2011-04-19T00:02:43 You disappoint me painfully slow internet. I give up. 2011-04-19T00:08:40 I bet I can beat you in a slowest internet competition 2011-04-19T00:09:29 Check out these numbas: http://www.speedtest.net/result/1258725911.png 2011-04-19T00:10:55 We just upgraded. 500 kb/s just wasn't enough anymore. Also that's evidently not a plan anymore. 2011-04-19T00:11:19 jaycobcoleman: i could probably do worse over serial cable 2011-04-19T00:11:30 i once connected a p2 laptop to the internets that way 2011-04-19T00:13:05 I once tried play Black Ops using my cell phone service. Terrible experience 2011-04-19T00:18:00 Where's sigh to brag about google's ridiculously fast internet? 2011-04-19T00:26:27 i lost my screenshot of my speed test 2011-04-19T00:28:37 is the 'owner' value that follows "a row col" going to be a character or integer value? 2011-04-19T00:31:21 I'm thinking integer. I'll just go with that. 2011-04-19T00:31:22 woot, portal 2 is out 2011-04-19T00:31:40 jaycobcoleman: I think it's a character but I'm not sure 2011-04-19T00:32:25 I'll play it later this week :) 2011-04-19T00:33:08 but I'm not sure about the character thing. I should think it would be consistent and the owner value for your own ants is 0 2011-04-19T00:36:43 ok, I'm really not sure. it should be easy enough to check by running the engine though 2011-04-19T00:44:11 *** brainiack has quit IRC (Read error: Connection reset by peer) 2011-04-19T00:48:17 *** Mathnerd314 has quit IRC (Ping timeout: 240 seconds) 2011-04-19T01:01:19 *** brainiack has joined #aichallenge 2011-04-19T01:08:22 *** Accoun has quit IRC (Ping timeout: 258 seconds) 2011-04-19T01:12:29 *** needsch has joined #aichallenge 2011-04-19T01:22:17 *** Cyndre has quit IRC (Ping timeout: 240 seconds) 2011-04-19T01:49:17 *** Ice_Harley has joined #aichallenge 2011-04-19T01:50:50 *** antimatroid has joined #aichallenge 2011-04-19T02:00:26 *** Eruonen has joined #aichallenge 2011-04-19T02:19:54 *** McLeopold has joined #aichallenge 2011-04-19T02:40:05 ergh 2011-04-19T02:40:40 i cannot find that xkcd comic about "just wanted to learn how to make games" 2011-04-19T02:44:45 :( 2011-04-19T02:45:01 i've never had so much trouble finding on in my life, and on one of the very rare instances i want it for a reference 2011-04-19T02:49:38 I'm not sure I know which one you're referring to. Do you remember anything else about it? 2011-04-19T02:54:38 *** ltriant has quit IRC (Read error: No route to host) 2011-04-19T02:58:45 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-04-19T03:04:46 the professor was saying something about finite state machines or something 2011-04-19T03:04:58 and the kid is sitting there thinking about how they just wanted to learn how to make games 2011-04-19T03:05:00 or something like that 2011-04-19T03:11:27 "i've never had so much trouble finding on in my life"? you doubt your studies are worth the time? 2011-04-19T03:11:48 antimatroid: you sure it was xkcd? 2011-04-19T03:12:01 pretty sure 2011-04-19T03:12:15 i have read all of them, but i don't think i saw that one 2011-04-19T03:12:24 there exists a comic along those lines 2011-04-19T03:12:38 either that or i can't differentiate from the dream world 2011-04-19T03:12:58 try this: http://www.google.com/cse?cx=012652707207066138651%3Azudjtuwe28q&ie=UTF-8&q=finite+state+machines&sa=Search&siteurl=xkcd.com%2F 2011-04-19T03:13:16 amstan: not sure if it exists, but can you make it easy for me to tell contest bot to spit out a link to the specs page? 2011-04-19T03:13:51 learn key as value 2011-04-19T03:14:01 contestbot: help learn 2011-04-19T03:14:01 amstan: (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. 