2011-06-23T00:01:23 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-06-23T00:06:34 *** Zeiris has quit IRC (Ping timeout: 244 seconds) 2011-06-23T00:07:04 *** Zeiris has joined #aichallenge 2011-06-23T00:13:30 *** Zeiris_ has joined #aichallenge 2011-06-23T00:16:29 *** Zeiris has quit IRC (Ping timeout: 252 seconds) 2011-06-23T00:18:29 *** Zeiris__ has joined #aichallenge 2011-06-23T00:21:22 *** Zeiris_ has quit IRC (Ping timeout: 240 seconds) 2011-06-23T00:33:18 *** Zeiris__ is now known as Zeiris 2011-06-23T00:41:02 *** Cyndre__ has joined #aichallenge 2011-06-23T00:44:24 *** Cyndre_ has quit IRC (Ping timeout: 240 seconds) 2011-06-23T01:33:23 *** stocha has joined #aichallenge 2011-06-23T01:33:39 hi 2011-06-23T01:33:48 hi 2011-06-23T01:34:01 the ranking seems to have gone wilder lately 2011-06-23T01:34:41 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-06-23T01:35:56 @seen Zaphus 2011-06-23T01:35:56 stocha: Zaphus was last seen in #aichallenge 16 hours, 54 minutes, and 49 seconds ago: off to eat some icecream and watch TV for a bit - hope to see that snake out there soon ! 2011-06-23T01:36:16 *** stocha has quit IRC (Client Quit) 2011-06-23T01:48:52 *** mcstar has joined #aichallenge 2011-06-23T01:49:35 pendactyl is incompatible with ff 5, so its one more reason no to use it 2011-06-23T02:37:20 *** okayzed is now known as okay 2011-06-23T02:42:17 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-23T02:44:38 *** Palmik has joined #aichallenge 2011-06-23T03:07:54 *** aerique has joined #aichallenge 2011-06-23T03:11:44 *** okay is now known as okayzed 2011-06-23T03:12:03 @seen Rabidus 2011-06-23T03:12:03 Rabidus: Rabidus was last seen in #aichallenge 1 hour, 38 minutes, and 24 seconds ago: hi 2011-06-23T03:12:19 @seen Rabidus 2011-06-23T03:12:19 Rabidus: Rabidus was last seen in #aichallenge 15 seconds ago: @seen Rabidus 2011-06-23T03:12:44 clearly a bug :) 2011-06-23T03:14:38 *** gobang has joined #aichallenge 2011-06-23T03:15:40 *** Kingpin13 has quit IRC (Quit: this used to be the life, but I don't need another one) 2011-06-23T03:32:24 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T03:37:39 *** gobang has joined #aichallenge 2011-06-23T03:40:55 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-06-23T04:05:17 score := s.Turn-1 2011-06-23T04:05:22 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T04:12:19 *** Naktibalda has joined #aichallenge 2011-06-23T04:29:00 *** george1914 has quit IRC (Ping timeout: 252 seconds) 2011-06-23T04:40:38 *** boegel has joined #aichallenge 2011-06-23T05:02:42 *** gobang has joined #aichallenge 2011-06-23T05:05:48 *** stocha has joined #aichallenge 2011-06-23T05:11:34 *** gobang has quit IRC (Ping timeout: 240 seconds) 2011-06-23T05:12:16 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-23T05:16:12 *** gobang has joined #aichallenge 2011-06-23T05:29:38 *** zwembroek has joined #aichallenge 2011-06-23T05:31:33 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T05:32:53 *** _0x47 <_0x47!~slash@p5B12F8A3.dip.t-dialin.net> has joined #aichallenge 2011-06-23T06:04:48 *** Zeiris_ has joined #aichallenge 2011-06-23T06:07:41 *** Zeiris has quit IRC (Ping timeout: 240 seconds) 2011-06-23T06:13:28 *** george1914 has joined #aichallenge 2011-06-23T06:15:28 *** george1914 has joined #aichallenge 2011-06-23T06:32:07 *** onensora has joined #aichallenge 2011-06-23T06:37:45 *** koola has joined #aichallenge 2011-06-23T06:48:04 *** _0x47 <_0x47!~slash@p5B12F8A3.dip.t-dialin.net> has quit IRC (Quit: Leaving.) 2011-06-23T06:55:19 *** koola has quit IRC (Read error: Connection reset by peer) 2011-06-23T06:55:57 *** gobang has joined #aichallenge 2011-06-23T07:04:52 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T07:05:17 *** gobang has joined #aichallenge 2011-06-23T07:06:07 *** stocha has joined #aichallenge 2011-06-23T07:06:23 The early game termination shoudn't give bonus points. 2011-06-23T07:06:49 especially not to the less than 10% cumulative ant count players. 2011-06-23T07:07:19 because if some other had a good fight, and were eliminated, they would then loose. this behavior favors hiding and blocking. 2011-06-23T07:07:42 here is sample game showing the bad final ranking : http://aichallengebeta.hypertriangle.com/visualizer.php?game=34074&user=226 2011-06-23T07:08:17 stocha: i somewhat agree 2011-06-23T07:08:22 i'll have a chat with mcleopold about it later 2011-06-23T07:08:25 maybe you *could* give the remaining player a percentage of food corresponding to their percent of ants. 2011-06-23T07:08:51 stocha: the idea of not giving the dominant player their "share" of the bonus is to encourage them to end the game before the cut off 2011-06-23T07:09:03 but simply not giving the bonus to anyone would also achieve this 2011-06-23T07:09:08 antimatroid: okay. The early termination is probably a good idea though. And in any case, as long as rule are known, there is no severe issues there (people will just make more shy bots then) 2011-06-23T07:09:21 yep pretty much :) 2011-06-23T07:09:22 antimatroid: yeah no giving the bonus is probably best in fact 2011-06-23T07:09:31 because the leading bot HAD to beat the lossing bots. 2011-06-23T07:09:38 i tend to agree right now, but there's always more things than i think of at one time 2011-06-23T07:09:46 (of course if there is perfect blocking, then it's a bit harsher, but still) 2011-06-23T07:10:10 i'm pretty keen to get rid of perfect blocking if possible for that reason 2011-06-23T07:10:27 antimatroid: what i find most probably not satisfying is to have the lossing shy bots receives such a huge bonus for doing nothing. 2011-06-23T07:10:54 the food bonus isn't given at max turns is it? 2011-06-23T07:11:06 so i'm happy with percent of total bonus given according to controled territory share, or ant share (if easier and cleaner) 2011-06-23T07:11:09 i think even with that it makes more sense just to not give it if we can the game early 2011-06-23T07:11:17 we wont do that 2011-06-23T07:11:37 it doesn't give any real incentive for the leading bot to end the game, just continue growing or whatever 2011-06-23T07:11:53 anyway start by talking to McLeopold about the problem. we shall see how it goes. 2011-06-23T07:12:26 i don't anticipate that i will be opposed to a solution :) 2011-06-23T07:12:53 he 2011-06-23T07:13:05 not i. why do i keep typing i, when i want to we. mystery. 