-10-06T16:44:56 *** Arthur_ has joined #aichallenge 2010-10-06T16:45:49 *** boegel has joined #aichallenge 2010-10-06T16:45:50 *** Meatkat has quit IRC (Ping timeout: 265 seconds) 2010-10-06T16:50:15 *** Mathnerd314 has joined #aichallenge 2010-10-06T16:50:38 *** Haggis has joined #aichallenge 2010-10-06T16:50:55 *** cfaftw has quit IRC (Quit: Leaving) 2010-10-06T16:51:09 *** boegel has quit IRC (Quit: Leaving) 2010-10-06T16:51:41 *** hexsprite has quit IRC (Quit: hexsprite) 2010-10-06T16:53:29 *** hexsprite has joined #aichallenge 2010-10-06T16:53:31 *** SANDIMUS has quit IRC () 2010-10-06T16:54:24 *** ratatatta has quit IRC (Quit: Page closed) 2010-10-06T16:54:49 Migi32: i havent done it yet but using something like tcp for local debugging would let you run pdb with live games 2010-10-06T16:55:29 pdb? you mean gdb? 2010-10-06T16:56:27 *** yasith__ has quit IRC (Ping timeout: 264 seconds) 2010-10-06T16:59:38 Top 10 players: ruilov(3906), dmj111(3847), davidjliu(3813), bix0r4ever(3802), Hazard(3783), felixcoto(3760), DocBot(3705), Arekku(3690), EBraun(3675), _iouri_(3660) 2010-10-06T17:02:48 *** agweber has joined #aichallenge 2010-10-06T17:03:09 *** Meatkat has joined #aichallenge 2010-10-06T17:04:57 Anything new on the servers? I wonder how well sandboxed the entries are. 2010-10-06T17:05:27 [Java] So I'm trying to make a basic script where each planet attacks the nearest planet to it. But all my planets are attacking the same planet, and I can't figure out why http://pastebin.com/6MfLdQ0x 2010-10-06T17:06:58 *** yasith__ has joined #aichallenge 2010-10-06T17:08:23 agweber: you're calculating the distance between the enemy planets, not between your planet and the enemy planet 2010-10-06T17:08:40 so the first planet in the list is set to target, and then it has a distance of 0 from itself... 2010-10-06T17:09:46 Ah! Thanks, not sure how I overlooked that 2010-10-06T17:09:48 you probably want p.X() - q.X() etc. 2010-10-06T17:10:28 Yep, that fixed it, thanks 2010-10-06T17:15:27 "Failed to move file from temporary to permanent location." When trying to upload. Is this a serverside or clientside problem? 2010-10-06T17:16:34 this is a known problem with uploading 2010-10-06T17:16:48 you probably won't be able to do so for a while 2010-10-06T17:17:12 serverside 2010-10-06T17:17:53 That's fine, just wanted to make sure I didn't screw up =] 2010-10-06T17:18:09 So what's the scale on these distances? How wide/long is a map typically? 2010-10-06T17:18:38 They keep all the entries in folders in the main folder, even the old ones. When the number of folders hits the filesystem limit (32000) it doesn't allow anymore to be added. They've patched this by making a script that takes the old entries out, and periodically they move the old ones over and you can submit. 2010-10-06T17:19:50 agweber: There are some maps that are relatively tightly packed, maybe 30x30 and some are much bigger, maybe 100x100 2010-10-06T17:19:51 *** felixcoto has quit IRC (Quit: Page closed) 2010-10-06T17:20:45 the map generator doesn't put any planets outside [0,24]^2 2010-10-06T17:20:49 *** Sluf has joined #aichallenge 2010-10-06T17:20:52 Thanks Meatkat, just trying to get a grasp on how large I should make the growthrate factor in 2010-10-06T17:21:28 hiya 2010-10-06T17:21:44 the sub dir of the server is broken again :C 2010-10-06T17:21:45 it won't upload 2010-10-06T17:25:16 yup 2010-10-06T17:27:14 j3camero: you around? 