2011-04-19T03:14:27 contestbot: help where 2011-04-19T03:14:27 amstan: (where ) -- Alias for "whatis $*". 2011-04-19T03:14:32 contestbot: help whatis 2011-04-19T03:14:32 amstan: (whatis [] [--raw] []) -- Looks up the value of in the factoid database. If given a number, will return only that exact factoid. If '--raw' option is given, no variable substitution will take place on the factoid. is only necessary if the message isn't sent in the channel itself. 2011-04-19T03:14:36 antimatroid: that? 2011-04-19T03:25:10 *** chris__0076 has joined #aichallenge 2011-04-19T03:27:32 *** chris_0076 has quit IRC (Ping timeout: 246 seconds) 2011-04-19T03:34:39 *** amstan_ has joined #aichallenge 2011-04-19T03:34:39 *** ChanServ sets mode: +o amstan_ 2011-04-19T03:37:26 *** amstan has quit IRC (Ping timeout: 276 seconds) 2011-04-19T03:40:15 amstan: ? 2011-04-19T03:40:20 ah he's gone 2011-04-19T03:40:27 a1k0n: my message got passed on then? :P 2011-04-19T03:42:40 *** j3camero_ has joined #aichallenge 2011-04-19T03:47:17 *** j3camero has quit IRC (*.net *.split) 2011-04-19T03:55:31 *** woudshoo has joined #aichallenge 2011-04-19T04:04:16 *** sigh has joined #aichallenge 2011-04-19T04:10:09 sigh: Have you fixed the python food overwriting issue? 2011-04-19T04:10:30 what issue was this? 2011-04-19T04:10:50 the one that bhasker brought up was not a problem with the engine, I thought? 2011-04-19T04:12:24 I think it was just a simple problem with the python starter's input parsing 2011-04-19T04:13:28 oh, the starter packs... I don't touch them... last time I did McLeopold reverted my change :D 2011-04-19T04:13:56 haha I was just asking because I'm pretty sure it's on the issue list and you're assigned to it 2011-04-19T04:14:03 *** FireFly has joined #aichallenge 2011-04-19T04:14:26 nope, it's not assigned to me 2011-04-19T04:14:44 evidently I'm crazy 2011-04-19T04:14:51 :P 2011-04-19T04:17:12 ooohhh, now I remember. McLeopold said you were the one to point out the issue. This would explain my confusion 2011-04-19T04:17:34 yeah, I pointed out that and the vision issue 2011-04-19T04:18:02 I believe the vision is fixed, isn't it? 2011-04-19T04:18:29 use c++ :P 2011-04-19T04:19:01 jaycobcoleman: nope 2011-04-19T04:19:31 Oh, well I've got working vision code if contribution is needed 2011-04-19T04:20:27 *shrug* talk to McLeopold about it 2011-04-19T04:20:32 animatroid: unfortunately I know slightly more about python than C++ :P But if I get a bot up in python that I'm satisfied with I'll probably doodle in C++ 2011-04-19T04:22:30 I've already perused through the code enough to have a basic inderstanding of what I'm doing, but little more than that 2011-04-19T04:26:36 *** berak has joined #aichallenge 2011-04-19T04:36:19 aichallenge: sigh epsilon * r6c63d32 / ants/ants.py : Update food bonus to include current food and ants - http://bit.ly/i5lLQ3 2011-04-19T04:42:15 *** kaemo has joined #aichallenge 2011-04-19T04:44:07 *** Naktibalda has joined #aichallenge 2011-04-19T04:46:41 *** antimatroid has quit IRC (Quit: Leaving.) 2011-04-19T04:46:55 *** antimatroid has joined #aichallenge 2011-04-19T04:52:02 aichallenge: janzert epsilon * rdfa4f39 / (7 files in 2 dirs): Advance the setup scripts closer to a fully working state - http://bit.ly/hzsv1g 2011-04-19T05:04:09 animatroid: What is the purpose of spawnradius2? 