2010-12-01T00:25:18 the games per minute has been droping like a rock lately lol 2010-12-01T00:27:00 but the games we play rising :) 2010-12-01T00:28:28 oh, player cutoff is nearly at 1000 2010-12-01T00:30:01 roughly 10games per hour now? 2010-12-01T00:32:17 should be close to that, yeah 2010-12-01T00:32:52 asavis has climbed quite a bit since yesterday 2010-12-01T00:33:08 i'm not helping^^ 2010-12-01T00:33:23 i think i have one of the most expensive bots 2010-12-01T00:33:27 if not the 2010-12-01T00:33:40 lets see.. to make the top10 ranks be fairly stable, with their current elo ratings as an indicator.. they need something like 20k games each? 2010-12-01T00:33:49 *** Rubicon-|-Cross has quit IRC (Read error: Connection reset by peer) 2010-12-01T00:34:00 which would be around a million games for the top 100 2010-12-01T00:34:04 i think the rankings have gone quite well 2010-12-01T00:35:09 which is roughly 10days of just the top100 playing? 2010-12-01T00:36:55 *** Rubicon-|-Cross has joined #aichallenge 2010-12-01T00:37:29 so the top10 bots, are going to be off by an average of probably around 5elo, and so most likely, 2 or 3 of them will be out of order 2010-12-01T00:37:35 since we only have a day left 2010-12-01T00:38:16 since the nature of the game, is subject to randomness.. this shouldn't matter though 2010-12-01T00:38:47 nope, it's close enough for me 2010-12-01T00:39:09 and the most important thing, the first place.. seems fairly obvious 2010-12-01T00:39:18 *** sanddbox has quit IRC () 2010-12-01T00:39:18 given the nature of the game and the number of bots, it's sort of got to be expected a little bit 2010-12-01T00:39:27 yep, imo 2 is obvious too 2010-12-01T00:39:45 Isn't a million too much? player combination x maps? Would it be 100 * 50 * 150 games in total for round robin across 150 maps? 2010-12-01T00:40:37 i beat slin and accoun in my last two games 2010-12-01T00:40:45 that playing each map once, isn't sufficient 2010-12-01T00:41:05 wait -- why would you want same pairings of bots to replay the same map? 2010-12-01T00:41:18 antimatroid: good for you! and for me, seeing that I have misguided goals of still reaching the top 10 :) 2010-12-01T00:41:20 well, ideally not the same map 2010-12-01T00:41:35 *** Rubicon-|-Cross has quit IRC (Client Quit) 2010-12-01T00:41:39 but a number of bots will not do the exact same thing in every game played 2010-12-01T00:41:42 *** Rubicon-|-Cross has joined #aichallenge 2010-12-01T00:42:13 davidjliu: i don't think you're knocking either of those two out of the top ten :P 2010-12-01T00:42:25 clearly you are :) 2010-12-01T00:42:36 i'm actually about on par with accouns bot 2010-12-01T00:42:43 that's why it's a round robin across all maps ;) -- if we take top 50, would it be 50 x 25 x 150? (50 x 25 for pairing people up, 150 for maps) 2010-12-01T00:42:56 1000000 games, for the top100, would be statistically approxiamtly the number of games required to make their scores converge(relative to each other), given an infinte supply of maps 2010-12-01T00:42:58 when it's just the two of us that is 2010-12-01T00:44:18 *** kincody has joined #aichallenge 2010-12-01T00:44:29 virdo.. how many spoons would you have to throw out the window.. to determine which side lands up more often(concave, or convex) 2010-12-01T00:45:16 *** smloh1 has joined #aichallenge 2010-12-01T00:45:40 oh, how many times do you need to slip a penny.. to determine if one side is heavier than the other? 2010-12-01T00:45:40 burny: I think it's fairly safe to say that if two bots play on the same map twice, they will almost always get the same result 2010-12-01T00:46:02 david, for most bots, yes, most likely 2010-12-01T00:46:09 burny: I'm not sure what you're getting at :) -- All I was saying is: if there are 100 maps... and you take top 50 bots... to play all bots against each other on every map... I'm not sure what issue there is here :) 2010-12-01T00:46:31 i think the point is we'd use more maps 2010-12-01T00:46:39 virdo, what I'm saying is.. that's not sufficient to make the scores converge 2010-12-01T00:46:48 yeah, if we used more maps, then that'd be an issue, but with 100, it's a limited population 2010-12-01T00:46:53 er, 150 2010-12-01T00:47:10 I'm not talking about scores / elos right now... I'm more thinking about pure wins / losses / draws calculation ;) 2010-12-01T00:48:17 Wouldn't it be enough to pit every top bot against every other top bot across a slew of maps and sort based on wins descending? 2010-12-01T00:48:38 not really.. 2010-12-01T00:48:38 Top 10 players: bocsimacko(3816), _iouri_(3620), Slin-.-(3576), jimrogerz(3568), _Astek_(3562), asavis(3546), GreenTea(3546), Accoun(3543), protocolocon(3538), george(3531) 2010-12-01T00:49:00 but the whole nature of how we're testing the bots, isn't truely indicitive.. but it's close enough that no one really cares 2010-12-01T00:49:10 in particular, the top bots, are not playing against many weak bots 2010-12-01T00:49:50 so if a bot wins against 99% of rank 1-100 bots, but only 80% of rank 101-4500 bots.. it'll win over one that beats 98% of everyone 2010-12-01T00:50:10 but why would top bots need to play weak bots? they already proved themselves, no? 2010-12-01T00:50:30 yes.. they proved themselves better then the weak bot.. 2010-12-01T00:50:43 but the didn't prove themselves better than other top bots, at beating weak bots 2010-12-01T00:51:17 if you beat EVERYone 60% of the time.. are you going to win? 2010-12-01T00:51:31 yeah, well, if we lived in a world with no resource limitations, I would agree with you... Reason round robin across all bots / 99999 maps isn't being played out is because of this limitation... 2010-12-01T00:51:34 clearly not.. but if you beat everyone 60% of the time.. who is better than you? 2010-12-01T00:53:59 *** smloh1 has quit IRC (Quit: Page closed) 2010-12-01T00:55:20 if some bot beats bosco 100% of the time.. but only beats the other rank 1-100 bots 49% of the time.. it's going to not be in the top100 2010-12-01T00:57:36 I'm not actually sure what your argument is -- are you saying that round robin across all players and all maps is unfair? or are you saying that including only top players in this round robin is unfair? 2010-12-01T00:57:55 no, I'm saying it's not statistically sufficient 2010-12-01T00:58:10 oh.. 2010-12-01T00:58:19 I guess, only top players, is slightly unfair 2010-12-01T00:58:41 but it probably doesn't matter much 2010-12-01T00:58:55 that it's not worth the processing time, to determine it's level of relevance 2010-12-01T00:59:40 Why wouldn't it be statistically sufficient? 2010-12-01T01:00:19 We're not talking about all these bot pairings playing on just one map... I agree, that would be biased... but 150 maps is still not enough? 2010-12-01T01:00:28 the same reason why it's hard to determine how many times you need to flip a coin, to determine which side is heavier 2010-12-01T01:01:38 What's your math here for it not being "statistically sufficient"? 2010-12-01T01:01:54 it's based on elo 2010-12-01T01:02:23 given 2 events, with 1 being 1% more likely than the other.. 2010-12-01T01:02:41 er.. 2 outcomes to an event 2010-12-01T01:03:25 if you run the event 100 times.. it's only 50% likely to end up with 1 of them, having more wins than the other 2010-12-01T01:03:59 Maybe a different analogy here will do better, rather than flipping coins :) -- Let's say it's people racing on a race track... Everyone races against everyone... How in the world isn't the person who wins most races the best? :) 2010-12-01T01:04:29 They might have had a bad run on some race track against another runner, but generally they're still the best... 2010-12-01T01:04:32 because of luck? 2010-12-01T01:04:48 ok, that's why there are 150 race tracks :) 2010-12-01T01:04:54 generall.. more likely to be the best, given the data available 2010-12-01T01:05:22 well, generaly in a race.. people are significantly stronger/weaker than the others 2010-12-01T01:06:08 but luck factor is also there -- liek not feeling well, spraining an ankle, whatever... 2010-12-01T01:06:51 if you look at the ELO's.. it right now suugests that if you take the 3rd place, and 10th place bots.. that the 4th place one will win roughly 56% of the games 2010-12-01T01:09:14 given a 56% chance of winning, the odds of winning 75 or less games, out of 150, is roughly.. 2% 2010-12-01T01:09:44 so there's a 2% chance, that the ranks will be off by up to 6 places 2010-12-01T01:10:07 I have noticed that players get pitched with the same opponent very often.. is this because the round robin confines you to play only with bots whose "Turns to next game" matches yours? Or is it because of the Rank belt that we get to play against?... 2010-12-01T01:10:43 kincody: it mostly matches you only against people with similar rank as yourself 2010-12-01T01:10:57 so generally teh same 100 or so opponents 2010-12-01T01:11:29 burny, you're still on the ELO -- I would say ELO did an awesome job getting us the top players -- I'm still not clear how you can claim round robin across all maps with top N players wouldn't be "good enough" -- but maybe I'm just stuck in my own world :) 2010-12-01T01:11:30 *** amstan has joined #aichallenge 2010-12-01T01:11:30 *** ChanServ sets mode: +o amstan 2010-12-01T01:11:58 kincody: it is possible for it to match you up with anyone, but the larger the rank difference the smaller the probability 2010-12-01T01:12:11 virdo, it's 'good enough', to not care 2010-12-01T01:12:16 as a result you play people near your rank much more often 2010-12-01T01:12:42 *** mega1 has joined #aichallenge 2010-12-01T01:12:43 just like, if you flip the coin 200 times, and get 101 to 99.. it's 'good enough' to not care which side is heavier 2010-12-01T01:13:12 so the "Turns to next game" Dose not effect your probability of playing an opponent? 2010-12-01T01:13:21 no, there is no turns to next game 2010-12-01T01:14:17 *** xxx has quit IRC (Ping timeout: 265 seconds) 2010-12-01T01:16:05 the pairing for final, takes the bot which has least games played.. not the bot which has played least recently 2010-12-01T01:17:39 but is there not a fixed no of players? so wont it be periodic? hence confining you to a group of players? 2010-12-01T01:18:18 no 2010-12-01T01:18:28 it doesn't take BOTH players who have played least 2010-12-01T01:18:47 takes the 1 player who has played least.. and gives them a random opponent, tha tis appropriate to various rules 2010-12-01T01:19:10 Oh ok.. thanks for the clarification.. 2010-12-01T01:19:13 so strictly speaking, it's possible that after a total of 3k games have been palyed.. one player had 2k total games 2010-12-01T01:19:43 *** amstan_ has joined #aichallenge 2010-12-01T01:19:43 *** ChanServ sets mode: +o amstan_ 2010-12-01T01:20:13 *** amstan has quit IRC (Read error: No route to host) 2010-12-01T01:22:24 after playing through an average of a 200 games per bot.. each bot is likely to have between about 199 and 203 games played 2010-12-01T01:25:46 *** Ken69267 has quit IRC (Ping timeout: 240 seconds) 2010-12-01T01:27:11 *** Ken69267 has joined #aichallenge 2010-12-01T01:34:15 *** sigh has quit IRC (Ping timeout: 240 seconds) 2010-12-01T01:44:23 *** kincody has left #aichallenge 2010-12-01T01:46:49 *** Appleman1234 has quit IRC (Ping timeout: 264 seconds) 2010-12-01T01:48:37 Top 10 players: bocsimacko(3813), _iouri_(3611), Slin-.-(3570), jimrogerz(3564), _Astek_(3561), asavis(3545), Accoun(3545), GreenTea(3541), protocolocon(3537), george(3523) 2010-12-01T01:52:23 *** davidjliu has quit IRC (Quit: Page closed) 2010-12-01T01:55:46 *** Appleman1234 has joined #aichallenge 2010-12-01T02:33:49 *** contestbot has joined #aichallenge 2010-12-01T02:33:49 *** ChanServ sets mode: +o contestbot 2010-12-01T02:36:32 *** sigh has joined #aichallenge 2010-12-01T02:48:39 Top 10 players: bocsimacko(3809), _iouri_(3609), Slin-.-(3567), jimrogerz(3565), _Astek_(3553), asavis(3545), Accoun(3541), GreenTea(3540), protocolocon(3539), george(3519) 2010-12-01T02:57:14 *** Qulinxao has quit IRC (Ping timeout: 265 seconds) 2010-12-01T03:47:07 *** contestbot has joined #aichallenge 2010-12-01T03:47:07 *** ChanServ sets mode: +o contestbot 2010-12-01T03:48:38 Top 10 players: bocsimacko(3813), _iouri_(3609), Slin-.-(3563), jimrogerz(3563), _Astek_(3547), Accoun(3540), asavis(3539), GreenTea(3538), protocolocon(3535), george(3516) 2010-12-01T03:52:38 *** meneky has joined #aichallenge 2010-12-01T03:53:51 http://ai-contest.com/visualizer.php?game_id=9480860 2010-12-01T03:53:57 i'd never seen such a flat line before :P 2010-12-01T03:57:01 *** fgump has joined #aichallenge 2010-12-01T04:15:02 *** Sylph has joined #aichallenge 2010-12-01T04:15:44 *** fgump has quit IRC (Quit: Ex-Chat) 2010-12-01T04:16:32 *** Sylph2 has quit IRC (Ping timeout: 265 seconds) 2010-12-01T04:24:22 *** px1704 has joined #aichallenge 2010-12-01T04:37:59 *** Sylph has quit IRC (Read error: Connection reset by peer) 2010-12-01T04:38:02 *** Sylph2 has joined #aichallenge 2010-12-01T04:42:03 *** bss03-AFK is now known as bss03 2010-12-01T04:42:37 *** Neverstu has joined #aichallenge 2010-12-01T04:48:39 Top 10 players: bocsimacko(3805), _iouri_(3606), jimrogerz(3559), Slin-.-(3557), _Astek_(3542), asavis(3540), Accoun(3539), GreenTea(3530), protocolocon(3528), george(3513) 2010-12-01T04:49:49 *** Neverstu has quit IRC (Quit: Leaving) 2010-12-01T04:50:02 *** bss03 is now known as bss03-AFK 2010-12-01T04:51:08 *** Sylph has joined #aichallenge 2010-12-01T04:53:52 *** Sylph2 has quit IRC (Ping timeout: 240 seconds) 2010-12-01T05:05:51 *** Epona has joined #aichallenge 2010-12-01T05:06:54 *** pyro- has joined #aichallenge 2010-12-01T05:21:15 *** Naktibalda has joined #aichallenge 2010-12-01T05:24:54 *** Frontier has joined #aichallenge 2010-12-01T05:35:21 *** jaspervdj has joined #aichallenge 2010-12-01T05:37:57 Xrillo managed to get 38 with UCT 2010-12-01T05:38:19 I have to read the source. 2010-12-01T05:39:10 results results, when when 2010-12-01T05:40:22 in about 13 hours I think 2010-12-01T05:42:37 right.. so it's 02 am here 2010-12-01T05:48:39 Top 10 players: bocsimacko(3805), _iouri_(3600), jimrogerz(3559), Slin-.-(3550), _Astek_(3537), Accoun(3530), asavis(3526), protocolocon(3521), GreenTea(3518), george(3506) 2010-12-01T05:49:36 rabidus_: it's only an estimate 2010-12-01T05:49:46 k 2010-12-01T05:49:50 thx anyway 2010-12-01T05:49:52 so thats it 2010-12-01T05:54:00 But now I think that the rankings are really beginning to be accurate 2010-12-01T05:54:19 *** medrimonia2 is now known as medrimonia 2010-12-01T05:55:19 mega1: is there a favourite thing of yours explaining how UCT is actually done? 2010-12-01T05:55:50 antimatroid1: not really. Error323 asked this question yesterday. 2010-12-01T05:56:14 i don't really get it :P, ah well 2010-12-01T06:00:01 how would one hash an arimaa board? i can't really get a number 64 digits long represented in base x, plus i think i'd want more than 10 bases 2010-12-01T06:00:32 i probably don't have the correct understanding of what a hash is, but i might 2010-12-01T06:01:14 perhaps you wanna take a look at bitboards 2010-12-01T06:05:48 antimatroid1: maybe look at something like this for hashing: http://en.wikipedia.org/wiki/Zobrist_hashing 2010-12-01T06:07:24 isn't that sort of just representing the board as a b based number like i said above? 2010-12-01T06:07:42 it's nothing like that o_O 2010-12-01T06:08:22 antimatroid1: yes sort of, you represent e.g. the moves of a single piece in a bitboard 2010-12-01T06:08:43 oh ok, you were talking to Error323 2010-12-01T06:08:46 which can be done in 64 bit 2010-12-01T06:08:51 8x8 2010-12-01T06:09:18 and then you can apply logical operators to apply moves quickly and determine states 2010-12-01T06:12:30 *** guyverthree has joined #aichallenge 2010-12-01T06:12:45 hey guy7s 2010-12-01T06:13:16 hey 2010-12-01T06:13:22 who are guyver one and two? 2010-12-01T06:15:27 have you not seen the manga guyver ??? 2010-12-01T06:15:44 when i was young it was the only user name that was free 2010-12-01T06:16:27 liol 2010-12-01T06:17:09 can't say i have 2010-12-01T06:17:28 although my first email address was something like girl_lover_3 so i can't talk aha 2010-12-01T06:17:40 O_o 2010-12-01T06:18:03 sigh: this was a good 10+ years ago :P 2010-12-01T06:18:26 i was but a young immature boy :) 2010-12-01T06:18:38 now i'm just a young immature man :) 2010-12-01T06:18:41 haha 2010-12-01T06:18:51 and you gave this address out to people? 2010-12-01T06:19:17 i would have been in like primary school, it would have been for registering to websites etc. 2010-12-01T06:19:21 i didn't know people to email 2010-12-01T06:19:52 :P 2010-12-01T06:21:13 yeah mine was when i was like 12 2010-12-01T06:21:22 i loved manga as a child 2010-12-01T06:22:59 i still do 2010-12-01T06:23:02 =] 2010-12-01T06:23:11 as a child 2010-12-01T06:25:05 woo for growing up with the internet :) 2010-12-01T06:25:15 i'd die without it 2010-12-01T06:25:21 well, permanently 2010-12-01T06:25:25 i can go away :P 2010-12-01T06:25:58 oh god, i was at a three day music festival over new years, and they had an internet cafe boothe, hardly the right place :\ 2010-12-01T06:26:55 mcleopold crashed against me 2010-12-01T06:27:19 sweet 2010-12-01T06:27:22 i'm climbing 2010-12-01T06:27:26 72 nwo 2010-12-01T06:28:41 i'm falling 2010-12-01T06:28:42 35 2010-12-01T06:38:33 I'm relatively stable 2010-12-01T06:38:44 between 17-19 2010-12-01T06:39:13 stable in what? 2010-12-01T06:39:32 rankings^^ 2010-12-01T06:40:21 ah I see, as their calculating everything? 2010-12-01T06:41:16 so its still not over ;_; 2010-12-01T06:48:40 Top 10 players: bocsimacko(3805), _iouri_(3592), jimrogerz(3551), Slin-.-(3543), _Astek_(3530), Accoun(3526), protocolocon(3516), asavis(3515), GreenTea(3513), george(3503) 2010-12-01T06:50:15 /msg mega1 looking good. are you managing to relax a little now? i suppose it isn't final yet, but it looks pretty certain to me :) 2010-12-01T06:50:22 heh 2010-12-01T06:50:49 /msg dmj111` yes, I'm relaxed 2010-12-01T06:50:57 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-12-01T06:51:12 213 points is a new record 2010-12-01T06:51:17 excellent. 2010-12-01T06:53:53 iouri looks safe as well 2010-12-01T06:54:12 but it's tightly packed below that 2010-12-01T06:59:41 yep. probably no one will come out of the group to get ahead of him. 2010-12-01T07:00:28 *** dmj111` is now known as dmj111 2010-12-01T07:02:53 i'm down at 460 2010-12-01T07:05:55 yeah 2010-12-01T07:05:59 it still amazes me 2010-12-01T07:06:02 the difference 2010-12-01T07:07:23 wow mega1 i just realized your binary is 48 MB 2010-12-01T07:07:25 that's insane 2010-12-01T07:07:36 Error323: that's the whole lisp 2010-12-01T07:07:48 it includes the vm 2010-12-01T07:08:01 it's a jvm+class files if you prefer 2010-12-01T07:08:41 the heap mmap'ed on startup so it's pretty quick 2010-12-01T07:08:48 *the heap is 2010-12-01T07:08:59 but the whole heap is in that binary 2010-12-01T07:09:26 ahh 2010-12-01T07:09:33 *** Apophis has quit IRC (Read error: Connection reset by peer) 2010-12-01T07:09:57 *** Apophis has joined #aichallenge 2010-12-01T07:09:57 *** rhinobot has joined #aichallenge 2010-12-01T07:10:45 it could also run from the fasl files (= class files) if it weren't bundled into one executable with the vm 2010-12-01T07:10:46 hmmm? 2010-12-01T07:10:49 *** amstan_ is now known as amstan 2010-12-01T07:11:18 oh no we woke up the beast 2010-12-01T07:11:54 argh! 2010-12-01T07:11:55 run! 2010-12-01T07:16:45 really weird 2010-12-01T07:16:56 i guess i broke the wall before me :P 2010-12-01T07:17:02 because i'm climbing 2010-12-01T07:18:46 just don't catch me :P 2010-12-01T07:19:04 *** Palmik has joined #aichallenge 2010-12-01T07:19:13 @time 2010-12-01T07:19:13 Error323: 07:19 AM, December 01, 2010 2010-12-01T07:19:23 11 hours left? 2010-12-01T07:19:41 oh well less then 24 :) 2010-12-01T07:19:45 It's closing at 18AM? 2010-12-01T07:20:08 I don't know, I heard "afternoon" and night 2010-12-01T07:20:18 we'll see 2010-12-01T07:21:04 Hum at least it'll be finished tomorrow morning (for our hour Error323) 2010-12-01T07:21:12 *** Frontier has quit IRC (Ping timeout: 240 seconds) 2010-12-01T07:21:17 freedom 2010-12-01T07:21:39 indd medrimonia 2010-12-01T07:23:42 contestbot is not server time 2010-12-01T07:23:43 amstan: User error -- Replace user. 2010-12-01T07:23:46 substract one from it 2010-12-01T07:24:03 OH 2010-12-01T07:24:06 oh* 2010-12-01T07:24:53 yep it's our hour-7 2010-12-01T07:26:56 sum [x | x<-[1..999], x `mod` 3 == 0 || x `mod` 5 == 0] 2010-12-01T07:26:57 ^^ that's neat :) 2010-12-01T07:26:59 i like haskell 2010-12-01T07:28:00 and by neat i mean concise :) 2010-12-01T07:28:21 antimatroid1: yeah, list comphrensions are awesome. 2010-12-01T07:28:41 so is spelling, I have heard. 2010-12-01T07:28:51 spelling is meh 2010-12-01T07:28:58 loool 2010-12-01T07:28:59 provided it's unambiguous 2010-12-01T07:29:08 lisp doesn't have list comphrensions right... 2010-12-01T07:29:17 but can the macro system add them? 2010-12-01T07:29:32 lol 2010-12-01T07:29:49 viral spelling errors... i cut and pasted that word 2010-12-01T07:30:05 two of us did it, must be right :) 2010-12-01T07:30:18 wikipedia would agree 2010-12-01T07:30:21 *** guyverthree has left #aichallenge 2010-12-01T07:30:45 i'd rather 2010-12-01T07:30:46 sum{x : x `elem`{1,...,999}, 3|x || 5|x} and `elem` be the proper relation that would be on my perfect keyboard 2010-12-01T07:31:40 there is an emacs mode somewhere to show haskell with the greek symbols and stuff. 2010-12-01T07:31:56 of course, there is an emacs mode for everything. 2010-12-01T07:32:07 there is an emacs mode where P=NP ! 2010-12-01T07:32:07 i should pick a better editor 2010-12-01T07:32:11 i just use codeblocks editor 2010-12-01T07:32:15 fibs = 0 : 1 : zipWith (+) fibs (tail fibs) <-- this is neat 2010-12-01T07:32:30 http://mult.ifario.us/p/emacs-haskell-mode-unicode-cuteness 2010-12-01T07:33:16 neat 2010-12-01T07:33:16 http://en.literateprograms.org/List_comprehension_%28Lisp%29 2010-12-01T07:33:27 pure awesomeness 2010-12-01T07:34:42 sigh: uhhh wth does it mean 2010-12-01T07:35:10 delt0r_: nice. 2010-12-01T07:35:15 Error323: to generate the list of fibonacci numbers add the list to the tail of itself :P 2010-12-01T07:35:25 seed with [0,1] 2010-12-01T07:35:48 thinking of going back to lisp 2010-12-01T07:36:11 but its been almost 15 years 2010-12-01T07:36:12 ah 2010-12-01T07:36:23 yeah i wanna look into lisp also 2010-12-01T07:36:31 delt0r_: i went and dug all of my books out that have been in the basement. 2010-12-01T07:36:37 i learned prolog during my bachelors 2010-12-01T07:36:39 *** phreeza has joined #aichallenge 2010-12-01T07:36:48 not lisp 2010-12-01T07:36:48 i have touched prolog during undergrad 2010-12-01T07:36:52 yes. 2010-12-01T07:36:54 lisp was my first "real" lanuage... ie not asm or basic 2010-12-01T07:36:55 not sure i'd say i learnt it 2010-12-01T07:37:06 lol 2010-12-01T07:37:17 well i was doing pretty well in it 2010-12-01T07:37:23 but it took a lot of time 2010-12-01T07:37:37 well my only problem with prolog is that its hard to reason the complexity level of some code 2010-12-01T07:37:44 our AI subject was taught in prolog 2010-12-01T07:37:57 it made me understand the concept of recursion to levels i never knew existed :P 2010-12-01T07:37:57 sigh: that's where i saw it 2010-12-01T07:38:26 recusion isn't as cool as induction 2010-12-01T07:38:34 mja it probably has to do with SWI prolog stemming from our country 2010-12-01T07:38:42 "show this happens for all n great tha..." bam induction 2010-12-01T07:38:52 =] 2010-12-01T07:38:52 induction is kinda recursion... 2010-12-01T07:38:56 indd 2010-12-01T07:39:09 they aren't really equivalent though :P 2010-12-01T07:39:16 what is the difference? 2010-12-01T07:40:24 induction starts at a and then assumes it hodls for n>=a, shows n+1 holds, then that says it holds for all n>=a 2010-12-01T07:40:25 induction just assumes the repeated process will just end at some point? 2010-12-01T07:40:32 i'm not seeing the equivalence :\ 2010-12-01T07:40:47 recursion is used for algorithms, induction for proofs 2010-12-01T07:41:34 both start from a base and the general case 2010-12-01T07:41:36 antimatroid1: sure, recursion can be infinite by design, but most times it has a base case 2010-12-01T07:42:34 it's a viscious cycle learning 2010-12-01T07:42:52 it just makes me want to learn more, but i can't keep up 2010-12-01T07:43:35 wow mega1 the difference is growing fast now 2010-12-01T07:43:42 is that due to the cutoff? 2010-12-01T07:44:04 the other bots are cowering in terror 2010-12-01T07:44:11 elos are dropping because range has decreased 2010-12-01T07:44:39 but mega1 's elo is dropping much slower 2010-12-01T07:45:37 *** mceier has joined #aichallenge 2010-12-01T07:48:40 Top 10 players: bocsimacko(3802), _iouri_(3585), jimrogerz(3545), Slin-.-(3539), _Astek_(3523), Accoun(3519), GreenTea(3509), protocolocon(3508), asavis(3508), george(3505) 2010-12-01T07:49:34 http://ai-contest.com/visualizer.php?game_id=9503048 win of raschi :D 2010-12-01T07:50:47 sloppy 2010-12-01T07:50:51 on both sides 2010-12-01T07:50:57 true ya 2010-12-01T07:51:14 but gj 2010-12-01T07:56:08 *** Migi32 has joined #aichallenge 2010-12-01T07:56:23 antimatroid1: induction is a proof because of repeated application of the fact that n+1 is true and that if its true for n=1 then its also true for any positive integer because i can repeatedly apply the n+1 2010-12-01T07:57:31 wow, these rankings... Are you writing a post-mortem yet, mega1? 2010-12-01T07:57:37 so its looping 2010-12-01T07:57:51 and looping == recursion 2010-12-01T07:57:54 qed 2010-12-01T07:57:58 ;) 2010-12-01T07:58:22 looping < recursion 2010-12-01T07:58:27 wow getting quite heavy snow now.... 2010-12-01T07:58:29 lol 2010-12-01T07:59:02 any loop can be written with recursion, but not every recursion can be written in a loop 2010-12-01T07:59:04 Error323: all recursions can be writen as loops and also the other way round 2010-12-01T07:59:13 yes it can 2010-12-01T07:59:17 oh? 2010-12-01T07:59:29 you just have an explicit stack 2010-12-01T07:59:47 ie you do the stack thing yourself 2010-12-01T07:59:49 ah mja 2010-12-01T07:59:51 Migi32: yes, I'm writing it 2010-12-01T08:00:01 mega1: eta? 2010-12-01T08:00:26 but recursion is cooler than loops 2010-12-01T08:00:44 Error323: tomorrow 2010-12-01T08:00:49 sweet 2010-12-01T08:01:36 *** nullkuhl has joined #aichallenge 2010-12-01T08:02:23 delt0r_: mmm, i don't yet see how one can simulate the unwinding 2010-12-01T08:02:32 but i probably need to think a bit harder 2010-12-01T08:02:34 unwinding? 2010-12-01T08:02:43 i avoid recursion when i can 2010-12-01T08:02:43 pop 2010-12-01T08:02:54 or not being lazy 2010-12-01T08:03:07 exactly pyro- 2010-12-01T08:03:35 push and pop--what function calling does under the hood 2010-12-01T08:03:45 doing some asm helps here... 2010-12-01T08:03:57 antimatroid1, Sometimes its the right tool for the job 2010-12-01T08:04:02 other times its not 2010-12-01T08:04:20 asm? 2010-12-01T08:04:24 I use it a lot because right now I am dealing with tree datastructures a lot 2010-12-01T08:04:26 assembly 2010-12-01T08:04:29 antimatroid1: you avoid recursion and you are learning haskell? 2010-12-01T08:04:29 you mean assembly? 2010-12-01T08:04:34 lol 2010-12-01T08:04:40 sigh: :P 2010-12-01T08:04:50 a lot of things i do don't make sense 2010-12-01T08:05:03 i don't dislike recursion 2010-12-01T08:05:06 in older assmblers at least you have to do the push and poping yourself... 2010-12-01T08:05:15 i've just been told it's bad to do that over the alternative where it's easy? 2010-12-01T08:05:17 is that wrong? 2010-12-01T08:05:23 yes 2010-12-01T08:05:23 more modern cpus have instructions for it 2010-12-01T08:05:37 ?? 2010-12-01T08:05:56 if a loop is more natrual...fine... but if recursion... then use it imo 2010-12-01T08:06:13 In java at least recursion has about the same performance as a loop 2010-12-01T08:06:15 ouch, my longest winning streak ended 2010-12-01T08:06:40 an iterative form rarely seems the most natural when you have an implementation of scheme 2010-12-01T08:06:46 don't know about other langs... would expect C/C++ stack frames to be quite light however 2010-12-01T08:06:56 x86 don't have a lot of registers 2010-12-01T08:07:29 recursion kills python though... that kind of sucks 2010-12-01T08:07:32 pyro-, yea---well lisp has "loop" 2010-12-01T08:07:38 * antimatroid1 doesn't know much past treating compilers like a big black box that do magic for me and what i learn about different languages 2010-12-01T08:08:03 recursion kills python? wat 2010-12-01T08:08:03 delt0r_, scheme has DO, but it doesn't see much use 2010-12-01T08:08:20 antimatroid1: it's more about understanding what a function call actually *does* 2010-12-01T08:08:36 but scheme must to the "tail recursive" thing right? 