2010-10-06T17:28:11 Migi32: i meant pdb cuz im using python.. but sure gdb perhaps 2010-10-06T17:28:56 Thanks for the help guys, I've gotta head out. But I'll be back, and you will all soon learn that I am to be loathed. Bwaha. 2010-10-06T17:29:06 *** agweber has quit IRC () 2010-10-06T17:29:48 *** mega1 has quit IRC (Ping timeout: 245 seconds) 2010-10-06T17:30:09 anyone know when the upload script will be up again? 2010-10-06T17:32:16 depends on how soon an admin becomes available to fix it 2010-10-06T17:32:37 so far as I know that has to be either amstan, danielvf or j3camero 2010-10-06T17:39:37 *** jimmerton has joined #aichallenge 2010-10-06T17:42:16 *** rogue780_ is now known as rogue780 2010-10-06T17:44:55 i bet the monte carlo approach could work but not the way i'm doing it. i tried to take a shortcut and got lost. 2010-10-06T17:45:25 the idea is so dumb i don't even want to talk about it 2010-10-06T17:46:07 :) it seems getting fast enough playouts yet with good enough quality will be hard to make MCTS work here 2010-10-06T17:48:49 *** RainCT has quit IRC (Remote host closed the connection) 2010-10-06T17:51:24 *** cfaftw has joined #aichallenge 2010-10-06T17:51:35 well simulating 200 turns can be done extremely cheaply 2010-10-06T17:51:49 but the move space is just too huge to explore fully 2010-10-06T17:52:02 lol 2010-10-06T17:52:07 yeah 2010-10-06T17:52:22 pruuuuuuuuune++ 2010-10-06T17:52:32 what are you talking about? (122 choose 22) isn't that many... :P 2010-10-06T17:52:48 *** fawek has joined #aichallenge 2010-10-06T17:52:53 (122 choose 22) what does that mean? 2010-10-06T17:53:03 right, hence needing quality playouts better than pure random so a reasonable part of the game space is explored 2010-10-06T17:53:07 so maybe beam search with some kind of smart heuristic. which might just be no better than greedily picking from your smart heuristic. 2010-10-06T17:53:25 a1k0n: i had exactly the same thoughtprocess :P 2010-10-06T17:54:41 122 choose 22 is the number of ways to choose 22 items from 122 of them 2010-10-06T17:54:49 permutations/combinations 2010-10-06T17:55:26 ah 122 over 22 we call that 2010-10-06T17:55:27 i'm curious about what rebelxt is using dynamic programming for, unless he's just doing dijkstra's on the map 2010-10-06T17:55:46 122 over 22 would be 5.5454545 where i come from :) 2010-10-06T17:56:23 yeah dutch can be weird sometimes 2010-10-06T17:58:19 someone else is already making a vortexbot, lol: http://72.44.46.68/getplayer?player=grahaminator 2010-10-06T17:59:29 haha nice 2010-10-06T17:59:39 Top 10 players: ruilov(3898), dmj111(3831), felixcoto(3791), bix0r4ever(3789), davidjliu(3787), Hazard(3770), DocBot(3692), Arekku(3684), EBraun(3653), goldman(3649) 2010-10-06T17:59:53 he's playing go instead of galcon and trying to surrround the opponent 2010-10-06T18:00:22 FYI, j3camero is on his way home from work and will fix the upload problem when he gets there 2010-10-06T18:00:39 *** sigh has joined #aichallenge 2010-10-06T18:01:06 *** smellyhippy has quit IRC (Ping timeout: 240 seconds) 2010-10-06T18:01:08 wow, vortex bot is not a good idea 2010-10-06T18:01:16 lol 2010-10-06T18:03:10 *** Meatkat has quit IRC (Ping timeout: 265 seconds) 2010-10-06T18:03:36 Galcon is played in the center 2010-10-06T18:03:53 The optimal bot favours the center slightly more than growth and distance, imo 2010-10-06T18:04:19 my vortexbot is a little different though: http://visualizer.naktibalda.lt/102 2010-10-06T18:04:34 vortex bot? 2010-10-06T18:04:42 just watch :P 2010-10-06T18:05:15 *** Blkt has quit IRC (Quit: Error: do not makunbound t please!) 