2011-04-19T05:04:42 it's the defines the distance at which food will become an ant 2011-04-19T05:05:25 but the ant will necessarily spawn on the location of the food? 2011-04-19T05:05:33 yes 2011-04-19T05:06:03 right, when a current ant gets within that distance 2011-04-19T05:06:05 phwew. Good. 2011-04-19T05:14:04 note that it is the actual radius squared 2011-04-19T05:14:05 hence the 2 2011-04-19T05:16:55 I nose 2011-04-19T05:31:41 *** Stocha has joined #aichallenge 2011-04-19T05:31:56 *** mceier has joined #aichallenge 2011-04-19T05:31:58 mggn ?* I have heard my anme. 2011-04-19T05:32:00 name 2011-04-19T05:32:18 Writting stuff for helping people improve the starter bot ? 2011-04-19T05:32:30 Mmgn. 2011-04-19T05:32:38 i'm not sure i understand the challenge :p 2011-04-19T05:32:57 I could explain step by step how to build my ref bot. 2011-04-19T05:33:09 but that seems a bit overdoing it. 2011-04-19T05:33:21 Most people will use A* it seems. And i don't. 2011-04-19T05:33:36 i'm not convinced on a* 2011-04-19T05:33:40 not for this 2011-04-19T05:33:42 So the basic step for improving the bot seems to be : do nothing 2011-04-19T05:33:53 then : go for food with A* somehow 2011-04-19T05:34:20 and then : send at most one ant to any food, starting with the closest ant, to the closest food. 2011-04-19T05:34:34 beside that i wouldn't have much to say about making stronger bots. 2011-04-19T05:34:58 well, none of my ideas past my refbot worked so far. 2011-04-19T05:35:48 it always degrades the strength with Damage rule one. Well, they don't try to fight , so maybe my tests are not the good tests. But still, i don't see anything to say past basic greedy food eating. 2011-04-19T05:36:21 So i'm not sure i'm the right person to tell people how they should improve the starter bots. 2011-04-19T05:36:27 Stocha: off topic, did you see the Go bot vs pro games last week? 2011-04-19T05:36:36 no ? 2011-04-19T05:36:39 how did it go ? 2011-04-19T05:36:49 what go bot was playing ? 19x19 ? 2011-04-19T05:37:02 9x9 and 19x19 handicap games 2011-04-19T05:37:18 did the 9x9 go well for the bot ? 2011-04-19T05:37:33 give me a link i guess :p 2011-04-19T05:37:50 pachi won a 7 stone handicap game against a 9p 2011-04-19T05:38:02 impressive. 2011-04-19T05:38:09 yeah, that game was awesome 2011-04-19T05:38:18 chased and killed a large group 2011-04-19T05:38:25 what was the pro comment ? was he used to bot play ? 2011-04-19T05:38:44 my understanding was that the pros were familiar with how bots play 2011-04-19T05:38:55 I haven't check bot progress for a long time. It's quite stuck for some years now :p 2011-04-19T05:39:12 I'm trying to find the link for the tourney 2011-04-19T05:39:16 okay 2011-04-19T05:39:55 its been making great progress the last few years, just that it has been more incremental so easier to overlook :) 2011-04-19T05:40:08 did it ? 2011-04-19T05:40:21 i totally overlooked it. there was very few article that i saw. 2011-04-19T05:40:32 yeah, from my perspective at least it's been pretty steady progress 2011-04-19T05:40:33 Beside the early mogo and crazystone. 2011-04-19T05:41:06 erica, zen, pachi seem to be the current generation of top contenders 2011-04-19T05:41:40 zen is totaly private as far as i know. I could even not use montecarlo, i wouldn't be able to know it for sure :p 2011-04-19T05:41:45 it 2011-04-19T05:41:47 pachi is open source, I think 2011-04-19T05:41:55 yes pachi is open 2011-04-19T05:42:09 zen is using mcts, I've seen that much on the computer-go list 2011-04-19T05:42:12 what is built in that make a difference with early crazystone/mogo ? 