2010-12-01T08:08:43 if you get that, then you will understand how to translate bw recursion and iteration 2010-12-01T08:08:43 Epona, the reference python interpreter is always running out of stack space when you write recursive functions 2010-12-01T08:08:55 delt0r_, yes, they have to optimise that 2010-12-01T08:09:01 example 2010-12-01T08:09:08 oh no, the flagcapper is gaining territory 2010-12-01T08:09:08 please 2010-12-01T08:09:40 Epona, do you know what a recursive function is? 2010-12-01T08:09:40 *** rhinobot has quit IRC (Quit: Page closed) 2010-12-01T08:09:41 Epona: in any lang 2010-12-01T08:09:43 @rankings 2010-12-01T08:09:44 amstan: Top 10 players: bocsimacko(3804), _iouri_(3585), jimrogerz(3547), Slin-.-(3538), _Astek_(3521), Accoun(3520), GreenTea(3510), protocolocon(3508), asavis(3504), george(3503) 2010-12-01T08:09:45 *** nullkuhl has quit IRC (Remote host closed the connection) 2010-12-01T08:10:04 function int bob() {return bob()} ;) 2010-12-01T08:10:17 :) 2010-12-01T08:10:33 if you can show me your problem in python i might be able to help 2010-12-01T08:10:56 Epona, we don't have a problem, we're complaining about the python interpreter 2010-12-01T08:11:08 Epona: you want an example of recursion killing python? 2010-12-01T08:11:22 if you need scalability --then converting recursion to loops is "accepted" wisdom 2010-12-01T08:11:25 since apparently its a bad problem, yes 2010-12-01T08:11:35 def s(x): return 1+s(x-1) if x > 0 else 0 2010-12-01T08:11:47 this is a simple way to reach the recursion limit easily 2010-12-01T08:11:57 delt0r_, generally not required in scheme 2010-12-01T08:12:06 (nor useful) 2010-12-01T08:12:24 well tail recursion is the same thing... no stack frames 2010-12-01T08:12:49 s(1000) is larger than the default recursion limit for python 2010-12-01T08:13:01 thats really really low 2010-12-01T08:13:05 r u sure? 2010-12-01T08:13:33 im forgot url to server stats page.. 2010-12-01T08:13:44 sigh, I don't see how that relates to python 2010-12-01T08:13:47 @where stats 2010-12-01T08:13:48 amstan: stats = http://ai-contest.com/game_stats.php 2010-12-01T08:13:48 you should use pastebin 2010-12-01T08:13:56 thank you! 2010-12-01T08:14:14 delt0r_: sys.getrecursionlimit() == 1000 on my machine 2010-12-01T08:14:32 Epona: you are right, it is related to the default implementation 2010-12-01T08:14:52 wow --same on mine 2010-12-01T08:15:37 even if you increase it, the frames for python are huge and no tail optimisation 2010-12-01T08:15:49 so often it is better to manage your own stack 2010-12-01T08:16:07 or use a better implementation ;-) 2010-12-01T08:16:25 sure :P 2010-12-01T08:16:54 *** Palmik has quit IRC (Remote host closed the connection) 2010-12-01T08:17:30 there is stackless python 2010-12-01T08:17:41 lol some "About Me's" are just funny and a bit weird at the same time: http://ai-contest.com/profile.php?user_id=12799 2010-12-01T08:17:46 eve online is/was writen in it 2010-12-01T08:18:41 *** Appleman1234 has joined #aichallenge 2010-12-01T08:24:46 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-12-01T08:25:19 another interesting read: http://reactorsoftware.com/blog/?p=24 2010-12-01T08:25:25 by neverstu 2010-12-01T08:26:44 *** Appleman1234 has joined #aichallenge 2010-12-01T08:31:44 *** Epona has quit IRC (Ping timeout: 265 seconds) 2010-12-01T08:41:19 http://ai-contest.com/visualizer.php?game_id=9508417 O_O 2010-12-01T08:43:53 ouch 2010-12-01T08:43:54 *** perror has joined #aichallenge 2010-12-01T08:45:27 *** Palmik has joined #aichallenge 2010-12-01T08:45:50 i love it when physcisists say there are only 4 dimensions 2010-12-01T08:47:44 why do you love it? 2010-12-01T08:48:42 Top 10 players: bocsimacko(3803), _iouri_(3586), jimrogerz(3546), Slin-.-(3540), Accoun(3520), _Astek_(3520), GreenTea(3510), george(3506), protocolocon(3506), asavis(3502) 2010-12-01T08:50:26 6:2 vs Accoun 2010-12-01T08:50:43 because they don't state that they're talking about space and time, just that there exists only 4 dimensions 2010-12-01T08:50:46 which is complete bullshit 2010-12-01T08:51:02 it makes me smile 2010-12-01T08:51:37 meh, usually the context makes it obvious 2010-12-01T08:51:37 hell, you can take as many spatial dimensions as you want 2010-12-01T08:52:00 nah, i've seen them tout it when it's not relevant to just that 2010-12-01T08:52:14 and argue with me that only 4 dimensions exist 2010-12-01T08:52:22 as a general blanket statement 2010-12-01T08:52:39 well, without any context that statement doesn't really make sense :S 2010-12-01T08:53:10 usually I've only seen random people who don't actually know anything make statements like that :P 2010-12-01T08:53:27 usually after watching that "imagining the 10th dimension" video 2010-12-01T08:54:09 hehe, it's a thread on askreddit, i'm pretty sure that's been posted 2010-12-01T08:54:13 i should just leave such things alone :P 2010-12-01T08:54:34 are you sure it's a physisist? 2010-12-01T08:54:45 *physicist 2010-12-01T08:55:30 probably not :P 2010-12-01T08:56:06 I've heard a lot of retarded statements regarding dimensions on reddit :P 2010-12-01T08:57:14 i've seen (and probably made) a lot of retarded comments on reddit :P 2010-12-01T08:58:16 :) 2010-12-01T08:59:22 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-12-01T09:00:36 * Error323 only just (through this channel) learned about reddit 2010-12-01T09:01:37 it's a trap! 2010-12-01T09:02:11 and its horrible, horrible layout 2010-12-01T09:02:20 no way i'm gonna read that 2010-12-01T09:02:23 phew 2010-12-01T09:02:23 haha 2010-12-01T09:03:06 anyway, time for me to sleep 2010-12-01T09:03:20 cya 2010-12-01T09:03:23 cya 2010-12-01T09:03:39 *** sigh has quit IRC (Quit: Leaving...) 2010-12-01T09:12:56 Erro323: the layout grows on you 2010-12-01T09:12:57 i love it 2010-12-01T09:13:09 Error323** 2010-12-01T09:13:32 I doubt that 2010-12-01T09:14:04 why not? it's wonderful 2010-12-01T09:14:11 it's simple and displays a lot of informations 2010-12-01T09:14:13 and looks neat 2010-12-01T09:16:07 *** mceier has quit IRC (Quit: leaving) 2010-12-01T09:20:47 *** nullkuhl has joined #aichallenge 2010-12-01T09:21:40 *** yasith has quit IRC (Ping timeout: 250 seconds) 2010-12-01T09:26:41 Erro323: how did you find out about the contest? 2010-12-01T09:26:55 i would guess that at least 90% of contestants are redditors :P 2010-12-01T09:27:25 i could be wrong, but i don't know where else it's been posted 2010-12-01T09:33:23 *** yasith has joined #aichallenge 2010-12-01T09:39:16 *** Zoo has quit IRC (Ping timeout: 240 seconds) 2010-12-01T09:43:13 *** nn has joined #aichallenge 2010-12-01T09:45:40 *** nn has quit IRC (Client Quit) 2010-12-01T09:45:52 *** antimatroid has joined #aichallenge 2010-12-01T09:46:52 *** antimatroid1 has quit IRC (Ping timeout: 255 seconds) 2010-12-01T09:48:43 Top 10 players: bocsimacko(3802), _iouri_(3586), jimrogerz(3539), Slin-.-(3536), _Astek_(3520), Accoun(3516), GreenTea(3509), protocolocon(3506), george(3504), asavis(3502) 2010-12-01T10:00:24 *** mischmo has joined #aichallenge 2010-12-01T10:00:28 Mistmanov: http://www.phdcomics.com/comics/archive.php?comicid=1354 2010-12-01T10:01:09 :o 2010-12-01T10:01:16 hey I totally started work on my thesis again 2010-12-01T10:01:29 I just... had.. to improve my bot again.. 2010-12-01T10:01:36 I'm near the horizon 2010-12-01T10:01:42 event horizon 2010-12-01T10:01:45 xD 2010-12-01T10:05:05 * antimatroid hasn't got far enough for thesises yet 2010-12-01T10:05:11 i have a small one next year 2010-12-01T10:13:29 *** superflit has quit IRC (Quit: superflit) 2010-12-01T10:16:21 *** mischmo has left #aichallenge 2010-12-01T10:17:11 *** Zoo has joined #aichallenge 2010-12-01T10:18:00 *** iouri has joined #aichallenge 2010-12-01T10:19:13 * phreeza didn't make the cutoff this time :( 2010-12-01T10:19:21 *** nullkuhl has quit IRC (Ping timeout: 245 seconds) 2010-12-01T10:23:04 *** nullkuhl has joined #aichallenge 2010-12-01T10:24:46 *** bhasker has joined #aichallenge 2010-12-01T10:24:52 *** JamesMG has joined #aichallenge 2010-12-01T10:25:07 *** bhasker has quit IRC (Client Quit) 2010-12-01T10:32:16 holy crap 2010-12-01T10:32:18 its iouri 2010-12-01T10:32:20 hey :) 2010-12-01T10:32:24 nice article 2010-12-01T10:32:32 *** Mistmanov has quit IRC (Ping timeout: 265 seconds) 2010-12-01T10:32:40 hey 2010-12-01T10:32:44 thanks! 2010-12-01T10:33:50 Neverstu also has a pretty neat article, describes things in a lot more detail than me: http://reactorsoftware.com/blog/?p=24 2010-12-01T10:33:59 yeah read that too 2010-12-01T10:34:21 mega1's (bocsimacko's) is comming out tomorrow 2010-12-01T10:34:21 =) 2010-12-01T10:34:37 I was concerned that if I started going in-depth into my source code, the article would be 5x as long as it is 2010-12-01T10:34:48 can't wait, would be an interesting read:) 2010-12-01T10:34:57 heh yeh, I examined your src code 2010-12-01T10:35:07 nice and clean 2010-12-01T10:35:13 well compared to my jibberish =) 2010-12-01T10:35:36 I liked it 2010-12-01T10:35:37 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-12-01T10:37:21 lol, "nice and clean"? I guess it might be to a degree.. It has a lot of useless stuff in it, out of date comments, and way too many #ifdef conditions 2010-12-01T10:38:03 well ifdefs mostly for submission 2010-12-01T10:39:03 I was always very curious as to how ppl evaluated actinos 2010-12-01T10:39:06 actions* 2010-12-01T10:39:12 as this was my biggest issue 2010-12-01T10:39:17 i tried alphabeta 2010-12-01T10:39:23 but couldn't get it right 2010-12-01T10:39:33 evaluation-wise 2010-12-01T10:39:47 Hm.. Yeah, I think tree-based searches just don't work as well in this game 2010-12-01T10:39:58 *** nullkuhl has joined #aichallenge 2010-12-01T10:40:15 I must admit that bocsimacko code is quite clean with really good comments 2010-12-01T10:40:37 iouri, I though your code was not bad either 2010-12-01T10:40:49 for c++ :p ( 2010-12-01T10:40:52 :D 2010-12-01T10:40:59 heh 2010-12-01T10:41:09 i wish bocsimacko's code was c++ hehe 2010-12-01T10:41:22 lisp is not so hard to read 2010-12-01T10:41:31 it is 2010-12-01T10:41:36 I hadn't done it for years... 2010-12-01T10:41:43 `,'((())))()()(())))(((()) 2010-12-01T10:41:45 # 2010-12-01T10:41:50 w.t.f. stuff 2010-12-01T10:41:51 =) 2010-12-01T10:41:51 yet could follow the code just by looking up the convections 2010-12-01T10:42:17 Error323, C++ does that too -- {[][]()[]((()))([])} 2010-12-01T10:42:31 less 2010-12-01T10:42:46 formatting convenctions are different... but easy to get use too 2010-12-01T10:42:51 but i'm not bothered by that by it that much 2010-12-01T10:43:03 just track the indentation 2010-12-01T10:43:09 what annoys me is the ',` stuff 2010-12-01T10:43:21 and view () as "making it easy for the parser" 2010-12-01T10:43:22 ',`#* even 2010-12-01T10:43:38 oh- well thats kind of important 2010-12-01T10:43:49 ye i figured that :P 2010-12-01T10:43:53 bit like all the stuff in C++ 2010-12-01T10:43:53 I didn't get around to looking at his code yet mostly because I'm too lazy to look at anything that's not browsable directly online 2010-12-01T10:44:26 and the & * stuff in C/C++ 2010-12-01T10:44:44 And the iterators are waay to wordy 2010-12-01T10:44:51 too much typing 2010-12-01T10:45:12 and frowned upon it seems 2010-12-01T10:45:31 ?? you mean STL iterators? 2010-12-01T10:45:36 yeah 2010-12-01T10:45:39 iouri: i got you a few times with my game tree :P 2010-12-01T10:45:50 and on tcp 2010-12-01T10:45:59 the final tree i uploaded is broken as all hell 2010-12-01T10:46:10 *** superflit has joined #aichallenge 2010-12-01T10:46:15 hm. There's a chance then that I didn't approach the game tree correctly 2010-12-01T10:46:36 I beat it 49 out of 150 :P 2010-12-01T10:47:03 in a couple of lines, my moves were vectors of fleets all with the same destination, but different sources and send times, then at each node i made 1 move per player, and sort of planed out moves for any point in time 2010-12-01T10:47:10 then i built the game tree off that 2010-12-01T10:47:32 iouri: you can browse mega1's source online at http://quotenil.com/git/?p=planet-wars.git 2010-12-01T10:47:32 it wasn't very good though :P 2010-12-01T10:48:00 yea, though i rlly hate that browser 2010-12-01T10:48:03 github ftw 2010-12-01T10:48:43 Top 10 players: bocsimacko(3800), _iouri_(3585), Slin-.-(3538), jimrogerz(3532), _Astek_(3520), Accoun(3515), GreenTea(3508), george(3507), asavis(3505), protocolocon(3499) 2010-12-01T10:48:59 well ones thing is for sure 2010-12-01T10:49:17 click on commit next to the final checkin and the important files are way down at the bottom of that page 2010-12-01T10:49:23 no matter what system is used --the winner was always going to be well determined 2010-12-01T10:49:34 delt0r_; yep 2010-12-01T10:49:49 hehe oh ye 2010-12-01T10:49:49 Error323: yeah, that browser is certainly not as nice as github 2010-12-01T10:50:14 even i went to the effort of getting amstan to put mine on github :P 2010-12-01T10:50:19 janzert: yea its clearly created by programmers alone :P 2010-12-01T10:50:57 gitweb iirc 2010-12-01T10:52:51 *** Utkarsh has quit IRC (Ping timeout: 272 seconds) 2010-12-01T10:53:20 i'm 5wins/1draw/1loss against accoun in the finals :P 2010-12-01T10:53:37 yet he insists he should be in second place 2010-12-01T10:53:37 *** rhinobot has joined #aichallenge 2010-12-01T10:54:01 he was doing pretty well on the TCP server 2010-12-01T10:54:12 i'm head of my country with 5 elo points... contest.must.stop.now. xD 2010-12-01T10:54:15 though there weren't nearly as many opponents there 2010-12-01T10:54:15 he was, but some people weren't there or always testing 2010-12-01T10:54:40 i tended to use more secretive names most of the time 2010-12-01T10:54:57 so people didn't know it was my retarded bot :P 2010-12-01T10:55:01 lol 2010-12-01T10:55:04 lol 2010-12-01T10:55:28 antimatroid: what names did you use? (unless it's still a secret) 2010-12-01T10:55:41 testbot the last couple of days 2010-12-01T10:55:55 others were crashbot, retardbot, pbot, hbot etc. 2010-12-01T10:56:30 lol 2010-12-01T10:56:55 i clinched first for australia right at the end from smiley and dimka too :P 2010-12-01T10:57:12 i hadn't put a bot up in about a month with an uncompilable submission uploaded 2010-12-01T10:58:44 http://ai-contest.com/visualizer.php?game_id=9521580 2010-12-01T10:58:48 *** Utkarsh has joined #aichallenge 2010-12-01T10:58:48 weird win 2010-12-01T10:59:01 I've got this strategy that just sends every ship when I attack (unless I was closest), because i found it to be very hard to predict what to send. 2010-12-01T10:59:24 Now darthmasta has an annoying antistrategy for it 2010-12-01T10:59:37 he just swaps planet with me -.- and wins because of that lol 2010-12-01T10:59:54 Error323: i sent exactly enough to take it if they sent everything 2010-12-01T11:00:17 same with expanding, on first round defense moves i think i just cancelled out their attacks though 2010-12-01T11:00:22 he does it after i send it 2010-12-01T11:00:52 http://ai-contest.com/visualizer.php?game_id=9509970 2010-12-01T11:00:53 i'd also plan my moves up to time "maxdistance" for any planet 2010-12-01T11:01:19 at one point mine was set up to send every possible ship from a planet only if the target was higher growth 2010-12-01T11:01:19 why send so much? 2010-12-01T11:01:33 intending to trade and then move back with the extra ships 2010-12-01T11:01:42 because at the moment of sending it becomes my new frontline -,- 2010-12-01T11:01:48 but then he flips planet 2010-12-01T11:01:50 admitidely my move selection functions are GROSS 2010-12-01T11:01:53 and its not anymore 2010-12-01T11:02:32 https://github.com/amstan/antimatroid/blob/master/GameState.cc look from line 229 2010-12-01T11:02:33 so gross :P 2010-12-01T11:03:37 *** xxx_ has joined #aichallenge 2010-12-01T11:04:55 I'm too ashamed to show mine 2010-12-01T11:05:34 i am so ashamed of the tree i uploaded at the end 2010-12-01T11:11:31 3 elo :S :S :S 2010-12-01T11:13:42 antimatroid: looking at your code... I wonder what would happen if i plug your pickMoves into my UCT bot :) 2010-12-01T11:14:07 *** rebelxt has joined #aichallenge 2010-12-01T11:14:36 that was my weakness i think... tried to come up with something sensible, but in the end i just sampled the entire space of possible moves uniformly 2010-12-01T11:14:49 phreeza: my pick move function is horrible inefficient 2010-12-01T11:14:57 which is quite terrible, but for some reason anything "smart" i tried was worse 2010-12-01T11:15:10 a lot better than mine... :) 2010-12-01T11:15:18 and i have nothing to make planets safe other than for incoming fleets 2010-12-01T11:15:49 the code for that is all fine except makeChildren in tree.cc is really really broken 2010-12-01T11:15:50 line 103: https://github.com/phreeza/planetwars_bot/blob/master/cpp/GameTree.cpp 2010-12-01T11:16:16 https://github.com/phreeza/planetwars_bot/blob/master/cpp/GameTree.cpp#L103 2010-12-01T11:16:33 iouri: nice blog post on your bot 2010-12-01T11:16:54 phreeza: did you use UCT? (i just joined chat) 2010-12-01T11:17:02 yep 2010-12-01T11:17:02 phreeza: where are you placed? 2010-12-01T11:17:12 phreeza: sweet 2010-12-01T11:17:15 rebelxt: thanks! 2010-12-01T11:17:19 terrible 2010-12-01T11:17:23 1300 i think 2010-12-01T11:17:25 just a sec 2010-12-01T11:17:35 rating is disappear 2010-12-01T11:17:41 on user pages.. 2010-12-01T11:17:42 xrillo is UCTing as well but he did not post source 2010-12-01T11:18:04 ah yes :( no more ratings 2010-12-01T11:18:12 phreeza: what i was trying to do at the end with my tree was only add moves for me if after checking their moves I could do at least as well as if i considered doing nothing 2010-12-01T11:18:16 but i messed it up terrible 2010-12-01T11:18:23 ratings reset incoming? ;-) 2010-12-01T11:18:44 i'm 1286 :D 2010-12-01T11:18:53 the biggest problem with my bot is the number of layers for me i generate is the max number of "actual" moves i'll make each turn, which is usually like 1 or 2, the rest of my fleets are all bluff carrying out other orders 2010-12-01T11:19:13 terribly* damnit! 2010-12-01T11:19:21 i made that mistake too many times in the last ten minutes 2010-12-01T11:19:59 http://ai-contest.com/visualizer.php?game_id=9523367 dohhhhhh 2010-12-01T11:20:04 oh, n_layers = n_moves? 2010-12-01T11:20:07 weird 2010-12-01T11:20:35 well i could have picked out 3 moves per "turn" at each node etc. but i couldn't work out how to correctly pick out moves that were jointly feasible 2010-12-01T11:20:44 and i liked having my tree "string" moves together 2010-12-01T11:21:17 but once a map has many claimable planets on the first turn, it's crippled 2010-12-01T11:21:31 anyone knows why the ranking positions have disappear from the profile pages? 2010-12-01T11:22:22 they are back 2010-12-01T11:22:22 they're back now, but frozen along with the leaderboard until final results are released this evening 2010-12-01T11:23:05 positions and elos have been frozen? 2010-12-01T11:23:09 janzert: you mean the contest has now stopped? or has become invisible? 2010-12-01T11:23:30 you just won't see any updates in the ranking or elos 2010-12-01T11:23:39 internally they are still being updated the same 2010-12-01T11:23:41 perfect torture 2010-12-01T11:23:46 duuuuuuude 2010-12-01T11:23:48 hahaha 2010-12-01T11:23:54 productivity++ 2010-12-01T11:24:08 you can still see the individual games as well 2010-12-01T11:24:26 sorry this was the compromise I worked with Jeff yesterday :P 2010-12-01T11:24:41 ok 2010-12-01T11:24:47 and so when will it stop theN? 2010-12-01T11:24:58 if he had his way you wouldn't get to see anything and could just sit and stew :) 2010-12-01T11:25:04 menekey_: it took some persuasion just to have the contest be live for this long :P 2010-12-01T11:25:09 but I didn't want to listen to all the screaming 2010-12-01T11:25:11 its 3 more elo between me and mistmanov if he wins 2010-12-01T11:25:14 i'll kill him 2010-12-01T11:25:15 xD 2010-12-01T11:25:32 hahaha 2010-12-01T11:25:53 my bot seems pretty happy at 35 2010-12-01T11:26:11 damn you manwe 2010-12-01T11:26:13 My bot seems to be comfortable at 18 ^^ 2010-12-01T11:26:13 my bot is more happy @ 1 2010-12-01T11:26:20 but no 2010-12-01T11:26:23 its at 75 :P 2010-12-01T11:26:26 and you medimonia! 2010-12-01T11:26:36 ^^ 2010-12-01T11:26:40 hehe, another win against bix0r 2010-12-01T11:26:47 i've had quite a few, i thought he'd be better 2010-12-01T11:27:14 win against protocolocon and jambachili for the two last, but I've lost a lot since the ranking aren't updated 2010-12-01T11:27:32 antimatroid: sure, but i'm lossing a lot of game now, and i'm really scared 2010-12-01T11:27:34 janzert: around what time (either EST or server time) would the final results likely be released? 2010-12-01T11:27:58 6pm server time 2010-12-01T11:28:15 @time 2010-12-01T11:28:15 antimatroid: 11:28 AM, December 01, 2010 2010-12-01T11:28:18 janzert: ok, thanks 2010-12-01T11:28:20 that's an hour fast^^ 2010-12-01T11:28:29 so that's in like 7.5 hours? 2010-12-01T11:28:37 :P 2010-12-01T11:28:40 it says pm 2010-12-01T11:28:43 and that's am 2010-12-01T11:28:47 so 19.5? 2010-12-01T11:29:22 23h for us Error323^^ 2010-12-01T11:29:28 7.5, i think 2010-12-01T11:29:31 the sooner the better for me, it's not going up :P 2010-12-01T11:29:35 about 7.5 hours from now 2010-12-01T11:29:40 oh, i am retarded :P 2010-12-01T11:29:41 :) 2010-12-01T11:29:49 :) 2010-12-01T11:30:03 i don't do arithmetic well :P 2010-12-01T11:30:27 hmm no about 01:00 tomorrow in fact 2010-12-01T11:30:27 i think i may be able to claim first place for smiley use too 2010-12-01T11:30:49 yes 2010-12-01T11:30:58 medrimonia: we are 7 hours ahead of servertime 2010-12-01T11:31:01 yep 2010-12-01T11:31:06 not 5^^ 2010-12-01T11:31:11 * Error323 hates timezones :P 2010-12-01T11:31:32 me too, why can't everyone use the same one and people just had different times for day/night :\ 2010-12-01T11:31:53 *** Meatkat has joined #aichallenge 2010-12-01T11:31:54 there is such a system, where a day is 1000 clicks 2010-12-01T11:32:03 but i forgot the naem 2010-12-01T11:32:04 Meatkat: you're as bad as the others 2010-12-01T11:32:05 name 2010-12-01T11:32:09 turning up in irc right at the end 2010-12-01T11:32:13 yeah wth 2010-12-01T11:32:22 contest over --> ppl getting in here 2010-12-01T11:32:23 =) 2010-12-01T11:32:25 Meow? 2010-12-01T11:32:33 7.5 hours? ;-) STOB = Stop Torturing Our Bots! ;) 2010-12-01T11:32:33 iouri: you get some of that hate ^^ too 2010-12-01T11:32:37 ROAAR 2010-12-01T11:32:45 Meatkat: hi! ;-) 2010-12-01T11:32:47 Sorry, used to hang out here a lot, but then had to study for the GREs.... 2010-12-01T11:32:58 oh god, i am going to do them next year 2010-12-01T11:33:08 how was the maths one? (assuming you did that one) 2010-12-01T11:33:18 people on reddit said it was brutalish 2010-12-01T11:33:26 Now I'm hoping I can cling in the top100, now at 81 or something like that 2010-12-01T11:33:32 GRE's? 2010-12-01T11:33:38 Maths dominates your face if you don't study 2010-12-01T11:33:39 what is that 2010-12-01T11:33:45 general exams before grad school 2010-12-01T11:33:45 I'm taking that one in April 2010-12-01T11:33:49 aha 2010-12-01T11:33:59 yeah i plan to do a fair bit of study 2010-12-01T11:34:05 apparently it has a lot more analysis than algebra :( 2010-12-01T11:34:06 this was just the general one though, I wanted to up my score so I memorized like 3000 vocab words before the test 2010-12-01T11:34:13 lol 2010-12-01T11:34:42 we have a very different system here 2010-12-01T11:34:54 I has EVERYTHING you did/could/should study in undergrad including topology, real analysis, linear algebra, complex analysis, and more 2010-12-01T11:34:59 Error323: we don't have them for australia, but i have dreams of mit :P 2010-12-01T11:35:14 Meatkat: i have/will have as well 2010-12-01T11:35:33 *** bhasker has joined #aichallenge 2010-12-01T11:35:38 bhasker: morning 2010-12-01T11:35:53 finally ducked into top 50 :) 2010-12-01T11:36:00 I took it once at the end of my undergrad and thought they'd be a joke like the SAT/ACT/General GRE but they pwn3d my face... and my face doesn't usually get pwn3d by standardized tests 2010-12-01T11:36:41 lol 2010-12-01T11:36:49 yeah, i have got the message to take it seriously 2010-12-01T11:37:12 antimatroid: I'd recommend picking up the Princeton Review for studying, when are you planning on taking it? 2010-12-01T11:37:30 next year, planning to grab that 2010-12-01T11:37:57 i just finished undergrad in cs, eco and maths, doing honours in maths next year, masters in maths after that, then i'm not even sure what to do a phd in if i do it 2010-12-01T11:38:02 something game theory related :P 2010-12-01T11:38:30 so you can pwn those ai contests ;) 2010-12-01T11:38:44 nah, i'm more interested in the pure side if i do that 2010-12-01T11:38:47 Cool, did my undergrad in math and cs, in the Peace Corps now, off to do a masters in Econ next year then a PhD in AI 2010-12-01T11:39:07 what kind of eco? 2010-12-01T11:39:21 micro theory is pretty much all i enjoy 2010-12-01T11:39:33 although i work as an ra doing programming stuff for a macroeconomist 2010-12-01T11:39:55 *** bhasker has quit IRC (Ping timeout: 255 seconds) 2010-12-01T11:40:01 Just general, it's a one year MSc program that I'm taking to strengthen my application for PhD 2010-12-01T11:40:20 undergrade=bachelor right? 2010-12-01T11:40:24 It's in Hong Kong, HKUST 2010-12-01T11:40:27 yeah 2010-12-01T11:40:46 thanks Meatkat =) 2010-12-01T11:40:54 take the post grad theory courses for micro and macro if they'll let you :) 2010-12-01T11:40:57 those will be the most fun 2010-12-01T11:41:31 i've done a post grad micro course, and i sat through a measure theory one this semester, although i sort of stopped paying much attention towards the end because of ai, so i need to go through and learn the material properly at some point 2010-12-01T11:41:32 Cool, I'll keep that in mind 2010-12-01T11:41:46 Haha 2010-12-01T11:42:30 i hate econometrics, cause i hate stats :P 2010-12-01T11:44:32 rebelxt: we just played... what are you ranked at? and hi! 2010-12-01T11:44:58 Meatkat: you can search users names on the right hand side 2010-12-01T11:45:47 *** bhasker has joined #aichallenge 2010-12-01T11:45:48 *** Aksenov239 has joined #aichallenge 2010-12-01T11:48:13 Meatkat: 58. we played a lot of games in the past i recall ;-) 2010-12-01T11:48:30 *** Aksenov239 has quit IRC (Client Quit) 2010-12-01T11:48:43 Top 10 players: bocsimacko(3802), _iouri_(3585), Slin-.-(3538), jimrogerz(3529), _Astek_(3521), Accoun(3513), george(3508), GreenTea(3508), asavis(3505), protocolocon(3496) 2010-12-01T11:49:45 *** davidjliu has joined #aichallenge 2010-12-01T11:52:05 oh, some asshole already took antimatroid on github 2010-12-01T11:52:10 grumble 2010-12-01T11:52:59 antimatroid: grr, we just played again 2010-12-01T11:53:05 rebelxt: Yeah, I seem to remember being ahead of you last time we met... apparently not anymore! Take six weeks off and look what happens! 2010-12-01T11:53:26 Meatkat: yeah you had been my target for a while! ;-)) 2010-12-01T11:54:29 rebelxt: Well good work! These last few hours are so nerve wracking! 2010-12-01T11:55:00 dmj111 and I are still locked in mortal combat for pythonic triumph 2010-12-01T11:55:04 yeah, it's a torture... hope we all meet again during the next contest ;-) 2010-12-01T11:55:12 davidjliu: i think i beat you both today :P 2010-12-01T11:55:20 you should go beat him some more :) 2010-12-01T11:55:21 Anyone have any guesses as to what the next game is going to be? 2010-12-01T11:55:26 rebelxt: i was here for tron and i'll be here again :) 2010-12-01T11:55:34 Meatkat: not decided yet :P 2010-12-01T11:55:37 davidjliu: heh yeah it's a real fight between you guys 2010-12-01T11:55:40 does anyone know why they aren't planning on lowering the cutoff more? 2010-12-01T11:56:12 cause 11/12/13 have slowly been inching their ways toward the top 10 2010-12-01T11:56:22 key word slowly 2010-12-01T11:56:37 They aren't planning on lowering at anymore? Sad. 2010-12-01T11:56:56 ratings are not updating 2010-12-01T11:57:08 GreenTea: no they froze it 2010-12-01T11:57:16 Why? 2010-12-01T11:57:23 Its gonna be so exciting, can't wait to see who'll be #1 2010-12-01T11:57:24 The End? ) 2010-12-01T11:57:24 :P 2010-12-01T11:57:41 why did ratings get frozen for next 7 hours? 2010-12-01T11:57:49 wait, they froze it for super-suspense? heh 2010-12-01T11:57:56 davidjliu: yes 2010-12-01T11:58:07 that would be my guess 2010-12-01T11:58:08 hoho 2010-12-01T11:58:15 it was originally meant to be total suspense since the end of submissions :P 2010-12-01T11:58:16 when was it frozen? 