2010-10-06T18:06:02 just a simple rule: be super aggressive and only send attacks with relative angle towards the middle of > 30� 2010-10-06T18:06:45 *** smellyhippy has joined #aichallenge 2010-10-06T18:06:45 *** smellyhippy has joined #aichallenge 2010-10-06T18:06:45 *** tapwater has joined #aichallenge 2010-10-06T18:06:51 send your fleets to the opposite planets your opponent sends them 2010-10-06T18:06:53 i see, like McLeopold's worm bot, but worse planet selection :P 2010-10-06T18:06:57 from the opposite planets 2010-10-06T18:07:16 Zannick: that's just copy bot? 2010-10-06T18:07:20 yes :P 2010-10-06T18:07:38 Migi32, let's see that against a real bot :) 2010-10-06T18:07:45 it fails 2010-10-06T18:07:47 badly 2010-10-06T18:08:06 it's fun to watch at least 2010-10-06T18:08:27 hm, i never though about measuring angles 2010-10-06T18:08:28 *** ShaneHudson has quit IRC (Quit: Lost terminal) 2010-10-06T18:08:40 (i don't think it's terribly necessary 2010-10-06T18:08:42 ) 2010-10-06T18:08:56 migi32 how does it play against itself? 2010-10-06T18:09:02 hahaha 2010-10-06T18:09:05 oh good idea Sluf 2010-10-06T18:09:08 they chase each other's tales 2010-10-06T18:09:19 I'd still like to see it :) 2010-10-06T18:11:25 it appears I still have a bug which makes them timeout when fighting eachother 2010-10-06T18:11:45 (this is very messy code written in 20 minutes, 15 of which were the angle calculation) 2010-10-06T18:12:38 but here's one where they made in 112 turns 2010-10-06T18:12:39 http://visualizer.naktibalda.lt/103 2010-10-06T18:13:12 what is the 30 degree angle relative to? 2010-10-06T18:13:17 the strongest enemy planet? 2010-10-06T18:13:17 the center 2010-10-06T18:13:21 oh, ofcause 2010-10-06T18:13:21 of the map 2010-10-06T18:13:22 nice! 2010-10-06T18:13:26 infinite chase 2010-10-06T18:13:51 now let me see if I can find that bug and let it play out longer 2010-10-06T18:13:58 *** nrub_ has joined #aichallenge 2010-10-06T18:14:01 let it play out for 1000 turns 2010-10-06T18:14:06 I 2010-10-06T18:14:25 I'm sure this has already been mentioned here but there seems to be a bug in the uploader 2010-10-06T18:14:48 yes 2010-10-06T18:15:01 is there an ETA for a fix? 2010-10-06T18:15:05 the bug is that there is no more space 2010-10-06T18:15:28 the bug is that the system could easily be fixed 2010-10-06T18:15:38 is that the bug or is it that the space is not freed after it's no longer needed? :P 2010-10-06T18:15:58 i think it saves all the uploads, nothing is overwritten 2010-10-06T18:21:22 i am disappoint 2010-10-06T18:23:35 *** Bobng has quit IRC (Quit: Leaving) 2010-10-06T18:26:06 *** fawek has quit IRC (Ping timeout: 250 seconds) 2010-10-06T18:26:31 *** Haggis has quit IRC (Quit: Page closed) 2010-10-06T18:26:43 *** irchs is now known as janstaunton 2010-10-06T18:31:54 *** Appleman1234 has joined #aichallenge 2010-10-06T18:33:37 *** davidjliu has joined #aichallenge 2010-10-06T18:34:29 *** Bobng has joined #aichallenge 2010-10-06T18:35:03 *** cfaftw has quit IRC (Remote host closed the connection) 2010-10-06T18:36:36 i can't believe the upload issue still exists 2010-10-06T18:37:21 well, it wasn't exactly solved 2010-10-06T18:37:45 *** felixcoto has joined #aichallenge 2010-10-06T18:38:10 yeah 2010-10-06T18:38:37 it's been like this for about 8 hours 2010-10-06T18:38:42 didn't they delete a bunch of stuff yesterday? 