2011-06-23T07:13:26 when i want to type HE. oh man. 2011-06-23T07:14:09 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T07:15:14 *** janzert has quit IRC (Read error: Connection reset by peer) 2011-06-23T07:16:17 *** janzert has joined #aichallenge 2011-06-23T07:17:28 *** moondust has quit IRC (Ping timeout: 244 seconds) 2011-06-23T07:18:11 *** moondust has joined #aichallenge 2011-06-23T07:21:43 *** orivej has joined #aichallenge 2011-06-23T07:30:30 *** Zaphus has joined #aichallenge 2011-06-23T07:31:59 Stocha: nice work on the snake 2011-06-23T07:35:55 *** FireFly has joined #aichallenge 2011-06-23T07:37:49 *** gobang has joined #aichallenge 2011-06-23T07:39:28 Zaphus: ah here you are ! 2011-06-23T07:39:47 Zaphus: i was like, i have to do that i less than 3 hours, before your return. And it took you 48 hours ! 2011-06-23T07:40:14 Yeah, life got in the way :-) 2011-06-23T07:40:26 Zaphus: maybe i'll add a bit of escape code to snake head, so it go the the tail if it feels in danger ? 2011-06-23T07:40:34 what do you think. 2011-06-23T07:40:49 Yeah, nice idea 2011-06-23T07:40:50 i don't want to work on it too much though, so i'll do only very easy stuff :p 2011-06-23T07:40:59 Like i had this idea of breeding new snakes etc. 2011-06-23T07:41:28 i still have to finish (or even start) my heat map bot :p 2011-06-23T07:41:34 Could the code cope with multiple snakes ? 2011-06-23T07:42:40 *** orivej has quit IRC (Ping timeout: 260 seconds) 2011-06-23T07:43:07 *** Rabidus has quit IRC (Ping timeout: 258 seconds) 2011-06-23T07:53:10 *** dr- has quit IRC (Ping timeout: 260 seconds) 2011-06-23T08:02:09 Zaphus: i don't know. I would have to modify a few thing most probably. 2011-06-23T08:03:03 Zaphus: but i didn't have much work to do to make the snake alive, because i had parts of bot librairies all around the place. So i found most of what i needed (and what i didn't found anything for, i didn't do : p ) 2011-06-23T08:03:35 Zaphus: my testing showed that the snake can break and do hawfull things. But in server's game mostly you don't see those. 2011-06-23T08:04:10 Zaphus: i uploaded the snake try to bit it's tail if it feels head menaced versions; We'll see if it looks as crapy as it does in my testings :p 2011-06-23T08:05:26 Zaphus: one sad thing about this contest, is that the official game is all about who can make it to the top. But i'm pretty sure there is much to expect from the ant game than just that. People having only account will make original works hard though. 2011-06-23T08:05:51 only one account 2011-06-23T08:06:59 *** dr- has joined #aichallenge 2011-06-23T08:07:11 i wonder if the snake bot i good for early testings :p With my setting it was beaten by my random bot pretty much every game (1vs1) 2011-06-23T08:09:58 this guys has made 144 submission. 2011-06-23T08:10:55 *** olexs has joined #aichallenge 2011-06-23T08:12:47 i don't see the be afraid features activates : / 2011-06-23T08:14:00 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T08:14:40 seeing how some people with very weak bots have made a lot of resubmission, i wonder if more in depth tutorial shouldn't be done. Yeah, i know, if i want it, i should do it. bleh. 2011-06-23T08:18:14 is there a hunter bot on the server ? 2011-06-23T08:18:17 *** gobang has joined #aichallenge 2011-06-23T08:18:51 *** stocha has quit IRC (Quit: Page closed) 2011-06-23T08:19:06 quick question... anyone know how to get playgame.py to run with identical food spawning? - it takes seed argument, but keeping seed fixed still produces different spawns 2011-06-23T08:19:40 note. my playgame.py was taken from git quite a few days ago 2011-06-23T08:22:14 george1914: is "--food symmetric" what you're looking for? 2011-06-23T08:22:26 you need to be using the symmetric maps my generator spits out 2011-06-23T08:22:51 *** stocha has joined #aichallenge 2011-06-23T08:22:52 "So you can see for an ant to survive the opponents it is facing it must have more focus than they do. If any opponent is equally or less occupied it will die." 2011-06-23T08:23:06 i don't get it ... 2011-06-23T08:23:12 https://github.com/aichallenge/aichallenge/wiki/Ants-Occupied-Battle-Resolution-Method 2011-06-23T08:23:12 antimatroid: my food -is- symmetric - but each time i run playgame.py, it appears in different places - a problem for debugging 2011-06-23T08:23:33 george1914: oh you want to set the seed? 2011-06-23T08:23:44 try --seed 40 but i'm guessing 2011-06-23T08:24:14 stocha: count up the number of enemy ants within the battle radius of each ant 2011-06-23T08:24:16 i guess mcstar and aerique would advise i switch to language with REPL :) 2011-06-23T08:24:32 antimatroid: ah, i was only using 'player seed' 2011-06-23T08:24:34 if there exists an enemy ant within the battle radius of yours with less than or equal to the same number of enemies in range, your ant dies 2011-06-23T08:24:37 will try that, thanks 2011-06-23T08:24:46 --engine_seed, no? 2011-06-23T08:24:50 antimatroid: is it what you do read in what i pasted ? 2011-06-23T08:24:54 i was guessing 2011-06-23T08:25:00 --engine_seed 2011-06-23T08:25:11 stocha: i'm just explaining the rule in terms i understand better 2011-06-23T08:25:21 *** dr- has quit IRC (Ping timeout: 264 seconds) 2011-06-23T08:25:34 imo the worded explanations are confusing, but apparently as a maths person i'm not in the majority with that opinion 2011-06-23T08:25:59 antimatroid: i my opinion the problem is not that it is confusing. It is that it is bugged. 2011-06-23T08:26:11 bugged? 2011-06-23T08:26:23 unless i read it wrong. So i would like other to confirm/unfirm it is wrong. 2011-06-23T08:26:26 oh, less focus? 2011-06-23T08:26:39 ok, --engine_seed worked, thanks for that folks :) 2011-06-23T08:27:12 stocha: does this read better? "So you can see for an ant to survive the opponents it is facing it must have less focus than they do. If any opponent is equally or less occupied it will die." 2011-06-23T08:27:33 with a , after the first survive 2011-06-23T08:27:50 antimatroid: it still is not consistent with the rule right ? 2011-06-23T08:27:58 it is then i think 2011-06-23T08:28:07 antimatroid: i really think you guys should use ONE term in the story. Not mix up many of them. 2011-06-23T08:28:18 i didn't write that 2011-06-23T08:28:19 so goes with focus, and stick to it. or with distracted and stick with that. 2011-06-23T08:28:32 antimatroid: it seems like nobody proofed read it. 2011-06-23T08:28:41 yet it was commited. 