2010-12-01T11:58:19 imagine how painful that would be 2010-12-01T11:58:27 davidjliu: like an hour ago maybe? 2010-12-01T11:58:40 hopefully after dmj's winning streak... heh 2010-12-01T11:59:19 so what's the reason partial suspense is needed? just curious since this can't affect me at all 2010-12-01T12:01:37 *** fgump has joined #aichallenge 2010-12-01T12:01:40 I still don't get why the cutoff isn't going to be lowered though 2010-12-01T12:02:15 well.. some of the players have high risk strategies that work well against good players but bad against worse players 2010-12-01T12:02:27 so imo its good the cutoff is as it is 2010-12-01T12:02:58 yeah, i'm fine with the cutoff 2010-12-01T12:03:10 I'd really like to see Xrillo's source. 2010-12-01T12:03:15 rabelxt: maybe for more fast playing of games. When ratings are recalculated all games should be stoped isn't it? 2010-12-01T12:03:30 mega1: same here 2010-12-01T12:03:36 I mean, I think taking the cutoff up to 250 wouldn't effect any of the top bots seeing as they don't play anyone outside of the top 250... and more games is always useful 2010-12-01T12:04:03 davidjliu: it does affect it 2010-12-01T12:04:07 #40 is quite high with uct 2010-12-01T12:04:28 isn't uct meant to be good? 2010-12-01T12:04:32 GreenTea: not sure if they ever stopped games to re-calculate ratings. bayeselo is fast 2010-12-01T12:04:35 everyone expects it to do badly? 2010-12-01T12:07:19 rebelxt: janzert said it takes about 8 mins to calculate 2010-12-01T12:08:57 Error323: ahh got it 2010-12-01T12:10:32 antimatroid: I gave upon it before starting to code. 2010-12-01T12:11:15 mega1: so how would one map UCT to this problem 2010-12-01T12:11:16 then maybe my bot manages to sneak back into top50 like it did every day before today ;-) 2010-12-01T12:11:51 Error323: I couldn't see a promising way that's why I'm interested in Xrillo's approach. 2010-12-01T12:12:54 right 2010-12-01T12:12:55 *** wh1teside has quit IRC (Read error: Connection reset by peer) 2010-12-01T12:13:18 *** wh1teside has joined #aichallenge 2010-12-01T12:13:22 *** aaa_ has joined #aichallenge 2010-12-01T12:14:24 i did monte carlo with semi-light playouts (using either rage bot or bully bot algorithm, for instance). but even giving it a large number of playouts (i.e. one minute per turn), it would not work well at all 2010-12-01T12:14:52 i.e. it would not challenge my best heuristic bot 2010-12-01T12:18:04 that's the hard part 2010-12-01T12:18:41 I had ideas of an randomish evaluation function that should converge on a likely playout but it didn't materialize 2010-12-01T12:21:14 http://www.google.nl/url?sa=t&source=web&cd=1&ved=0CBkQFjAA&url=http%3A%2F%2Fciteseerx.ist.psu.edu%2Fviewdoc%2Fdownload%3Fdoi%3D10.1.1.102.1296%26rep%3Drep1%26type%3Dpdf&rct=j&q=Bandit%20based%20Monte-Carlo%20Planning&ei=94P2TJGVN8SVOvm8xJkI&usg=AFQjCNHSi9UuIyNSgXMuHoK-HkNPACea9Q 2010-12-01T12:21:23 Xrillio put up his code i think 2010-12-01T12:21:24 best UCT paper i found so far 2010-12-01T12:21:26 on the post 2010-12-01T12:21:27 didnt he? 2010-12-01T12:21:44 no only that he did UCT 2010-12-01T12:23:19 bhasker: he did not it appears 2010-12-01T12:24:56 Has anyone said that they've been using anything other than a mess/majority of heuristics? 2010-12-01T12:25:34 whilest reading some more about minimax during these months and about naish equilibrium, isn't minimax a special method of computing the naish equilibrium? 2010-12-01T12:26:20 i tried alphabeta 2010-12-01T12:26:39 but if I only could read those blogs beforehand it would've worked :P 2010-12-01T12:26:47 I'd say that mini-max isn't assured to find a max equilibrium, but it is possible that it finds it. Depends on the complexity/rules of the game 2010-12-01T12:26:48 couldn't come up with a good eval functino 2010-12-01T12:27:17 Meatkat: why isn't it assured? 2010-12-01T12:27:27 assuming it can go all the way down of course 2010-12-01T12:28:50 Error323: that paper is basically where UCT was born 2010-12-01T12:29:08 (two hungarian authors) 2010-12-01T12:29:27 yeah I read that 2010-12-01T12:29:29 cool :) 2010-12-01T12:29:40 thats why you grok it? =) 2010-12-01T12:30:56 it seems like a combination between reinforcement learning and minimax 2010-12-01T12:31:20 Error323: can't think of a counter-example off the top of my head... just doesn't seem like it would be true, Nash equilibria aren't very easy to find, especially when things get increasingly complex 2010-12-01T12:34:37 but *if* you can assume that your mini-max can cover the whole search space... maybe? it seems like there would be a chance of accidentally pruning something. Also not many non-trivial problems are left that mini-max could step through in a reasonable amount of time 2010-12-01T12:34:55 minimax doesn't prune 2010-12-01T12:35:02 *** rou has joined #aichallenge 2010-12-01T12:35:35 and if alphabeta can explore all the way to the leafnodes its equal to minimax 2010-12-01T12:36:08 You look at your next possible moves, choose the state with the max utitily, look at your opponents possible moves (given that you've made that move), choose the one that minimizes your utility function 2010-12-01T12:36:20 correct? 2010-12-01T12:36:30 and so on.... 2010-12-01T12:36:38 you chose on the recursive unwind 2010-12-01T12:36:42 not on the way down 2010-12-01T12:38:37 the way I implemented it, it followed each top level node to the end and then presented that final game state 2010-12-01T12:38:52 chose the best out of that bunch 2010-12-01T12:39:52 it was a nasty recursive method, it took a lot of time to execute 2010-12-01T12:40:26 well i got my implementation in the "alphabeta" branch on github, its pretty fast 2010-12-01T12:41:04 *** rou has quit IRC (Quit: Page closed) 2010-12-01T12:43:10 *** mceier has joined #aichallenge 2010-12-01T12:46:29 *** rhinobot has quit IRC (Quit: Page closed) 2010-12-01T12:48:43 Top 10 players: bocsimacko(3802), _iouri_(3585), Slin-.-(3538), jimrogerz(3529), _Astek_(3521), Accoun(3513), george(3508), GreenTea(3508), asavis(3505), protocolocon(3496) 2010-12-01T12:48:47 Meatkat: it's possible to exploit minimax 2010-12-01T12:48:50 Error323: You're right, in zero-sum games the minimax solution is the same as the Nash equilibrium. 2010-12-01T12:48:52 looking at matching pennies 2010-12-01T12:49:24 cool :) i'm not crazy then =) 2010-12-01T12:49:28 i tried to do a simultaneous game where i'd pick out pure strategy nash equilibria or try to approximate a mixed strategy nash equilibria if no such pure one existed, but wasn't going to happen 2010-12-01T12:50:02 unless you correctly randomise over choices in your tree 2010-12-01T12:50:33 but my game tree is the best i came to anything non pure heuristics 2010-12-01T12:51:06 i could release my old code if anyone was interested? that has the pure strategy stuff but there's LOTS of files 2010-12-01T12:51:37 simultaneous game* stuff 2010-12-01T12:51:46 How were you abstracting to the point where a tree was possible? 2010-12-01T12:52:51 in my tree, a single move was a vector of fleets all with the same destination, but multiple sources and send times 2010-12-01T12:53:07 in my zero sum game, a pure strategy was a vector of such jointly feasible moves 2010-12-01T12:53:31 i was using decision trees to generate pure strategies for the zero sum game 2010-12-01T12:53:36 so you attacked/defended one planet per turn 2010-12-01T12:53:50 Error323: no, i would pick my strings of moves out 2010-12-01T12:53:58 but there was very few moves per turn yes 2010-12-01T12:54:12 *** wh1teside_ has joined #aichallenge 2010-12-01T12:54:16 the zero sum game was meant to try and fix that by sampling the best nodes from the decision trees 2010-12-01T12:54:25 then making the payoff matrix from that, but it just wasn't working 2010-12-01T12:55:29 at the start of each turn i would run a round of defensive moves over planets and at the end redistribute, but other than that my bot makes like 1-2 rounds of moves to a planet each turn 2010-12-01T12:55:42 * Error323 kinda curious what would happen if he were to implement a proper eval function and better actions using alphabeta 2010-12-01T12:55:44 *** Frontier has joined #aichallenge 2010-12-01T12:55:59 my errored attempt at the end was meant to try and get the tree a bit deeper, but i just broke it badly 2010-12-01T12:56:00 mega1: you tried multiple plies in later games right? 2010-12-01T12:56:19 4 plies in the opening phase 2010-12-01T12:56:42 mega1: is that sort of how you did your zero sum game? 2010-12-01T12:57:09 antimatroid: sorry, I don't understand the question. 2010-12-01T12:57:26 you tried implementing a simultaneous game in a payoff matrix at some point yeah? 2010-12-01T12:57:30 yes 2010-12-01T12:57:33 then finding nash equilibrium 2010-12-01T12:57:40 *** wh1teside has quit IRC (Ping timeout: 255 seconds) 2010-12-01T12:57:43 yes, didn't improve things markedly 2010-12-01T12:57:49 but was too slow to bear 2010-12-01T12:57:52 i described sort of how i went about it above, i'm curious if it's the same way you went about it :P 2010-12-01T12:58:22 * Error323 is having a "I should have" syndrom :/ 2010-12-01T12:58:56 Error323: i don't think any of my approaches were right for this 2010-12-01T12:59:02 the thing is I discussed alphabeta/minimax with some friends/collegues and they talked me out of it >_> 2010-12-01T12:59:05 none of them covered enough moves 2010-12-01T12:59:18 ofc it was my decision 2010-12-01T12:59:24 but ye... bummer 2010-12-01T12:59:30 i didn't use alpha beta 2010-12-01T12:59:42 *** fgump has quit IRC (Quit: Ex-Chat) 2010-12-01T12:59:44 as i've said many times, i don't like it when extending my tree, and evaluating it was almost free 2010-12-01T12:59:56 antimatroid: a move in my nash setup was a real move: a set of orders without restriction on target planets 2010-12-01T13:00:17 what do you mean restriction on target planet? 2010-12-01T13:00:32 you "sent" fleets without sending them anywhere? 2010-12-01T13:00:40 like with the chance of losing it? 2010-12-01T13:00:52 antimatroid: you said that a move in your approach was to a single planet 2010-12-01T13:01:02 ah yeah okay 2010-12-01T13:01:03 you mean you computed every possible move @ start mega? 2010-12-01T13:01:09 such moves corresponded to rows and column of the payoff matrix, right? 2010-12-01T13:01:15 i had a vector of those moves for a pure strategy in my zero sum game 2010-12-01T13:01:20 yep 2010-12-01T13:01:44 okay, I took the N most promising moves for each player and those were my rows/columns 2010-12-01T13:01:58 i see 2010-12-01T13:02:00 basically i made 2 decision tree's for each player, so they could string moves together, then picked out the moves to get to each terminal node (or best x terminal nodes) and those formed my pure strategies for each player 2010-12-01T13:02:02 where most promising is a 1 ply search with my normal eval function 2010-12-01T13:02:38 i think that's the same? 2010-12-01T13:02:41 pretty much? 2010-12-01T13:02:58 is the ranking page not being updated anymore? 2010-12-01T13:03:03 is your 1 ply just a decision tree? 2010-12-01T13:03:08 bhasker: nope 2010-12-01T13:03:10 wait till results from here 2010-12-01T13:03:15 oh 2010-12-01T13:03:16 hmm 2010-12-01T13:03:28 no wonder i have been losing all morning and it shows the sme rank 2010-12-01T13:03:40 march of the red army http://ai-contest.com/profile_games.php?user_id=5822 2010-12-01T13:03:56 wow ouch 2010-12-01T13:04:06 antimatroid: I'm not sure I follow. 2010-12-01T13:04:50 mega1: Did you do any special treatment of the first turn to find the best order in which to take neutral planets to maximise ships at a specific future turn? 2010-12-01T13:05:12 Frontier: not really 2010-12-01T13:05:17 what do you do with a 1 ply search? my decision tree goes like layer one {no moves, move to best planet 1, ...}, then i make a tree with just my moves like that 2010-12-01T13:05:36 Frontier: the opening phase is special in the sense that it's using 4 ply alpha beta 2010-12-01T13:05:44 that includes the first turn. 2010-12-01T13:05:55 mine did that too 2010-12-01T13:06:02 it had quite a dramatic improvement 2010-12-01T13:06:03 yeah, selection of planets on the first turn is much more important than maximising return 2010-12-01T13:06:50 antimatroid: and what do you get out of that tree? 2010-12-01T13:07:52 moves that i plan to make, or for the zero sum game, my pure strategies 2010-12-01T13:08:01 zero sum == simultaneous** 2010-12-01T13:09:03 *** aaa_ has quit IRC (Quit: Page closed) 2010-12-01T13:09:05 does that tree contain orders for the future? 2010-12-01T13:09:39 i don't increment any states at any point, my fleets can contain moves sent from a planet that i don't even own yet 2010-12-01T13:09:55 so yeah 2010-12-01T13:09:57 ok, I understand 2010-12-01T13:10:25 the biggest problem with doing the game tree like that was it'd pick out moves for them, respond to them and then send those orders 2010-12-01T13:10:28 that's putting a lot of faith in ur tree =) 2010-12-01T13:10:32 I have a function called generate-moves-from-steps that does kind of what your decision tree does. 2010-12-01T13:10:37 but i managed to eliminate that by sending fleets only when they were needed 2010-12-01T13:10:53 Error323: my whole bot is my tree :P 2010-12-01T13:11:10 managed to mostly* eliminate that 2010-12-01T13:11:11 it orders steps by scores and tries to form composite moves 2010-12-01T13:11:30 yep, that was the idea 2010-12-01T13:11:34 mega1: by composite you mean synchronized? 2010-12-01T13:11:45 no 2010-12-01T13:12:00 it means take planet a, then consider the reward from that when taking b 2010-12-01T13:12:12 at least that's my understanding 2010-12-01T13:12:29 Error323: no, each step is a synchronized attack to a single planet 2010-12-01T13:12:42 a move is just a set of steps. 2010-12-01T13:12:43 ah ok 2010-12-01T13:13:17 so I order steps by evaluating them with the usual eval function 2010-12-01T13:13:30 What time do we get results? 2010-12-01T13:13:34 @time 2010-12-01T13:13:34 Meatkat: 01:13 PM, December 01, 2010 2010-12-01T13:13:47 in like 6 hours and 45 minutes? 2010-12-01T13:13:55 why are the bots still playing if the leaderboard is frozen? 2010-12-01T13:14:11 mega1: the leaderboard is still running, just no longer public 2010-12-01T13:14:16 oh 2010-12-01T13:14:17 it's meant to give suspense 2010-12-01T13:14:36 *** BaronTrozo has joined #aichallenge 2010-12-01T13:14:36 Man that's 1am here... lame 2010-12-01T13:14:47 it's 5:15am here atm :P 2010-12-01T13:14:47 same here ye 2010-12-01T13:15:12 All nighter? or early riser? 2010-12-01T13:15:20 all nighter 2010-12-01T13:15:24 it's sort of the norm of late 2010-12-01T13:15:28 i really need to kick the habit 2010-12-01T13:15:40 suspense, huh? ok, time for some coffee 2010-12-01T13:16:04 Meatkat: i have never heard of your country :$ 2010-12-01T13:16:21 neither 2010-12-01T13:16:41 oh in Africa 2010-12-01T13:17:10 *** GreenTea_ has joined #aichallenge 2010-12-01T13:18:14 what country? 2010-12-01T13:18:22 Burkina Faso 2010-12-01T13:19:07 i've been stuck in australia most of my life 2010-12-01T13:19:14 i went to new zealand when i was like 4 2010-12-01T13:19:24 *** Meatkat has quit IRC (Ping timeout: 245 seconds) 2010-12-01T13:19:25 Do you know Surinam? 2010-12-01T13:19:34 I have hear from it but the only thing i know is the name of the capital 2010-12-01T13:19:51 nope 2010-12-01T13:20:01 But in your profile put netherlands 2010-12-01T13:20:54 Well I've heard of Burkina Faso, but I don't know much about it. If anyone asked me, I would think of deserts and dictators maybe :p 2010-12-01T13:21:28 my bots doing rather badly atm 2010-12-01T13:24:57 uufff, 2010-12-01T13:25:20 I am netherlands BaronTrozo , were discussing someone else :) 2010-12-01T13:25:46 I am dutch I mean... -.- 2010-12-01T13:25:57 anyway, i might knock off so i can wake up in time for the end 2010-12-01T13:26:21 when does the contest end 2010-12-01T13:26:48 in 5 hours and 40 minutes 2010-12-01T13:26:51 antimatroid: http://ai-contest.com/visualizer.php?game_id=9521580 2010-12-01T13:27:16 I would like to discuss the french influence in west-africa, blood diamonds and hopes of the next generation, but Meatkat is gone :( 2010-12-01T13:28:02 I went to Burkina Faso Meatkat 2010-12-01T13:28:31 My parents have adopted (I don't know it's right) my young sister there 2010-12-01T13:28:45 I was in amsterdam four years ago, on January, very cold 2010-12-01T13:28:45 *don't know if 2010-12-01T13:29:15 BaronTrozo: saw that, quite the fluke i pulled off :) 2010-12-01T13:29:22 BaronTrozo: oh yeah, =). If you remember that you didn't do it right ;) 2010-12-01T13:30:24 Did the games just stopped? 2010-12-01T13:30:32 http://ai-contest.com/game_stats.php 2010-12-01T13:30:37 *** ulope has joined #aichallenge 2010-12-01T13:30:46 ohw interesting 2010-12-01T13:31:13 i went to amsterdam, but with my wife pregnant so i didn't do any special except visiting the city 2010-12-01T13:31:40 nice Van Gogh Museum 2010-12-01T13:31:44 BaronTrozo: I understand :) it's a cliche really that ppl goto amsterdam for drugs etc -.- quite annoying 2010-12-01T13:32:13 i find it funny how many people go there and smoke weed, yet consider it a terrible thing done in their own country 2010-12-01T13:32:24 I visit a coffe shop but i only have a ber 2010-12-01T13:32:32 ber-> beer 2010-12-01T13:32:43 BaronTrozo: I bike by it every day, never visited it, I should probably 2010-12-01T13:32:48 they sell mushrooms there too yeah? 2010-12-01T13:32:52 i'm surprised at that 2010-12-01T13:32:59 not anymore 2010-12-01T13:33:08 too much fatalities of tourists -.- 2010-12-01T13:33:11 ahh okay, those things are way harder 2010-12-01T13:33:29 they grow here, one can just go pick them during the winter when it's rained enough 2010-12-01T13:33:30 apparently cocaine + mushrooms + beer + weed is fatal... who knew 2010-12-01T13:33:35 antimatroid: The funny thing in Europa is that the country where smoking weed is the less repressed are the country where there's the less weed "addicts" 2010-12-01T13:33:35 hehe 2010-12-01T13:33:41 o that was amazing, every body in bike, while it was raining, very impresive for a spanish guy 2010-12-01T13:33:47 hahaha 2010-12-01T13:34:34 BaronTrozo: speaking of drugs I thought Netherlands was #1 at that but I visited the Boom festival in Portugal this year 2010-12-01T13:34:56 and omg... NL is a walk in the park compared to the stuff I've seen there... 2010-12-01T13:35:23 people were taking lsd like sigarettes 2010-12-01T13:35:35 the country with most coaine per person i think is Spain, at leat on european rankins 2010-12-01T13:35:46 wowa 2010-12-01T13:35:50 didn't know that 2010-12-01T13:36:07 I never visited spain :/ 2010-12-01T13:36:10 BaronTrozo: Is cocaine consumption tolerated in Spain? 2010-12-01T13:36:20 I really doubt that :P 2010-12-01T13:36:24 no is illegal 2010-12-01T13:36:27 *** MikeTV has joined #aichallenge 2010-12-01T13:36:36 Error323: In Switzerland it's tolerated 2010-12-01T13:36:45 medrimonia: really?? 2010-12-01T13:36:54 You can't sell cocaine but there's even place of drugs consumption 2010-12-01T13:37:06 here no drugs except tobaco an alcohol is tolerated 2010-12-01T13:37:15 It's forbidden to buy or sell drugs 2010-12-01T13:37:24 is chis contest a drug????? ;) 2010-12-01T13:37:30 chis -> this 2010-12-01T13:37:35 But there's very few drugs really prohibited 2010-12-01T13:37:40 lsd is better than a lot of other drugs 2010-12-01T13:37:50 you can't easily overdose etc. 2010-12-01T13:37:56 i don't know, i dont use them 2010-12-01T13:38:06 perhaps one beer 2010-12-01T13:38:15 antimatroid: its not, lsd can be very dangerous also 2010-12-01T13:38:31 it can activate a psychosis if you are sensitive to that 2010-12-01T13:38:37 aswell as mushrooms 2010-12-01T13:38:41 *** Ken69267 has left #aichallenge 2010-12-01T13:38:43 Error323: most drugs have issues like that 2010-12-01T13:38:44 so do peanuts 2010-12-01T13:38:54 i'm not saying they're healthy 2010-12-01T13:39:07 but i'd rather have someone try that than something like meth 2010-12-01T13:39:33 Error323: It has been medicaly proved that there's no addiction to lsd 2010-12-01T13:39:41 medrimonia: yes that's true 2010-12-01T13:39:52 but it can be really dangerous 2010-12-01T13:40:01 I never dared trying it 2010-12-01T13:40:22 well, this was a competition, drugs are not 2010-12-01T13:40:33 yeah, topic has gone way off topic 2010-12-01T13:40:38 convo* 2010-12-01T13:40:45 ye srry :) 2010-12-01T13:40:54 that always happens when people talk about the netherlands 2010-12-01T13:41:16 bah, 5 45am, night :) 2010-12-01T13:43:08 ye well the irony is i kinda started it 2010-12-01T13:43:15 *** ulope has left #aichallenge 2010-12-01T13:43:21 why did they stop updating the ranking page though:( 2010-12-01T13:43:26 it was a good distraction! 2010-12-01T13:43:38 so BaronTrozo where should one go if one were to visit Spain? 2010-12-01T13:45:05 *** Mistmanov has joined #aichallenge 2010-12-01T13:45:23 *** Accoun has quit IRC () 2010-12-01T13:47:21 Mistmanov: the tension is killing ey? =) 2010-12-01T13:47:40 eh 2010-12-01T13:47:41 :p 2010-12-01T13:47:45 :P 2010-12-01T13:48:22 it's a bit dissapointing that we won't be getting random games within top100 2010-12-01T13:48:24 but oh well 2010-12-01T13:48:29 can't have everything =) 2010-12-01T13:48:43 i've had a pretty good go at most bots in the top 100 i think 2010-12-01T13:48:44 Top 10 players: bocsimacko(3802), _iouri_(3585), Slin-.-(3538), jimrogerz(3529), _Astek_(3521), Accoun(3513), george(3508), GreenTea(3508), asavis(3505), protocolocon(3496) 2010-12-01T13:48:45 not disappointing to me :P 2010-12-01T13:48:53 i've certainly had quite a few games with top 10 bots 2010-12-01T13:49:02 and won a surprising number of them 2010-12-01T13:52:32 i just wish they would update ranking 2010-12-01T13:52:36 this mystery is killing me now 2010-12-01T13:52:44 it was reassuring to see it hold 50 2010-12-01T13:52:45 :( 2010-12-01T13:52:52 5 more hours :P 2010-12-01T13:53:02 I wish they would start a completely awesome competition in january or something 2010-12-01T13:53:09 januari? 2010-12-01T13:53:10 +1 2010-12-01T13:53:17 January 2010-12-01T13:53:25 I think :P 2010-12-01T13:54:00 tbh, most of the ideas discussed didn't seem really interesting to me 2010-12-01T13:54:14 (I might not have read all the discussion, but eh..) 2010-12-01T13:54:19 so what do you find interesting? 2010-12-01T13:54:21 racing, rts, weird chess lke games.. 2010-12-01T13:54:28 those are all bad :p 2010-12-01T13:54:31 lol 2010-12-01T13:54:35 rts ain't bad :P 2010-12-01T13:54:59 maybe it's because I've played rts too seriously back in the day 2010-12-01T13:55:14 it kinda burned me out, all the stress of managing all the stuff at the same time.. 2010-12-01T13:55:16 so what would you like? 2010-12-01T13:55:21 and sure, a computer would be doing it now, but still :p 2010-12-01T13:55:27 yeah I dunno ^^ 2010-12-01T13:55:33 i like simple 2 player simultaneous games 2010-12-01T13:55:41 but minimax gets kinda old 2010-12-01T13:55:42 *** smloh1 has joined #aichallenge 2010-12-01T13:55:43 rts is fun 2010-12-01T13:55:52 but it would be complicated for most people 2010-12-01T13:55:54 bhasker: yeah you were still holding on to top50! we kinda swapped places yesterday ;-) 2010-12-01T13:55:55 i think the reason this was fun 2010-12-01T13:55:59 are the rankings frozen for suspense? 2010-12-01T13:56:04 smloh1: indeed 2010-12-01T13:56:11 was it was simple to get a bot up and running 2010-12-01T13:56:20 an RTS bot will require a lot of boiler plate code 2010-12-01T13:56:29 true 2010-12-01T13:56:30 so the starter packages will have to be a lot more comprehensive 2010-12-01T13:56:35 ah... 2010-12-01T13:56:50 i think the game should be a simple one to get into 2010-12-01T13:57:05 otherwise we won't see many takers 2010-12-01T13:57:07 nah, imo we should evolve =) 2010-12-01T13:57:19 you can make a simple interface for a complex game 2010-12-01T13:57:33 yea so the interface will be critical to get right 2010-12-01T13:58:14 I personally liked that you could just approach this game like "ok.. so what would be a good move to make?" 2010-12-01T13:58:25 like... you could fully understand what a bot is doing 2010-12-01T13:58:38 many of the more advanced AI tactics don't have that property I think 2010-12-01T13:58:47 or at least, your average Joe won't be able to think of it 2010-12-01T13:58:47 *** Accoun has joined #aichallenge 2010-12-01T13:59:00 yeah but you also don't want the same "class" of problems each time 2010-12-01T13:59:08 *** krypton has joined #aichallenge 2010-12-01T13:59:24 true. But I really liked the "ok, so I want my bot to only attack when I know for sure that it can take it", and then just doing simple calculations to get the numbers 2010-12-01T13:59:25 *** krypton is now known as _krypt0n_ 2010-12-01T13:59:27 but I do agree this game was very well chosen 2010-12-01T14:00:31 as was tron 2010-12-01T14:00:40 too bad i didn't know of it back then =) 2010-12-01T14:01:32 but an rts can be made easy or complex very simple 2010-12-01T14:01:43 ehhhh very gradually i mean 2010-12-01T14:03:14 janzert: you could reactivate the countdown to countdown to when the final results will be released 2010-12-01T14:03:34 *** Meatkat has joined #aichallenge 2010-12-01T14:05:11 well, the release of results depends a bit on how quickly Jeff makes it home from work and can get everything together to make the release so may not be quite as firm and I'd rather not countdown to "dead air" :/ 2010-12-01T14:06:23 I am pretty hopeful that it will happen pretty close to the scheduled time though 2010-12-01T14:06:43 dammit janzert this mystery is killing me :p we all know rank 1 anyway 2010-12-01T14:06:54 all these losses are making a nervous wreck out of me:p 2010-12-01T14:06:54 http://ai-contest.com/profile.php?user_id=5822 2010-12-01T14:06:56 :( 2010-12-01T14:07:07 an extremely dumbed down RTS could work. Maybe with only one resource, one unit type and one building type 2010-12-01T14:07:29 iouri: exactly 2010-12-01T14:07:56 *** BaronTrozo has quit IRC (Quit: Page closed) 2010-12-01T14:08:10 would be hard to do it on a map that would fit into a browser screen though 2010-12-01T14:08:32 janzert: that's alright if the countdown ends and results aren't released... it'll just result in thousands of voices crying out in agony, and the server going down from millions of refreshes 2010-12-01T14:08:56 smloh1: yeah, exactly :P 2010-12-01T14:09:07 where is the countdown? 2010-12-01T14:09:23 iouri: how so? just zoom out enough? 2010-12-01T14:09:49 Error323: good point 2010-12-01T14:10:01 supcom strategic view 4tw 2010-12-01T14:10:10 3d tron ftw 2010-12-01T14:10:17 supcom was awesome, i wish they hadn't dumbed it down in supcom2 2010-12-01T14:10:26 supcom2 was so freaking lame compartively 2010-12-01T14:10:28 bhasker: there isn't one, smloh1 was asking my to turn the one used to countdown to the start of the finals back on 2010-12-01T14:10:40 ah 2010-12-01T14:10:58 bhasker: you should really try TASpring 2010-12-01T14:11:04 i tried it long back 2010-12-01T14:11:08 but supcom > TASpring 2010-12-01T14:11:12 its improved quite a lot 2010-12-01T14:11:15 no! =) 2010-12-01T14:11:16 i have supcom and the supcom expansion 2010-12-01T14:11:20 and supcom2 2010-12-01T14:12:01 does it have a good multiplayer group? 2010-12-01T14:12:08 cause i don't like playing against AI 2010-12-01T14:12:13 yes 2010-12-01T14:12:14 i find it very boring 2010-12-01T14:12:17 to beat RTS AI 2010-12-01T14:12:22 try mine :P 2010-12-01T14:12:24 very few are challenging enough 2010-12-01T14:12:27 its very agressive 2010-12-01T14:12:34 i liked sorian's AI in Supcom 2010-12-01T14:12:34 it does cheat though 2010-12-01T14:12:41 yea cheating AI's are lame 2010-12-01T14:12:52 sorian's AI in supcom would do interesting stuff 2010-12-01T14:12:56 like transport drops and stuff 2010-12-01T14:13:03 even the supcom2 AI is quite interesting 2010-12-01T14:13:06 does really annoying stuff 2010-12-01T14:13:12 like sending engineers behind lines 2010-12-01T14:13:15 to capture resource spots 2010-12-01T14:13:19 mass spots etc 2010-12-01T14:13:21 what would be interesting is a game where you played, but you could write your own code to give you certain information 2010-12-01T14:13:22 very very annoying 2010-12-01T14:13:31 or you could tell it to pick moves from a particular type of tree etc. 2010-12-01T14:13:44 so which spring RTS should i install? 