2011-04-19T05:42:22 in pachi i mean. 2011-04-19T05:42:31 for one it's a lot better at handicap games 2011-04-19T05:42:39 yeah. why ? 2011-04-19T05:42:41 due to dynamic komi, i think 2011-04-19T05:43:03 e.g. in the 7H games it imposes a large komi on itself to be more aggressive 2011-04-19T05:43:11 that's quite an old idea. I didn't knew it was successful 2011-04-19T05:43:18 and linearly scales it down during the game 2011-04-19T05:43:19 *** berak has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2011-04-19T05:43:30 yeah, otherwise the bots jus think they are winning and play like crap 2011-04-19T05:43:38 yeah, dynamic komi seems to be one of the small improvements that several groups have gotten to actually work recently 2011-04-19T05:43:43 it's hard to tell what "crap" is. 2011-04-19T05:44:19 Stocha: and sometimes it is quite easy 2011-04-19T05:44:28 But if dynamic komi works, that then says that without it moves were objectivilly crap, after all :p 2011-04-19T05:44:49 janzert: have you been making anything lately? 2011-04-19T05:44:57 Well, a lot of good player see the bot and say they play crap. but it's beyond me. 2011-04-19T05:44:57 including working on your arimaa bot 2011-04-19T05:45:13 antimatroid: for the contest? or something else? 2011-04-19T05:45:21 anything cool 2011-04-19T05:45:27 since people are talking about cool stuff 2011-04-19T05:45:28 Stocha: it's more obvious when you look at bots playing endgame... you really see them play a good endgame 2011-04-19T05:45:49 how do you define "good endgame" ? 2011-04-19T05:45:53 either they are ahead, and they play crap because they can get away with it... other they are behind and have resigned already or are playing desperate moves 2011-04-19T05:45:56 nothing cool :( just checked in some more work on the contest setup scripts that kept me occupied the last few days 2011-04-19T05:45:57 near optimal play 2011-04-19T05:46:00 mogo was notoriously strong at end game. 2011-04-19T05:46:05 :P 2011-04-19T05:46:13 optimal in the sense of the game that try to maximize score ? 2011-04-19T05:46:21 Stocha: they are good at endgame, except you rarely see them exhibit it 2011-04-19T05:46:38 yes, optimal in the sense of maximising score 2011-04-19T05:47:04 bots don't care about maximising score, only about winning 2011-04-19T05:47:05 (score being stone / eyes count at the end of the game. Not percent of winning or anything alike :p ) 2011-04-19T05:47:20 sigh: depends on how the bot is written :P 2011-04-19T05:47:34 antimatroid: sure, I'm talking about the current implementations 2011-04-19T05:47:35 If the game goal is winning. Then carying about winning makes sense. 2011-04-19T05:47:59 Stocha: not disputing that... I'm not saying that the bots should play differently... just stating a fact 2011-04-19T05:48:39 bots do play a very strong endgame, it's just not optimal in terms of maximising score so some of the moves look like 30k moves 2011-04-19T05:48:51 unless the game is really close 2011-04-19T05:48:53 yeah, the fact is the bot play crap with handicap game i guess. 2011-04-19T05:49:06 yes, and dynamic komi is an attempt to fix that 2011-04-19T05:49:13 and it seems to work quite well 2011-04-19T05:49:18 I think they also have trouble with seki. Maybe even in the endgame. Although it might have changed. 