2011-06-23T08:29:19 I'm afraid it will be VERY confusing in it's current state. Which was probably not the goal. 2011-06-23T08:29:30 actually, you're right, it's really wrong, how about... 2011-06-23T08:30:06 an ant dies if it is equally or more distracted that any of it's opponents. 2011-06-23T08:30:16 yes 2011-06-23T08:30:26 so each enemy ant must be more occupied 2011-06-23T08:30:39 .... 2011-06-23T08:30:41 *** dr- has joined #aichallenge 2011-06-23T08:30:56 "So you can see for an ant to survive, each enemy ant must be more occupied. If any opponent is equally or less occupied it will die." 2011-06-23T08:30:58 is that better? 2011-06-23T08:31:03 *** Rabidus has joined #aichallenge 2011-06-23T08:31:46 occupy = to dwell or reside in. 2011-06-23T08:31:55 how is that related to distraction ? :p 2011-06-23T08:32:07 occupied := number of enemy ants occupying your battle efforts 2011-06-23T08:32:09 antimatroid: are you english native speaker ? 2011-06-23T08:32:14 i am 2011-06-23T08:32:29 i'm not. 2011-06-23T08:32:36 so i used google to find meaning :p 2011-06-23T08:32:54 is that logically correct now though? 2011-06-23T08:33:45 i don't think so. 2011-06-23T08:34:08 an ant is as distracted as the number of ennemy in range i believe. 2011-06-23T08:34:22 ah distracted=occupied 2011-06-23T08:34:27 *** olexs has quit IRC (Quit: Leaving.) 2011-06-23T08:34:28 "So you can see for an ant to survive, each enemy ant must be more occupied. If any opponent is equally or less occupied your ant will die." that might be clearer 2011-06-23T08:34:40 so an ant is as occupied as the number of ennemy in range then 2011-06-23T08:34:45 yep 2011-06-23T08:34:57 *** gobang has quit IRC (Ping timeout: 264 seconds) 2011-06-23T08:34:58 and if you're less occupied than all of your enemies in range, you survive 2011-06-23T08:35:09 so it's probably logically correct. But a bit beyond my ability to understand :) 2011-06-23T08:35:20 do you understand the rule in general? 2011-06-23T08:35:34 occupy just don't hit anything in me :) 2011-06-23T08:35:34 it's really funny, it's hard to understand, but once you do i think it's quite simple 2011-06-23T08:35:58 your time can be occupied by something 2011-06-23T08:36:02 it's a fairly common english word 2011-06-23T08:36:04 i think i understand it with both "ennemy in range" and "distracted" 2011-06-23T08:36:13 but i certainly have trouble with occupied. 2011-06-23T08:36:28 i think of it as |enemies(a)| < |enemies(b)| forall b \in enemies(a) 2011-06-23T08:36:34 which seems understandable when i look at google for occupied. 2011-06-23T08:37:04 *** gobang has joined #aichallenge 2011-06-23T08:37:06 occupied is more often used in the millitary meaning 2011-06-23T08:37:19 for link on google being how germany occupied france. 2011-06-23T08:37:41 first link on google 2011-06-23T08:37:58 google is different for each person 2011-06-23T08:38:14 it's even different for the same person on the same account on a different computer/location 2011-06-23T08:38:27 distracted though refers to "mentally confused" which i feel goes well with dying. for some reason :) 2011-06-23T08:38:51 antimatroid: you are sort of wrong. because the main stream stay coherent. 2011-06-23T08:39:03 i use it for example, for checking spellings. 2011-06-23T08:39:09 it works well for that; 2011-06-23T08:39:32 albeit results may vary a bit. i always can distinguish good spelling from bad speelings. 2011-06-23T08:39:41 speelings 2011-06-23T08:39:43 (in french) 2011-06-23T08:40:10 en de twa cut sanc sis :P 2011-06-23T08:40:14 can i spell well? 2011-06-23T08:40:34 it could be for a french one :D 2011-06-23T08:40:35 "bad spelling" = 2,800,000 hit on google 2011-06-23T08:40:52 "bad spellings"= 24,700 hits on google 2011-06-23T08:41:07 so "bad spelling" is more frenquent than "bad spellings" 2011-06-23T08:41:31 those actual numbers may vary everyone will get it consistent with the conclusion. 2011-06-23T08:41:36 try bad spells 2011-06-23T08:41:37 where do you learn french numbers antimatroid ? 2011-06-23T08:41:49 dr-: primary school, a long time ago :P 2011-06-23T08:41:57 i can count up to 39 very poorly maybe? 2011-06-23T08:42:08 yep, i can see that 2011-06-23T08:42:10 antimatroid: it can be understood. if being imaginativ. But the spelling is not 100% 2011-06-23T08:42:30 un deux trois quatre cinq six 2011-06-23T08:42:36 i know that, bon jeur, au revoir and pleteau (not sure how to spell that, meant to be somewhat like "fuck"?) 2011-06-23T08:43:01 bonjour. aurevoir ... what ? 2011-06-23T08:43:09 still, french ppl learning english at college have a very bad english :/ 2011-06-23T08:43:21 pleteau ? this one must have been twisted hard ... 2011-06-23T08:43:26 yeah i have no foreign language skills 2011-06-23T08:43:33 stocha: would not surprise me 2011-06-23T08:43:33 yeah, i don't understand this one too 2011-06-23T08:43:56 the pronounciation i have heard reminds me of the cartoon dog 2011-06-23T08:43:57 dr : i think it was putain. some 1000 years ago. and evolved to plateau :p 2011-06-23T08:44:25 what an evolution 2011-06-23T08:44:56 one would hope it plateaued there :P 2011-06-23T08:45:08 dr i may be wrong also. maybe it's an exotic bad word, that french people tend to forget. Like "diantre" maybe. badly pronounced. 2011-06-23T08:45:41 i think i mean pluta? more, i would expect putain to be right if that's an "offensive" word? 2011-06-23T08:45:57 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T08:46:02 antimatroid: yeah that one means bitch. 2011-06-23T08:46:08 i always wonder what the Merovingian says in Matrix: Revolutions 2011-06-23T08:46:10 almost. 2011-06-23T08:46:24 mcstar: what is Matrix:Revolutions? 2011-06-23T08:46:24 sorry, its was Reloaded 2011-06-23T08:46:28 what's that? 2011-06-23T08:46:30 antimatroid: not talking to you 2011-06-23T08:46:36 mcstar: what do they say ? 2011-06-23T08:46:44 stocha: did you see that part? 2011-06-23T08:46:58 when he curses in french, at the dinner table 2011-06-23T08:47:11 mcstar: probably. when it first came out. And this is not a film i remember particularly well ... 2011-06-23T08:47:14 I'm french stocha, and i was also looking for a word near "plateau" 2011-06-23T08:47:21 although i do kindof remember the dinner 2011-06-23T08:47:50 dr : i had bet that you were french dr. or native french speaker at least. 