2010-12-01T14:14:02 http://springrts.com/wiki/Download 2010-12-01T14:14:38 i like the idea of a game where players have weapons, but players make their own weapons based on certain guidelines 2010-12-01T14:14:42 haven't played it for quite a while though 2010-12-01T14:14:57 and weapons get sent off like bots, with pre programmed characteristics 2010-12-01T14:15:05 and run their own instances 2010-12-01T14:15:21 evolution RTS looks quite nice 2010-12-01T14:15:38 * Error323 was hardcore XTA player 2010-12-01T14:16:22 balanced annihilation is played most by far though 2010-12-01T14:20:56 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-12-01T14:23:18 balanced annihilation? mod for spring? 2010-12-01T14:23:23 yes 2010-12-01T14:23:34 *** davidjliu has quit IRC (Ping timeout: 265 seconds) 2010-12-01T14:23:56 hmm so like i download the engine 2010-12-01T14:23:58 then download the mods 2010-12-01T14:23:59 ? 2010-12-01T14:24:10 yes 2010-12-01T14:24:11 <-- on a mac will ahve to do it on my desktop 2010-12-01T14:24:21 should work on a mac too 2010-12-01T14:24:28 but you might have to compile 2010-12-01T14:26:22 looks lke a lot of work to get it to run in mac 2010-12-01T14:26:46 *** Palmik has quit IRC (Ping timeout: 255 seconds) 2010-12-01T14:26:47 possibly, I don't own a mac 2010-12-01T14:26:50 i will just take the easy way out and install it on my windows laptop 2010-12-01T14:26:55 oops i mean desktop 2010-12-01T14:27:06 i bought settlers hd for ipad yesterday a good fun rts 2010-12-01T14:27:10 simple but fun 2010-12-01T14:27:15 reminds me of some old RTS games 2010-12-01T14:32:05 *** perror has quit IRC (Quit: Bye all !) 2010-12-01T14:32:29 *** lavalamp has joined #aichallenge 2010-12-01T14:35:22 Settlers of Catan would be freaking sweet, though getting the rights would be difficult 2010-12-01T14:35:37 Oh I love that game yes 2010-12-01T14:36:20 but that's quite a complex game for newcommers 2010-12-01T14:36:23 lots of rules 2010-12-01T14:36:45 i say we take civ 1 2010-12-01T14:36:48 or civ2 2010-12-01T14:36:48 *** _krypt0n_ <_krypt0n_!d030e307@gateway/web/freenode/ip.208.48.227.7> has quit IRC (Quit: was late to the party, congrats to all.) 2010-12-01T14:36:50 ^ 2010-12-01T14:37:04 simplify the city construction though 2010-12-01T14:37:09 not so many structures 2010-12-01T14:37:16 maybe barracks/granary and research building 2010-12-01T14:37:24 per city 2010-12-01T14:37:39 I believe MIT had a rts programming contest let me find it 2010-12-01T14:38:15 *** Mjiig has joined #aichallenge 2010-12-01T14:38:36 http://www.battlecode.org/info/ 2010-12-01T14:42:56 *** Mjiig has quit IRC (Client Quit) 2010-12-01T14:45:30 oh or what about a towerdefense game 2010-12-01T14:45:48 janzert: gonna participate next contest? =) 2010-12-01T14:48:35 I probably won't be around much at all for the next contest, either as a participant or organizer 2010-12-01T14:48:44 Top 10 players: bocsimacko(3802), _iouri_(3585), Slin-.-(3538), jimrogerz(3529), _Astek_(3521), Accoun(3513), george(3508), GreenTea(3508), asavis(3505), protocolocon(3496) 2010-12-01T14:48:46 i see 2010-12-01T14:50:08 A tower defense game would be be pretty cool, especially if you were able to influence the creeps for the enemy tower maze 2010-12-01T14:51:23 it would also be pretty simple 2010-12-01T14:54:07 *** McLeopold has joined #aichallenge 2010-12-01T14:54:25 could* 2010-12-01T14:54:28 =0 2010-12-01T14:58:03 janzert: is there any place we can change the "About me:" in our profile? 2010-12-01T14:59:05 it's something I would like to change but there currently is not, sorry. 2010-12-01T14:59:21 ok, thanks. 2010-12-01T15:14:05 *** px1704 has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2010-12-01T15:15:22 a tower defense game would be fun to do, if it was complicated enough 2010-12-01T15:16:53 If you had a consistantly changing environment it would help, that would stop people from using set maze techniques 2010-12-01T15:17:11 So each map has a slightly different space for you to build your maze 2010-12-01T15:17:38 *** phreeza has quit IRC (Read error: Connection reset by peer) 2010-12-01T15:18:12 Then to make it more head-to-head instead of just a one player thing you could have the option to spend money to increase the power/number of the enemies 'creeps' (creatures going through the maze) 2010-12-01T15:18:41 we should write the game :) 2010-12-01T15:18:46 enemys' 2010-12-01T15:18:50 I' 2010-12-01T15:19:02 I'd be down, though maybe a specification first 2010-12-01T15:19:41 I'd play it human vs human actually, sounds fun 2010-12-01T15:20:14 Yeah, used to play a lot of tower defense custom mods in WC3, there were several like that 2010-12-01T15:20:21 hella fun, but a bit hard to balance 2010-12-01T15:20:39 Speaking of which... how would we balance it enough to be worthy of selection for this competition? 2010-12-01T15:20:55 symmetry for starters 2010-12-01T15:21:09 as far as making every game not a landslide victory, I have no idea 2010-12-01T15:21:11 On the map terran, definately 2010-12-01T15:21:14 definitely* 2010-12-01T15:22:11 We could use some basic heuristics to give a cost comparison to make things about equal 2010-12-01T15:22:38 it's hard for me to predict how balanced things would be without seeing it, sadly 2010-12-01T15:22:46 *** protocolocon has joined #aichallenge 2010-12-01T15:22:53 also adding compounding interest to unspent income adds another element, though we might be getting too complex 2010-12-01T15:23:45 it seems that without some way to target the other player's weak spot(weakly defended planets in this case) 2010-12-01T15:23:54 once a bot was winning it wouldn't lose 2010-12-01T15:24:42 Different creep types - ie flying, immune to slowing effects, reduced dmg vs certain dmg types 2010-12-01T15:25:15 oh, good idea 2010-12-01T15:25:18 Say we use three types of towers, with a slowing tower and a magic tower that does low dmg but always does that damage 2010-12-01T15:25:26 so five types total 2010-12-01T15:25:35 or maybe you can simplify that down 2010-12-01T15:25:44 but in any case that would add an element of weakness 2010-12-01T15:25:59 you have to analyse your enemies maze and see what creep you want to buff for the next wave 2010-12-01T15:26:17 would you allow maze mods during a wave? 2010-12-01T15:26:26 ...every creep you kill gets added to the queue for your opponent 2010-12-01T15:26:31 *** rebelxt has left #aichallenge 2010-12-01T15:26:33 or just have a setup phase between waves? 2010-12-01T15:27:10 building during allows for cycles to be created... ie the creeps get trapped in the maze going back and forth as the player sells and builds 2010-12-01T15:27:48 so probably in between? but the entire cost of the tower would need to be represented in the cost - usually it's cost + build time 2010-12-01T15:28:00 which would simplify things 2010-12-01T15:28:21 and then we don't have to deal with timing over stdout 2010-12-01T15:28:35 yup 2010-12-01T15:28:55 and we can give them a certain amount of time to set build orders and enemy creep buff/command orders 2010-12-01T15:29:13 then it's deterministic from there 2010-12-01T15:29:31 unless we do random tower damage, which we should maybe avoid? 2010-12-01T15:29:33 just have to set the limit low enough to prevent exhaustive tree searches 2010-12-01T15:29:44 yeah, I'd say stay away from randomness 2010-12-01T15:30:05 makes for unhappy programmers... why did my tower only do 1 dmg 10 times in a row! 2010-12-01T15:30:20 plus then Accoun will call everyone chiters some more :) 2010-12-01T15:30:31 ROFL 2010-12-01T15:30:55 3. Canadian reasons 2010-12-01T15:30:56 ... 2010-12-01T15:31:14 or was it special, Canadian only reason? idk 2010-12-01T15:31:21 either way that one made me crack up 2010-12-01T15:31:26 haha, yeah 2010-12-01T15:31:51 was the planetwars spec/gameplay engine open source before the contest? 2010-12-01T15:32:09 Planetwars is based on Galcon which is not open source 2010-12-01T15:32:21 I think they had to ask the creator if they could mod the idea for the contest 2010-12-01T15:32:29 ah 2010-12-01T15:32:50 I played Galcon on Steam months before this contest 2010-12-01T15:33:04 it feels so slow after watching bots play hundreds of games :) 2010-12-01T15:33:42 One other question though.... how about target selection/prioritization? Or I guess that could be part of the AI, individual tower programming algorithms 2010-12-01T15:33:51 but then you run into problems during gametime 2010-12-01T15:34:00 the waves are no longer deterministic 2010-12-01T15:34:07 right 2010-12-01T15:34:10 but tower target selection is pretty important 2010-12-01T15:34:39 being able to target with the towers efficiently would make a huge difference... 2010-12-01T15:35:32 and then you're back to user input for every step in the wave, which could work, but makes games MUCH longer 2010-12-01T15:35:50 maybe ditch the "wave" idea and go with a constant stream of creeps 2010-12-01T15:35:55 (read: more computationally expensive) 2010-12-01T15:35:56 and a timed step/turn 2010-12-01T15:36:02 :] 2010-12-01T15:36:17 but then yeah, games take a very long time to run 2010-12-01T15:36:32 yay, beat #13 2010-12-01T15:36:51 when was the contest going to end again? 2010-12-01T15:37:05 middle of the night for me probably (it's 21.30 here) 2010-12-01T15:37:24 1am 2010-12-01T15:37:30 damn 2010-12-01T15:37:30 I think it ends in 4 hours or so? 2010-12-01T15:37:43 that's too early.. might be tempted to stay up :p 2010-12-01T15:37:50 Haha 2010-12-01T15:37:55 me 2 =) 2010-12-01T15:38:53 Is the targetting problem a tower-defense killer? or is there a way around it? 2010-12-01T15:38:53 anyone would like an sport based game, as a simplified soccer or basket 2010-12-01T15:38:58 ? 2010-12-01T15:39:35 hm 2010-12-01T15:39:39 that could probably be interesting 2010-12-01T15:39:45 if we didn't have to support so many languages I'd say let the bot provide a targetting algo that gets run in realtime 2010-12-01T15:39:55 that is very(!) complez 2010-12-01T15:40:00 multiagent 2010-12-01T15:40:15 for each tower type 2010-12-01T15:40:57 JamesMG: Yeah... giving them a rather simple targetting priority rules and just saying tough for the rest 2010-12-01T15:41:14 if you had a sports game that just included players of a certain radius, a ball, 2 goals, and passing the ball to a certain location, it would be pretty cool 2010-12-01T15:41:17 or provide a handful of priorities to choose from 2010-12-01T15:41:33 closest creep to end zone, weakest creep, etc 2010-12-01T15:41:43 anyone knows liquidwar? 2010-12-01T15:41:52 YES! 2010-12-01T15:41:55 it's a very challenging real time geometric game 2010-12-01T15:42:14 could be interesting for an ai engine 2010-12-01T15:42:27 I forget how to take over other peoples liquid though, do you surround it? or is it something different 2010-12-01T15:42:45 *** tapwater has joined #aichallenge 2010-12-01T15:42:48 JamesMG: Yeah... defining those could get a bit tricky, but it would be doable 2010-12-01T15:43:27 this tower idea is already multiples more complex than planetwars I think 2010-12-01T15:43:36 nop, there is something like a frontier pressure, the more pressure, the more one liquid "gets into the other" 2010-12-01T15:44:48 it's available under most linux distros 2010-12-01T15:44:48 JamesMG: That's true. Maybe bring it down to two tower types, black and white, keep the interest, and use money to buff/change the black/white balance of the opponents creeps? 2010-12-01T15:45:12 apt-get install liquidwar 2010-12-01T15:45:36 JamesMG: Maybe they don't have to be towers at all! Maybe you're just assigning black/white damage to certain squares... and when a creep walks on those squares it takes damage 2010-12-01T15:45:55 JamesMG: then you'd have to be able to build mazing walls to force the creeps in 2010-12-01T15:46:08 I like that square defense idea 2010-12-01T15:46:10 http://www.gnu.org/software/liquidwar6/videos/liquidwar6-0.0.2beta-demo-400x300.mpeg 2010-12-01T15:46:31 protocolocon: Yeah... coming back to me now, apt-getting it now, it's a fun game... how are you going to issue commands though 2010-12-01T15:47:22 you only have to move a pointer (center of gravity of your liquid) 2010-12-01T15:47:28 around the map 2010-12-01T15:47:35 *** aaa_ has joined #aichallenge 2010-12-01T15:47:40 i think it should be simplified 2010-12-01T15:47:46 JamesMG: There has to be a way to force the creeps to walk over a square though, their movement should be governed by a very simple set of rules 2010-12-01T15:47:57 very cpu intensive 2010-12-01T15:48:15 protocolocon: do you think it could easily scale down? 2010-12-01T15:48:40 Meatkat, i used to play against my old job teamates, and it's really challenging 2010-12-01T15:48:46 Top 10 players: bocsimacko(3802), _iouri_(3585), Slin-.-(3538), jimrogerz(3529), _Astek_(3521), Accoun(3513), george(3508), GreenTea(3508), asavis(3505), protocolocon(3496) 2010-12-01T15:48:56 Meatkat, maybe water possitions could be discretized 2010-12-01T15:49:24 i think there was a win2 port 2010-12-01T15:49:24 looks cool 2010-12-01T15:50:11 protocolocon: I would agree that it's very challenging, and yeah, if you cut the playing field into discrete units, that would help a lot of the calculations... 2010-12-01T15:50:11 *** choas has joined #aichallenge 2010-12-01T15:50:43 it would also take away some of its awesomeness though =) 2010-12-01T15:50:52 but still might be cool 2010-12-01T15:52:12 JamesMG: or maybe you have an nxn tile board, and some of the tiles are black, some are white. Every iteration you can change the color of a certain amount of tiles. At the same time your enemy gets to choose the pathway for the single creep to use your maze after you've finished. the creep can either be black or white and takes dmg if it steps on an opposite square 2010-12-01T15:52:30 *** smloh1 has quit IRC (Ping timeout: 265 seconds) 2010-12-01T15:52:52 JamesMG: or maybe you are using the same board, one is black one is white, and you both get to choose an equal number of tiles to change color then find a route accross? 2010-12-01T15:52:56 That is really simple 2010-12-01T15:53:11 so a combination of walking a maze and building one 2010-12-01T15:53:42 it allows for some pretty neat AI I think 2010-12-01T15:54:07 that does sound pretty interesting 2010-12-01T15:54:49 I guess your score is the number of your creeps that get all the way to the other side? 2010-12-01T15:55:18 its like dynamic go =) 2010-12-01T15:55:19 Or if you are only sending one, the number of enemy tiles you need to step on to get there 2010-12-01T15:55:53 You could make it more interesting by allowing tile changes while the creep is on the tiles 2010-12-01T15:56:10 just add a tile to your "resources" every so often? 2010-12-01T15:56:15 so you and your enemy have a creep that needs to get the opposite side that you can control and there are two stages: change and move 2010-12-01T15:56:22 *** rebelxt has joined #aichallenge 2010-12-01T15:56:36 yeah, you can change a tile every certain number of moves 2010-12-01T15:56:55 maybe you can say that your creep can't even walk on the enemy tiles 2010-12-01T15:57:13 and make the score the number of moves it takes to reach the end? 2010-12-01T15:57:19 and instead of adding a tile in wherever, you push the whole row/column down or up, and add another tile 2010-12-01T15:57:19 higher being worse 2010-12-01T15:57:26 yup 2010-12-01T15:59:02 *** Frontier has quit IRC (Ping timeout: 250 seconds) 2010-12-01T16:01:02 I think that could be interesting... you start on an n by n randomized symmetric board with black starting in one corner and white in the other. white then gets to move a column up or down or a row left or right and add a white tile in the newely created space. Then both players move (legal moves are up/down/left/right and only to a square of your color). Then black gets to manipulate the maze and they both move again. Repeat 2010-12-01T16:02:51 It's like opposite maze vs The Cube 2010-12-01T16:02:58 or meets rather 2010-12-01T16:05:04 Whoever gets to the opposite corner first wins 2010-12-01T16:05:25 or whoever is Euclidian closest after 200 turns 2010-12-01T16:05:35 or so 2010-12-01T16:06:06 I think that would be fun 2010-12-01T16:06:16 if we write it, are we disqulified from participating? :) 2010-12-01T16:06:27 yes xD 2010-12-01T16:07:08 Lame sauce 2010-12-01T16:07:43 *** Cold-Phoenix has joined #aichallenge 2010-12-01T16:09:40 Well I'll at least try and float it up to someone who could program it... it's really simple and has the potential to be pretty interesting 2010-12-01T16:15:52 And according the contest eligibility requirements I don't think it would disqualify us... it states that only contest admins are DQed, not contributors. 2010-12-01T16:16:05 http://code.google.com/p/ai-contest/wiki/ContestEligibility 2010-12-01T16:16:38 we should code it :) 2010-12-01T16:17:53 *** bss03-AFK is now known as bss03 2010-12-01T16:18:50 contributors are not dq'ed 2010-12-01T16:19:00 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-12-01T16:19:04 nor committers 2010-12-01T16:19:13 I don't see why we would be DQ'd, since the code would be available to all anyways 2010-12-01T16:19:23 *** nullkuhl has joined #aichallenge 2010-12-01T16:20:32 *** GreenTea_ has quit IRC (Ping timeout: 265 seconds) 2010-12-01T16:21:05 Cool... we should probably start out doing a specifications document first, make sure we have all our ducks in a row 2010-12-01T16:21:30 if you throw something in Google docs and share it I'll go over it also 2010-12-01T16:21:33 *** Kingpin13 has joined #aichallenge 2010-12-01T16:22:29 *** davidjliu has joined #aichallenge 2010-12-01T16:23:29 *** bhasker has quit IRC (Quit: bhasker) 2010-12-01T16:23:56 *** aaa_ has quit IRC (Quit: Page closed) 2010-12-01T16:26:14 kk, une moment, SVP 2010-12-01T16:28:37 JamesMG: Do you have a catchy sounding name? 2010-12-01T16:28:43 I'd say Black and White 2010-12-01T16:28:54 *** nibalizer has quit IRC (Ping timeout: 276 seconds) 2010-12-01T16:29:03 but I think that's already taken ;) 2010-12-01T16:29:11 *** nibalizer has joined #aichallenge 2010-12-01T16:29:12 Maze Warrior 2010-12-01T16:29:20 BW Maze Wars 2010-12-01T16:29:21 Black and White was a fun game :) 2010-12-01T16:29:41 A Mazing Game 2010-12-01T16:29:46 First game built around AI as well 2010-12-01T16:29:48 haha, I like that 2010-12-01T16:29:54 iouri: hahaha 2010-12-01T16:30:13 *** fgump has joined #aichallenge 2010-12-01T16:31:15 *** GreenTea_ has joined #aichallenge 2010-12-01T16:31:30 *** pgpaskar has quit IRC (Ping timeout: 276 seconds) 2010-12-01T16:31:34 *** pgpaskar_ has joined #aichallenge 2010-12-01T16:32:25 Alright 'A Mazing Game' it is... might change later... iouri you're a contributer now! 2010-12-01T16:32:39 lol 2010-12-01T16:32:58 so what's the game about? I just opened the page back and saw talk about mazes and black and white 2010-12-01T16:33:53 (also, maybe "A Mazing Battle" would be better?) 2010-12-01T16:34:18 *** rebelxt has left #aichallenge 2010-12-01T16:34:33 that sounds a bit... corny... if you know what I mean (hehe?) 2010-12-01T16:34:40 *** Epona has joined #aichallenge 2010-12-01T16:34:41 I PMed the watered down description to you iouri 2010-12-01T16:34:41 yep 2010-12-01T16:34:43 *** politik has quit IRC (*.net *.split) 2010-12-01T16:35:00 is bocs bot ready yet? o.o 2010-12-01T16:35:35 davidjliu: concur, it's quite corny 2010-12-01T16:35:37 Yay, you can now edit your "About me" section of your profile 2010-12-01T16:35:59 C-up C-left, C-right, C-up, C-left, C-right 2010-12-01T16:36:00 Thanks janzert ! 2010-12-01T16:36:17 anyone? 2010-12-01T16:36:26 lol 2010-12-01T16:36:32 win 2010-12-01T16:36:35 haven 2010-12-01T16:36:41 't heard that in a while 2010-12-01T16:37:05 JamesMG: thanks. Looks interesting.. though would it look as cool as Tron or Planet Wars? 2010-12-01T16:37:24 *** politik has joined #aichallenge 2010-12-01T16:37:40 iouri: it came from discussing making a tower defense style game for the next challenge 2010-12-01T16:37:56 ahh, gotcha 2010-12-01T16:38:48 I don't think it'd have the same awesome fireworks planetwars does though 2010-12-01T16:39:26 I think if you were looking at a replay it would be pretty interesting to watch, but yeah, not even close the level of epic star battles 2010-12-01T16:39:52 Though I think it's a more exciting twist on maze games in general 2010-12-01T16:40:11 Maybe we could add some element of explosions or guns or something 2010-12-01T16:40:20 tower defense is aweful 2010-12-01T16:40:38 well, that's a clear opinion :p 2010-12-01T16:40:39 It's no longer recognizable as tower defense 2010-12-01T16:41:09 Has anyone seen the movie "The Cube?" 2010-12-01T16:41:17 I played bloons, realized theres only one or 2 ways to beat it on the hardest difficulty, and never looked back 2010-12-01T16:41:17 it's more like a maze version of go or Othello 2010-12-01T16:41:20 it's like that... sort of... but without the danger 2010-12-01T16:42:19 however danger in the form of monsters/explosions/traps/etc... could be added in 2010-12-01T16:42:25 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-12-01T16:42:25 Meatkat: what do we want to happen if the white and black "avatars" collide? 2010-12-01T16:42:31 So I take it we are rejecting that name then? 2010-12-01T16:43:09 I kinda like the name, even if it barely fits 2010-12-01T16:43:14 Well they could never occupy the same square since black can only more on black squares and white on white squares... 2010-12-01T16:43:35 Alright... well we can float a couple other ones just in case 2010-12-01T16:43:52 sigh: Better: fibs = let tfibs = 1 : zipWith (+) fibs tfisb in 0 : tfibs 2010-12-01T16:45:17 *** Mathnerd314 has joined #aichallenge 2010-12-01T16:45:39 Meatkat: I saw the cube 2010-12-01T16:45:47 all of them, good movies :) 2010-12-01T16:48:46 Top 10 players: bocsimacko(3802), _iouri_(3585), Slin-.-(3538), jimrogerz(3529), _Astek_(3521), Accoun(3513), george(3508), GreenTea(3508), asavis(3505), protocolocon(3496) 2010-12-01T16:49:06 2 hours and 15 minutes remaining! 2010-12-01T16:49:41 *** bhasker has joined #aichallenge 2010-12-01T16:49:46 how many more hours i wonder 2010-12-01T16:49:54 2 hours and 15 minutes remaining! 2010-12-01T16:50:29 lol excellent timing 2010-12-01T16:50:52 go bot go! 2010-12-01T16:51:17 I kinda liked the sports idea -_- 2010-12-01T16:51:26 :) 2010-12-01T16:51:29 http://ai-contest.com/visualizer.php?game_id=9543971 turn 67,68 2010-12-01T16:51:30 there's a football competition for robot dogs right? 2010-12-01T16:51:36 the true power of synchronized attacks 2010-12-01T16:51:42 it never bores me somehow 2010-12-01T16:54:37 Mistmanov: yes 2010-12-01T16:54:44 *** Appleman1234 has joined #aichallenge 2010-12-01T16:55:44 Mistmanov: http://www.youtube.com/watch?v=RbE3qhPcx4g&feature=related 2010-12-01T16:55:57 omg 2010-12-01T16:55:59 genius 2010-12-01T16:56:30 I think coding something like that would be awesome 2010-12-01T16:56:41 even without the element of a cuty clumsy robot executing your code 2010-12-01T16:56:42 *** bss03 is now known as bss03-AFK 2010-12-01T16:56:46 you have no idea of its complexity =) 2010-12-01T16:56:52 true :p 2010-12-01T16:57:04 but that's a good thing right 2010-12-01T16:57:07 *** sigh has joined #aichallenge 2010-12-01T16:57:12 I think that for simple behavior 2010-12-01T16:57:18 multiagent, computervision, robotics 2010-12-01T16:57:20 it should be pretty easy, depending on the exact rules 2010-12-01T16:57:33 all of these field could occupie you for a lifetime :P 2010-12-01T16:58:23 true, but for the AI contest.. just need a program hat can move in a straight line towards a ball, and pass it 2010-12-01T16:58:23 these are teamefforts 2010-12-01T16:59:26 the actual dogs are way more complicated because they need to... actually kick the ball and stuff 2010-12-01T17:00:04 but if you just had a function of kickBall(power, direction).. 2010-12-01T17:00:05 and with 3 or 4 agents per team (not 11 as real soccer)? 2010-12-01T17:00:11 and simplified rules 2010-12-01T17:00:36 anyone have code out that has a neural net or mutational algorithms? 2010-12-01T17:00:40 id love to see it 2010-12-01T17:02:03 I'm almost feeling an urge to go code such a ball kicking game myself.. 2010-12-01T17:02:42 damnit i'm losing more 2010-12-01T17:02:42 ..which really isn't good considering I promised my supervisor to hand in some new progress on my thesis this sunday 2010-12-01T17:02:49 heh 2010-12-01T17:02:51 oef 2010-12-01T17:03:03 i think he knows you by now :P 2010-12-01T17:04:05 http://ai-contest.com/profile.php?user_id=5822 ./cry 2010-12-01T17:04:10 i am going to be in the 70-80s 2010-12-01T17:04:25 nah 2010-12-01T17:04:30 it doesn't change that fast 2010-12-01T17:04:58 )$#########($% another loss 2010-12-01T17:05:05 NEIN NEIN NEIN NEIN NEIN 2010-12-01T17:05:24 * Mistmanov highfives the chiting pairing-algorithm 2010-12-01T17:05:36 lmao 2010-12-01T17:06:36 but yeah, we were used to the ranking significantly changing with every few games. 2010-12-01T17:07:01 so it feels kinda weird now that a lot of games can have almost no impact 2010-12-01T17:08:01 however, at the moment of the "freeze" 2010-12-01T17:08:07 we had an elo difference of 3 2010-12-01T17:08:22 so it kinda makes me scary 2010-12-01T17:08:46 yeah it's basically luck now 2010-12-01T17:08:46 when did the freeze happen? 2010-12-01T17:08:51 2 more hours left 2010-12-01T17:08:53 5 min per game 2010-12-01T17:09:00 ehh each 5 min a game 2010-12-01T17:09:01 Mistmanov: did you explain somewhere how your bot works? 2010-12-01T17:09:21 the final bot doesn't have a lot of description, no 2010-12-01T17:09:37 but there isn't really a big overarching strategy 2010-12-01T17:10:48 how come it performs remarkably constantly against oppositions of different levels? 2010-12-01T17:11:13 I think it's because it's very weak at expanding 2010-12-01T17:11:19 I think because its good at certain map 2010-12-01T17:11:21 maps 2010-12-01T17:11:27 but pretty good at conquering 2010-12-01T17:11:46 so if a map requires good expansion, I'm going to lose 2010-12-01T17:12:05 if a map requires direct fighting with the enemy, I can win against a wide range of opponents 2010-12-01T17:12:29 at least, that's my theory 2010-12-01T17:12:44 mega1: will you also enter the next contest? 2010-12-01T17:12:48 or give us a chance? 2010-12-01T17:12:50 ^_^ 2010-12-01T17:13:12 Mistmanov: Error323 tells me that you won 1/3 against my bot in local testing 2010-12-01T17:13:17 *** iouri has quit IRC (Quit: Page closed) 2010-12-01T17:13:20 *** choas has quit IRC (Quit: leaving) 2010-12-01T17:13:22 yeah I read that.. but I somehow doubt it :p 2010-12-01T17:13:47 that would make my bot like... the best against yours? 2010-12-01T17:13:54 I really, really doubt that 2010-12-01T17:14:21 that's much better than anything else I saw and don't think there are so many non-expansion maps 2010-12-01T17:14:30 yes, by far 2010-12-01T17:14:40 (I wanted to test your bot against mine on my computer, but it seems there is no sbcl compiler for cygwin..) 2010-12-01T17:15:04 re-running 2010-12-01T17:15:17 JamesMG: alright made the document, how do I get you an invite? 2010-12-01T17:15:34 you can download a windows binary http://www.sbcl.org/platform-table.html and hope it works 2010-12-01T17:15:39 perhaps i did sth wrong 2010-12-01T17:19:21 *** amstan has quit IRC (Ping timeout: 245 seconds) 2010-12-01T17:21:24 Number of player 1 wins: 8 2010-12-01T17:21:24 Number of player 2 wins: 26 2010-12-01T17:21:24 Number of draws: 3 2010-12-01T17:21:32 results so far 2010-12-01T17:21:38 Mistmanov: being player 1 2010-12-01T17:22:24 :/ 2010-12-01T17:22:37 that's already quite high Mistmanov 2010-12-01T17:22:51 yeah.. mega1 isn't timing out, or crashing, or whatever? 2010-12-01T17:23:00 nope 2010-12-01T17:23:03 or my bot is hogging all the cpu/memory/etc? 2010-12-01T17:23:05 sigh 3 more losses 2010-12-01T17:23:21 this is depressing :-\ 2010-12-01T17:23:27 Mistmanov: amazing I won like 8 games against you on official now 2010-12-01T17:23:38 yeah, that's a bit sucky 2010-12-01T17:23:49 whilest my chance of winning is sth like 40% 2010-12-01T17:24:00 I'm not a lucky person 2010-12-01T17:24:06 * Error323 high fives the random generator 2010-12-01T17:24:20 -_- 2010-12-01T17:24:48 *** amstan has joined #aichallenge 2010-12-01T17:24:48 *** ChanServ sets mode: +o amstan 2010-12-01T17:24:57 to illustrate: with the end-of-year lottery, you have a ~40% chance to at least win something, right? they draw like 4 final numbers that make you win something like $10 2010-12-01T17:25:15 we've gotten a ticket every year for as long as I can remember, and we've never won a thing 2010-12-01T17:25:26 lol 2010-12-01T17:25:51 calculate those probabilities and you'll never enter again 2010-12-01T17:25:53 if ur sane :P 2010-12-01T17:26:01 oh, my dad gets them from his work 2010-12-01T17:26:09 buying lottery tickets is indeed not our thing :p 2010-12-01T17:26:27 75 wins,92 losses, 15 draws today 2010-12-01T17:26:28 :-\ 2010-12-01T17:26:40 Mistmanov: does your bot use/need the 1s? 2010-12-01T17:26:51 1s? 2010-12-01T17:26:53 the probability of me winning 8 times in a row from you mist is 0.4^8 lol 2010-12-01T17:27:14 Mistmanov: you know, the 1 second time limit per turn :-) 2010-12-01T17:27:15 which 1s if I may ask? 2010-12-01T17:27:20 oh 1 second :p 2010-12-01T17:27:21 no 2010-12-01T17:27:27 something like 30 ms max 2010-12-01T17:28:10 http://ai-contest.com/profile.php?user_id=7347 2010-12-01T17:28:23 look like I'me on a winning streak^^ 2010-12-01T17:28:29 wow 2010-12-01T17:29:42 playing the lottery = voluntarily paying extra taxes 2010-12-01T17:30:04 unless you're lucky ;) 2010-12-01T17:30:19 luck has nothing to do with it.. 2010-12-01T17:30:31 you coulda just got together with a bunch of random people, and played craps 2010-12-01T17:30:36 I approach it a bit like going to the casino.. you should only do it if the fun you get from it is worth the money you enter 2010-12-01T17:30:36 canadian chiting? 