2010-10-06T18:38:47 1200 files 2010-10-06T18:39:06 *** Bobng has quit IRC (Client Quit) 2010-10-06T18:39:27 there are probably a lot of people uploading fairly often 2010-10-06T18:39:35 yeah, 1200 files didn't last long :D 2010-10-06T18:39:42 1200 submissions is not a lot when you consider 32000 was the limit that was hit 2010-10-06T18:40:17 wait, if 1200 fill up in less than a day, then 32000 fills up in under a month 2010-10-06T18:40:31 yeah, the server needs redesign to not store them all 2010-10-06T18:40:57 as it is, this is a DOS 2010-10-06T18:41:37 It's quite simple 2010-10-06T18:41:42 each user can has 1 entry at a time 2010-10-06T18:41:47 when they upload, the older file is deleted 2010-10-06T18:42:13 that way, the server can support 32000 users uploading simultaneously (in theory) 2010-10-06T18:42:13 haha... what's even funnier about these limits: 2010-10-06T18:42:14 #aichallenge.2010-10-04.log:2010-10-04T22:15:43 We need somebody to write a genetic algorithm that auto-submits bots once every hour to test different parameters. That would be ballin. 2010-10-06T18:42:15 sigh: Run as fast as you can and don't look back. 2010-10-06T18:42:50 add # to the list of command signals :P 2010-10-06T18:42:58 amstan really should disable that 2010-10-06T18:43:58 # 2010-10-06T18:47:16 *** Meatkat has joined #aichallenge 2010-10-06T18:48:53 *** felixcoto has quit IRC (Quit: Page closed) 2010-10-06T18:56:24 *** Naktibalda has joined #aichallenge 2010-10-06T18:59:40 Top 10 players: ruilov(3887), dmj111(3823), felixcoto(3786), bix0r4ever(3786), davidjliu(3778), Hazard(3763), savchenko(3695), DocBot(3684), Arekku(3677), EBraun(3646) 2010-10-06T19:04:04 *** info-del has quit IRC (Quit: Page closed) 2010-10-06T19:04:57 *** Sluf has quit IRC (Quit: Page closed) 2010-10-06T19:08:55 *** Meatkat has quit IRC (Ping timeout: 265 seconds) 2010-10-06T19:10:07 *** Meatkat has joined #aichallenge 2010-10-06T19:11:30 *** jarfhy has joined #aichallenge 2010-10-06T19:14:39 *** Migi32 has quit IRC (Quit: http://boinc.berkeley.edu) 2010-10-06T19:17:07 *** Meatkat has quit IRC (Ping timeout: 265 seconds) 2010-10-06T19:42:46 *** iris1 has joined #aichallenge 2010-10-06T19:45:42 *** amstan has joined #aichallenge 2010-10-06T19:45:42 *** ChanServ sets mode: +o amstan 2010-10-06T19:48:25 *** amstan has quit IRC (Remote host closed the connection) 2010-10-06T19:55:06 *** ichti has quit IRC (Ping timeout: 240 seconds) 2010-10-06T19:58:01 *** mceier has quit IRC (Quit: leaving) 2010-10-06T19:58:11 *** Zeiris_ has joined #aichallenge 2010-10-06T19:59:38 *** Titankiller has joined #aichallenge 2010-10-06T19:59:40 Top 10 players: ruilov(3875), dmj111(3815), felixcoto(3772), davidjliu(3759), Hazard(3741), bix0r4ever(3732), savchenko(3686), DocBot(3670), Arekku(3664), EBraun(3637) 2010-10-06T20:14:16 *** McLeopold has joined #aichallenge 2010-10-06T20:15:30 *** jaspervdj has quit IRC (Quit: NEVER GONNA GIVE YOU UP NEVER GONNA LET YOU DOWN) 2010-10-06T20:16:41 *** yellowseed has joined #aichallenge 2010-10-06T20:18:23 hmm 2010-10-06T20:19:26 6 games per minute, 3000 bots /sigh 2010-10-06T20:19:50 hmm? 2010-10-06T20:19:51 oh 2010-10-06T20:20:08 lol did not mean to divide by you, sigh ;-) 2010-10-06T20:20:16 :) 2010-10-06T20:20:58 I wish there was 3000 bots/sigh... I could do with some more bots :P 2010-10-06T20:21:16 3000 bots per sigh 2010-10-06T20:21:20 mcleopolds bots are interesting 2010-10-06T20:21:24 there is 1 sigh, so there are 3000 bots 2010-10-06T20:21:34 though i can't figure out his latest scotbot 2010-10-06T20:23:57 is mcleopold just having fun with some unusual strategies? 