2011-06-23T08:47:53 he says something pretty long, and it supposed to be full of cursing 2011-06-23T08:48:22 antimatroid: sry, but you always ask those questions(i like all 3 of them) 2011-06-23T08:48:23 mcstar: can you find the scene on youtube or something ? 2011-06-23T08:48:31 maybe 2011-06-23T08:48:48 mcstar: the existence should not be recognised :P 2011-06-23T08:48:59 why did you think that stocha, because of my bad english ? ;) 2011-06-23T08:49:18 antimatroid: if someone do like the film, it can be taken personnaly. understably. 2011-06-23T08:49:55 dr: because of your reacting at antimatroid counted. And vague impression that you somehow were flaged french in the past. 2011-06-23T08:49:55 hmmm fair call, people should care less about what other people think though :P 2011-06-23T08:50:20 antimatroid: i find your hatred funny though :p 2011-06-23T08:50:34 i'm an equal opportunist hater 2011-06-23T08:50:45 antimatroid: what do you think if highlander 2 ? I'm yet to find someone that is offensed by another dislike of it. 2011-06-23T08:50:59 haven't heard of it? :P 2011-06-23T08:51:00 so i always talk high of it. 2011-06-23T08:51:10 antimatroid: you know highlander ? 2011-06-23T08:51:14 the film. 2011-06-23T08:51:17 nope 2011-06-23T08:51:34 antimatroid: that would explain your not knowing the second opus then. 2011-06-23T08:52:18 http://en.wikipedia.org/wiki/Highlander_(film) 2011-06-23T08:52:23 http://www.youtube.com/watch?v=knFmjVf0jVk&feature=related 2011-06-23T08:52:32 the first opus is a classic. 2011-06-23T08:52:53 the second one ... well it's certainly the WORST second oppus in second opus history. 2011-06-23T08:53:06 stocha: i wasn't even born when that came out :P 2011-06-23T08:53:16 troll 2 is the best bad film ever 2011-06-23T08:53:29 antimatroid: does that means you can't watch it ? 2011-06-23T08:53:36 that means i love it 2011-06-23T08:53:42 oh or highlander? 2011-06-23T08:54:05 doesn't mean i can't, but i don't really get down to old movies 2011-06-23T08:55:00 mcstar: loooll :) 2011-06-23T08:55:16 so, what does he say? 2011-06-23T08:55:25 mcstar: let me try to give you a french text sample, then you'll try to translate it with google or whatever you want 2011-06-23T08:55:36 heh? 2011-06-23T08:55:38 nonono 2011-06-23T08:55:45 i want Your translation 2011-06-23T08:55:51 "nom de dieu de putain de border de merde" 2011-06-23T08:56:08 "de saloperie de connard d'enculer de ta mère" 2011-06-23T08:56:12 god, fuck, shit? 2011-06-23T08:56:32 i'm not sure i know how to translate it ... 2011-06-23T08:56:47 try your best 2011-06-23T08:56:56 nom de dieu is like bloody hell. 2011-06-23T08:57:04 (the name of the god) 2011-06-23T08:57:14 putain de border de merde ... 2011-06-23T08:57:16 literaly 2011-06-23T08:57:30 let's speak in private though ... 2011-06-23T08:57:47 putain de bordel de merd 2011-06-23T08:57:51 it's not border, it's bordel 2011-06-23T09:00:23 it would be really sweet if we could keep bears as pets without getting mauled to death 2011-06-23T09:00:45 antimatroid: ? 2011-06-23T09:01:07 stocha: http://24.media.tumblr.com/tumblr_ln8d4v1rAx1qzfutho1_500.jpg 2011-06-23T09:01:16 how sweet would that thing be as a pet if it didn't kill you 2011-06-23T09:02:31 *** gobang has joined #aichallenge 2011-06-23T09:02:53 *** UncleVasya has joined #aichallenge 2011-06-23T09:04:13 lol, just seen a comedist on comedy central who was speaking of pocket sized bears 2011-06-23T09:05:00 hi guys. 2011-06-23T09:06:11 UncleVasya: hi :) 2011-06-23T09:07:15 antimatroid: nice to see that you already have a bot :) 2011-06-23T09:07:38 UncleVasya: yeah it's not really doing anything about battles though 2011-06-23T09:11:20 antimatroid: is that you? in bear costume? 2011-06-23T09:11:31 he is smiling i think 2011-06-23T09:12:47 i beleive stocha lost his connection 2011-06-23T09:12:59 i hate this, 5 lines, and he wont read them 2011-06-23T09:13:06 oh no 2011-06-23T09:14:19 mmgng ? 2011-06-23T09:15:59 mcstar: hany éves vagy ? 2011-06-23T09:16:12 hehe 2011-06-23T09:16:14 yes 2011-06-23T09:17:08 is yes the answer ? 2011-06-23T09:17:22 hehe, i thought i will confuse you for the fun of it 2011-06-23T09:17:32 sorry 2011-06-23T09:17:33 25 2011-06-23T09:17:48 ah i though use hungarian words :p 2011-06-23T09:18:04 okay :) 2011-06-23T09:18:14 *** Zaphus has quit IRC (Quit: Bye) 2011-06-23T09:20:17 my hungarian level is very very low. 2011-06-23T09:20:33 Especially when my two litle girls are learning it. 2011-06-23T09:22:03 DEJA FING VU 2011-06-23T09:22:16 come on we were discussing this a couple of weeks ago 2011-06-23T09:23:09 and the plateau stuff too 2011-06-23T09:23:12 do you remember? 2011-06-23T09:23:32 at the time i first appeared on this channel, well after some time that 2011-06-23T09:26:06 *** UncleVasya has quit IRC (Ping timeout: 260 seconds) 2011-06-23T09:26:47 mcstar: we're like the guy off 50 first dates 2011-06-23T09:26:56 except our memory lasts a little bit longer 2011-06-23T09:27:56 ive seen it but i cant remember wtf was it about. the girl had amnesia? i mean did she forget every day what happened last day? 2011-06-23T09:29:25 that's exactly that mcstar 2011-06-23T09:29:32 like the movie Memento 2011-06-23T09:29:48 theres a SG1 episode Memento Mori 2011-06-23T09:29:54 i havent seen Memento 2011-06-23T09:30:06 that does not mean the same thing :p 2011-06-23T09:30:47 its the closest title i know 2011-06-23T09:31:12 you should see it, and all other Nolan's movies 2011-06-23T09:32:03 oh, carrie-ann 2011-06-23T09:32:54 ive seen Prestige and Inception 2011-06-23T09:33:05 firt one i liked, second one nooo 2011-06-23T09:34:03 The Prestige is very nice 2011-06-23T09:34:08 the book is better ;) 2011-06-23T09:34:24 you have time to read? good for you 2011-06-23T09:34:38 Inception i good to see, if you are one of those people that can influence their own dream. I am one. I found it enjoyable. 2011-06-23T09:35:07 mcstar: you never take mass transport ? 2011-06-23T09:35:34 i yes, then you have time to read too :p 2011-06-23T09:35:35 when i did, i read about nonstandard analysis 2011-06-23T09:35:43 mcstar: ah. 2011-06-23T09:36:00 mcstar: you don't like the diagonal process of cantor ? 2011-06-23T09:36:11 only to realize that i dont ever want to read abstract math :D 2011-06-23T09:36:40 stocha: are you a Lucid Dreamer? 