2011-04-19T05:49:42 I think there was a problem with Seki+ko also. 2011-04-19T05:49:51 before the pros pachi was being tuned against some of the high dans, and 7-8d were having trouble beating it on 6 stones 2011-04-19T05:50:12 yeah. 2011-04-19T05:50:15 like, pachi won more often than not 2011-04-19T05:50:19 that make it 1d. not very impressive. 2011-04-19T05:50:27 pachi is 3d on KGS 2011-04-19T05:50:33 comparing to early crazystone i mean. 2011-04-19T05:51:15 That doesn't sound like a revolution. More like they are trying to work around some weakness by heavy engineering. i doubt that will lead past 6d 2011-04-19T05:51:41 the difference between dan ranks is huge in terms of the effort required 2011-04-19T05:52:01 i think early montecarlo /winning crazy stone went beyond dan rank. some years back on kgs. 2011-04-19T05:52:16 sigh : ? 2011-04-19T05:52:21 what do you mean ? 2011-04-19T05:52:49 i have been told by some that they had found a problem at 4k, but that 1d and 3d felt the same to them. 2011-04-19T05:53:06 i mean that to improve a single rank at dan level is difficult and gets progressively harder 2011-04-19T05:53:09 Well someone told me also that 4k and 1d felt the same to him. 2011-04-19T05:53:27 it certainly get hard when you go past 5d on aga. 2011-04-19T05:53:29 definelty disagree on the 4k-1d comment 2011-04-19T05:53:39 but your first point does depends on the person 2011-04-19T05:54:12 it's certainly are the same if a person is DDK 2011-04-19T05:54:13 The thing is, a was 4k and sundenly became 1d. And had no idea what changed in his playing :p 2011-04-19T05:54:15 I shouldn't single out dan ranks... at any rank it gets progressivly harder to improve as you get stronger 2011-04-19T05:54:41 Past 5d with aga, you begin to work the asymptotic behavior of the ranking :p 2011-04-19T05:54:45 Stocha: ah, that can happen when you overcome some sort of roadblock in your play :) 2011-04-19T05:54:47 maybe it's not aga 2011-04-19T05:54:55 i mean the system that is used in france. 2011-04-19T05:55:14 sigh: yeah but why don't the person realize that he did ? 2011-04-19T05:55:29 so 4k and 1d in different systems? 2011-04-19T05:55:39 Stocha: because a lot of go is intuitive 2011-04-19T05:55:50 i have some evidence that my own level can go from 4k to 1d and back to 4k. 2011-04-19T05:55:57 Naktibalda: it can be, but I hope we are under the assumption that we are looking at the same system 2011-04-19T05:56:53 human doesn't fit well to ranking systems :p 2011-04-19T05:56:55 Stocha: maybe you are right... but from my perspective I can definitely say 1d are harder to beat 2011-04-19T05:57:03 they never play the same conditions. you can't set their seed or anything. 2011-04-19T05:57:07 4k has a very low chance to beat 1d 2011-04-19T05:57:10 you can't make play a 100 games to be sure. 2011-04-19T05:57:20 1d should have 50%, it's a huge difference 2011-04-19T05:57:35 I'm not talking about winning chances. I'm talking about quality of play. 2011-04-19T05:57:44 wich is a sort of parallel mettric. 2011-04-19T05:58:26 problem with winning chance, is that given enough time, you are supposed to find a 25K that beats a pro in a 3 game series. 2011-04-19T05:58:56 sure, it's only an approximation... it breaks down 2011-04-19T05:59:10 Worst : if a pro do one game against every human living, i will losse against a 25k :p 2011-04-19T05:59:16 but for ranks that aren't too far away it is good enough 2011-04-19T05:59:21 he 2011-04-19T05:59:39 i dunnow. it's only a metric. 