2010-10-06T20:24:33 *** janstaunton has quit IRC (Quit: janstaunton) 2010-10-06T20:25:18 yea 2010-10-06T20:25:33 he explained that he was writing bots that are cool to watch and can explore weaknesses in his other bots 2010-10-06T20:26:45 *** jimmerton_ has joined #aichallenge 2010-10-06T20:26:46 *** jimmerton has quit IRC (Read error: Connection reset by peer) 2010-10-06T20:26:47 *** jimmerton_ is now known as jimmerton 2010-10-06T20:27:16 hehe yeah some fun to watch games 2010-10-06T20:28:07 *** EfTwelve_ has joined #aichallenge 2010-10-06T20:28:08 *** jimmerton has quit IRC (Read error: Connection reset by peer) 2010-10-06T20:28:14 *** jimmerton has joined #aichallenge 2010-10-06T20:28:47 *** Mathnerd314 has quit IRC (Ping timeout: 252 seconds) 2010-10-06T20:29:17 *** EfTwelve has quit IRC (Ping timeout: 255 seconds) 2010-10-06T20:29:19 *** EfTwelve_ is now known as EfTwelve 2010-10-06T20:38:19 *** jimmerton_ has joined #aichallenge 2010-10-06T20:40:19 *** jimmerton has quit IRC (Ping timeout: 240 seconds) 2010-10-06T20:40:19 *** jimmerton_ is now known as jimmerton 2010-10-06T20:44:28 *** keynes has joined #aichallenge 2010-10-06T20:49:02 I Want to write a bot that spells things out. 2010-10-06T20:49:19 Like, at turn 54 "hi" shows up or something. 2010-10-06T20:49:52 you need special map for it :) 2010-10-06T20:51:03 Nah you wouldn't. I bet you could approximate most of it You could do it with one planet. It may be a little wonky looking. The more starting positions you had the higher your resolution. 2010-10-06T20:51:26 A few outside planets could do a pretty good job. 2010-10-06T20:51:45 *** jimmerton_ has joined #aichallenge 2010-10-06T20:51:45 *** jimmerton has quit IRC (Read error: Connection reset by peer) 2010-10-06T20:51:46 *** jimmerton_ is now known as jimmerton 2010-10-06T20:55:23 *** Titankiller2 has joined #aichallenge 2010-10-06T20:55:55 *** amstan has joined #aichallenge 2010-10-06T20:55:55 *** ChanServ sets mode: +o amstan 2010-10-06T20:57:47 *** Titankiller has quit IRC (Ping timeout: 265 seconds) 2010-10-06T20:59:34 *** jimmerton_ has joined #aichallenge 2010-10-06T20:59:34 *** jimmerton has quit IRC (Read error: Connection reset by peer) 2010-10-06T20:59:36 *** jimmerton_ is now known as jimmerton 2010-10-06T20:59:41 Top 10 players: ruilov(3860), dmj111(3809), felixcoto(3751), davidjliu(3744), savchenko(3741), Hazard(3723), bix0r4ever(3717), Arekku(3642), DocBot(3641), EBraun(3618) 2010-10-06T21:01:29 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-10-06T21:13:58 *** Accoun_ has joined #aichallenge 2010-10-06T21:15:04 *** Accoun has quit IRC (Ping timeout: 264 seconds) 2010-10-06T21:19:15 *** bhasker has quit IRC (Quit: bhasker) 2010-10-06T21:24:34 *** jimmerton_ has joined #aichallenge 2010-10-06T21:24:35 *** jimmerton has quit IRC (Read error: Connection reset by peer) 2010-10-06T21:24:36 *** jimmerton_ is now known as jimmerton 2010-10-06T21:25:08 *** pjreddie has joined #aichallenge 2010-10-06T21:34:26 *** keynes has quit IRC (Ping timeout: 272 seconds) 2010-10-06T21:34:56 *** Titankiller2 has quit IRC (Quit: Nettalk6 - www.ntalk.de) 2010-10-06T21:37:47 *** bhasker has joined #aichallenge 2010-10-06T21:41:27 *** Accoun_ has quit IRC (Ping timeout: 240 seconds) 2010-10-06T21:47:41 *** Accoun_ has joined #aichallenge 2010-10-06T21:52:53 *** keynes has joined #aichallenge 2010-10-06T21:59:04 server ok? 2010-10-06T21:59:41 Top 10 players: ruilov(3842), dmj111(3790), davidjliu(3735), felixcoto(3730), Hazard(3705), bix0r4ever(3698), savchenko(3693), Arekku(3633), DocBot(3621), jambachili(3590) 2010-10-06T22:02:11 