2011-06-23T09:36:44 dr-> qu'est ce tu penses de l'élève ducobu ? 2011-06-23T09:37:18 mcstar: i guess i was just that. All my friend were when i was 15 (arguably i knew two other people who were, and happened to be my friends :p ) 2011-06-23T09:37:34 mcstar: although i think there as degree to that. 2011-06-23T09:37:44 *** gobang has quit IRC (Ping timeout: 258 seconds) 2011-06-23T09:37:48 mcstar: i'm a standard case of lucid dreamer :) 2011-06-23T09:37:56 i read about this, but i cant really control my dreams 2011-06-23T09:38:07 i cant punch or run in my dreams 2011-06-23T09:38:10 mcstar: it's not about controling. 2011-06-23T09:38:21 mcstar: although i sometime did just that. 2011-06-23T09:38:32 mcstar: it's more about them being lucid. 2011-06-23T09:38:45 *** gobang has joined #aichallenge 2011-06-23T09:38:46 let me rephrase: i dont recognise that im dreaming 2011-06-23T09:38:48 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T09:39:30 mcstar: The only case i really "controlled" them hard, was when i was about to dye in them. The big fall case being recurent. So i would just twist the dream, in order not to wake up :) 2011-06-23T09:40:13 i dreamt once that im falling off the roof of a long house, i fell off repeatedly like in a computer game 2011-06-23T09:40:46 mcstar: i think i did recognize them because of silly contingency. When i was three (or six ?), i dreamed i was in the toilet .. and when i woke up, i realized it was a dream. So i develop this habit of being very carefull i was not in a dream was going to the toilet :p 2011-06-23T09:40:55 *** gobang has joined #aichallenge 2011-06-23T09:41:41 mcstar: from the techniques i read for having lucid dreams, that's probably one important part of the affair 2011-06-23T09:42:13 the other one being that since the age of three, my father told me that dreams were mine, and that i could do whatever i want in them. 2011-06-23T09:42:16 have you seen Waking Life? 2011-06-23T09:42:24 mcstar: no 2011-06-23T09:42:59 i suggest you do, its a cartoonish movie(from the director-or-whatever of the man who did A Scanner Darkly) 2011-06-23T09:43:18 it deals with this sort of stuff, quite good actually, even if your are not a lucid dreamer 2011-06-23T09:44:35 i dont remember that my parent ever talked to me about dreams 2011-06-23T09:44:38 s 2011-06-23T09:45:03 mcstar, once i happened to be dreaming of me being pursued. Of course i was able to fly (i had trouble outrunning well at this time), but i was annoyed. So i wanted to commit suicide to escape the dream 2011-06-23T09:45:12 and i dont really have nightmares, and when i do, i usually enjoy them 2011-06-23T09:45:22 fighting with aliens/robots being my favourite 2011-06-23T09:45:25 then i had this terrible doubt : am i 100% sure that this a dream (okay i can fly in there, but you never know ...) 2011-06-23T09:45:46 so i decided to just push hard on a wall, very hard (the roof wall in fact) 2011-06-23T09:45:55 and then it worked. I woke up. 2011-06-23T09:46:07 hm 2011-06-23T09:46:15 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T09:46:17 then i hear someone bang at the door. I oppen, and it was the pursuer ! 2011-06-23T09:46:29 well, my argument is: you can do whatever in a dream, never worry. 2011-06-23T09:46:37 only i couldn't fly any more for some reason. That's really troubling when this happens :) 2011-06-23T09:46:40 if you are AWAKE you would always know 2011-06-23T09:46:44 After a while i woke up naturally. 2011-06-23T09:47:02 mcstar: it's not always that clear. 2011-06-23T09:47:24 well, just think it through 2011-06-23T09:47:26 mcstar: with time passing lucid dreams transformed into something else. 2011-06-23T09:47:36 dont apply any sort of safeguards 2011-06-23T09:47:46 mcstar: you don't have your full awaked thinking habits. 2011-06-23T09:47:50 it's more like you retain some. 2011-06-23T09:48:15 while most probably big parts of your brain have other duty to fulfil 2011-06-23T09:48:29 ok, so what i mean, is that i dont second guess myself when i dream(at least that is what i remember) 2011-06-23T09:49:04 A few month ago (i think) i had this dream, and there was magnificent view. I said to myself, wow. That's a dream. I should really try to memorize that viewing. 2011-06-23T09:49:21 :) 2011-06-23T09:49:41 but then the dream went on, and the fact that it was a dream was less and less important to me. I was all in the story. Which was quite complex at the end. 2011-06-23T09:49:56 Mostly i don't remember my dream though. 2011-06-23T09:50:46 what i try sometimes before i try to fall asleep, is to visualize something in details, for example my room, or anything material 2011-06-23T09:50:56 I think a few years back, i managed to run really really fast in a dream. It felt very good. I don't think there was someone pursuing me this time. But when i woke up, it was like : now they will never catch me back again :) 2011-06-23T09:51:00 sometimes for a second its like i see all the details 2011-06-23T09:51:16 as if i was holding that thing in my hands 2011-06-23T09:51:26 mcstar: when i was younger, i wanted to decide the content of my dreams before hand. That never worked for me. 2011-06-23T09:51:54 maybe you should just train harder, i dont know 2011-06-23T09:51:59 mcstar: the human brain and imagination seems to be a vast place indeed though. Dream being an important aspect of it. 2011-06-23T09:52:21 mcstar: with years passing, i got totally disinterested with that. I have the luck of never having nightmares. 2011-06-23T09:52:32 what i dont want is to get lost in dreaming, it can interfere with your real life i think 2011-06-23T09:52:40 good dreams for me, are simply good stories. Or even feeling good when i wake up. 2011-06-23T09:53:13 I don't need remembering them in most cases. (except if something really important happens. Like running fast at last, or something very impressive) 2011-06-23T09:53:15 *** gobang has joined #aichallenge 2011-06-23T09:54:04 i think its very important to remember them, cause its pointless to have them if you have no memory of them 2011-06-23T09:54:12 mcstar: i don't think you could let dream interfere with your own life voluntarily. but who knows. 2011-06-23T09:54:34 mcstar: i found out this was not true for me. 2011-06-23T09:54:47 *** foRei has joined #aichallenge 2011-06-23T09:54:55 totally accepting not to remember them, be they lucid or not, offers lots of freedom. 