2010-12-01T17:30:57 sorry, Haskell fans but yay steve yegge posted on his blog :) 2010-12-01T17:31:00 http://steve-yegge.blogspot.com/2010/12/haskell-researchers-announce-discovery.html 2010-12-01T17:32:25 LOL 2010-12-01T17:33:40 bot finally posted on the forum... 2010-12-01T17:33:49 I think dmj has me beat though 2010-12-01T17:33:57 mega1: how is your blog comming along? 2010-12-01T17:34:06 it's ready 2010-12-01T17:34:13 mega1: sweeeeeeet 2010-12-01T17:34:14 a git push away 2010-12-01T17:34:34 your gonna wait another 1.5 hours aren't you :P 2010-12-01T17:34:45 mega1: I'm assuming your blog is about your bot dev? 2010-12-01T17:35:00 your blog is also on git? 2010-12-01T17:35:01 suddenly he goes on a thirty game losing streak 2010-12-01T17:35:02 ... 2010-12-01T17:35:05 and probably still ends up 1st 2010-12-01T17:35:07 is the pairing cutoff going to change any further? 2010-12-01T17:35:11 Error323, Meatkat, bhasker: yes 2010-12-01T17:35:30 *** meduza has joined #aichallenge 2010-12-01T17:35:34 *** rebelxt has joined #aichallenge 2010-12-01T17:35:36 thats a first people putting their blog under revision control 2010-12-01T17:35:46 =] 2010-12-01T17:35:55 my thesis is under git aswell 2010-12-01T17:36:08 Error323: probably more because I'm going to bed now 2010-12-01T17:36:10 looks liek first and 2 are quite decided upon 2010-12-01T17:36:15 mega1: nooo 2010-12-01T17:36:19 :/ 2010-12-01T17:37:17 bhasker: a considerable part of my home dir is also in git 2010-12-01T17:37:27 interesting 2010-12-01T17:37:43 i tried understanding git but too complicated till i try it out a bit 2010-12-01T17:37:49 will try it for my next project 2010-12-01T17:37:55 er, um.. 2010-12-01T17:38:00 it is too complicated indeed 2010-12-01T17:38:03 the 'games played' is going down instead of up? 2010-12-01T17:38:07 http://ai-contest.com/game_stats.php 2010-12-01T17:38:16 its 24 hours 2010-12-01T17:38:28 first 1000 use heavy computation 2010-12-01T17:38:38 so less games overall 2010-12-01T17:38:42 but more among the 1000 2010-12-01T17:38:45 games are being unplayed ;-) 2010-12-01T17:39:01 oh.. ya, didn't notice the 24hours there 2010-12-01T17:39:35 82 minutes left, correct? 2010-12-01T17:39:52 yes 2010-12-01T17:39:55 well 2010-12-01T17:39:56 good 2010-12-01T17:40:07 assuming jeff is here by then 2010-12-01T17:40:23 ;-) 2010-12-01T17:40:27 82mins till what? 2010-12-01T17:40:27 hmmm mega1, Mistmanov its 22 wins for mistmanov and 6 draws 2010-12-01T17:40:51 out of 75 or 150? 2010-12-01T17:40:52 oh phew my bot ifnally won a game 2010-12-01T17:40:56 out of 150 2010-12-01T17:41:07 that's not quite the 49 I remembered 2010-12-01T17:41:12 i don't know how I said 49 2010-12-01T17:41:13 :S 2010-12-01T17:41:13 My bot has won 13 games over the 15 lasts =) 2010-12-01T17:41:18 mja 2010-12-01T17:41:31 but still double the next best 2010-12-01T17:41:31 <-- 93 or 94 losses today 2010-12-01T17:41:33 sorry, strange 2010-12-01T17:41:34 and like 76 wins 2010-12-01T17:41:35 :-\ 2010-12-01T17:41:39 medrimonia: heroic mode? 2010-12-01T17:41:54 Number of player 1 wins: 22 2010-12-01T17:41:55 Number of player 2 wins: 122 2010-12-01T17:41:55 Number of draws: 6 2010-12-01T17:42:01 rebelxt: Yep I'll give all I can until the end! 2010-12-01T17:42:23 could somebody test me against mega? I don't know how to run fancy lisp stuff... ready to be embarassed 2010-12-01T17:42:38 where is your source 2010-12-01T17:42:45 thats a nice page to look at medri 2010-12-01T17:42:45 medrimonia snatches the python title at the last moment 2010-12-01T17:42:46 But the funniest is that my two losses are against people out of top 30 2010-12-01T17:42:50 um, it's in the thread 2010-12-01T17:42:57 http://ai-contest.com/forum/viewtopic.php?f=17&t=1168&start=50 2010-12-01T17:42:59 Error323: what happens when you rerun one of the maps that I "won" and watch it in a visualizer? 2010-12-01T17:43:03 *** Frontier has joined #aichallenge 2010-12-01T17:43:06 davidjliu: http://ai-contest.com/visualizer.php?game_id=9543971 2010-12-01T17:43:19 davidjliu: I think it's too late for my bot to wake up 2010-12-01T17:43:45 you and dmj have too much elo points for me 2010-12-01T17:43:52 amstan: yeah, I'm something like 0-5 against him on the server... 2010-12-01T17:44:05 Mistmanov: these are the games: http://fhuizing.no-ip.org/2010-12-01-23-11-50/ 2010-12-01T17:46:02 hmm 2010-12-01T17:46:16 * Mistmanov can't think of anything deeper to say 2010-12-01T17:46:35 you can use the planetwars paste thingie 2010-12-01T17:47:12 @pastewars 2010-12-01T17:47:13 amstan: pastewars = http://visualizer.naktibalda.lt/ 2010-12-01T17:47:27 ty 2010-12-01T17:48:46 Top 10 players: bocsimacko(3802), _iouri_(3585), Slin-.-(3538), jimrogerz(3529), _Astek_(3521), Accoun(3513), george(3508), GreenTea(3508), asavis(3505), protocolocon(3496) 2010-12-01T17:49:13 davidjliu: its running 2010-12-01T17:49:32 (cowers in fear) 2010-12-01T17:49:36 hopefully I win 1 2010-12-01T17:49:44 you did already :) 2010-12-01T17:49:47 don't be shy lol 2010-12-01T17:49:53 #12th == good 2010-12-01T17:49:55 how exciting. heh 2010-12-01T17:52:12 http://ai-contest.com/visualizer.php?game_id=9558283 #$&%#)$^*&#()$*&%#$% 2010-12-01T17:52:50 Ooookay I am home from work. Time to do this thang. amstan, janzert, you guys here? 2010-12-01T17:53:00 ;d 2010-12-01T17:53:03 j3camero: i'll be leaving soon 2010-12-01T17:53:14 What about all your karma? :-) 2010-12-01T17:53:15 zomg 2010-12-01T17:53:21 so the final is over in about an hour? or they're changing the cutoff at that point? 2010-12-01T17:53:28 burny: no its over 2010-12-01T17:53:32 j3camero: ugh.. how much will it take? 2010-12-01T17:53:37 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-12-01T17:53:41 The plan is to release at 7. 2010-12-01T17:53:55 yeah, i'm leaving at 6, going to hamilton 2010-12-01T17:53:56 looks like I'm goan be anywhere from 89th to 96th then 2010-12-01T17:53:59 won't get there before 7 2010-12-01T17:54:04 I might have to submit it and ask the IRC channel for upboats and steal all the sweet sweet karma! 2010-12-01T17:54:12 i am probably going to be in the 60's 2010-12-01T17:54:16 lost too many games today 2010-12-01T17:54:17 :-\ 2010-12-01T17:54:23 and still losing 2010-12-01T17:54:26 :-( 2010-12-01T17:55:12 bhasker: recent losses barely matter since you have lots of games played. might lose a couple of spots but not 10 for sure 2010-12-01T17:55:28 * sigh grabs popcorn and watches the drama unfold 2010-12-01T17:55:36 * Error323 joins sigh 2010-12-01T17:55:38 i hope you are right rebel lol 2010-12-01T17:55:57 all we need is Accoun's conspiracy theories to make this complete 2010-12-01T17:55:58 i am like going 4 losses 1 win 2010-12-01T17:56:09 oh yeah that would be epic 2010-12-01T17:56:13 actually 2010-12-01T17:56:14 *** Appleman1234 has joined #aichallenge 2010-12-01T17:56:15 g2g now 2010-12-01T17:56:16 cya guys 2010-12-01T17:56:18 good luck 2010-12-01T17:56:19 bye 2010-12-01T17:56:21 cu 2010-12-01T17:56:37 j3camero: ur on ur own 2010-12-01T17:56:51 what about janzert? 2010-12-01T17:57:23 must find bots with worse loss runs than me 2010-12-01T17:57:28 dammit every bot has more green than red:( 2010-12-01T17:58:01 that's what you get for not being bocsimacko 2010-12-01T17:58:07 haha 2010-12-01T17:58:08 Just committed one last change to the game engine. if (username == "bhasker") {lose_game();}. That should do it. 2010-12-01T17:58:17 Oh uh, I mean, uh.... 2010-12-01T17:58:22 looooooool 2010-12-01T17:58:23 win_game() would be nice 2010-12-01T17:58:25 lol:p 2010-12-01T17:58:32 *** Error323 is now known as Accoun_ 2010-12-01T17:58:36 Canadian chiters 2010-12-01T17:58:39 Canadian chiters 2010-12-01T17:58:41 though with the loss ratio i doubt i will get up there 2010-12-01T17:58:42 heh 2010-12-01T17:58:43 *** Accoun_ is now known as Error323 2010-12-01T17:59:01 but i will surely beat accoun 2010-12-01T17:59:02 you only repeated yourself twice, and you didn't even change the message 2010-12-01T17:59:09 TIL Accoun and Error323 are the same person. 2010-12-01T17:59:21 *** smloh1 has joined #aichallenge 2010-12-01T17:59:22 shit 2010-12-01T17:59:25 busted =) 2010-12-01T17:59:30 doh i'm streaking red... 2010-12-01T17:59:34 j3camero change it to if (username=="bhasker" and enemyname=="Accoun") { win_game_for_bhasker(); } 2010-12-01T17:59:42 welcome to my club smloh 2010-12-01T17:59:57 and in the game matching code 2010-12-01T18:00:07 j3camero: gonna do it now or wait till 6pm? 2010-12-01T18:00:14 "if username=="account" { enemy_user = "bhasker" } ' 2010-12-01T18:00:17 so that he only plays me 2010-12-01T18:00:20 and loses all his games 2010-12-01T18:00:34 lol account 2010-12-01T18:00:35 =) 2010-12-01T18:00:37 lol 2010-12-01T18:00:40 i thought the engine was all in python, what's with the braces and semi-colons? 2010-12-01T18:00:52 *** medrimonia has quit IRC (Ping timeout: 265 seconds) 2010-12-01T18:01:18 *** medrimonia has joined #aichallenge 2010-12-01T18:02:25 total 522 games played, 240 wins, 245 losses, 37 draws 2010-12-01T18:02:45 make that 246 losses 2010-12-01T18:02:47 :- 2010-12-01T18:02:55 shouldn't jsut about everyone outside of the top30 have a similar record? 2010-12-01T18:03:21 not sure 2010-12-01T18:03:42 i wish the game page provided the data in form a single flat file 2010-12-01T18:03:50 of* 2010-12-01T18:04:40 *** mega1 has quit IRC (Ping timeout: 255 seconds) 2010-12-01T18:05:07 games per minute dropping fast, is it over? 2010-12-01T18:05:59 nooooooo not with this red streak on my page 2010-12-01T18:06:32 It's too late bhasker. Accoun is right. It's all a conspiracy. We were never going to let you win anyways. 2010-12-01T18:06:34 looks like it's been 0games for about 3mins now 2010-12-01T18:06:35 :-) 2010-12-01T18:06:35 bhasker: do you really want more red color? ;) 2010-12-01T18:06:45 now that is a good question rebel 2010-12-01T18:06:50 hmm 2010-12-01T18:06:57 gambler's fallacy at work 2010-12-01T18:07:05 i have had so many red so luck has to change 2010-12-01T18:07:10 it has to turn green now 2010-12-01T18:07:11 !! 2010-12-01T18:07:16 Who knows. All the red ink could be because you won a game against a high ranked person, moved up in the rankings, and started losing. Could be a good sign! 2010-12-01T18:07:49 so camero have you shut down the instances? 2010-12-01T18:07:56 looksl ike all numbers went to zero 2010-12-01T18:07:57 just urs :P 2010-12-01T18:07:57 :-\ 2010-12-01T18:08:05 aww all servers down 2010-12-01T18:08:06 *** Mistmanov has quit IRC (Quit: sleepy, oh so sleepy... good luck everyone =)) 2010-12-01T18:08:06 *** amstan has quit IRC (Ping timeout: 245 seconds) 2010-12-01T18:08:09 http://www.ai-contest.com/game_stats.php 2010-12-01T18:08:11 :-\ 2010-12-01T18:08:18 aw, my last game was a loss :( 2010-12-01T18:08:18 gone 2010-12-01T18:08:24 mine too 2010-12-01T18:08:27 mine too 2010-12-01T18:08:28 my last X were all losses lol 2010-12-01T18:08:43 davidjliu: 8 wins so far =) 2010-12-01T18:08:46 against mega1 2010-12-01T18:08:51 8/10 losses in last 10 games 2010-12-01T18:09:03 ouch 2010-12-01T18:09:30 but I bet i'm more worried 2010-12-01T18:09:36 error323: how many games played? 2010-12-01T18:09:42 last 20 games 2010-12-01T18:09:44 3 elo diff with mistmanov before freeze 2010-12-01T18:09:45 :S 2010-12-01T18:09:49 13 losses/6 wins 1 draw 2010-12-01T18:09:50 :-\ 2010-12-01T18:09:56 i really wonder wether i've won of my dutch nemesis 2010-12-01T18:09:59 roughyl 400k games total? 2010-12-01T18:10:11 davidjliu: 72 so far 2010-12-01T18:10:19 heh, that's pretty good :) 2010-12-01T18:10:21 138868 2010-12-01T18:10:29 going for a strong second-half effort 2010-12-01T18:10:31 0 0 0 (0%) 138868 1000 2010-12-01T18:10:31 New users New submissions Successful submissions Games played Pairing cutoff 2010-12-01T18:10:34 .. 2010-12-01T18:10:53 8/2/64 w/d/l davidjliu 2010-12-01T18:11:06 well what's a couple of places up or down anyway 2010-12-01T18:11:12 yay! more than 10% :) 2010-12-01T18:11:33 j3camero so are you going to update the final rankings now? 2010-12-01T18:12:54 bhasker: i think i beat you on that worst last 20 games record 2010-12-01T18:13:00 That's a beautiful end for me^^ 2010-12-01T18:13:08 damn you medri 2010-12-01T18:13:30 oh wow smloh 2010-12-01T18:13:34 you do beat me! 2010-12-01T18:13:39 lol 2010-12-01T18:13:40 Nah. Gonna try to get them out by 7 EST. 2010-12-01T18:13:47 *** Epona has quit IRC (Quit: Leaving) 2010-12-01T18:13:55 ah k 2010-12-01T18:13:57 http://www.phdcomics.com/comics/archive.php?comicid=1323 haha 2010-12-01T18:14:06 on the other side medri my bot beat you all 3 times! 2010-12-01T18:14:12 too bad we weren't able to get back to rank #11, eh medrimonia 2010-12-01T18:14:21 yep smloh1! 2010-12-01T18:14:25 *** meduza has quit IRC (Quit: Page closed) 2010-12-01T18:14:29 you just might? 2010-12-01T18:14:33 then we might have been cursed all the way back to #30! 2010-12-01T18:14:43 since rankings haven't been updated for 5 hrs 2010-12-01T18:14:54 that wwould have been fun 2010-12-01T18:15:04 bhasket: if you want to get rid of your red: javascript:els = (function(){els=document.getElementsByTagName('td'); for(i=0;i would* 2010-12-01T18:15:14 *** JamesMG has quit IRC (Quit: home) 2010-12-01T18:15:21 *bhasker 2010-12-01T18:15:28 I think im somewhere in {60,...,80} 2010-12-01T18:15:30 :P 2010-12-01T18:15:35 *** meduza has joined #aichallenge 2010-12-01T18:15:36 let me try that sigh 2010-12-01T18:15:54 LOOOL 2010-12-01T18:15:58 sigh: niiice one 2010-12-01T18:16:13 woohoo 2010-12-01T18:16:20 hmm but it still shows Loss 2010-12-01T18:16:23 need to change the innerText 2010-12-01T18:16:25 of the td 2010-12-01T18:16:31 to make everything a win 2010-12-01T18:16:58 damn i can't remember my javascript skillz 2010-12-01T18:17:54 javascript:els = (function(){els=document.getElementsByTagName('td'); for(i=0;i wins for all! 2010-12-01T18:18:45 javascript:els = (function(){els=document.getElementsByTagName('td'); for(i=0;i hehe 2010-12-01T18:19:19 haha, I have a stupid els= at the start 2010-12-01T18:19:19 thanks :), really the best tool of the contest 2010-12-01T18:19:28 refactoring one liners is eror-prone :P 2010-12-01T18:19:31 damn that looks awesome 2010-12-01T18:19:35 *error-prone 2010-12-01T18:19:50 all green 2010-12-01T18:20:00 So moral of the story is that ranking won't be up for another 12 hours ish? 2010-12-01T18:20:15 Which means I can sleep? 2010-12-01T18:20:16 Meatkat: noooh their time 7 2010-12-01T18:20:17 7 EST 2010-12-01T18:20:20 aren't they already up? 2010-12-01T18:20:21 i guess he means 7pm 2010-12-01T18:20:21 it means 2 am? 2010-12-01T18:20:25 so another 40 minutes? 2010-12-01T18:20:30 oh 2010-12-01T18:20:32 its 3:20 PST here 2010-12-01T18:20:40 EST woudl be 6:20 right now? 2010-12-01T18:20:47 OH 7PM! 2010-12-01T18:20:51 w00t 2010-12-01T18:20:59 sigh: awesome! lol 2010-12-01T18:21:03 I think in an hour and 40 minutes 2010-12-01T18:21:11 so used to 24 time 2010-12-01T18:21:19 me too 2010-12-01T18:21:21 hmm in 40 minutes? 2010-12-01T18:21:25 i think 2010-12-01T18:21:28 not an hour and 40 2010-12-01T18:21:45 well they said it ends at 18:00 servertime 2010-12-01T18:21:52 Wed, Dec 1, 2010, 6:21PM EST 2010-12-01T18:21:54 which is 7 hours behind our time 2010-12-01T18:21:58 @time 2010-12-01T18:21:58 Error323: 06:21 PM, December 01, 2010 2010-12-01T18:22:01 thats the current time 2010-12-01T18:22:05 *** BaronTrozo has joined #aichallenge 2010-12-01T18:22:06 server time is cst 2010-12-01T18:22:14 hmm no 2010-12-01T18:22:17 looks like EST 2010-12-01T18:22:37 oh wait it might be CST 2010-12-01T18:22:41 last game played is at 5pm 2010-12-01T18:23:03 17:02:34 2010-12-01T18:23:06 *** Rubicon-|-Cross has joined #aichallenge 2010-12-01T18:23:13 its like new years countdown for geeks 2010-12-01T18:23:14 :D 2010-12-01T18:23:19 :P 2010-12-01T18:23:19 *** aaaaaaaaa has joined #aichallenge 2010-12-01T18:23:35 * Error323 grabs the champange 2010-12-01T18:23:38 aaaaaaaaa: quite a nick 2010-12-01T18:23:39 AI challenge is way bigger than NY's 2010-12-01T18:23:48 rebelxt: true xD 2010-12-01T18:23:49 Error323: Yes but with a lot more of emotion 2010-12-01T18:24:01 I'll shed tears yes 2010-12-01T18:24:03 *** aaaaaaaaa has quit IRC (Client Quit) 2010-12-01T18:24:07 if I lose to mistmanov 2010-12-01T18:24:08 :P 2010-12-01T18:24:18 eitherway 2010-12-01T18:24:23 I'll need alcohol :P 2010-12-01T18:24:26 haha 2010-12-01T18:24:57 I've to put a beer in the fridge! 2010-12-01T18:25:04 Error323: imagine you come to the store and everything is sold out (to Mistmanov) 2010-12-01T18:25:14 stock up 2010-12-01T18:25:37 I _will_ hunt him down 2010-12-01T18:25:44 and put his head on a stick 2010-12-01T18:25:53 gently of course 2010-12-01T18:26:00 haha 2010-12-01T18:26:00 the thing I don't understand.. is, aren't teh results ALREADY up? why's it gona take an hour? 2010-12-01T18:26:13 canadians chiting 2010-12-01T18:26:18 lmao 2010-12-01T18:26:18 it takes its time 2010-12-01T18:26:25 burny: prepare the site 2010-12-01T18:26:29 etc 2010-12-01T18:26:32 burny the rankings haven't been updated in last 5 hrs 2010-12-01T18:26:32 what's chiting mean? 2010-12-01T18:26:32 burny, results need to be manipulated by hand a bit.. lots of people requested this 2010-12-01T18:26:45 and yea accoun needs to be moved 9999 2010-12-01T18:26:45 prepare the clowns 2010-12-01T18:26:48 manipulated by hand? 2010-12-01T18:26:52 burny: yes 2010-12-01T18:27:02 yea it was decided that people most active on IRC should wint he competition 2010-12-01T18:27:14 esp ppl from canadia 2010-12-01T18:27:18 unless they are already nr 1 2010-12-01T18:27:19 ahh 2010-12-01T18:27:20 so everyone is being ranked by amount of time spent in IRC 2010-12-01T18:27:21 yes, you should have filled out a rating adjustment request form 2010-12-01T18:27:24 *** felixcoto has joined #aichallenge 2010-12-01T18:27:34 contestbot keeps track of time spent 2010-12-01T18:27:35 bhasker: No! 2010-12-01T18:27:38 it was mailed to those in irc 2010-12-01T18:27:41 lol 2010-12-01T18:27:47 lol contestbot 2010-12-01T18:27:47 ;) 2010-12-01T18:27:50 lmao 2010-12-01T18:27:52 FelixCoto: Great competion. 2010-12-01T18:27:52 @timespent 2010-12-01T18:27:53 Meatkat: No! 2010-12-01T18:28:03 competition, sorry 2010-12-01T18:28:05 @timewasted 2010-12-01T18:28:06 rebelxt: No! 2010-12-01T18:28:09 @timespentwastingmylifeinIRC 2010-12-01T18:28:10 Meatkat: I'm sorry Dave, err Meatkat; I cannot 'timespentwastingmylifeinIRC'. 2010-12-01T18:28:26 @chiting 2010-12-01T18:28:27 Error323: I do not know about 'chiting', but I do know about these similar topics: 'closing' 2010-12-01T18:28:42 its not the same 2010-12-01T18:28:55 Felicidades a ti tambien :P Congratulations!!! 2010-12-01T18:29:11 oohh, i think the results are public on wikileaks 2010-12-01T18:29:21 BaronTrozo: haha 2010-12-01T18:29:31 @learn chiting=if country!="canadian" : rank="-999999" 2010-12-01T18:29:31 Error323: (learn [] as ) -- Associates with . is only necessary if the message isn't sent on the channel itself. The word 'as' is necessary to separate the key from the value. It can be changed to another word via the learnSeparator registry value. 2010-12-01T18:29:32 :) 2010-12-01T18:29:39 Accoun -> 437 2010-12-01T18:30:11 @predict 2010-12-01T18:30:11 I predict that superflit will win! 2010-12-01T18:30:22 @predict 2010-12-01T18:30:22 I predict that bss03-AFK will win! 2010-12-01T18:30:31 @predict 2010-12-01T18:30:31 I predict that xxx_ will win! 2010-12-01T18:30:37 @predict 2010-12-01T18:30:37 I predict that pgpaskar_ will win! 2010-12-01T18:30:37 ... 2010-12-01T18:30:42 @predict 2010-12-01T18:30:42 I predict that Tobu will win! 2010-12-01T18:30:42 @predict 2010-12-01T18:30:43 I predict that pyro- will win! 2010-12-01T18:30:44 I feel like I'm coming up next 2010-12-01T18:30:45 @predict 2010-12-01T18:30:45 I predict that delt0r_ will win! 2010-12-01T18:30:47 @predict 2010-12-01T18:30:47 I predict that chl_ will win! 2010-12-01T18:30:47 @predict 2010-12-01T18:30:48 I predict that GreenTea_ will win! 2010-12-01T18:30:51 @predict 2010-12-01T18:30:52 I predict that Migi32 will win! 2010-12-01T18:30:55 @predict 2010-12-01T18:30:55 I predict that Tobu will win! 2010-12-01T18:31:03 @learn winner bhasker 2010-12-01T18:31:04 bhasker: (learn [] as ) -- Associates with . is only necessary if the message isn't sent on the channel itself. The word 'as' is necessary to separate the key from the value. It can be changed to another word via the learnSeparator registry value. 2010-12-01T18:31:20 @learn winner as bhasker 2010-12-01T18:31:20 bhasker: I come to serve. 2010-12-01T18:31:22 @winner 2010-12-01T18:31:22 bhasker: lol 2010-12-01T18:31:22 bhasker: winner = bhasker 2010-12-01T18:31:25 @learn predict "as I predict Error323 will win" 2010-12-01T18:31:25 Error323: (learn [] as ) -- Associates with . is only necessary if the message isn't sent on the channel itself. The word 'as' is necessary to separate the key from the value. It can be changed to another word via the learnSeparator registry value. 2010-12-01T18:31:26 lol 2010-12-01T18:31:27 :D 2010-12-01T18:31:33 @learn predict as "I predict Error323 will win" 2010-12-01T18:31:33 Error323: Ready to serve. 2010-12-01T18:31:37 @predict 2010-12-01T18:31:37 I predict that CIA-14 will win! 2010-12-01T18:31:40 crap 2010-12-01T18:31:42 lol 2010-12-01T18:31:45 @winner 2010-12-01T18:31:45 bhasker: winner = bhasker 2010-12-01T18:31:47 :) 2010-12-01T18:31:53 @contestbot 2010-12-01T18:31:54 medrimonia: I have no idea what you mean. 2010-12-01T18:31:54 @winner 2010-12-01T18:31:55 bhasker: winner = bhasker 2010-12-01T18:32:07 What he doesn't know even who he is 2010-12-01T18:32:15 @predict 2010-12-01T18:32:15 I predict that janzert will win! 2010-12-01T18:32:20 fail 2010-12-01T18:32:26 go janzert go! 2010-12-01T18:32:37 @learn predict as "I predict bhasker will win" 2010-12-01T18:32:37 bhasker: I think that worked... 2010-12-01T18:32:40 @predict 2010-12-01T18:32:40 I predict that Cyndre_ will win! 2010-12-01T18:32:43 dammit 2010-12-01T18:32:47 stupid bot 2010-12-01T18:32:48 @predict 2010-12-01T18:32:48 I predict that reportingsjr will win! 2010-12-01T18:33:16 @bribe 100USD 2010-12-01T18:33:17 Meatkat: I have no idea what you mean. 2010-12-01T18:33:26 ;-) 2010-12-01T18:33:39 @contestbot word 2010-12-01T18:33:40 Meatkat: User error, it's not my fault. 2010-12-01T18:33:55 Meatkat: i can guarantee you a top100 spot for 100$ 2010-12-01T18:33:59 @learn chiting as "rank(x) = country(x) == canadian ? +9999 : -9999" 2010-12-01T18:33:59 Error323: OK 2010-12-01T18:34:13 @learn chiting as "rank(x) = country(x) == canada ? +9999 : -9999" 2010-12-01T18:34:13 Error323: Ready to serve, my lord. 2010-12-01T18:34:34 rebelxt: top 1 spot and you've got a deal, you have to make it look legit though 2010-12-01T18:35:21 *** mceier has quit IRC (Quit: leaving) 2010-12-01T18:35:21 Meatkat: this is ez. a database crash due to alien invasion to Canada 2010-12-01T18:35:47 Meatkat: my last game I lost to you >_> 2010-12-01T18:35:52 alien invasion to canada?.. you mean FROM canada? :P 2010-12-01T18:35:58 *** ebrahim has joined #aichallenge 2010-12-01T18:36:02 ;-) 2010-12-01T18:36:21 *** Neverstu has joined #aichallenge 2010-12-01T18:36:57 we need to get over 100 people (err, bots) in the room 2010-12-01T18:37:19 Error323: that you did, it was a pretty odd game 2010-12-01T18:37:48 @chiting 2010-12-01T18:37:49 bhasker: chiting = (#1) rank(x) = country(x) == canadian ? +9999 : -9999, or (#2) rank(x) = country(x) == canada ? +9999 : -9999 2010-12-01T18:38:13 The good news: I'm ranked #1 in my country! (Burkina Faso) 2010-12-01T18:38:35 Are you the son of the president??? 2010-12-01T18:38:38 i'm ranked #1 among people with my player id! 2010-12-01T18:38:46 bad news, ranked last in your country? 2010-12-01T18:38:56 bhasker: also rank = +9999 is bad... 2010-12-01T18:39:09 @burny correct... last time I checked 2010-12-01T18:39:10 Meatkat: An error occured while trying to show the previous error. 2010-12-01T18:39:18 I hope to be ranked #1 of my country too ^^ 2010-12-01T18:39:25 burny: correct... last time I checked* 2010-12-01T18:39:42 BaronTrozo: nope, I'm in the Peace Corps! haha 2010-12-01T18:40:44 If you are on the Peace Corps why your bot is so violent??? 2010-12-01T18:40:57 hahaha 2010-12-01T18:41:10 :) 2010-12-01T18:41:10 must be a believer in peace through universal conquest 2010-12-01T18:41:41 @unlearn 2010-12-01T18:41:42 Error323: I'm sorry Dave, err Error323; I cannot 'unlearn'. 2010-12-01T18:41:43 Your bot migth first try to dialoge 2010-12-01T18:41:46 @remove 2010-12-01T18:41:46 Error323: Error: The command "remove" is available in the Alias, BadWords, Dunno, Herald, Later, News, Quote, RSS, Scheduler, Success, Todo, and Topic plugins. Please specify the plugin whose command you wish to call by using its name as a command before "remove". 2010-12-01T18:42:18 @predict 2010-12-01T18:42:18 I predict that CIA-14 will win! 2010-12-01T18:42:24 hahah 2010-12-01T18:42:29 wth 2010-12-01T18:42:39 @predict 2010-12-01T18:42:39 I predict that BaronTrozo will win! 2010-12-01T18:42:46 Yuju 2010-12-01T18:42:48 love it 2010-12-01T18:42:59 nice one 2010-12-01T18:43:14 =) 2010-12-01T18:43:36 *** xxx_ has quit IRC (Ping timeout: 265 seconds) 2010-12-01T18:43:39 no more pushing F5, what will I do with my life ;_; 2010-12-01T18:44:00 it's over 2010-12-01T18:44:02 There's still only 17 minutes to wait 2010-12-01T18:44:11 @predict 2010-12-01T18:44:11 I predict that yasith will win! 2010-12-01T18:44:46 Error323: Press Alt+F4? 2010-12-01T18:45:09 that's fun just once 2010-12-01T18:45:12 =) 2010-12-01T18:45:31 @predict 2010-12-01T18:45:31 I predict that chl_ will win! 2010-12-01T18:45:36 would be interesting if they calculate total lines of code for all submissions.. must be way over a million 2010-12-01T18:46:15 rebelxt: could be a nice project.. 2010-12-01T18:46:16 I'd like to see a 'dirty hax' rating for the entries as well... I wonder if that could be calculated 2010-12-01T18:46:36 i think the bot code size does not reflect the huge effort invested in the contest 2010-12-01T18:46:48 *** medrimonia1 has joined #aichallenge 2010-12-01T18:47:15 the scm deltas might 2010-12-01T18:47:16 mine has about 1000 lines 2010-12-01T18:47:23 1700 here 2010-12-01T18:47:48 *** medrimonia has quit IRC (Quit: Leaving.) 2010-12-01T18:47:59 500 here 2010-12-01T18:48:01 but maybe the lowest rate of lines/hour in my life 2010-12-01T18:48:41 (F5 pushing must be included in the stats :P) 2010-12-01T18:48:42 ~1K 2010-12-01T18:48:46 Top 10 players: bocsimacko(3802), _iouri_(3585), Slin-.-(3538), jimrogerz(3529), _Astek_(3521), Accoun(3513), george(3508), GreenTea(3508), asavis(3505), protocolocon(3496) 2010-12-01T18:48:52 *** medrimonia1 is now known as medrimonia 2010-12-01T18:49:03 i wonder if we could have had a problem that was directly translatable to a real-world problem 2010-12-01T18:49:26 like, we design bots for a game, but the algorithms can be directly used to, say, most-efficiently route traffic 2010-12-01T18:49:38 like stock market trading? 2010-12-01T18:49:57 but we can't have a competition for that, too boring and non-violent, has to be disguised as a game 2010-12-01T18:51:07 davidjliu: 2010-12-01T18:51:08 Number of player 1 wins: 23 2010-12-01T18:51:08 Number of player 2 wins: 124 2010-12-01T18:51:08 Number of draws: 3 2010-12-01T18:51:26 hmm, wow. bocsimacko uses a bit vector of ship counts. i don't get it. 2010-12-01T18:51:29 Can you run the test for me too Error323? =P 2010-12-01T18:51:34 huh?.. traffic routing is non violent? 