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-10-06T22:04:29 *** espes has quit IRC (Ping timeout: 276 seconds) 2010-10-06T22:09:25 *** amstan_ has joined #aichallenge 2010-10-06T22:09:25 *** amstan has quit IRC (Ping timeout: 252 seconds) 2010-10-06T22:09:25 *** ChanServ sets mode: +o amstan_ 2010-10-06T22:26:25 *** jimmerton has quit IRC (Quit: jimmerton) 2010-10-06T22:34:42 *** JCS^ has joined #aichallenge 2010-10-06T22:39:00 *** Frontier has left #aichallenge 2010-10-06T22:42:33 *** Frontier has joined #aichallenge 2010-10-06T22:43:23 *** rogue780 has quit IRC (Quit: Leaving) 2010-10-06T22:44:21 Hey Lua could make for a good addition to a C++ bot during the tunig phase. One could tweak scoring funtions on the fly and repeat a certain turn without recompiling or even stopping the debugger. But it would be a lot of work to get something useful. 2010-10-06T22:59:42 Top 10 players: ruilov(3828), dmj111(3766), felixcoto(3739), savchenko(3703), davidjliu(3701), Hazard(3683), bix0r4ever(3675), Arekku(3612), DocBot(3608), adumlah(3574) 2010-10-06T23:06:03 *** nann has joined #aichallenge 2010-10-06T23:11:03 *** Apophis has quit IRC (Read error: Connection reset by peer) 2010-10-06T23:11:05 *** Accoun_ has quit IRC (Ping timeout: 272 seconds) 2010-10-06T23:11:12 *** millertime has quit IRC (Ping timeout: 272 seconds) 2010-10-06T23:11:13 *** CIA-78 has quit IRC (Ping timeout: 272 seconds) 2010-10-06T23:12:15 *** CIA-78 has joined #aichallenge 2010-10-06T23:12:41 *** Accoun_ has joined #aichallenge 2010-10-06T23:13:13 *** millertime has joined #aichallenge 2010-10-06T23:15:17 Frontier: how do you change lua on the fly? 2010-10-06T23:18:04 *** jarfhy has quit IRC (Ping timeout: 264 seconds) 2010-10-06T23:20:27 *** MarcAndrePC has joined #aichallenge 2010-10-06T23:21:08 *** FlagCapper has joined #aichallenge 2010-10-06T23:23:58 *** McLeopold has joined #aichallenge 2010-10-06T23:26:59 *** davidjliu has quit IRC (Quit: Page closed) 2010-10-06T23:30:35 You could simply edit a .lua file and have your bot check the modification date of the file. I have only used lua once for something very simple (parsing a formula from stdin), but it should be easy to exchange the code. 2010-10-06T23:31:53 meh 2010-10-06T23:32:05 i don't see how that's simpler than having any other code compile 2010-10-06T23:32:08 You could do this in Java as well if you are crazy enough. As long as the JDK is installed you can compile strings that resemble valid Java code to new classes (in memory!) and load them into the VM. Any script language can do this I guess. 2010-10-06T23:32:28 even C can do this, you read a file for a variable.. then remember it 2010-10-06T23:32:59 new bot nearly done... O_O 2010-10-06T23:33:07 Sure, but lua can read formulas not only values 2010-10-06T23:33:25 so can C, actually :P 2010-10-06T23:33:36 but you have to write an interpreter 2010-10-06T23:33:55 or use TCC or something :D 2010-10-06T23:34:21 jmcarthur: Sure, thats much simpler and faster ^^. 2010-10-06T23:34:59 I think most of us developers have once written a simple math parser for the standard operators + - * / 2010-10-06T23:35:36 a scheme interpreter isn't much harder 2010-10-06T23:35:53 well, a subset of scheme, of course 2010-10-06T23:36:08 *** krokokrusa has joined #aichallenge 2010-10-06T23:36:13 but you can get all the way to a fairly convenient, turing complete language 2010-10-06T23:36:42 I just want to share my idea of an ideal world. Keep the debugger running, change a script, replay the turn. See how the change turns