2011-06-23T09:55:25 i have some problem with concentration/tiredness, so i definitely dont want to start this sort of dreaming now, i want to be in bed as little as possible 2011-06-23T09:55:27 although it's probably good to remember a bit, or the whole thing from time to time. (like once in year) 2011-06-23T09:56:24 mcstar: dreaming (when asleep) is a very active period of like. The most active in fact. (brainwise) 2011-06-23T09:56:32 life 2011-06-23T09:56:43 rem sleep is, yes 2011-06-23T09:56:43 So one has to respect that. 2011-06-23T09:57:23 People tend to want control over there dreams. But control is not good. You have to let thing go. Be you lucid or not. 2011-06-23T09:57:50 Although a litle bit of contral *can* be fun. Like trying to go through wall (or glass). I never did that very well. 2011-06-23T09:57:52 my only wish right now is that i wake up refreshed 2011-06-23T09:58:10 Then ask for that that before sleeping. 2011-06-23T09:58:12 unfortunately its rarely the case 2011-06-23T09:58:27 hey you, do whatever you want, i'll appreciate if i wake up refreshed :) 2011-06-23T09:58:55 stocha: im not sort of guy who tricks himself with self-pep-talk 2011-06-23T09:59:02 I also never use beeper to wake up. I just ask : it would be good if i woke at 8:00 tomorow. 2011-06-23T09:59:27 mcstar: it's a way to be clear with yourselve. But their might be other ways. 2011-06-23T09:59:31 thats ok, i can wake up on time, but ill get sleepy soon 2011-06-23T09:59:39 mcstar: i use it when i want to wake up at a precise time. 2011-06-23T10:00:06 mcstar: is it easy to fall asleep for you ? 2011-06-23T10:00:30 very easy usually(when im not on coffeine, but sometimes even then) 2011-06-23T10:00:49 How much time do sleep then ? 2011-06-23T10:01:02 depends, 6-7-8 2011-06-23T10:01:38 maybe you should juste give you a bit more time then. 2011-06-23T10:01:52 or have a nap. 2011-06-23T10:01:56 no, it doesnt help, tried it 2011-06-23T10:02:20 I'll say : do some yoga. might help :) 2011-06-23T10:02:39 i never tried it myself though. 2011-06-23T10:02:54 but from what i have read, it seems to be efficient. 2011-06-23T10:03:03 i dont want to do yoga, maybe tai-chi 2011-06-23T10:03:13 ah i did a bit of tai-chi 2011-06-23T10:03:20 i think it's quite good. 2011-06-23T10:03:40 but what is good, is running i liked that one, unfortunately my left leg is not in very good shape 2011-06-23T10:03:50 but i think i'll do a bit of aikido first. in a few years i might try tai-shi again. 2011-06-23T10:04:14 running is very hard on bones and muscles it seems. 2011-06-23T10:04:26 but its the best thing, really 2011-06-23T10:04:33 mentally and physically 2011-06-23T10:04:47 stoSnake is 60th 2011-06-23T10:05:01 right, lets program 2011-06-23T10:05:14 mcstar: i think taishi (and also aikido) are good for people that spend long time on a chair (or worst) 2011-06-23T10:06:06 sure 2011-06-23T10:06:16 its good for everyone 2011-06-23T10:08:47 *** stocha has quit IRC (Quit: Page closed) 2011-06-23T10:24:19 package main 2011-06-23T10:24:22 2011-06-23T10:24:24 import ( 2011-06-23T10:24:24 "os" 2011-06-23T10:24:26 "strconv" 2011-06-23T10:24:26 //~ "image" 2011-06-23T10:24:29 //"fmt" 2011-06-23T10:24:31 ) 2011-06-23T10:24:31 2011-06-23T10:24:38 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T10:25:02 *** gobang has joined #aichallenge 2011-06-23T10:49:15 *** aerique has quit IRC (Quit: ...) 2011-06-23T11:01:03 *** sigh has joined #aichallenge 2011-06-23T11:27:57 *** boegel has quit IRC (Read error: Operation timed out) 2011-06-23T11:28:38 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-06-23T11:31:57 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-23T11:33:44 *** sigh has joined #aichallenge 2011-06-23T11:34:35 *** sigh has quit IRC (Client Quit) 2011-06-23T11:37:30 *** janzert1 has joined #aichallenge 2011-06-23T11:37:51 *** janzert has quit IRC (Ping timeout: 255 seconds) 2011-06-23T11:47:30 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T11:49:05 *** gobang has joined #aichallenge 2011-06-23T11:56:17 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T11:56:40 *** gobang has joined #aichallenge 2011-06-23T11:57:18 *** sigh has joined #aichallenge 2011-06-23T12:05:20 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-23T12:07:06 *** onensora has quit IRC (Ping timeout: 276 seconds) 2011-06-23T12:08:42 *** sigh has joined #aichallenge 2011-06-23T12:16:15 *** onensora has joined #aichallenge 2011-06-23T12:28:53 *** nux67 has joined #aichallenge 2011-06-23T12:34:35 *** choas has joined #aichallenge 2011-06-23T12:35:49 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T12:36:10 *** gobang has joined #aichallenge 2011-06-23T12:36:50 *** gobang has joined #aichallenge 2011-06-23T12:43:47 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T12:47:28 *** _0x47 <_0x47!~slash@p5B12F8A3.dip.t-dialin.net> has joined #aichallenge 2011-06-23T12:58:07 *** _0x47 <_0x47!~slash@p5B12F8A3.dip.t-dialin.net> has quit IRC (Read error: Connection reset by peer) 2011-06-23T12:58:30 *** _0x47 <_0x47!~slash@p5B12F8A3.dip.t-dialin.net> has joined #aichallenge 2011-06-23T13:07:12 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-23T13:07:59 hi as high 2011-06-23T13:10:25 *** gobang has joined #aichallenge 2011-06-23T13:12:16 *** FireFly has quit IRC (Read error: Connection reset by peer) 2011-06-23T13:15:15 *** bhasker has joined #aichallenge 2011-06-23T13:27:02 *** FireFly has joined #aichallenge 2011-06-23T13:39:01 *** davidjliu has joined #aichallenge 2011-06-23T13:49:08 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T13:49:31 *** gobang has joined #aichallenge 2011-06-23T13:54:24 *** davidjliu has quit IRC (Ping timeout: 252 seconds) 2011-06-23T13:57:24 *** McLeopold has joined #aichallenge 2011-06-23T13:57:40 janzert1: can I get you to stop the workers? 