2010-12-01T18:51:38 Error323: a strong second half :) 2010-12-01T18:51:42 danke 2010-12-01T18:51:55 medrimonia: I tested against you earlier 2010-12-01T18:52:02 um 2010-12-01T18:52:07 medrimonia: sure 2010-12-01T18:52:15 I won 79, I lost 52, drew 19 2010-12-01T18:52:24 medrimonia: where is ur src? 2010-12-01T18:52:27 davidjliu: thanks 2010-12-01T18:52:33 traffic routing is pretty well researched (VRP class of problems), but there are still tons of interesting problems 2010-12-01T18:52:36 apparently we like to traw 2010-12-01T18:52:38 draw 2010-12-01T18:53:04 http://ai-contest.com/forum/viewtopic.php?f=3&t=1168&start=20#p7858 2010-12-01T18:53:48 davidjliu: yep that makes more than 10 % of draws 2010-12-01T18:53:57 yeah, if we could have an interesting productive problem for the next competition, then it'd be a double-win 2010-12-01T18:54:25 smloh1: agreed. real world problem would be cool 2010-12-01T18:54:30 the end is nigh... 2010-12-01T18:54:51 so is it now a 10 min wait or 1h and 10 min wait 2010-12-01T18:55:07 5 minutes remaining^^ 2010-12-01T18:55:09 *** Appleman1234 has quit IRC (Ping timeout: 276 seconds) 2010-12-01T18:55:23 * Error323 opens a beer 2010-12-01T18:55:24 Mine history is 315 wins 203 losses 13 Draws, over the final contest 2010-12-01T18:55:43 * medrimonia follows Error323 2010-12-01T18:55:44 what about the next challenge? does anyone know similar competitions? =) 2010-12-01T18:56:04 BaronTrozo: that's pretty good, maybe it reflects an unlucky start so you spend a lot of time winning and climbing back up 2010-12-01T18:56:37 I think that i have good luck over the hole contest ;) 2010-12-01T18:56:49 how do you have these stats BaronTrozo? 2010-12-01T18:57:05 manually count from the stats 2010-12-01T18:57:25 waouw 2010-12-01T18:57:29 ouch 2010-12-01T18:57:53 hahaha 2010-12-01T18:58:08 grep ftw 2010-12-01T18:58:48 I have cut and paste the history on UltraEdit (i'm windows user) 2010-12-01T18:59:00 is there an easy way to find out what the worst rank you lost to is? 2010-12-01T18:59:01 ouch 2010-12-01T18:59:03 ^_^ 2010-12-01T18:59:09 wingrep ft 2010-12-01T18:59:16 304wins, 224losses, 3draws 2010-12-01T19:00:02 what's the highest rank, with more losses than wins? 2010-12-01T19:01:19 *** aaaaaaaaaaaa has joined #aichallenge 2010-12-01T19:01:27 i think it wasn't too bad for my bot climbing up since the late bloomers kind of clustered 2010-12-01T19:01:32 *** mogron has joined #aichallenge 2010-12-01T19:01:41 Anyone have any idea what the link will be for the official results 2010-12-01T19:01:49 here is command i've been using: curl http://ai-contest.com/profile_games.php?user_id=5853&page=1 | grep -o Win | wc -w 2010-12-01T19:01:51 &p.m. canadian time and the server shows same stats???? 2010-12-01T19:01:56 change Win/Loss/Draw 2010-12-01T19:02:19 but you have to search also on page 2 and 3 2010-12-01T19:02:34 yeah, 3 pages 2010-12-01T19:02:50 was too lazy to right 5 lines to do it all ;) 2010-12-01T19:02:55 write* 2010-12-01T19:03:26 It is the time and no ranking still available 2010-12-01T19:03:40 *** lavalamp has quit IRC (Quit: Page closed) 2010-12-01T19:04:38 263 wins here 2010-12-01T19:05:05 265 win 252 loss / 5 draw 2010-12-01T19:05:46 ./cry 2010-12-01T19:05:50 my losses are more 2010-12-01T19:05:51 :(( 2010-12-01T19:06:11 bhasker: doesn't mean anything 2010-12-01T19:06:17 they're up!!! 2010-12-01T19:06:20 i guess elo matters 2010-12-01T19:06:21 !!! 2010-12-01T19:06:37 Upvote me! http://www.reddit.com/r/programming/comments/eesxh/google_ai_challenge_final_rankings_released/ 2010-12-01T19:06:41 NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO 2010-12-01T19:06:50 <- 53 2010-12-01T19:06:54 haha rebel 2010-12-01T19:06:54 Lost by 1 elo 2010-12-01T19:06:55 1 elo diff 2010-12-01T19:06:57 FUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUUCK 2010-12-01T19:07:05 i am behind you by 1 elo 2010-12-01T19:07:13 urg, dmj111 beat me :) 2010-12-01T19:07:16 bhasker: wow just saw it 2010-12-01T19:07:26 but otoh i beat all the bots from india 2010-12-01T19:07:26 lol 2010-12-01T19:07:28 for you hacker news types http://news.ycombinator.com/item?id=1960338 2010-12-01T19:07:30 28! even a tie for it 2010-12-01T19:07:33 I haven't moved 2010-12-01T19:07:35 the top one is 1 rank below me 2010-12-01T19:07:37 *** ebrahim has quit IRC (Remote host closed the connection) 2010-12-01T19:07:37 *** superflit has quit IRC (Quit: superflit) 2010-12-01T19:07:39 yep nice job 2010-12-01T19:07:41 and 2 elo below me 2010-12-01T19:08:08 i'm amazed at how close these scores are 2010-12-01T19:08:14 http://www.reddit.com/r/programming/comments/eesxh/google_ai_challenge_final_rankings_released/ upvote the reddit thread so that the word gets out! 2010-12-01T19:08:20 ah wagstaff swapped places with me 2010-12-01T19:08:22 Oh you have got to be kidding me with this 2010-12-01T19:08:26 :( 2010-12-01T19:08:28 Error323... we tied! 2010-12-01T19:08:34 *** fgump has quit IRC (Ping timeout: 255 seconds) 2010-12-01T19:08:41 http://ai-contest.com/country_profile.php?country_id=167 :( 2010-12-01T19:08:50 Meatkat: indd :) 2010-12-01T19:08:55 71, better than expected :) 2010-12-01T19:08:55 Error323: Mistmanov came back from behind?.. 2010-12-01T19:09:12 WITH 1 F*%$ING ELO POINT 2010-12-01T19:09:19 #2 in canada! 2010-12-01T19:09:24 Been ahead of him all the time 2010-12-01T19:09:25 grats neverstu 2010-12-01T19:09:27 15 no bad for a Baron of last century 2010-12-01T19:09:28 * Neverstu holds hands with _iouri_ 2010-12-01T19:09:30 thanks! 2010-12-01T19:09:46 you are #3 according to final ranks 2010-12-01T19:09:53 mangara seems to be ahead 2010-12-01T19:09:53 whoops #3, sorry mangara! 2010-12-01T19:10:11 too much caffeine for me 2010-12-01T19:10:24 * Error323 dies 2010-12-01T19:10:30 e323 there there 2010-12-01T19:11:00 woo i tied for 949 2010-12-01T19:11:07 davidjliu: i can't believe how often you and dmj111 were ranked side by side over the course of the competition 2010-12-01T19:11:55 http://www.reddit.com/r/programming/comments/eesxh/google_ai_challenge_final_rankings_released/ upvoooootes! 2010-12-01T19:12:01 ohhh Meatkat ur last game vs me causes me pain :P 2010-12-01T19:12:04 =) #3 of python 2010-12-01T19:12:09 davidjliu: it always made me do a double-take 2010-12-01T19:12:09 haha I am such a karma whore. 2010-12-01T19:12:27 j3camero: don't have an account srry 2010-12-01T19:12:27 Error323... yeah our bots weren't doing really well 2010-12-01T19:12:29 it's pretty exciting. even though I lost 2010-12-01T19:12:38 Error323: I just made one :) 2010-12-01T19:12:45 250win 11tie 251loss 2010-12-01T19:12:55 so much chiting 2010-12-01T19:13:10 j3camero: upvoted 2010-12-01T19:13:16 Accoun doesn't have an account? 2010-12-01T19:13:33 rebelxt: thanks! I must have ALL the karmas... 2010-12-01T19:13:35 Meatkat: we will have our revenge 2010-12-01T19:13:37 =) 2010-12-01T19:13:39 *** eburnette has joined #aichallenge 2010-12-01T19:13:49 Congratulations everyone! 2010-12-01T19:14:02 bhask: you had more losses than wins? what rank? 2010-12-01T19:14:04 Yes congratulations! 2010-12-01T19:14:25 I hope you all enjoyed the contest. You'll all welcome to help build the next one! We generally just add anyone that asks to the project committers so don't be shy. 2010-12-01T19:14:27 and thx devs! j3camero, janzert, amstan etc 2010-12-01T19:14:39 Thanks to all organizers! 2010-12-01T19:14:41 see you all in next contest !! a pleasure to know you all !! 2010-12-01T19:14:43 yeah, thanks!! 2010-12-01T19:14:43 aaaaaaawesomenes 2010-12-01T19:14:49 thanks guys for great contest! 2010-12-01T19:14:54 well, congratulations All! this was fun 2010-12-01T19:15:06 @rainbow FIREWORKS 2010-12-01T19:15:06 Error323: FIREWORKS 2010-12-01T19:15:06 j3camero: thanks for nice time and great contest! 2010-12-01T19:15:14 yay :) 2010-12-01T19:15:19 thanks everyone for participating 2010-12-01T19:15:22 bdumzz! 2010-12-01T19:15:32 and yeah janzert, j3camero, you guys rock! 2010-12-01T19:15:40 Janzert is the man by the way. 2010-12-01T19:15:40 FelixCoto: Congratulations, you really don't know how much i suffer trying to win your bot and reaching 2º place of spanish rank 2010-12-01T19:15:48 lol at fireworks 2010-12-01T19:16:01 @rainbow Congratulations! 2010-12-01T19:16:01 eburnette: Congratulations! 2010-12-01T19:16:02 I'm off for a bit of a break, I'll check back in and make sure the server hasn't melted or anything a little later 2010-12-01T19:16:14 cya! 2010-12-01T19:16:24 I'm gonna cry myself to sleep 2010-12-01T19:16:27 54 burny 2010-12-01T19:16:31 knowing i got second by one elo 2010-12-01T19:16:33 lalala 2010-12-01T19:16:35 congrats all 2010-12-01T19:16:36 felixcoto: you were CCCF on TCP server, right? 2010-12-01T19:16:41 @BaronTrozo Thanks, your bot make me suffer at final 2010-12-01T19:16:42 felixcoto: An error occured while trying to show the previous error. 2010-12-01T19:16:59 bhasker: 54 is still quite good right? 2010-12-01T19:17:02 nice 2010-12-01T19:17:11 yes i am happy 2010-12-01T19:17:11 rebelxt: yes i was cccf at TCP 2010-12-01T19:17:11 medrimonia: you beat me! but i'm glad i managed to stay in the to 20. 2010-12-01T19:17:28 felixcoto: cool, we guessed it right ;-) 2010-12-01T19:17:37 this was crazily competitive. i'd say top 100 is all pretty amazing 2010-12-01T19:17:38 149 ftw! 2010-12-01T19:17:39 smloh1: Yep at least none of us has gone outside of top 20 =) 2010-12-01T19:17:40 medrimonia: next time we will overcome the rank #11 curse! 2010-12-01T19:17:49 mega1 post your blog entry now 2010-12-01T19:17:51 congrats:) 2010-12-01T19:17:55 Next time we'll make it to the top 10 together =P 2010-12-01T19:18:00 haha 2010-12-01T19:18:08 the contest has been incredibly adictive, my congrats to the organizers 2010-12-01T19:18:09 I will make awesomeness next time 2010-12-01T19:18:22 Addictive by design, for sure. 2010-12-01T19:18:23 next time we will all write in lisp 2010-12-01T19:18:27 *** Cold-Phoenix has quit IRC () 2010-12-01T19:18:33 lol 2010-12-01T19:18:36 :) 2010-12-01T19:18:42 Im too) 2010-12-01T19:18:43 not lisp, forth 2010-12-01T19:18:44 *** xxx has joined #aichallenge 2010-12-01T19:18:51 or Haskell.. 2010-12-01T19:18:51 *** mogron has quit IRC (Quit: Page closed) 2010-12-01T19:18:57 next time I'll use structures!!! 2010-12-01T19:19:02 bhasker: i was hoping we could sneak into top50 :( oh well, pretty close... chiting FTL ;) 2010-12-01T19:19:12 xD 2010-12-01T19:19:13 Dude it took all the maturity I have to keep from making a subtle backhanded crack about Lisp in the announcement text. Would have loved to see the outrage that would have generated on Reddit. 2010-12-01T19:19:14 hah 2010-12-01T19:19:15 i started learning haskell yesterday 2010-12-01T19:19:17 *** aaaaaaaaaaaa has quit IRC (Quit: Page closed) 2010-12-01T19:19:19 just for the next contest 2010-12-01T19:19:22 or the binary language of moisture evaporators 2010-12-01T19:19:23 :P 2010-12-01T19:19:28 next time I'll use "for" loops! 2010-12-01T19:19:33 yea i was hoping the same rebel 2010-12-01T19:19:37 actually i was tied with dabino 2010-12-01T19:19:45 for and hazard multiple times at 49 2010-12-01T19:19:47 yesterday 2010-12-01T19:19:53 sadly string of losses today screwed it up 2010-12-01T19:20:24 oh well but only 6 elo diff between us and 49 2010-12-01T19:20:29 so i am not too sad :) 2010-12-01T19:20:40 yeah, crazy close 2010-12-01T19:20:52 yup 2010-12-01T19:20:54 well, congratulations to all people, I see you in the next contest 2010-12-01T19:21:10 probably with CCCF as name :P 2010-12-01T19:21:16 yep cya soon 2010-12-01T19:21:28 When't the next contest? 2010-12-01T19:21:33 whats the background behind that name though felix 2010-12-01T19:21:41 *** Migi32 has quit IRC (Quit: www.ai-contest.com) 2010-12-01T19:21:54 It's a joke with my friends 2010-12-01T19:22:03 *** yasith has quit IRC (Ping timeout: 272 seconds) 2010-12-01T19:22:04 Centro cultural casa Felix 2010-12-01T19:22:13 yeah, good competing with you, felixcoto. i was always happy whenever my bot managed to beat yours 2010-12-01T19:22:20 =) 2010-12-01T19:22:36 protocolocon: congrats for 1 spanish 2010-12-01T19:23:03 Who were oldest and youngest participants, anyone know? 2010-12-01T19:23:06 protocolocon: also congratulations 2010-12-01T19:23:27 The next contest is as soon as possible. Whenever we can get the next one up it will be here. If you want it to be sooner, you can help! It's all open source. 2010-12-01T19:23:37 #8 python bot yay 2010-12-01T19:23:37 felixcoto, thanks, greetings from enigmatic Rufes Band :p 2010-12-01T19:23:45 cool 2010-12-01T19:23:54 So more often than yearly j3camero? 2010-12-01T19:24:03 Rufes Band 1º group, Yuju 2010-12-01T19:24:16 *** felixcoto has quit IRC (Quit: Page closed) 2010-12-01T19:25:24 davidjliu: that is really too close to call. nice work. 2010-12-01T19:25:32 congrats everyone. 2010-12-01T19:25:44 :) 2010-12-01T19:25:50 dmj111: heh, thanks. nice job to you too, Mr. python master 2010-12-01T19:25:58 *** Appleman1234 has joined #aichallenge 2010-12-01T19:26:45 seriously, when is everyone going to figure out we are the same person? 2010-12-01T19:26:54 bahahahah..... 2010-12-01T19:27:09 dmj111: i knew it!!! 2010-12-01T19:27:13 lol 2010-12-01T19:27:14 loool 2010-12-01T19:27:19 oO 2010-12-01T19:28:20 so UW organized it.. and got 0 people in the top100? for shame :P 2010-12-01T19:28:39 Congrats everyone! See you next time 2010-12-01T19:28:41 wait, we are? I was unaware :) 2010-12-01T19:29:10 oh, right... sorry, i forgot :) again, 4 points just means i am lucky. 2010-12-01T19:29:10 Mr. Smith, I presume? 2010-12-01T19:29:47 are there any teams in top10? looks like individual only 2010-12-01T19:31:01 *** Meatkat has quit IRC (Quit: Leaving) 2010-12-01T19:31:15 here is too late and i will try to recover all lost dreams hours, so congratulations to all 2010-12-01T19:32:01 time to celebrate. later all 2010-12-01T19:32:11 Thanks to all who write a bot which I can win) Especially to you dmj111. So long time you was on first places before bocsi.. 2010-12-01T19:32:32 Good luck in next competition! 2010-12-01T19:32:43 yeah who pioneered supply lines? i just imitated those shamelessly! :) 2010-12-01T19:32:53 oh yeah me too 2010-12-01T19:33:02 kudos to hiim =) 2010-12-01T19:33:06 *** row_ has joined #aichallenge 2010-12-01T19:33:06 or her 2010-12-01T19:33:13 ) 2010-12-01T19:33:29 *** BaronTrozo has quit IRC (Quit: Page closed) 2010-12-01T19:33:32 *** rebelxt has left #aichallenge 2010-12-01T19:33:39 yeah it's one of the few fundamentally dominant strategies 2010-12-01T19:33:56 indeed 2010-12-01T19:34:06 next to synchronized attacks/captures 2010-12-01T19:34:13 also very nice 2010-12-01T19:34:27 Makes pretty pictures too 2010-12-01T19:35:07 *** GreenTea_ has quit IRC (Quit: Page closed) 2010-12-01T19:35:07 still 2010-12-01T19:35:10 one elo 2010-12-01T19:35:16 hehe 2010-12-01T19:35:18 this will haunt me forever 2010-12-01T19:35:29 *** yasith has joined #aichallenge 2010-12-01T19:35:30 GreenTea_: thanks. it was nice competing with everyone. 2010-12-01T19:35:39 * eburnette gives E323 1 elo 2010-12-01T19:35:46 :D 2010-12-01T19:35:49 haha 2010-12-01T19:36:25 http://www.reddit.com/r/programming/comments/eesxh/google_ai_challenge_final_rankings_released/ 2010-12-01T19:36:29 Moar upvotes! 2010-12-01T19:36:45 Gotta get to the front page of Reddit so that we get more people competing next time! 2010-12-01T19:36:51 j3camero: you had me at reddit. first time i logged on to my account in about three years. 2010-12-01T19:37:20 You must be an old Reddit user! I have probably only been on for less than a year. 2010-12-01T19:37:42 yikes. showing my age. i think i even used it when it was still lisp. 2010-12-01T19:37:50 hahah 2010-12-01T19:37:51 I should write another article on this 2010-12-01T19:38:16 j3camero: you sure your servers will appreciate that? =) 2010-12-01T19:38:25 They'll be fine. 2010-12-01T19:38:32 eburnette: link to an earlier article? 2010-12-01T19:39:40 still why is reddit so horribly ugly 2010-12-01T19:39:44 lol 2010-12-01T19:40:17 down vote everything above googleaicontest ^_^ 2010-12-01T19:40:41 :} 2010-12-01T19:40:45 Looks like I got to play 3 of the top 10 in the last couple of days, 1 win 2 losses: http://ai-contest.com/visualizer.php?game_id=9494648 , http://ai-contest.com/visualizer.php?game_id=9448802 , and http://ai-contest.com/visualizer.php?game_id=9057777 2010-12-01T19:41:44 Previous article was http://www.zdnet.com/blog/burnette/its-youth-vs-experience-in-the-google-ai-challenge-and-the-kids-are-winning/2123 2010-12-01T19:42:21 So I was wondering who were oldest and youngest competitors, I heard of one 14yr old 2010-12-01T19:43:29 um, I'm 17 2010-12-01T19:44:03 there goes my twin idea. 2010-12-01T19:44:09 hahaha 2010-12-01T19:44:23 i'm 27 2010-12-01T19:44:28 -,- 2010-12-01T19:44:40 I feel very... immature now. heh 2010-12-01T19:45:33 I'm 47, does that make me the oldest? (prays not) 2010-12-01T19:45:48 phew. now I can admit that I am 35. 2010-12-01T19:45:56 lol 2010-12-01T19:45:59 lol 2010-12-01T19:46:05 wait, its not 1998 anymore? 2010-12-01T19:47:16 want:: http://shop.angrybirds.com/collections/frontpage/products/8-red-bird 2010-12-01T19:48:01 *** row_ has quit IRC (Quit: Page closed) 2010-12-01T19:48:47 Top 10 players: bocsimacko(3765), _iouri_(3565), Slin-.-(3524), _Astek_(3501), jimrogerz(3500), Accoun(3498), george(3494), GreenTea(3489), asavis(3480), bix0r4ever(3476) 2010-12-01T19:49:04 *** Neverstu has quit IRC () 2010-12-01T19:49:08 man.. in the top 50, the bots that I beat the most were both top10 2010-12-01T19:49:55 * eburnette thinks burny missed his calling as a PR spin expert 2010-12-01T19:50:30 just like bocs lost to someof the 200-400 rank bots.. 2010-12-01T19:50:51 my bot did bad against the bots that're worse than top10.. 2010-12-01T19:51:03 cearly I needed better competition to prevail :) 2010-12-01T19:51:36 Anybody know real affiliation of bocsimacko, iouri, Astek, Account, george, asavis, and bix0r4ever? Web site just says "other". 2010-12-01T19:52:08 rock! 2010-12-01T19:52:14 congrats everybody! 2010-12-01T19:53:13 eburnette: I'm 15 at your youngest competitor thing 2010-12-01T19:53:37 cool, is this your first contest? 2010-12-01T19:53:48 Second 2010-12-01T19:53:58 Did tron before? 2010-12-01T19:54:10 Yup, got disqualified due to a timeout on that one 2010-12-01T19:54:13 Whoa 15 that's awesome. 2010-12-01T19:54:17 91st here 2010-12-01T19:54:54 nice, self taught programmer or did you have classes? 2010-12-01T19:54:56 *** tapwater has quit IRC (Quit: tapwater) 2010-12-01T19:54:59 Self-taught 2010-12-01T19:55:21 I have a programming class at school but all we do is visual basic 2010-12-01T19:56:14 makin' guis to track ip addresses? 2010-12-01T19:56:30 Exactly. 2010-12-01T19:57:06 * Zannick happy with his 949th place and a credit on the thanks page :) 2010-12-01T19:57:45 Any android enthusiasts here? 2010-12-01T19:58:45 Anybody besides me get in trouble for spending *way* too much time on the contest? 2010-12-01T19:58:59 I got in trouble with myself for not spending enough helping out 2010-12-01T19:59:03 But that's a different story. 2010-12-01T19:59:04 probably anyone married 2010-12-01T19:59:12 i'm pretty sure there were a few 2010-12-01T19:59:23 someone said my name 2010-12-01T19:59:39 Haha, my wife just walked in and says "I thought that thing was over!" 2010-12-01T19:59:41 how are you jm 2010-12-01T19:59:42 Cyndre_: /lastlog -hilight 2010-12-01T19:59:52 nevermind, xchat, dunno how that works 2010-12-01T20:00:02 ha ha eburn :P 2010-12-01T20:00:31 oh that was contestbot talking out his ass - I think Cyndre_ will win 2010-12-01T20:00:32 lol 2010-12-01T20:01:17 Hey, how did that GA bot do? 2010-12-01T20:01:21 my startup took off a few weeks into the contest and didnt do anything on it since :( 2010-12-01T20:01:31 *** flag has quit IRC (Quit: Lost terminal) 2010-12-01T20:01:59 *** flag has joined #aichallenge 2010-12-01T20:02:27 i'm 33, and i would sold my soul to have a contest like this when i was 15, very impressive flag 2010-12-01T20:02:39 no kidding meneky - I am 30 2010-12-01T20:02:59 same here meneky 2010-12-01T20:03:19 when I was in school, they mostly only had math contests :\ 2010-12-01T20:03:23 meneky: did you hear about last years? 2010-12-01T20:03:29 meneky: thanks 2010-12-01T20:03:44 burny: top 100 in canada and top 25 in alberta during grade 9 math tests 2010-12-01T20:04:09 vn flag, ranked 95 myself (that is 1195) 2010-12-01T20:04:13 how many different ones did you do? 2010-12-01T20:04:23 Cyndre_ : no, this is my first ai contest, and began too late :( 2010-12-01T20:04:27 burny: only the grade 9 one 2010-12-01T20:04:39 ah.. I did a whole bunch of them 2010-12-01T20:04:42 I started tron late, but started this one on time but drug away for last month of contest 2010-12-01T20:04:52 Are you in grade 9 Cyndre? 2010-12-01T20:05:01 nope, 30 flag 2010-12-01T20:05:07 Oh, lol 2010-12-01T20:05:11 II thought I was already young with my 20^^ 2010-12-01T20:05:16 in todays day and age I would probably not even place top 500 though 2010-12-01T20:05:32 I found a picture of mega1 at http://media.linkedin.com/mpr/mpr/shrink_80_80/p/3/000/053/091/12b232f.jpg if anybody's interested 2010-12-01T20:06:10 Oct 3rd 18:57:01 was my last upload - good or bad that Im 1100 now? 2010-12-01T20:06:20 well 1200 2010-12-01T20:06:32 October, wow, I'd say 1100 was not too shabby 2010-12-01T20:07:18 ooo they added ships per turn to the graph 2010-12-01T20:08:41 But flag, you've got programmation classes at school and you're 15? 2010-12-01T20:08:43 lmao 9 uploads sept 30, to 1 upload oct 3rd and then dead 2010-12-01T20:09:51 Introductory class in highschool, only teaches us visual basic and how to use dreamweaver 2010-12-01T20:09:59 i remember when we had programming competitions when i was in school... in Basic 2010-12-01T20:10:04 It's really quite useless 2010-12-01T20:10:36 Before university I hadn't any programation classes 2010-12-01T20:11:00 Just a very short introduction to mathlab but that's all 2010-12-01T20:11:09 my firs tupload was nov22 :( 2010-12-01T20:11:26 I don't think they start to teach anything complex until university here anyway, but they give us a start I suppose 2010-12-01T20:11:40 i uploaded sep 8th, sep 11th, and nov 23rd 2010-12-01T20:11:42 I'd rather they had an electronics class tbh 2010-12-01T20:11:59 (i uploaded a bot i hadn't touched in months in the last week. oh well) 2010-12-01T20:12:31 how'd it do a1k0n? 2010-12-01T20:12:38 In fact, I haven't really learn anything complexe in my univeristy 2010-12-01T20:12:56 #441. and i'm not doing anything very different from iouri. it's amazing how the details matter. 2010-12-01T20:12:59 man.. I only worked on the thing for 4days :\ 2010-12-01T20:13:13 eburnette: my wife is more subtle, she said "Let's celebrate!" but i'm pretty sure it means the same thing... 2010-12-01T20:13:41 have you told her about the next contest yet smloh1? 2010-12-01T20:13:47 i have a couple other versions of the bot which might have done better in the final contest but i'm not sure. it's weird how A beats B beats C beats A 2010-12-01T20:14:06 smloh1 and eburnette: My girlfriend is sleeping but I'm sure that she's happy that the contest has ended 2010-12-01T20:14:55 its over? 2010-12-01T20:14:57 alk0n_ try this for fun: click on any profile, look at their last lost game, click on the profile of who they lost to, and repeat. 2010-12-01T20:15:04 next contest? heheh i'm not sure this is a good time to broach that topic 2010-12-01T20:15:11 my days are finally going back to 8 hours and I could have time to work on the contest now 2010-12-01T20:15:30 *** Mathnerd314 has quit IRC (Ping timeout: 260 seconds) 2010-12-01T20:15:36 There's an Arimaa tourney coming up in Jan 2010-12-01T20:15:57 I'll have free time to.... work for my exams 2010-12-01T20:16:18 arimaa is the one that uses chess pieces? 2010-12-01T20:16:34 burny: yep 2010-12-01T20:16:59 eburnette: heh, I did that when I was bored 2010-12-01T20:17:04 er, the clicking thing 2010-12-01T20:17:27 now Im watching arimaa 2010-12-01T20:17:31 Gábor Melis has participated in several other computer AI contests including the Computer Games Olympiad. http://webdocs.cs.ualberta.ca/~hayward/hex/cgo/beijing.pdf 2010-12-01T20:17:50 *** McLeopold has quit IRC (Quit: Page closed) 2010-12-01T20:18:05 you from edmonton eburnette? 2010-12-01T20:18:50 North Carolina 2010-12-01T20:19:00 cool 2010-12-01T20:19:05 so who won? 2010-12-01T20:19:26 "The level of Hex play has improved signicantly since the previous Hex competition in 2006, when Six (Melis' program) soundly defeatedWolve and HexKriger for the gold medal at the 11th Computer Games Olympiad in Torino." 2010-12-01T20:20:28 *** virdo has joined #aichallenge 2010-12-01T20:20:47 I'm interested in doing some AI for GO.. but don't have enough spare time for that kind of thing 2010-12-01T20:23:12 More on mega1's Hex program at http://six.retes.hu/ . Looks like he won the gold in 2003, 2004, and 2006. 2010-12-01T20:23:14 I wanted to do Go for this AI contest, but people shot it down hard. 2010-12-01T20:23:39 go is WAY too hard for something liek this(not to mention, it's already studied) 2010-12-01T20:24:00 j3camero because it's been done so much? 2010-12-01T20:24:21 I started coding the engine, but the biggest problem I encountered was the victory condition. There is actually no victory condition. 2010-12-01T20:24:24 if go was a brand new game.. it could be appropriate though 2010-12-01T20:24:26 I started coding the engine, but the biggest problem I encountered was the victory condition. There is actually no victory condition. 2010-12-01T20:24:30 *** bhasker has quit IRC (Quit: bhasker) 2010-12-01T20:24:39 j3camero how did you get involved in this contest anyway? 2010-12-01T20:24:44 In general, there is no way to end the game unless both players agree to end the game. 2010-12-01T20:25:01 lmao - http://steve-yegge.blogspot.com/2010/12/haskell-researchers-announce-discovery.html 2010-12-01T20:25:05 And there is no good starter package strategy for that. 2010-12-01T20:25:16 eburnette: I invented it :-) 2010-12-01T20:25:58 Was tron the first one? 2010-12-01T20:26:34 LOL Cyndre_ 2010-12-01T20:26:50 funniest post I have read in a while 2010-12-01T20:26:58 eburnette: no. There were two before Tron. 2010-12-01T20:27:22 eburnette: Minesweeper Flags, Rock Paper Scissors, Tron, and Planet Wars. 2010-12-01T20:27:28 brb 2010-12-01T20:27:47 Moar upvotes ppl! http://www.reddit.com/r/programming/comments/eesxh/google_ai_challenge_final_rankings_released/ 2010-12-01T20:27:49 Is there any plan on when the next contest will be? 2010-12-01T20:27:58 ...Rock Paper Scissors? 2010-12-01T20:28:01 flag: asap. Help us code it! 2010-12-01T20:28:08 Oh yeah I've been meaning to ask, how do you program an AI for RPS? Isn't it random? 2010-12-01T20:28:15 First step is to choose a game :-) 2010-12-01T20:28:43 Alright, I'll probably have some time during the winter break, I'll see what I can do 2010-12-01T20:28:47 statego 2010-12-01T20:29:03 did people seriously try rps?.. or was there some sort of twist to it? 2010-12-01T20:29:03 risk would be fun to 2010-12-01T20:29:34 er stratego not statego 2010-12-01T20:29:57 sports game please, soccer or the like, or at least something funny to replay 2010-12-01T20:30:02 How about a tower defense game 2010-12-01T20:30:07 burny: RPS is harder than you think. It's not random. It's iterated RPS. 2010-12-01T20:30:24 it must be funny to replay and addictive, its key to success 2010-12-01T20:30:39 meneky, agreed 2010-12-01T20:30:47 why not a bot battle - with shooting the enemy into submission 2010-12-01T20:31:21 iterated? 2010-12-01T20:31:30 timed grenades that go off x turns after placing and you only start with 10 2010-12-01T20:31:32 Angry Birds? Or maybe a tank game like iShoot. 2010-12-01T20:32:01 ishoot looks like a total ripoff of scorced earth 2010-12-01T20:32:07 i think another key to success is an small learning curve, newbies must be able to make a bot fast 2010-12-01T20:32:17 starter packs help there 2010-12-01T20:32:28 more like atari tank with mines or something 2010-12-01T20:32:58 Yeah, atari tanks rock, good idea 2010-12-01T20:33:10 Cyndre_: very good idea 2010-12-01T20:33:11 2 track speeds, 360 degree turret, fire gun, lay mine for x turns 2010-12-01T20:33:28 oh.. the atari tank game, with bouncy walls, and invisible tanks could be quite cool 2010-12-01T20:33:37 and fun to watch 2010-12-01T20:33:55 the only problem with tank.. would be processing time/game duration 2010-12-01T20:34:02 There's a nice Wii version of the tank game I like 2010-12-01T20:34:03 could have x health, y ammo, and z mines with random popups 2010-12-01T20:34:35 destroyable walls? different kinds of ammo? 2010-12-01T20:35:00 depends how far you want to go - I was thinking just simple gun and mines that you program to go off in x turns 2010-12-01T20:35:03 a game with a total time limit, in addition to a per-turn time limit, could be nice 2010-12-01T20:35:14 60 second timelimit and you lose 1 health per turn 2010-12-01T20:35:19 start with 60 health 2010-12-01T20:36:26 real time play might be a problem 2010-12-01T20:36:32 i think is better to keep the concept as simple as possible, as few rules as possible to make it complex enought to be a challenge 2010-12-01T20:37:03 well its not realtime play, they can move x distance for every 1 second turn (and maybe just tank with most health after 60 turns wins) 2010-12-01T20:37:06 * eburnette votes for predetermined maps 2010-12-01T20:37:43 premade maps, with random/unknown starting spot! 2010-12-01T20:38:14 Yeah, I think the maps should be hand drawn, not generated too. 2010-12-01T20:38:58 or capture the flag would be fun and simple 2010-12-01T20:39:16 I'd like non symmetric maps 2010-12-01T20:39:21 an AI challenge that would be kind of interesting(from teh AI perspective).. 