2011-04-19T05:59:44 like QI 2011-04-19T06:00:06 you have to have some way of ranking people... and this is quite good 2011-04-19T06:00:15 you seem to dislike uncertainty :P 2011-04-19T06:02:05 Stocha: here is the game where pachi won: http://files.gokgs.com/games/2011/4/14/nutngo1-pachi2.sgf 2011-04-19T06:02:58 a human 3d would not have been able to kill that group :P 2011-04-19T06:03:26 Stocha: chances of pro loosing against 25k is 0.(0) with infinite number of zeroes 2011-04-19T06:03:39 *** Sotcha has joined #aichallenge 2011-04-19T06:03:53 go isn't a poker 2011-04-19T06:06:11 *** Stocha has quit IRC (Ping timeout: 252 seconds) 2011-04-19T06:11:48 okay so it's a win at 7 stone. (how many cores/clusters were playing ?). Maybe it's the famous 1 over 1000000 game the 25k will win against the pro at 7 stone :p 2011-04-19T06:12:49 64 processors I think 2011-04-19T06:13:00 regardless of anything else, that was an impressive kill 2011-04-19T06:13:02 sigh: you wouldn't have killed the group ? you would have lost the game ? 2011-04-19T06:13:13 I don't think I could have killed that group 2011-04-19T06:13:22 form where ? 2011-04-19T06:14:17 say move 70 2011-04-19T06:14:56 against another 3d, you wouldn't be confident ? 2011-04-19T06:15:31 at move 70 I would not be able to read a kill... I would attack though 2011-04-19T06:15:48 maybe the bot was only trying to get a few sente moves. 2011-04-19T06:15:54 *** jaycobcoleman has quit IRC (Quit: Page closed) 2011-04-19T06:15:56 he didn't got them. 2011-04-19T06:16:02 he killed the group instead :p 2011-04-19T06:16:24 anyway, my comment was in regard to playing a pro 2011-04-19T06:16:29 naktibalda: what's that about go not being poker ? you mean poker is a sure thing, and go is all about chances of winning ? 2011-04-19T06:16:51 no, poker has a lot of luck involved 2011-04-19T06:17:01 and't it was proven in Sweden's court :) 2011-04-19T06:17:03 Still it's only one game. I wonder about when they 'll show bot winning two game in a row. 2011-04-19T06:17:37 even if you play 10000000000 poker game between a pro and a rookie, you have randomness built in ? 2011-04-19T06:17:52 then go is like poker. 2011-04-19T06:18:02 rookie can win 20% or 40% of games 2011-04-19T06:18:03 in that respect. 2011-04-19T06:18:12 sure... bots aren't there quite yet to play pros on 7H... but that kill was above 3d level against a pro 2011-04-19T06:18:12 in go he could win 1, when pro is drunk 2011-04-19T06:18:24 a poker game = 1 go move. 2011-04-19T06:18:27 I mean totally drunk 2011-04-19T06:18:30 if by game you mean hand. 2011-04-19T06:18:44 pros get stronger when they are drunk :P 2011-04-19T06:18:44 a rooking playing one good move, is not beyond imagining. 2011-04-19T06:19:09 i know a young french 3d that told me i was stronger when drunk. 2011-04-19T06:19:24 i don't know if the beer is clouding his judgment or whatever. 2011-04-19T06:19:53 :) 2011-04-19T06:20:16 Still you can't compare a one pocker hand and a full 600 moves long go game. 2011-04-19T06:20:33 go games don't last 600 moves o_O 2011-04-19T06:20:40 don't they ? 2011-04-19T06:20:51 what scoring do you use ? 2011-04-19T06:20:59 i use stone count + eye count :p 2011-04-19T06:21:07 poor you 2011-04-19T06:21:17 o_O 2011-04-19T06:21:35 that's call chineese rule. and is compatible with french scoring :p 2011-04-19T06:21:56 yeah... but it doesn't make the game significantly longer 2011-04-19T06:22:07 just enough to fill in the dame 2011-04-19T06:22:24 fill 2011-04-19T06:22:24 No you also feel the inside. 