2011-06-23T13:58:18 sure, stop as in a restart or stop as in take them down for a while 2011-06-23T13:58:33 take down so they can be updated after I push new code 2011-06-23T13:58:42 ok 2011-06-23T14:00:03 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T14:01:16 I've got some nice histograms from before the new cutoff code was added and after now 2011-06-23T14:01:43 just waiting to get a few more games post-cutoff before making a post on the forums 2011-06-23T14:02:29 the cutoff is certainly a great help with lowering game lengths though 2011-06-23T14:02:33 aichallenge: McLeopold epsilon * r7e277a0 / (10 files in 5 dirs): added cutoff options to playgame and ants, tracking winning and ranking turn in db - http://bit.ly/kzy3Bb 2011-06-23T14:02:53 *** bhasker has quit IRC (Remote host closed the connection) 2011-06-23T14:03:33 workers are down 2011-06-23T14:03:38 *** janzert1 is now known as janzert 2011-06-23T14:03:44 janzert: now you are going to get more juicy stats 2011-06-23T14:03:59 :) 2011-06-23T14:04:15 *** sigh has joined #aichallenge 2011-06-23T14:05:14 the one I really still want to make, but is going to be a bit of work, is to find the move where the rankings are in the final order and both do a direct histogram of that and also a histogram of how much longer the game lasted 2011-06-23T14:06:41 *** gobang has joined #aichallenge 2011-06-23T14:07:11 *** george1914 has quit IRC (Quit: Page closed) 2011-06-23T14:08:33 janzert: testing changes... 2011-06-23T14:09:00 *** davidjliu has joined #aichallenge 2011-06-23T14:09:06 janzert: you have winning_turn and ranking_turn in the db and replay now 2011-06-23T14:09:33 so, the work should be easy now 2011-06-23T14:10:28 what is the definition of those 2? 2011-06-23T14:10:29 plus a cutoff flag 2011-06-23T14:10:43 winning turn is the turn that the eventual winner last took the lead 2011-06-23T14:10:54 ranking_turn is the turn that the eventual rank was finalized 2011-06-23T14:11:28 based on score, not population 2011-06-23T14:11:43 run playgame now and you can see the stats listed 2011-06-23T14:12:28 or checkout the screen session right now on beta 2011-06-23T14:12:33 ahh, great 2011-06-23T14:12:49 uh oh, bad php code :( 2011-06-23T14:13:05 oops 2011-06-23T14:13:25 shift-g 2011-06-23T14:13:31 jumps to end of file 2011-06-23T14:13:43 oh, you're watching :p 2011-06-23T14:13:47 :) 2011-06-23T14:13:56 I wanted to see those new stats ;) 2011-06-23T14:13:58 $winning_turn 2011-06-23T14:15:22 afk, for a bit 2011-06-23T14:15:27 k 2011-06-23T14:21:51 aichallenge: McLeopold epsilon * r7913818 / (website/api_game_result.php worker/worker.py): fix to api_game_result - http://bit.ly/kbjnve 2011-06-23T14:22:47 *** choas has quit IRC (Quit: leaving) 2011-06-23T14:25:34 janzert: If I can determine that 100% of the users have resubmitted since a bots last game, then I want to increase sigma to the starting value 2011-06-23T14:25:57 but, what if there mu is at 25? their skill would then be -25 2011-06-23T14:26:04 25 - 3 * 16.667 2011-06-23T14:26:55 I could make sure that low and high stay within 0 to 100, but then bot ranked high and low won't ever get as much sigma increase 2011-06-23T14:29:43 what's the current value for initial mu ? 2011-06-23T14:30:30 50 2011-06-23T14:30:37 the display seems to already be limited to 0 anyway 2011-06-23T14:30:46 true... 2011-06-23T14:30:54 at least I'm pretty sure I saw one a few days ago that should have been -2 or so 2011-06-23T14:31:00 but on the top end, it might go above 100 2011-06-23T14:31:19 not from increasing sigma 2011-06-23T14:31:20 I remember putting in a min function so people wouldn't get depressed :) 2011-06-23T14:31:46 anyway I don't think it's a problem increasing the sigma the same at the limits 2011-06-23T14:32:08 I feel it might be an issue of the bell curve doesn't stay in the 0-100 range 2011-06-23T14:32:38 *** okayzed is now known as okay 2011-06-23T14:32:45 in which case I feel it's already too spread out for the number of players we have 2011-06-23T14:33:03 you mean, go ahead and limit the increase to stay within the bounds? 2011-06-23T14:33:10 no 2011-06-23T14:33:31 I mean the current mu spread is probably to large 2011-06-23T14:33:44 what's our ranges? 2011-06-23T14:34:11 98.81 to -7.06 2011-06-23T14:34:56 wow, how did he go negative? 2011-06-23T14:35:34 that's the mu not skill 2011-06-23T14:35:45 I know 2011-06-23T14:36:01 still, I thought trueskill kept it in bounds 2011-06-23T14:36:31 there isn't any theoretical limit, just the farther out it is the less likely it is 2011-06-23T14:37:01 *** amstan has joined #aichallenge 2011-06-23T14:37:01 *** ChanServ sets mode: +o amstan 2011-06-23T14:37:07 adjusting the 80% chance to win distance will effect that 2011-06-23T14:37:24 trying to remember what that variable is called ;) 2011-06-23T14:37:35 okay sigma increase = (new_submission_users / active_users) * (16.666 - sigma) 2011-06-23T14:37:48 tau or epsilon 2011-06-23T14:37:54 beta 2011-06-23T14:38:05 are you adjusting the sigma to get a bigger spread? 2011-06-23T14:38:14 okay: no 2011-06-23T14:38:16 not quite 2011-06-23T14:38:29 want to make the numbers look larger? 2011-06-23T14:38:40 no, just closer to 50 2011-06-23T14:39:09 janzert: can you do a histogram of the mu's? 2011-06-23T14:40:00 from what data? 2011-06-23T14:40:08 just as observed at the end of each game? 2011-06-23T14:40:32 janzert: okay, the code works, fire up the workers 2011-06-23T14:40:39 make sure to update 2011-06-23T14:40:53 they'll be coming up from scratch :) 2011-06-23T14:40:59 I'm thinking just from the current leaderboard 2011-06-23T14:41:10 janzert: geeze, why can't we just update some code? 2011-06-23T14:41:32 actually easier this way, and we know everything still works from scratch too 2011-06-23T14:41:37 unless you just like compiling golang every day 2011-06-23T14:41:49 we don't compile it at all anymore 2011-06-23T14:42:13 how long does setup take now? 2011-06-23T14:42:20 7-10 minutes 2011-06-23T14:43:15 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T14:43:15 *** _0x47 <_0x47!~slash@p5B12F8A3.dip.t-dialin.net> has quit IRC (Quit: Leaving.) 2011-06-23T14:43:38 also in case you haven't noticed, the manager script has to be restarted after mysql is 2011-06-23T14:43:51 I didn't notice 2011-06-23T14:44:04 yeah, it crashes when mysql goes away 2011-06-23T14:44:15 I wondered what the problem was 2011-06-23T14:44:24 although they do always occur when I'm doing stuff 2011-06-23T14:44:38 :) 2011-06-23T14:46:46 *** gobang has joined #aichallenge 2011-06-23T14:47:16 new workers started about 2 minutes ago 2011-06-23T14:47:36 I'm increasing viewradius to 77 and food_visible to 2-4 2011-06-23T14:48:00 It's an attempt to make the game more skillful 2011-06-23T14:48:01 can you wait another 1000 games? 