2010-12-01T20:39:47 Hey maybe the US army could co-sponsor :) 2010-12-01T20:39:59 for the tank wars? 2010-12-01T20:40:19 yeah 2010-12-01T20:40:35 is limited to a total of 1meg of internet traffic, 1meg of code/data. and given a list of words, and 30seconds. pick a word from the list, plus any other one word, and get the most google hits possible 2010-12-01T20:40:42 eburnette: in all honesty whats the difference if its a sensor saying obstacle in front or a function check?> 2010-12-01T20:41:40 well hopefully you can look around more than just in front 2010-12-01T20:42:20 but fog of war would be interesting twist, limited knowledge of what's going on away from you (instead of perfect planetwars knowledge) 2010-12-01T20:42:26 yea, but meant if you check for stuff with a camera or by asking the server for the map - really its the same thing, your just making decisions with other peoples info 2010-12-01T20:43:00 some of the stuff they are doing with the kinect is awsome fyi 2010-12-01T20:43:27 sure, of course if you're driving a physical object you have innacuracies to deal with 2010-12-01T20:44:04 something with lego mindstorm perhaps? 2010-12-01T20:44:18 and who else watched the video jeff hawkins (I am pretty sure thats who) did on neural networks? 2010-12-01T20:45:03 In tank game would your own mines blow you up? 2010-12-01T20:45:27 yes, they could either blow up after x turns, or blow up when moved onto them (undetectable) 2010-12-01T20:45:33 so remember where you place yours 2010-12-01T20:45:51 *** winexception has joined #aichallenge 2010-12-01T20:46:09 http://www.numenta.com/htm-overview/education/jeff-2010-smithgroup-lecture.php this was an awsome video 2010-12-01T20:46:21 *** protocolocon has quit IRC (Quit: Leaving) 2010-12-01T20:46:51 *** meduza has quit IRC (Ping timeout: 265 seconds) 2010-12-01T20:47:18 *** Cyndre_ is now known as Cyndre 2010-12-01T20:47:31 the audio quality is pure win 2010-12-01T20:47:58 mtbd 2010-12-01T20:48:07 thats the encoding they used for it 2010-12-01T20:48:47 Top 10 players: bocsimacko(3765), _iouri_(3565), Slin-.-(3524), _Astek_(3501), jimrogerz(3500), Accoun(3498), george(3494), GreenTea(3489), asavis(3480), bix0r4ever(3476) 2010-12-01T20:49:00 in case your wondering what mtbd is - much to be desired 2010-12-01T20:50:53 Why do you want the minds to blow up on their own? strategy could involve building a mine fortress that stays around 2010-12-01T20:51:12 *** smloh1 has quit IRC (Ping timeout: 265 seconds) 2010-12-01T20:51:52 because I feel most mines are indiscriminate on who they blow up 2010-12-01T20:52:41 ok so they blow up anybody but don't time out, how about that 2010-12-01T20:52:48 can you shoot mines? 2010-12-01T20:52:49 sounds fair 2010-12-01T20:53:28 has anybody done an ai for Battleship? 2010-12-01T20:53:48 no, but it needs SOME sort of twist, to make it be interesting 2010-12-01T20:54:05 eg. opponent can move ships, or multiple shots at a time, etc. 2010-12-01T20:54:58 battleship is fairly borring - hunt and peck for ship, find direction, then continue both ways till miss 2010-12-01T20:55:29 in RL battleship has interestign strategy elements 2010-12-01T20:55:38 eg. how far apart do you think they spaced the ships? 2010-12-01T20:56:01 I mix it up - I would place my ships randomly I think 2010-12-01T20:56:19 if each turn though, you get a total of 5 moves(or one per ship.. or whatever).. 2010-12-01T20:56:23 how many contestants were there for planetwars and how did that compare with the previous contests? 2010-12-01T20:56:29 and a move can be: shoot at enemy field, or move a ship 1 space 2010-12-01T20:56:34 it could be rather interesting 2010-12-01T20:57:20 4619? 2010-12-01T20:58:20 *** winexception has quit IRC (Quit: Bye) 2010-12-01T20:58:34 eburn.. I would argue tha tonly about 3k 'really' competed atall 2010-12-01T20:58:41 *** winexception has joined #aichallenge 2010-12-01T20:59:01 submiting the default ai.. or soemthing that crashes every time.. doesn't really count 2010-12-01T20:59:31 how many people have more then 15 uploads? 2010-12-01T20:59:34 when you can turn the default bot, into a top500 bot in about 5minutes if you're good at math, and understand the game 2010-12-01T21:02:15 I was thinking of basically writing a really simple MMORTS. Basically a simple Starcraft with an infinite map and thousands of players. I would make it so you could start playing it through the browser. Then slap an API onto it so people can bot it. 2010-12-01T21:02:34 just take the default.. make it only attack when it can capture.. and make it send as many attacks as it wants.. 2010-12-01T21:02:50 choose planet to attack absed on a function of population, distance, and growth 2010-12-01T21:03:27 j3camero: i'd love to create a test bot for it :) 2010-12-01T21:03:33 j3, what happens with people who enter the game much later?.. how is the metric for scoring done? 2010-12-01T21:04:51 ai for something like gauntlet could be interesting 2010-12-01T21:05:09 but no clue on how to make an ai challenge from a MMORTS (a fair one) 2010-12-01T21:05:31 random level, random 3 team-mates.. and the goal is to make your bot, work well with all other bot types 2010-12-01T21:06:47 but the problem with more than 2 players is that you dont really depend on your actions and one enemy, algo depends on the balance of quality between the other two players 2010-12-01T21:07:14 or am i missing something obvious? 2010-12-01T21:07:39 "Think Lisp is a dead language? Not according to Gábor Melis, who was just announced as the winner of the PlanetWars Google AI challenge." 2010-12-01T21:07:43 meneky ya.. it would be different ai :P 2010-12-01T21:07:54 I was thinking every time someone conquered more than 50% of the map they would "win" and the game would start over again. The final tournament would consist of one or more games of this sort. 2010-12-01T21:08:07 So it just automatically starts over once somebody wins. 2010-12-01T21:08:21 that's not really an 'mmo' then :P 2010-12-01T21:08:28 eburnette where did you get that? 2010-12-01T21:08:40 it'd be hilareous though.. 2010-12-01T21:08:42 he's writing it 2010-12-01T21:08:44 when the winning bot.. 2010-12-01T21:08:44 :) 2010-12-01T21:08:48 is 3 asian guys in a closet 2010-12-01T21:08:49 I'm writing an article, posting here for feedback 2010-12-01T21:10:10 j3: ok, but the problem remains the same, lets say i am a good player and near me there are one average and two crap bots, it the two crap bots let their territory be taken by the average i'm i a clear disavantage 2010-12-01T21:10:26 *** wh1teside has joined #aichallenge 2010-12-01T21:10:46 "His bot "bocsimacko" dominated a field of over 4600 contestants, but was one of only 33 programmed using the 50 year old language." 2010-12-01T21:10:55 just to be surrounded by better players than the average player 2010-12-01T21:10:56 meneky.. have you ever played FFA rts games? 2010-12-01T21:11:01 noes 2010-12-01T21:11:22 well, I played warcraft 3 a bunch.. 2010-12-01T21:11:27 and there's 4-6 people per map 2010-12-01T21:11:41 and the top players, win like 80-90% of the time 2010-12-01T21:11:59 I won about 50% of the time 2010-12-01T21:12:26 there are luck elements.. 2010-12-01T21:12:36 but it is highly skill based, if the game is balanced right 2010-12-01T21:12:47 planetwars doesn't work, because of the method of balance 2010-12-01T21:13:07 With multiple players though -- wonder if you should worry about collusion more. 2010-12-01T21:13:13 you can't "make defenses" in planetwars, for instance 2010-12-01T21:13:40 ok, i said nothing, then 2010-12-01T21:13:54 no experience in online multiplayer rts 2010-12-01T21:13:58 only solo mode 2010-12-01T21:14:01 *** wh1teside_ has quit IRC (Ping timeout: 255 seconds) 2010-12-01T21:14:33 j3camero: hmm, i should do that with my little infinite map engine 2010-12-01T21:14:35 if the game is fair i think it can be a great game 2010-12-01T21:14:55 http://a1k0n.net/code/jscroll/b/ 2010-12-01T21:15:00 it needs to be fair: and also balanced correctly(to remain interesting) 2010-12-01T21:15:56 next contest should be minecraft AI 2010-12-01T21:16:01 meneky, the thing that makes "real RTS's" be pheasible for FFA's.. is that 2 players, can bring the exact same army into battle.. and one loses 25/25 units, while the other loses 3/25 2010-12-01T21:16:05 a game far more interesting that chess go an the like 2010-12-01T21:17:11 and an exampel of a game with poor balance.. is the original starcraft.. 2010-12-01T21:17:23 I could win 6vs1 against the AI in that game 2010-12-01T21:17:30 You know what? I think you should post something to forums so people can discuss this ;) -- How many players per game/map? How long for it to be "over"? Victory conditions? etc :) 2010-12-01T21:17:33 could never manage to win 7vs1 though 2010-12-01T21:17:54 Holy crap, you kicked so much SC butt? or are you talking about FFA specifically? 2010-12-01T21:18:06 no.. 6vs1, not ffa 2010-12-01T21:18:24 starcraft ai wasn't terribly good at playing 2010-12-01T21:18:36 starcraft's AI definitely kicked my ass 2010-12-01T21:18:38 It whooped my ass ;) 2010-12-01T21:18:41 it could beat most people though 2010-12-01T21:18:51 You're sure you're not thinking of FFA, right? 2010-12-01T21:18:57 right 2010-12-01T21:19:20 6vs 1 I'd kill 4 of them, by time 2 of them teamed up on me, and they'd take my base, but I'd have 3 others 2010-12-01T21:19:37 but they're teamed up from the get go burny :) 2010-12-01T21:19:44 that's why I'm thinking you're talking about FFA :) 2010-12-01T21:20:19 *** rebelxt_ has joined #aichallenge 2010-12-01T21:20:29 *** Appleman1234 has quit IRC (Ping timeout: 272 seconds) 2010-12-01T21:20:32 oh, ya maybe.. the ai in that might have ALWAYS attacked player first.. I forget 2010-12-01T21:20:44 but they didn't attack each other, in whatever mode it was 2010-12-01T21:21:07 Well, FFA is a different game... Thought you really did mean 6v1, which I think would be... incredibly tough, if even possible :) 2010-12-01T21:21:11 Oh, they didn't attack each other? 2010-12-01T21:21:39 *** rebelxt_ has quit IRC (Client Quit) 2010-12-01T21:21:43 ya, I forget if it was "team mode", or if the AI just ALWAYS attacked players before AI's.. 2010-12-01T21:21:52 but either way.. the ai didn't attack eachother 2010-12-01T21:22:25 Ok, but you're giving confusing information... If you played team mode, 6 v 1 -- such that 6 AIs vs just you -- and you managed to fight them off, that's awesome! :) 2010-12-01T21:22:49 yes, it was 6AI players, against me 2010-12-01T21:23:02 I'd win about 40% of the time 2010-12-01T21:23:05 but from the sounds of you, you're either talking about team games (3v3/4v4?) or FFA, which makes it quite different :) 2010-12-01T21:23:10 1vs3 I'd win about 99% of the time 2010-12-01T21:23:30 ya, in warcraft3.. I played FFA( 4-6 players) 2010-12-01T21:23:39 and win about 50% of the time.. the best players won 80-90% 2010-12-01T21:24:11 *** pyro- has left #aichallenge 2010-12-01T21:28:54 *** xxx has quit IRC (Ping timeout: 265 seconds) 2010-12-01T21:28:55 so for an FFA rts game, it's be pretty easy to give it decent level of balance, just by making themap layout, be so that bots are paired together(and basically have to kill each other, before expanding further).. then they meet another bot who won a mini best of 2, before they expand into a larger area, etc. 2010-12-01T21:29:43 How many bots per map/game? 2010-12-01T21:29:57 could do as many as ya want really 2010-12-01T21:30:00 I recall someone mentioning MMO, but.. 2010-12-01T21:30:15 ya.. I don't think MMO would work, without massive game tweaks 2010-12-01T21:30:42 Well, how fast does a game resolve itself? Current planet wars was what, more or less 200 seconds max? 2010-12-01T21:31:10 however fast you make it be? 2010-12-01T21:32:39 First of all, do we really mean "RTS" when we talk about this.. I mostly ask about the first 2 characters of that abbreviation ;) 2010-12-01T21:32:59 Okay I have had enough contest for the time being. I am going to check out this .p2p thing. 2010-12-01T21:33:04 a 'frame' is a turn :P 2010-12-01T21:33:18 j3camero: which / what p2p thing? 2010-12-01T21:33:28 I think we should all go over and help get the .p2p project going. Wouldn't it be awesome to have the entire DNS system out of government control? 2010-12-01T21:33:42 burny: RT stands for real time :) 2010-12-01T21:34:10 real time just means you don't get infinite time to make descision 2010-12-01T21:34:26 j3camero: what initiative are you talking about? p2p is broad, so maybe a link here would help :) 2010-12-01T21:35:18 dns? 2010-12-01T21:35:50 we should have a hearts ai contest 2010-12-01T21:35:58 boo :) 2010-12-01T21:38:11 I did hearts ai before.. it's a bit too easy I think 2010-12-01T21:38:30 spades would be a bit more interesting 2010-12-01T21:38:56 or sheepshead. do you pick partners in spades? i forgot 2010-12-01T21:39:06 spades is a fixed 2vs2 2010-12-01T21:39:48 sheepshead is kinda somewhere between spades and hearts, best played with 5. it's played a lot in wisconsin, and not a lot elsewhere 2010-12-01T21:40:26 the worst AI game that doesn't seem bad at first.. 2010-12-01T21:40:27 j3camero: this? http://www.theinquirer.net/inquirer/news/1929499/pirate-bay-start-p2p-dns ? I'm not sure if DNS is the only thing blocking from "uncensored Internet"... Makes me wonder how freenet and tor handle DNS ;) 2010-12-01T21:40:28 is conenct4 2010-12-01T21:40:30 virdo: search for "dot p2p" you should find it. 2010-12-01T21:40:35 trivial to make an invincible bot 2010-12-01T21:40:40 yeah 2010-12-01T21:41:09 j3camero: I'm behind on tech news... haven't read /. in over 2 months... sniff sniff :/ 2010-12-01T21:44:22 er.. p2p dns isn't publiclly solved yet? 2010-12-01T21:44:53 oh, hm.. guess it's harder than the problem I was thinking of at first 2010-12-01T21:45:17 I know people who've done p2p dns, for subsets of the internet, but never the whole internet 2010-12-01T21:45:38 *** winexception has quit IRC (Quit: Bye) 2010-12-01T21:46:47 burny: what do you mean "p2p dns" though? run their own dns server? run it for their domain? I think this talks about something different ;) 2010-12-01T21:47:27 basically.. a private internet, for a specialized service 2010-12-01T21:48:36 "private internet" -- what do you mean? running a website / web service that's not reachable externally / through the internet? 2010-12-01T21:48:53 Top 10 players: bocsimacko(3765), _iouri_(3565), Slin-.-(3524), _Astek_(3501), jimrogerz(3500), Accoun(3498), george(3494), GreenTea(3489), asavis(3480), bix0r4ever(3476) 2010-12-01T21:49:41 like.. 2010-12-01T21:50:07 a TV service, via internet 2010-12-01T21:51:40 but what does a tv service through internet have to do with dns or p2p? :) 2010-12-01T21:51:52 or specifically, this dot p2p 2010-12-01T21:52:03 because it doesn't use isp dns 2010-12-01T21:54:30 What does it use? The only solution to not using dns is by knowing the IP address ahead of time, no? 2010-12-01T21:55:06 *** superflit has joined #aichallenge 2010-12-01T21:55:37 use p2p, to keep track of servers, hosts, etc. 2010-12-01T21:56:51 Well, unless it guesses other p2p clients (scans entire networks), it probably connects to some kind of server first 2010-12-01T21:57:06 but anyways, I'll shut up :) 2010-12-01T22:01:08 to conenct to the service initially requires a known server(or port scanning) 2010-12-01T22:01:27 once conencted.. it uses dns like service, to remain in the loop 2010-12-01T22:02:05 a simple method is like.. 2010-12-01T22:02:26 when you connect, keep a list of the peers you connect to that have good connections 2010-12-01T22:02:36 and when you want to re-conenct.. conenct to them first 2010-12-01T22:02:54 >> After seeing Gábor's performance, many participants commented that they would be taking a look at the 50 year old language. "Bocsimako is the proof that lisp is best for AI algorithms," wrote one player. 2010-12-01T22:03:11 >> "I'll have to learn more Lisp just to understand your code," said another, after Melis uploaded his source code at the contest's end. If you speak Lisp, you can view it online here. 2010-12-01T22:03:38 another strategy.. make 'friends' and make offers with your friends of "I'll keep track of what's going on from 6am to 6pm.. you keep track frmo 6pm to 6am, and we'll talk to each other at 6am and 6pm every day 2010-12-01T22:11:00 I only see one issue with the p2p dns thing. Whats stopping an isp from setting up a node, scanning all banned sites p2p dns and then blocking ips? 2010-12-01T22:11:50 "Bocsimako is the proof that lisp is best for AI algorithms," haha, someone actually said this? 2010-12-01T22:12:11 lol 2010-12-01T22:12:15 stupid press 2010-12-01T22:12:36 because anything is 'the best'.. 2010-12-01T22:12:49 Where is this article that you're reading? 2010-12-01T22:13:21 eburnette is writing an article and posting extracts here 2010-12-01T22:13:54 >> Bocsimacko debuted in the official contest ratings on October 22nd. Two days later, it was in the top spot and never looked back. 2010-12-01T22:14:31 sigh: ok 2010-12-01T22:14:44 "2010-11-22T18:25:07 bosimacko is the proof that lisp is best for ai algorithms :P" 2010-12-01T22:15:08 "The contest quickly became a race for second place" 2010-12-01T22:15:13 ah, the ":P" gives the statement a whole new meaning 2010-12-01T22:16:18 maybe you should just write it and then paste a link here? 2010-12-01T22:16:20 ;) 2010-12-01T22:17:01 "2010-10-22T18:11:17 Top 10 players: Hazard(4090), rsergio(4059), felixcoto(4052), asavis(4030), cfaftw(4001), sin_sun(3977), dmj111(3972), bocsimacko(3917), davidjliu(3893), RebelXT(3889)" was first appearance of bocsimacko I could find on irc logs 2010-12-01T22:17:28 Whatever happened to Hazard? He was doing pretty well at first. 2010-12-01T22:17:51 tortoise and the hare? 2010-12-01T22:17:55 *** Appleman1234 has joined #aichallenge 2010-12-01T22:18:07 * burny is the hare :( 2010-12-01T22:18:16 i jest though 2010-12-01T22:18:27 oh jmcarthur, what happened to your awesome algorithm? 2010-12-01T22:18:51 in fps's, and lots of other things.. I dominate for the first few minutes, and then get owned as other people get comfortable with the new situation 2010-12-01T22:20:47 sigh: i don't believe i ever claimed awesomeness 2010-12-01T22:20:51 i didn't finish it 2010-12-01T22:20:54 i quit early 2010-12-01T22:21:11 ah too bad 2010-12-01T22:21:19 this was taking too much of my time when i could have been focusing on more productive things 2010-12-01T22:21:31 what were you actually planning. I just remember you being very secretive :P 2010-12-01T22:21:37 i believe i did explain what i was doing at one point 2010-12-01T22:21:41 flag I added your line, thanks 2010-12-01T22:21:55 >> When asked about his language choice, Gábor responded "I'm thankful to have been programming in Lisp professionally for 6 years." He says he tried Java but found it "incredibly boring". 2010-12-01T22:22:01 i was being secretive not because i thought it would work but because i just wanted something unique ;) 2010-12-01T22:22:14 :) 2010-12-01T22:23:07 anyway, it was just a partical swarm optimization thing with a couple novel ideas about how to map strategies to positions in the search space and making the thing play nice with some game theory concepts 2010-12-01T22:23:12 >> He is a senior developer for RavenPack, a privately held firm specializing in computer algorithms that parse news feeds to create models for stock trading programs. 2010-12-01T22:23:56 eburnette: no problem 2010-12-01T22:24:18 the search space was basically "what proportion of my available ships do i want to send from planet A to planet B on turn N?", with a heuristic for figuring out what planets to send the ships *from* during the fitness test, which was basically a complete run of the game 2010-12-01T22:24:31 there were two swarms going simulataneously trying to beat each other 2010-12-01T22:25:15 hmm... sounds complicated o_O 2010-12-01T22:25:17 and the update functions were probabilistic based on mixed strategies computed from running all the positions of one swarm against all the positions of the other 2010-12-01T22:25:27 yeah that's why it was taking so much of my time :\ 2010-12-01T22:26:08 i had implemented most of the parts in isolation but quite before i got them integrated together 2010-12-01T22:26:14 s/quite/quit/ 2010-12-01T22:26:26 afternoon 2010-12-01T22:26:27 ah, too bad 2010-12-01T22:26:31 congrats people 2010-12-01T22:26:31 hey antimatroid 2010-12-01T22:26:51 i did expect that it would do... decently 2010-12-01T22:27:03 assuming i could run enough simulations in one second 2010-12-01T22:27:20 the first several turns would be bad though 2010-12-01T22:27:23 and those are critical 2010-12-01T22:27:32 so i would have probably had to implement some heuristics for those 2010-12-01T22:27:47 and then switch over to the PSO AI once it had enough meaningful iterations 2010-12-01T22:27:47 you can always hack the first few turns with something like knapsack and then switch :P 2010-12-01T22:27:51 yeah 2010-12-01T22:28:24 antimatroid: so you working on aarima now? 2010-12-01T22:28:51 sigh: the comp starts in a month, i think i'm going to slowly work on one for the following year 2010-12-01T22:28:59 i can't afford to spent another month doing nothing else 2010-12-01T22:29:07 haha 2010-12-01T22:29:12 just make one in 4days :P 2010-12-01T22:29:44 antimatroid: also, I saw that post you responded to about dimensions... no way that guy was a physicist :P 2010-12-01T22:29:51 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-12-01T22:29:55 no, i know :P 2010-12-01T22:30:38 that whole thread was a bit painful to read, though :( 2010-12-01T22:31:26 *** nullkuhl has joined #aichallenge 2010-12-01T22:32:03 yeah, i stopped looking :P 2010-12-01T22:33:28 *** nullkuhl has quit IRC (Read error: No route to host) 2010-12-01T22:34:16 *** nullkuhl has joined #aichallenge 2010-12-01T22:39:39 >> According to contest organizer Jeff Cameron, PlanetWars is the 4th AI contest sponsored by the University of Waterloo Computer Science Club. 2010-12-01T22:40:07 >> The first one was Minesweeper Flags, followed by Rock Paper Scissors, and then Tron. Yes, they actually had a contest writing programs to play Rock Paper Scissors. "RPS is harder than you think," says Jeff. "It's not random. It was iterated RPS." 2010-12-01T22:42:02 you probably should ask jeff what he meant by iterated 2010-12-01T22:42:17 trials were not independent 2010-12-01T22:42:28 right 2010-12-01T22:42:36 ie. you would play 200-2000 games in the same process 2010-12-01T22:42:43 so you could "track" your enemy and predict them etc 2010-12-01T22:42:45 and amstan won 2010-12-01T22:42:54 amstan won rps yeah? 2010-12-01T22:43:03 Yeah 2010-12-01T22:43:13 i don't see how that's anything other than who has the best random number generator 2010-12-01T22:43:19 which one did he go with? 2010-12-01T22:43:20 rock? 2010-12-01T22:43:23 :P 2010-12-01T22:43:24 *** Azrathud has joined #aichallenge 2010-12-01T22:43:28 Hey. 2010-12-01T22:43:30 jmcarthur: did you ever get an entry together at all? 2010-12-01T22:43:35 So it's basically a contest that depends on other people being predictable 2010-12-01T22:43:36 Agh. Wrong channel. 2010-12-01T22:43:55 antimatroid: because randomness wins you 33% of the games 2010-12-01T22:43:58 ties are not wins 2010-12-01T22:44:06 you have to be /better/ tahn random. 2010-12-01T22:44:07 basically.. a random bot, will just always be at 50% wins.. 2010-12-01T22:44:08 how do you accidentally pick #aichallenge? 2010-12-01T22:44:16 if you want to do better.. you have to play not just random 2010-12-01T22:44:21 and there in lies the contest 2010-12-01T22:44:25 It won't be 50% wins 2010-12-01T22:44:30 there are 3 outcomes: loss, win, tie 2010-12-01T22:44:33 ties are NOT wins 2010-12-01T22:44:38 whatever.. 50% score 2010-12-01T22:44:47 ...that's not right though 2010-12-01T22:44:50 ._. 2010-12-01T22:45:10 if I play rps pure random, and you play rps with ANY sort of strategy.. 2010-12-01T22:45:21 we will win, tie, and lose, 33% of the time 2010-12-01T22:46:03 but you're not pure random 2010-12-01T22:46:05 that's the point 2010-12-01T22:46:21 uh ya.. that's what I was saying... 2010-12-01T22:46:30 but its not random vs random 2010-12-01T22:46:32 random is 50%, and thus, random can't possibly be the best strategy 2010-12-01T22:46:45 thus, people will make SOME sort of strategy.. 2010-12-01T22:46:58 *** Rubicon-|-Cross has quit IRC (Read error: Connection reset by peer) 2010-12-01T22:47:00 and your bot should be trying to solve the opponents strategy, better than they solve yours 2010-12-01T22:47:09 wait hang on, was it rps or repeated rps against the same opoonent? 2010-12-01T22:47:16 iterated rps! 2010-12-01T22:47:26 so how is that competition of the "better random number generator" 2010-12-01T22:47:32 ah okay, i was thinking it was like this where you don't know your opponent each turn 2010-12-01T22:47:38 um.. it isn't? 2010-12-01T22:47:48 oh thats what flag said 2010-12-01T22:47:53 carry on. 2010-12-01T22:47:55 * pgpaskar_ whistles 2010-12-01T22:48:00 thought you claimed that 2010-12-01T22:48:00 a1k0n_: no, when i quit, i quit hard. no regrets. i've been so much more productive since i quit 2010-12-01T22:48:09 heh. 2010-12-01T22:48:26 i work in an office now, it's harder to procrastinate with ai contests 2010-12-01T22:48:35 i'm feeling very over minimax for the moment 2010-12-01T22:48:48 Top 10 players: bocsimacko(3765), _iouri_(3565), Slin-.-(3524), _Astek_(3501), jimrogerz(3500), Accoun(3498), george(3494), GreenTea(3489), asavis(3480), bix0r4ever(3476) 2010-12-01T22:48:52 i wish i got minimax to work. i'm really curious about mega1's bit vector thing. 2010-12-01T22:48:59 *** Rubicon-|-Cross has joined #aichallenge 2010-12-01T22:49:02 i might try something more spectacular next time and flop more awesomly :) 2010-12-01T22:49:03 is he evaluating ranges of fleet sizes in parallel? 2010-12-01T22:49:26 also i shouldn't have submitted the version of my bot overtrained on the old maps 2010-12-01T22:49:31 when it comes to games.. VERY few games, are actually based on raw luck 2010-12-01T22:49:44 if it is, it's luck on the first move 2010-12-01T22:49:50 or both bots just suck 2010-12-01T22:49:55 even something as simple as crazy 8's, can be won 80%+ of the time 2010-12-01T22:50:55 pgpaskar_: so what was the winning strategy for repeated rps? tit for tat? 2010-12-01T22:51:23 A stupidly complicated prediction-antiprediction-prediction-randomness bot 2010-12-01T22:51:29 everyone copied some canonical algorithm with their own tweaks (hey, just like in tron) 2010-12-01T22:51:41 antimatroid: how can you tit-for-tat in rps? 2010-12-01T22:51:42 It had prediction techniques, prediction of prediction techniques, and a good random number generator 2010-12-01T22:51:56 and then it mix-and-matched these with meta-predictors to determine which would work best 2010-12-01T22:52:02 it's not like the prisoner's dilemma :P 2010-12-01T22:52:25 sigh: you're right i guess, you could return optimally to some increment of their past strategy or something 2010-12-01T22:52:31 or would sort of be like it 2010-12-01T22:52:53 yeah, i make some really stupid comments sometimes :) 2010-12-01T22:52:57 :) 2010-12-01T22:54:27 *** MikeTV has quit IRC (Ping timeout: 265 seconds) 2010-12-01T22:55:26 bah, am i remembering right that I can't do templates in c++ and actually define all my functions inside a .cpp file? 2010-12-01T22:55:29 *** Mathnerd314 has joined #aichallenge 2010-12-01T22:55:44 unless you only use the tmeplate isnide the t cpp, correct 2010-12-01T22:55:47 *** Prillicy has joined #aichallenge 2010-12-01T22:55:53 bah crap 2010-12-01T22:55:54 so, quick question...when's the next challenge start? 