2011-04-19T06:22:26 longest go game is like 400 moves 2011-04-19T06:22:31 why the inside? 2011-04-19T06:22:39 that's ooooold rules, i think 2011-04-19T06:23:01 well it's the same as counting inside that is sure territory, and dame. 2011-04-19T06:23:20 but you can fill the inside up to when there is only stone and eyes. And count that for each players. 2011-04-19T06:23:30 why do you fill in the inside? 2011-04-19T06:23:38 because it doesn't change anything 2011-04-19T06:23:48 and make counting easier :p 2011-04-19T06:24:08 Well, 600 hundred moves is an approximation for length of go game played by bot that don't forfeit. 2011-04-19T06:24:19 that's hardly part of the game... that's just part of counting... 2011-04-19T06:24:23 bot games don't last that long 2011-04-19T06:24:36 historically they did. 2011-04-19T06:24:43 first mogo versions did that :p 2011-04-19T06:24:53 it juste filled all empty point, until the opponent passed. 2011-04-19T06:24:57 fair enough... but not anymore :) 2011-04-19T06:25:08 or there was no empty point one point eye left. 2011-04-19T06:25:18 ah, new days are bad days. 2011-04-19T06:25:38 okay so let's say a go game is 100 moves. 2011-04-19T06:25:56 and let me rephrase : it's not fair comparing a poker hand with a full 100 moves go game. 2011-04-19T06:25:57 there are 361 points on a board, it's hard to make more than 400 moves 2011-04-19T06:26:25 Naktibalda : it would. if there was not that silly capturing rule. 2011-04-19T06:26:26 Stocha: when you were talking about counting eyes, I thought you were talking about group tax http://senseis.xmp.net/?GroupTax 2011-04-19T06:26:40 which is why I said it was oooooold 2011-04-19T06:26:49 I designed a very nice scheme for a go bot once. And ended up with a good idea of an hex bot. there is no silly capture rule in hex. 2011-04-19T06:27:04 Naktibalda: longest pro game is a bit over 400 moves 2011-04-19T06:27:07 411 i think 2011-04-19T06:27:50 yup, 411 http://www.msoworld.com/mindzine/news/orient/go/special/records/longest.html 2011-04-19T06:28:17 group tax would make the game a litle less weird i guess :p. But you don't need that for the counting i proposed. 2011-04-19T06:28:39 yeah, I was confused as to what you were talking about :) 2011-04-19T06:29:40 you can make almost arbritrarilly long game, without playing in one point eyes. by using kos. 2011-04-19T06:30:13 i'm not very gifted with phrasing. lately. 2011-04-19T06:30:29 you can extend the game a lot by capturing large group then playing inside them 2011-04-19T06:31:04 yeah. ko are nice also :p because you can do that in more reasonable terms that putting your stone on 100% own territory :p 2011-04-19T06:31:28 doesn't have to be your own territory 2011-04-19T06:31:37 like this fun problem: http://senseis.xmp.net/?LargestCaptureThatStillCantLive 2011-04-19T06:31:54 if it's 100% owned ennemy is the same. That's what i meant. owned. 2011-04-19T06:33:28 i still have to find a replay in a 16 captured slot that is not 100% owned in an official game :p 2011-04-19T06:33:46 well let's talk about ant then ... 2011-04-19T06:33:57 :) 2011-04-19T06:33:59 i don't understand this game.Ant i mean. 2011-04-19T06:34:22 *** chris___0076 has joined #aichallenge 2011-04-19T06:34:44 What's more complex. go space, or ant space ? 2011-04-19T06:35:07 how many legal ant game is there up to 256x256 maps with two players starting with only one ant ? 