out. 2010-10-06T23:36:59 in my ideal world my bot writes itself 2010-10-06T23:37:26 me too 2010-10-06T23:37:28 You are one of those genetic programming guys, huh? 2010-10-06T23:37:35 I shout the idea and it write the code 2010-10-06T23:37:45 preferably it would also proves some theroems like P=NP along the way 2010-10-06T23:37:53 sigh : better version now http://72.44.46.68/getplayer?player=bhaskerv20.10 2010-10-06T23:37:55 that would be impossible, it reduces to the halting problem 2010-10-06T23:38:02 I'm writing my bot with a backdoor. If you send 25 fleets to my home on turns 1 & 2, I shut down and do nothing. 2010-10-06T23:38:02 solved some basic problems with my previous iterations 2010-10-06T23:38:07 in my ideal world my bot improved itself to the point where it was bored with the contest 2010-10-06T23:38:16 so it has to write a second bot that competes in the competition 2010-10-06T23:38:17 still need to make it a little more conservative when acquiring neutrals 2010-10-06T23:38:57 *** jarfhy has joined #aichallenge 2010-10-06T23:39:10 bhasker: have you uploaded to the main site yet? 2010-10-06T23:39:18 not the new version 2010-10-06T23:39:22 older version is at rank 77 2010-10-06T23:40:04 http://ai-contest.com/profile.php?user_id=5822 2010-10-06T23:40:12 bhasker: it's mainly that you go after neutrals to close to the enemy 2010-10-06T23:40:20 yea need to fix that 2010-10-06T23:40:27 fixed my first move too 2010-10-06T23:40:33 :) 2010-10-06T23:40:34 made it into a 0-1 knapsack based solution 2010-10-06T23:40:41 so no more picking bad planets at start 2010-10-06T23:40:46 at least in most cases 2010-10-06T23:40:46 yeah, I should do that too 2010-10-06T23:40:56 is your bot in python? 2010-10-06T23:41:03 nope, c++ 2010-10-06T23:41:11 bhasker: how do you explain this? 2010-10-06T23:41:11 hmm here let me give you some java code 2010-10-06T23:41:12 http://72.44.46.68/canvas?game_id=349075 2010-10-06T23:41:16 http://jhave.org/learner/misc/knapsack/knapsack.shtml 2010-10-06T23:41:41 my bot is stupid in someways, i have some if then's that screw up 2010-10-06T23:41:43 Surley the 60 in the center was better than the 79 at the back? 2010-10-06T23:41:47 bhasker: yeah, the algorithm itself is easy... just haven't been bothered 2010-10-06T23:42:10 my profit function probably 2010-10-06T23:42:21 i calculate profit as # of ships it can generate 20 turns from now 2010-10-06T23:42:30 or so 2010-10-06T23:42:37 the 79 was probably a bit closer 2010-10-06T23:42:49 i have to adjust the cost function 2010-10-06T23:42:49 a bit 2010-10-06T23:42:57 rather the profit function 2010-10-06T23:43:03 but the basic algo is in place 2010-10-06T23:43:22 earlier it would do much worse and pick weird planets 2010-10-06T23:43:25 McLeopold: can you reveal the logic behind scotbot? 2010-10-06T23:43:26 when it was purely heuristic based 2010-10-06T23:44:29 let me see why it picked those planets mcleopold 2010-10-06T23:45:32 McLeopold: when will you send a new bot to the main server? 2010-10-06T23:45:44 sigh: you mean the reason I send invalid order in half my games? 