2010-12-01T22:56:08 Prillicy: some time earlyish next year is the best guess 2010-12-01T22:56:15 I can't wait 2010-12-01T22:56:28 will it be announced on the same site? 2010-12-01T22:56:39 i have a game theory library in c++, i was going to separate it so headers were a hell of a lot cleaner, i guess not 2010-12-01T22:56:42 Do it over Christmas break -- some of us will have more time during this period :P 2010-12-01T22:57:00 and then after Xmas.. cause some of us won't have jobs/familys anymore :P 2010-12-01T22:57:03 virdo: i am on summer break atm :) but it's not the right time for them 2010-12-01T22:57:29 Yeah, probably not enough tiem to prepare everything in time (framework / api / all the other goodies they're talking about) 2010-12-01T22:57:31 i like 2 player simultaneous games, but i don't want another one simply dominated with minimax 2010-12-01T22:57:38 I'll keep checking in here...and I'd lose my life if I played minecraft. 2010-12-01T22:57:38 i want a new strategy to play around with 2010-12-01T22:57:45 *** Prillicy has quit IRC (Client Quit) 2010-12-01T22:57:55 minecraft! 2010-12-01T22:58:07 haha, ultimate joke on Prillicy -- minecraft bot! :) 2010-12-01T22:58:08 bocs wasn't minimax was it? 2010-12-01T22:58:17 no 2010-12-01T22:58:27 we should do something with 16 players at a time 2010-12-01T22:58:29 tron was though right? 2010-12-01T22:58:44 I agree, multi player is tons more interesting 2010-12-01T22:58:46 i'm down for that, but as soon as you have more than 2 players, the games become a lot harder 2010-12-01T22:58:51 it uses minimax to some extent, I think 2010-12-01T22:58:51 exactly 2010-12-01T22:58:55 (bocs' bot) 2010-12-01T22:59:00 sigh: only on the first turn 2010-12-01T22:59:08 you either have difficulty or you have minimax 2010-12-01T22:59:08 should also give some sort of minimal form in which the bots can communicate 2010-12-01T22:59:12 to some extent :P 2010-12-01T22:59:53 i think a game with hidden state would be good (you don't always know what the opponent has or is capable of) 2010-12-01T23:00:07 fog of war? ;) 2010-12-01T23:00:12 ya, that can be good 2010-12-01T23:00:25 the first ai thing I ever did, had that.. 2010-12-01T23:00:32 that shooter game i suggested a week or two ago had both many players and hidden information 2010-12-01T23:00:37 and I had the winning bot in under an hour 2010-12-01T23:00:56 which shooter game? 2010-12-01T23:01:04 +1 for private information 2010-12-01T23:01:05 h/o 2010-12-01T23:01:16 bayesian games :) 2010-12-01T23:01:29 hmm, not realy 2010-12-01T23:01:32 another stupid comment :) 2010-12-01T23:01:44 https://docs.google.com/viewer?a=v&pid=explorer&chrome=true&srcid=0B3_Nay8FIDhVNWMxNjk1OTItMWMwMi00NDMyLWFjNzMtMGI2ZjI0MmU2Yzhj&hl=en 2010-12-01T23:01:53 you could let them choose their starting private information, then it'd be bayesian 2010-12-01T23:02:05 the hidden information part was the first bullet point under Variations 2010-12-01T23:02:10 next term's contest logo: http://lamezone.net/dogaday/img/dogs/dog146.png 2010-12-01T23:02:44 basically if an opponent goes behind a wall and the emerges later you don't necessarily know how much charge they have 2010-12-01T23:02:47 *and then 2010-12-01T23:04:03 jmcarthur: i like that, although why not do it all continously without turn time limits? 2010-12-01T23:04:12 let players call for adjustments and fire as they please 2010-12-01T23:04:14 summary: asteroids -- but against other bots and walls / other stuff on the map? ;) 2010-12-01T23:04:40 antimatroid: it just gets tougher that way, and some people would cry foul because their language is slow than others... 2010-12-01T23:04:41 http://www.textmodegames.com/download/snipes.html 2010-12-01T23:04:48 multiplayer shooter 2010-12-01T23:04:48 antimatroid: i personally would be all for it as long as the engine is fair 2010-12-01T23:05:02 can one do clock cycles and make it fair for languages? 2010-12-01T23:05:13 i have no idea about those kind of things 2010-12-01T23:05:29 not sure what you mean. you mean give each bot exactly the same number of clock cycles? 2010-12-01T23:05:32 it's hard 2010-12-01T23:05:32 languages are fair.. 2010-12-01T23:05:43 ease of deveopment vs performance, etc. 2010-12-01T23:05:57 if onelanguage is better.. us eit 2010-12-01T23:06:06 i like the idea of self programmed weapons that meet some conditions 2010-12-01T23:06:10 I wonder --- instead of giving a second, maybe you give: you have max 200 seconds left... and communicate that back to the bot... so, if a bot realizes they're slow, they can do less stuff per turn? 2010-12-01T23:06:13 so you don't even know exactly what their weapons will do 2010-12-01T23:06:28 antimatroid: yeah i was trying to think of a sensible way to put something like that into my idea 2010-12-01T23:06:40 *** davidjliu has quit IRC (Quit: Page closed) 2010-12-01T23:07:06 jmcarthur: i'd say let people write some kind of "missile script" that the weapons start their own instance of and that then does it's own thing 2010-12-01T23:07:27 weapon with AI? 2010-12-01T23:07:33 yeah 2010-12-01T23:07:38 I think a good method to make cpu time work.. 2010-12-01T23:07:43 is a per turn, and per game, limit 2010-12-01T23:07:44 like writing your own program for your red shell in mario kart 2010-12-01T23:07:47 so now you are meta-programming in your AI :P 2010-12-01T23:07:54 and the shell follows the program you wrote for it before expiring 2010-12-01T23:08:01 so each turn is limited to 1second, and the full game is limited to 10seconds, or some such 2010-12-01T23:08:13 antimatroid: you mean you can give the weapon a different program every time you fire? 2010-12-01T23:08:20 because that would rock 2010-12-01T23:08:22 jmcarthur: sure 2010-12-01T23:08:40 you could write 10 different scripts for it if you wanted, then dependent on the state send out the weapon you want 2010-12-01T23:08:41 sounds hard :) 2010-12-01T23:08:45 exactly :) 2010-12-01T23:08:57 however, I do think that simple rules are better for this kind of contest 2010-12-01T23:09:22 yes, simple rules, local testing, instant gratification! 2010-12-01T23:09:31 what if there were always 2 tournys in the future 2010-12-01T23:09:36 i agree to an extent 2010-12-01T23:09:40 very similar games, but one simplified a lot for beginners? 2010-12-01T23:09:43 eh 2010-12-01T23:10:00 i'm not a fan of the idea. i think beginners would never want to play in the beginner bracket 2010-12-01T23:10:05 I had the same idea anti 2010-12-01T23:10:17 s/think/expect/ 2010-12-01T23:10:18 burny: i think i was the one saying it was bad to you :P 2010-12-01T23:10:28 and i still semi agree, but i want HARD games damnit :( 2010-12-01T23:10:34 yeah me too 2010-12-01T23:10:35 I'd say the way to do it..is make the 'beginner bracket' have simplified rules, that guide them to tuning a portion of their ai 2010-12-01T23:10:51 antimatroid: would you say planetwars was easy or hard? 2010-12-01T23:10:53 antimatroid: wasn't this a HARD game? 2010-12-01T23:11:05 virdo: yes 2010-12-01T23:11:10 eg. if it's a 2vs2 game.. it will tell you which player is the 'leader' and 'follower'.. and you're intended to write both modes, and test them independantly 2010-12-01T23:11:11 sigh: hard 2010-12-01T23:11:21 but in the 'real' game.. it' sup to the bots to decide who is leading 2010-12-01T23:11:28 anyway, rule complexity doesn't have much corrolation with game difficulty 2010-12-01T23:11:34 sigh: exactly! 2010-12-01T23:11:43 if map sizes were smaller for this some much cooler programs would have been made 2010-12-01T23:11:50 but the games would have been less cool 2010-12-01T23:12:06 did anyone ever get a general solution for 3 planet maps? 2010-12-01T23:12:25 but simple rules has a lot of benefits, low barrier for entry, easier to create a balanced game, easier to make tools (such as visualisers) 2010-12-01T23:12:34 I think the best would have been if it wasn't symetrical.. and each pair played 2 games.. and the bot that wins the fastest(or wins both), wins 2010-12-01T23:12:36 i had like a page of algebra at one point trying to set up players expected utilities and equating etc. and it was just so messy with all the parameters 2010-12-01T23:12:36 general solution? depends on location of planets, no? and their growth / ship counts? :) 2010-12-01T23:12:39 Did google sponsor the RPS and Minesweeper contests too? j3camero 2010-12-01T23:13:17 virdo: a general solution would just assign general parameters for those values 2010-12-01T23:13:37 you might need some kind of composite thing for different types of games 2010-12-01T23:13:49 but a general solution in mixed strategies exists 2010-12-01T23:14:32 so you're saying a starting planet for both players, and a central planet? and that's it? 2010-12-01T23:14:47 yep, any kind of isosceles triangle 2010-12-01T23:14:53 if planetwars was about winning fast.. rather than just winning.. it would have required much more interesting strategies 2010-12-01T23:15:09 burny: it was already tough enough as it was :) 2010-12-01T23:15:11 antimatroid: So basically, wait till enemy tries to take neutral and steal it? 2010-12-01T23:15:21 flag: not always 2010-12-01T23:15:26 yeah, not always... 2010-12-01T23:15:40 what if growth > ship count? :) 2010-12-01T23:15:48 it depends on whether them sniping you leaves them vulnerable to you then sniping their home 2010-12-01T23:15:50 or ^^ 2010-12-01T23:16:08 hmm, i guess that can't happen 2010-12-01T23:16:16 Yeah... 2010-12-01T23:16:31 ya it can 2010-12-01T23:16:31 which? growth > ship count? of course it can? 2010-12-01T23:16:39 no, my example 2010-12-01T23:16:42 yours is fine 2010-12-01T23:17:02 you can get some complicated 3 planet maps with that feature, but not as many 2010-12-01T23:17:07 In virdo's case then you would go for the neutral 2010-12-01T23:17:16 flag: depends where centre is 2010-12-01T23:17:32 Unless of course the enemy could then kill your home 2010-12-01T23:17:33 well, the only other case I see is this... if distance to neutral planet x 2 == distance between players... you could send 100% fleet to neutral, and still make it back in time to defend your home base, right? 2010-12-01T23:17:38 if you send everything at the neutral leaving your home planet vulnerable, you lose 2010-12-01T23:17:59 Basically, take as much as you can without leaving anything vulnerable 2010-12-01T23:18:29 hmm, the real question is, does there exist 3 planet rps situations? 2010-12-01T23:18:42 non-equal games, with metrics on who won by how much: makes for very interesting AI imo 2010-12-01T23:19:06 burny: i agree, but the process of ranking bots also gets harder 2010-12-01T23:19:27 rank based on how much they won, vs how much they were expected to win by 2010-12-01T23:19:48 So if his_home_number_ships > your_ships + distance_to_enemy*your_growth_rate - the_fleet_youre_sending : don't send the fleet 2010-12-01T23:20:29 or, if neutral_num_ships > neutral_growth_rate and enemy_incoming[neutral] < neutral_num_ships : don't send the fleet 2010-12-01T23:20:48 otherwise: ATTACK! 2010-12-01T23:21:01 ? 2010-12-01T23:21:01 not sure about that second part being equivalent? 2010-12-01T23:21:07 I don't think this contest had any truely interesting ai's.. 2010-12-01T23:21:24 interesting stuff would be like.. 2010-12-01T23:21:29 So if the enemy isn' 2010-12-01T23:21:49 t attacking the neutral, and the number of ships at it is greater than the growth rate, don't attack it either 2010-12-01T23:21:51 a planet has 40 ships, but you know the opponent will snipe it if you hit it.. but it's the best move you can make.. 2010-12-01T23:21:57 Also don't attack if your home planet can be sniped 2010-12-01T23:22:14 attack it with 20, expecting the enemy will then decide to finish it off, and claim the growth.. 2010-12-01T23:22:26 *** MikeTV has joined #aichallenge 2010-12-01T23:22:32 Or your opponent could just wait 2010-12-01T23:22:37 and then send your attack the next frame, so they won't get to temporarily own it 2010-12-01T23:22:42 And you'd lose at 200 moves due to less ships 2010-12-01T23:22:51 doh. 2010-12-01T23:22:52 if i go for a neutral, i send either 1 more than they could possibly send in that time, or everything i can if it's equal to what they can 2010-12-01T23:23:22 but i wont do that last move unless neutal ships <= neutral growth 2010-12-01T23:23:38 Are we still talking about the map with 3 planets? 2010-12-01T23:23:47 don't think so 2010-12-01T23:23:52 that's what my strategy would do 2010-12-01T23:24:05 and it'd make a tree with both players doing either options 2010-12-01T23:24:24 My strategy would wait for the opponent to attack the planet neutral and go from there 2010-12-01T23:24:25 although my tree is broken, fix up that and my bot does really really well on maps with small numbers of planets 2010-12-01T23:24:26 3planet map is pretty trivial to solve 2010-12-01T23:24:44 what if hte planet is closer to the opponent than yourself? 2010-12-01T23:24:44 burny: That's what I'm saying 2010-12-01T23:24:52 Then it's not a fair map? 2010-12-01T23:24:59 i spent time tweaking my bot for small maps such that i couldn't beat it in human mode even playing perfectly 2010-12-01T23:25:11 so? life isn't fair 2010-12-01T23:25:15 antimatroid: unless it's RPS type of situation? 2010-12-01T23:25:27 virdo: then it's up for anyone :P 2010-12-01T23:25:34 What if I start with 1000 ships and you start with 4? 2010-12-01T23:25:40 :) 2010-12-01T23:25:47 i originally had my tree tie braking randomly, but as the game was already so messy, i got sick of a bot making different moves all the time 2010-12-01T23:26:09 flag: there was a game posted on the forum where someone came back from having 1 ship on 1 planet :P 2010-12-01T23:26:18 not deterministic? mine wasn't either, even though I tried to sort by planet IDs and stuff ;) 2010-12-01T23:26:20 that's amazing 2010-12-01T23:26:25 what if you star with 1000ships and I start with 4.. but I start with 20grow, and you start with 0 2010-12-01T23:26:41 Then I send everything I have at you and win 2010-12-01T23:26:49 depends on distance 2010-12-01T23:26:55 antimatroid: he said 0 growth 2010-12-01T23:26:56 but it's a pretty clear strategy depending on parameters 2010-12-01T23:26:57 send it all :) 2010-12-01T23:27:03 Unless the distance between me and you is greater than (1000 - 4)/20 2010-12-01T23:27:10 Or something like that 2010-12-01T23:27:17 this is the game: http://ai-contest.com/visualizer.php?game_id=8584510 2010-12-01T23:27:20 (move 26) :P 2010-12-01T23:27:40 what if you start beside a 1cost 5grow planet?.. you wouldn't take that first? 2010-12-01T23:27:46 i think symmetric maps are fine 2010-12-01T23:28:03 ya, symmetry is ok.. but I think non symm is a lot more interesting 2010-12-01T23:28:08 that game is awesome 2010-12-01T23:28:17 burny: Well if you keep changing the map then yeah, my bot would do different things. 2010-12-01T23:28:50 I never said or sugested the map has only 2 planets 2010-12-01T23:29:24 I think we should brainstorm on the next challenge -- or we can rehash planet wars more ;) - different growth planets, fog of war, sacrificing your troops to increase growth production, multiple players, moving planets, 3d, ... :D 2010-12-01T23:29:24 1000+0 vs 4+20, could be an interesting cat and mouse game.. 2010-12-01T23:29:33 what would be interesting is if every neutral was non-snipable on the first turn :P 2010-12-01T23:29:35 the 4+20 is tryng to buidl the best defensive path.. 2010-12-01T23:29:39 including centre planets 2010-12-01T23:29:45 while hte opponent is trying to build the best attack path 2010-12-01T23:29:48 Ok guys, the article is finally finished. 2010-12-01T23:30:00 i wanted to construct a map that would result in bots just chasing each other around a circle of planets 2010-12-01T23:30:19 if center planets werne't snipable on first turn, they would probably too superior to the rest of the map 2010-12-01T23:30:19 Gosh these things take a long time to write. Anyway thanks for your input... 2010-12-01T23:30:35 Url is http://www.zdnet.com/blog/burnette/hungarian-lisp-developer-walks-away-with-google-ai-contest/2131 2010-12-01T23:30:42 causign the primary strategy to be, deciding which panets to hit on first turn, and how many to send to each 2010-12-01T23:31:15 but whoever expands to the middle with a sufficiently strong force first, will win most certainly 2010-12-01T23:31:23 Feel free to vote it up, submit it to digg and reddit, etc. :) 2010-12-01T23:31:35 *** iamluck has quit IRC (Read error: Connection reset by peer) 2010-12-01T23:31:55 *** iamluck has joined #aichallenge 2010-12-01T23:32:12 digg? :P 2010-12-01T23:32:16 if enough bots were sufficently good.. this game would devolve into RPS 2010-12-01T23:32:51 if enough bots are sufficiently good, then chess is a trivial game :P 2010-12-01T23:33:05 :) 2010-12-01T23:33:45 true.. but it's not pheasible to be good enough at chess, currently 2010-12-01T23:33:57 that's the point :P 2010-12-01T23:34:35 oh ya.. I forgot, that it's not general concensus, that this game is solvable 2010-12-01T23:35:12 watch burny create a bot that beats bocsi tonight :) 2010-12-01T23:35:15 depends what you mean by solvable 2010-12-01T23:35:20 why bother? 2010-12-01T23:35:23 burny, RPS alone suggests it isn't? 2010-12-01T23:35:31 chess, however, is :) 2010-12-01T23:35:45 virdo: there exists a solution in mixed strategies 2010-12-01T23:36:04 if you devolve it into RPS, and th eopponent doesn't.. then you win 2010-12-01T23:36:20 ... on average 2010-12-01T23:36:35 well going by the previous conversation that's not the whole story... optimal mixed strategy for rps is pure randomness 2010-12-01T23:36:55 for rps with no information about your opponent, you mean 2010-12-01T23:36:57 but people did better, because other opponents weren't random 2010-12-01T23:36:57 ??? whole idea is A > B > C > A -- so, if you do A, and enemy does B, you win... but if they do C, they win... ? 2010-12-01T23:37:04 sigh: that's a different game 2010-12-01T23:37:07 sigh: but then by definition it's not solvable? :) 2010-12-01T23:37:14 pure random is NOT the best(nor CAN be the best) 2010-12-01T23:37:17 there is a big difference between repeated games with known and unknown opponents 2010-12-01T23:37:27 unknown is equivalent to single instances of a game pretty much 2010-12-01T23:37:36 if you're doing that, then you can't do better than complete randomisation 2010-12-01T23:37:39 fair enough 2010-12-01T23:37:47 uhm, but we're tlaking here about Planet Wars, yes? So, given a RPS styled map, and you're playing with an opponent, how can it be solvable? 2010-12-01T23:37:54 *** Azrathud has quit IRC (Ping timeout: 276 seconds) 2010-12-01T23:37:58 or wait, I take it back. 2010-12-01T23:38:00 virdeo: you use a mixed strategy 2010-12-01T23:38:28 l8r guys, maybe cya next contest 2010-12-01T23:38:29 yes, you optimise *expected* outcome 2010-12-01T23:38:33 virdo, the thing is if the opponent DOESN"T know, that it is rps 2010-12-01T23:38:56 eburnette: l8r! 2010-12-01T23:38:58 eburnette: see ya :) 2010-12-01T23:39:01 still reading your post 2010-12-01T23:39:20 *** eburnette has quit IRC (Quit: Page closed) 2010-12-01T23:39:41 for instance, if I ask you for the square root of 3.. and the answer field is an integer.. 2010-12-01T23:39:53 integers aren't a field :P 2010-12-01T23:39:59 you can basically choose 1, or 2, or can choose to not give an answer 2010-12-01T23:40:06 * sigh hides 2010-12-01T23:40:17 hehe 2010-12-01T23:40:42 and if one of those 3 at random, is the 'right' answer.. and my bot knows this, and yours doesn't.. 2010-12-01T23:40:59 then I can guess your answer is most likely to be either 1 or 2.. 2010-12-01T23:41:07 burny: you need multiplicative inverses for a field :) 2 * x = 1 doesn't have a solution in your set 2010-12-01T23:41:12 and then choose whatever answer doesn't lose to the 1 and 2 2010-12-01T23:41:31 :P 2010-12-01T23:41:36 I don't know math stuff 2010-12-01T23:41:48 sorry, I was just making a joke 2010-12-01T23:41:58 ya, I didn't get it till anti said his bit 2010-12-01T23:41:59 yeah, and i was painfully explaining :) 2010-12-01T23:43:08 What's solvable to you anyways? Guaranteed a win or stalemate at worst? or, knowing that if I do A and enemy does B, I will win, but if I do A and enemy does C, I will lose? 2010-12-01T23:43:37 solvable is finding a nash equilibrium 2010-12-01T23:43:54 well for these 2 player zero sum games 2010-12-01T23:44:05 antimatroid: so... the first part of what I wrote? or the second? :) 2010-12-01T23:44:20 breakign it down to an at-worst, paper rock scissors scneario 2010-12-01T23:44:26 *** davidjliu has joined #aichallenge 2010-12-01T23:44:46 which is probably the same thign as what anti just said 2010-12-01T23:44:51 virdeo: either picking your pure strategy, or assigning a probability distribution over your pure strategies that is playing a nash equilibrium 2010-12-01T23:44:59 a game is solved if you have a general algorithm for finding an optimal strategy. bonus points if the algorithm actually terminates before you die of old age 2010-12-01T23:45:09 :) 2010-12-01T23:45:16 jmcarthur: i disagree :P 2010-12-01T23:45:22 a game is solved once you get the strategy 2010-12-01T23:45:38 i could write you a program do find the optimal strategy in chess, does that mean i've solved it? 2010-12-01T23:45:39 antimatroid: i was talking about the more general case (different maps, etc.) 2010-12-01T23:45:55 jmcarthur: but can a planet wars RPS like situation be considered solvable? if not, that would make planet wars not solvable, quite simply, right? :) 2010-12-01T23:45:58 antimatroid: a specific map is solved by a strategy 2010-12-01T23:46:11 so chess is solved? 2010-12-01T23:46:19 virdo: it's certainly *solvable*. just not necessarily within time constraints... 2010-12-01T23:46:47 no.. it's solved when the solution is deterministic, and can be solved in O(n), where n is game duration 2010-12-01T23:46:56 jmcarthur: but what is solvable, I come back... in RPS there is no optimum strategy -- as it depends on luck? therefore, it makes the whole planet wars not solvable? am I grasping it correctly here? :) 2010-12-01T23:47:13 virdo: there is an optimal strategy when you consider mixed strategies 2010-12-01T23:47:16 ok, so planetwars is not solvable due to RPS like maps. 2010-12-01T23:47:17 burny: why O(n)? 2010-12-01T23:47:30 if you just play through the finite number of possible non-repeat moves.. that's not 'solving' it, unless there are sufficiently few possible moves(liek tic tac toe) 2010-12-01T23:47:55 tic tac toe is hard to solve still when you blow dimensions out 2010-12-01T23:47:56 antimatroid: i was offering a more general definition of solvable for the purposes of a game with varying parameters. tricking me into conflating my own meaning with the traditional meaning is kind of a trap 2010-12-01T23:48:02 that's my test game on my game theory library trees 2010-12-01T23:48:10 antimatroid: luckily, you can't have more than 4 2010-12-01T23:48:26 jmcarthur: i'm a maths student, I get ripped to pieces on things like that, it just comes naturally to me :) 2010-12-01T23:48:42 antimatroid: i won't fall for it :P 2010-12-01T23:48:44 virdo, if we are playing a game of aim the sniper rifle.. and I have the game devolved down to a case of RPS.. and you have obscure huristics and what not.. 2010-12-01T23:48:50 Top 10 players: bocsimacko(3765), _iouri_(3565), Slin-.-(3524), _Astek_(3501), jimrogerz(3500), Accoun(3498), george(3494), GreenTea(3489), asavis(3480), bix0r4ever(3476) 2010-12-01T23:48:51 my analysis lecturer said she would take marks off us if we tried to bullshit her in the exam 2010-12-01T23:48:53 then I will win, or tie, in all cases 2010-12-01T23:49:55 because if you happen to play an R or P or S.. then it's random.. but if you pick an inferior move.. I can alter my original RPS choice, to not lose 2010-12-01T23:49:57 burny: but for RPS planetwars, you can't guarantee win or tie... you can lose too... that's the whole point of RPS :) 2010-12-01T23:50:23 burny: you mean, by going into the past and changing your move to counter mine? wtf ;) 2010-12-01T23:50:25 virdo: you can guarentee an expected outcome for a randomised strategy 2010-12-01T23:50:35 and optimise that 2010-12-01T23:50:38 antimatroid: as long as i'm allowed to bind a meaning to a word without being flagged for overriding an existing definition i don't think i'm making an error here. it was merely a suggestion to adopt a more useful definition for the purposes of a game with observable but not-given-up-front parameters, that's all 2010-12-01T23:50:54 if I pick rock.. and you pick didn't pick paper.. 2010-12-01T23:51:01 or rock, or scissors.. 2010-12-01T23:51:03 then I win 2010-12-01T23:51:04 and perhaps i should have simply restricted it to algorithms with certain running times 2010-12-01T23:51:04 jmcarthur: sorry, i wasn't trying to offend :) 2010-12-01T23:51:14 otherwise.. it wasn't an RPS situation 2010-12-01T23:51:18 ok guys, wikipedia never lies: "A solved game is a game whose outcome can be correctly predicted from any position when each side plays optimally."... You can't play optimally on RPS map. 2010-12-01T23:51:21 although i still don't like the part where it says you've solved it if you have an algorithm that will solve it in finite time 2010-12-01T23:51:36 yes you can.. 2010-12-01T23:51:44 optimally in RPS sitation.. is random between R P and S 2010-12-01T23:51:49 or.. 2010-12-01T23:51:53 antimatroid: yeah i could have said it had to have a specific bound rather than simply be finite 2010-12-01T23:51:58 UHM, UHM... That makes it UNSOLVABLE :) 2010-12-01T23:51:59 :) 2010-12-01T23:52:01 to use knowledge of your bot.. to choose the beter of R P and S 2010-12-01T23:52:09 virdo: look up what mixed strategies are 2010-12-01T23:52:12 You cannot predict the outcome, making it not a solved game, BY DEFINITION... 2010-12-01T23:52:32 virdo: you can come up with an *expected* outcome 2010-12-01T23:52:38 virdo: "on average" 2010-12-01T23:52:51 Nash provied every non-cooperative finite n-player simultaneous game has at least one mixed strategy nash equilibrium back in the 40's 2010-12-01T23:52:53 a solution exists 2010-12-01T23:53:00 checkers is solved.. should red resign on first move? 2010-12-01T23:53:21 no, because if both play perfectly, it's a draw :D 2010-12-01T23:53:27 no 2010-12-01T23:53:31 no? 2010-12-01T23:53:40 yes.. 2010-12-01T23:53:46 but the opponent is NOT playing perfectly 2010-12-01T23:53:58 virdo: read the second paragraph of "Perfect play" section in the wikipedia page you quoted yourself 2010-12-01T23:54:02 hasn't it been shown that one of the players in checkers can always force a win? 2010-12-01T23:54:04 http://en.wikipedia.org/wiki/Solved_game#Perfect_play 2010-12-01T23:54:07 "A solved game is a game whose outcome can be correctly predicted from any position when each side plays optimally." -- "each side plays optimally" 2010-12-01T23:54:07 it talks about rps 2010-12-01T23:54:12 yes.. black wins in checkers 2010-12-01T23:54:24 virdo: ^^ 2010-12-01T23:55:07 virdo: you can play the optimal mixed strategy at each choice you have 2010-12-01T23:55:25 hence each side plays optimally at each choice 2010-12-01T23:55:48 there's actually a whole line of research about picking strategies that give you better results when you believe your opponent is imperfect, so if you find yourself playing as red in a game of checkers that might be a good place to start 2010-12-01T23:56:02 it wont give the same result for every instance of the game, but your expected payoff is optimised for the game as a whole, meaning in an infinitely repeated process with an unknown enemy, you will do optimally 2010-12-01T23:57:13 having said that, finding mixed strategy nash equilibrium is a pita 2010-12-01T23:57:49 you can use simplex for 2 player zero sum games, above that and you need to use things like newtons method to try and find them all or grobner bases (Which i don't understand) 2010-12-01T23:57:52 not that bad once you have a game matrix 2010-12-01T23:58:02 okay yeah, for 2 player games 2010-12-01T23:58:24 i'm planning to do my thesis next year on the later :) 2010-12-01T23:58:31 burny: from that wiki page, it says "From the standard starting position, both players can guarantee a draw with perfect play" for checkers :P 2010-12-01T23:58:33 i wasn't under the impression that non-zero sum was much harder. is it really? 2010-12-01T23:58:39 i never tried it 2010-12-01T23:58:44 yes it is 2010-12-01T23:58:50 hmm, okay 2010-12-01T23:59:09 in zero sum, what i gain you lose 2010-12-01T23:59:17 that adds some symmetry in there somewhere 2010-12-01T23:59:29 i'm mostly self taught with game theory though, so i don't know as much as i would like 2010-12-01T23:59:55 symmetry probably wasn't hte right word there 2010-12-01T23:59:55 *** dmj111 has quit IRC (Remote host closed the connection)