2011-04-19T06:36:15 depends on how food is spawned 2011-04-19T06:36:36 counting legal game takes into account all possible food spawning. 2011-04-19T06:36:47 obviously you need a super ko rule also. 2011-04-19T06:36:49 when you have 100 ants you have 4^100 variations for next move, approximatelly, correct? 2011-04-19T06:36:54 *** chris__0076 has quit IRC (Ping timeout: 246 seconds) 2011-04-19T06:37:13 that's sound like a lot. 2011-04-19T06:37:23 that's an upper bound, yes 2011-04-19T06:37:46 yeah, it's a lot 2011-04-19T06:37:53 how many ant with 4 moves can you put on a 256x256 map ? :p 2011-04-19T06:38:01 planet wars had a ridiculous number of possible moves too 2011-04-19T06:38:15 because of cardinality ? 2011-04-19T06:38:25 like sending 56 then 1 different from sending 1 then 56 ? 2011-04-19T06:38:41 even without that 2011-04-19T06:39:00 N planet x N-1 planet ? :p 2011-04-19T06:39:10 say at the start, choosing which ships to send to which planets resulted in a huge number of moves 2011-04-19T06:39:21 no, you have to decide number of ships too 2011-04-19T06:39:21 at the start ? 2011-04-19T06:39:34 yes, even at the start 2011-04-19T06:39:35 one ship has only N-1 planet to go to. 2011-04-19T06:39:51 you have 100 ships 2011-04-19T06:39:52 so that's a cardinality thing. 2011-04-19T06:40:05 that's still only 100xN 2011-04-19T06:40:14 100^N if you are naive 2011-04-19T06:40:20 ah. 2011-04-19T06:40:24 i am. 2011-04-19T06:40:33 N^100 2011-04-19T06:40:35 is it with two players ? 2011-04-19T06:40:36 wrong way 2011-04-19T06:40:40 yes 2011-04-19T06:41:06 one ship has N possible dest doesn't it have ? 2011-04-19T06:41:08 I meant naive meaning that N^100 counts the moves naively as it contains a lot of duplicates 2011-04-19T06:41:51 mggn 2011-04-19T06:41:57 i have to go back to work anyway. 2011-04-19T06:42:11 how is google ? 2011-04-19T06:42:24 mm, i don't remember who was the new google staff member. 2011-04-19T06:42:28 was it you sigh ? 2011-04-19T06:42:32 or someone else maybe. 2011-04-19T06:42:40 yeah, me 2011-04-19T06:42:54 still setting everythng up and get my head around things :P 2011-04-19T06:43:04 free food everywhere! 2011-04-19T06:43:22 i have heard you are encouraged to work on your project and can have free time for that. is it true ? :p 2011-04-19T06:43:34 your own projects 2011-04-19T06:43:41 like go bots, ant bots or whatever. 2011-04-19T06:44:17 20% time... I don't think it can be completely anything 2011-04-19T06:44:22 and that you live in like in a campus. 2011-04-19T06:44:36 *** brainiack has quit IRC (Ping timeout: 246 seconds) 2011-04-19T06:44:37 I'm living in a hotel at the moment :P 2011-04-19T06:44:39 20% time. is that like 2,4 hour per day ? :p 2011-04-19T06:44:50 1 day a week 2011-04-19T06:44:53 no, that's 4.8 2011-04-19T06:45:09 after you finished your other tasks :) 2011-04-19T06:45:22 is sleeping working or not. 2011-04-19T06:45:31 :P 2011-04-19T06:45:46 I do a lot when i sleep. and not much when i don't. 2011-04-19T06:45:57 i really have to go. 2011-04-19T06:46:04 ok, cya 2011-04-19T06:46:08 *** Sotcha has quit IRC (Quit: Page closed) 2011-04-19T06:48:25 *** chris___0076 has quit IRC (Read error: Operation timed out) 2011-04-19T06:50:17 *** chris___0076 has joined #aichallenge 2011-04-19T06:51:32 *** delt0r___ has joined #aichallenge 2011-04-19T06:52:50 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-04-19T06:55:38 *** antimatroid has quit IRC (Ping timeout: 246 seconds) 2011-04-19T07:09:29 *** antimatroid has joined #aichallenge 2011-04-19T07:43:36 *** antimatroid