2010-10-06T23:45:50 hah there is a bug in my code !:( 2010-10-06T23:45:52 need to fix that 2010-10-06T23:46:15 McLeopold: no, just the general idea behind how scotbot works, if you willing to say 2010-10-06T23:46:45 *if you are 2010-10-06T23:46:49 *** Arthur_ has quit IRC (Ping timeout: 252 seconds) 2010-10-06T23:46:54 thanks for pointing it out McLeopold ...i shal endeavour to fix it! 2010-10-06T23:47:00 grow fast. I've got a homebrew knapsack solution 2010-10-06T23:47:29 ah, yup... makes sense, thanks 2010-10-06T23:48:03 it was slightly less obvious than worm or snake :P 2010-10-06T23:48:03 the official version is to dumb to know it's behind on growth 2010-10-06T23:48:21 McLeopold: same 2010-10-06T23:48:23 worm is doing quite well, I'm surprised 2010-10-06T23:48:33 *** foucist has joined #aichallenge 2010-10-06T23:48:55 *** Arthur_ has joined #aichallenge 2010-10-06T23:49:06 There something to be said for throwing caution to the wind and going for the throat 2010-10-06T23:49:21 McLeopold: why? 2010-10-06T23:49:22 that's should be obvious to everyone from looking at rage :) 2010-10-06T23:50:10 well, sometimes the maps are such that you should just go for the opponent 2010-10-06T23:50:19 worm does well on those 2010-10-06T23:50:32 i am also planning to do some convex hull detection 2010-10-06T23:50:36 to identify edge planets 2010-10-06T23:50:43 does it ever beat your main bot, McLeopold? 2010-10-06T23:50:45 wondering if it will be useful 2010-10-06T23:50:52 worm also doesn't get it's planets poached 2010-10-06T23:51:05 cause it's always sending more if it needs it or not 2010-10-06T23:51:24 sigh, it won't when i fix my bugs 2010-10-06T23:51:56 right now it does on some close start maps or linear maps 2010-10-06T23:52:08 i see 2010-10-06T23:52:59 bhasker: that sound complicated 2010-10-06T23:53:44 well the idea is if i can identify a convex hull i would detect planets being taken over by enemy in middle of my base 2010-10-06T23:53:58 case in point this game http://72.44.46.68/canvas?game_id=349165 2010-10-06T23:54:11 i reacted too slowly to an enemy planet in middle of my base 2010-10-06T23:54:14 and lost the game 2010-10-06T23:54:50 so, your just looking for enemy planets within you borders? 2010-10-06T23:54:55 *** Arthur_ has quit IRC (Ping timeout: 240 seconds) 2010-10-06T23:55:07 pretty much 2010-10-06T23:55:53 at turn 26 you launch 55 fleets away from the battle 2010-10-06T23:55:59 after his launch started 2010-10-06T23:56:18 at didn't even take the neutral 2010-10-06T23:56:51 well i do synced launches so the 55 was probably to be followed 2010-10-06T23:56:55 but since he attacked 2010-10-06T23:56:59 it aborted the remaining attack 2010-10-06T23:57:08 at turn 42, you can see a 42 and 10 fleet taking a neutral on the back line 2010-10-06T23:57:09 or probably didnt have enough ships to follow through 2010-10-06T23:57:41 I've had so much trouble with my synced launches as well 2010-10-06T23:57:51 *** Arthur_ has joined #aichallenge 2010-10-06T23:57:54 yea those are all synced launches which had to be aborted 2010-10-06T23:58:00 because he probably attacked and i lost the ships to complete it 2010-10-06T23:58:20 the 2 problems i am having are a) when not to take aneutral 2010-10-06T23:58:27 my bot is just too aggresive and will do stupid things 2010-10-06T23:58:41 and 2) first turn planet selection 2010-10-06T23:58:55 problem 2) is more or less solved with a couple of bugs that you pointed out with the knapsack algo 2010-10-06T23:59:11 yea, that game you started out strong 2010-10-06T23:59:12 1) is much harder to solve, still trying to devise a good heuristic to decide when not to take a neutral 2010-10-06T23:59:36 also trying to see if i should just tell my bot not to acquire neutrals too rapidly once it gets to midgame 2010-10-06T23:59:42 Top 10 players: ruilov(3825), dmj111(3762), felixcoto(3733), savchenko(3721), davidjliu(3690), Hazard(3682), bix0r4ever(3670), Arekku(3629), DocBot(3592), adumlah(3572)