2011-06-23T14:48:13 why? 2011-06-23T14:48:28 so I can get good stats on just the cutoff change without mixing in those changes 2011-06-23T14:49:15 okay, done 2011-06-23T14:49:26 we also need to start thinking about switching to "serial" mode games sometime soon 2011-06-23T14:49:28 you might have 2 or 3 games different 2011-06-23T14:49:35 *** racko has joined #aichallenge 2011-06-23T14:49:37 *** amstan has quit IRC (Read error: Connection reset by peer) 2011-06-23T14:49:38 I was looking at the setting, still at off 2011-06-23T14:49:38 or abandon the idea 2011-06-23T14:49:53 didn't you get 2 at a time working? 2011-06-23T14:50:08 *** amstan has joined #aichallenge 2011-06-23T14:50:08 *** ChanServ sets mode: +o amstan 2011-06-23T14:50:17 yep, when we do turn it on just set it to 2 for the current workers 2011-06-23T14:50:25 that runs 2 bots at a time 2011-06-23T14:50:35 want to do that now or wait for 1000 games? 2011-06-23T14:51:00 probably best to wait 2011-06-23T14:51:17 although it shouldn't have that large of an effect 2011-06-23T14:51:49 that's waiting about 10 hours 2011-06-23T14:52:43 I should try and figure out a rough number on how much it's going to slow down our game rate 2011-06-23T14:53:36 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T14:53:45 first rough guess is between 1/2 and 1/3 ? 2011-06-23T14:54:02 what is "it" 2011-06-23T14:54:10 i.e. the new game rate will be between 1/2 and 1/3 of the current rate 2011-06-23T14:54:17 due to what change? 2011-06-23T14:54:25 setting serial=2 2011-06-23T14:54:28 oh 2011-06-23T14:54:51 you could base it on average players per game 2011-06-23T14:55:01 *** gobang has joined #aichallenge 2011-06-23T14:55:24 yes, I with an average of 6 players per game it would be 1/3 if all the players were active for the full game 2011-06-23T14:55:43 but since players will drop out throughout the course of the game it should be higher 2011-06-23T14:56:02 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-23T14:56:06 once there are 2 players left it will played at the same rate 2011-06-23T14:58:39 *** Naktibalda_ has joined #aichallenge 2011-06-23T14:58:59 *** sigh has joined #aichallenge 2011-06-23T14:59:13 why does a submission have a worker_id? 2011-06-23T14:59:44 aichallenge: McLeopold epsilon * rddd7f5f / ants/ants.py : incorrect cutoff value in replay - http://bit.ly/kmp6cP 2011-06-23T15:00:00 that is the worker that did the compile 2011-06-23T15:00:18 and functional test 2011-06-23T15:00:30 ok 2011-06-23T15:00:49 so, want to recreate workers for that change :) 2011-06-23T15:01:01 McLeopold; what do you think about giving no bonus if the game ends early? 2011-06-23T15:01:03 for that one I'll go ahead and just update them :} 2011-06-23T15:01:13 antimatroid: already did it 2011-06-23T15:01:19 :) 2011-06-23T15:02:04 a split bonus only punishes killed bots 2011-06-23T15:02:10 we want to punish the leader 2011-06-23T15:02:55 yes 2011-06-23T15:03:09 meaning a bot that tries to farm that came from behind will be sorely dissapointed 2011-06-23T15:03:09 stocha also made a valid comment about punishing someone who just got knocked out 2011-06-23T15:03:14 *** george1914 has joined #aichallenge 2011-06-23T15:03:15 and then giving a bonus to someone 2011-06-23T15:03:17 *** amstan has quit IRC (Remote host closed the connection) 2011-06-23T15:03:36 *** amstan has joined #aichallenge 2011-06-23T15:03:36 *** ChanServ sets mode: +o amstan 2011-06-23T15:05:26 *** stocha has joined #aichallenge 2011-06-23T15:05:37 McLeopold: what's food visible ? 2011-06-23T15:05:49 @seen McLeopold 2011-06-23T15:05:49 stocha: McLeopold was last seen in #aichallenge 2 minutes and 39 seconds ago: meaning a bot that tries to farm that came from behind will be sorely dissapointed 2011-06-23T15:06:01 stocha: the amount of food a bot will see on the first turn 2011-06-23T15:06:09 the range right now is 1 - 3 2011-06-23T15:06:23 later we will increase it to 2 - 4 2011-06-23T15:06:39 how does increasing the view radius make the game more challanging? 2011-06-23T15:06:42 more information? 2011-06-23T15:07:04 It means good path finders have more of an advantage 2011-06-23T15:07:23 ... because they're struggling to beat non-path finding bots? :P 2011-06-23T15:07:42 McLeopold: is still don't understand what is food_visible. 2011-06-23T15:07:48 once everyone does pathfinding, we want to be able to separate good from bad 2011-06-23T15:07:59 stocha: the number of food squares visible to your ant on the first turn 2011-06-23T15:08:06 McLeopold: i though a food is visible if in view radius 2011-06-23T15:08:07 stocha: you get 1 ant to start with, and you can see only a small part of the map 2011-06-23T15:08:30 McLeopold: yeah. So you can see food if the ant can see it. 2011-06-23T15:08:41 the game places starting food at the beginning, some of that food will be place in the ants vision at the start 2011-06-23T15:08:50 McLeopold: what's happening with the spawn rate? 2011-06-23T15:09:02 i still think there's too many ants and the maps are flooding 2011-06-23T15:09:05 staying the same, I think it is good 2011-06-23T15:09:06 but that could just be crap bots 2011-06-23T15:09:45 McLeopold: so that is the number of food in the first ant vision radius ? 2011-06-23T15:09:45 i'd prefer more strategic battles than just a flood of ants going at it :P 2011-06-23T15:09:55 if the spawn rate is too slow at the start, it seems to make the game have more luck 2011-06-23T15:10:02 <_flag> If there's too little food there's little incentive to do battles over simply running away and hoping other bots kill each other 2011-06-23T15:10:08 stocha: yes 2011-06-23T15:10:14 McLeopold: why is this thing needed ? 2011-06-23T15:10:42 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T15:10:51 stocha: so that bots have a good start, it's to reduce the luck factor of chosing the wrong first direction to explore 2011-06-23T15:10:53 antimatroid: isn't that what's diversity is for ? 2011-06-23T15:11:12 McLeopold: okay. 2011-06-23T15:11:26 McLeopold: it's all about map generation. 2011-06-23T15:11:50 antimatroid: janzert was arguing for more food than I was. I wouldn't mind dropping the rate more 2011-06-23T15:12:05 Arguably with 3 ants, good bot will do better than bad bots. Where with only one, they might do the same. 2011-06-23T15:12:31 McLeopold: isn't the rate random per game ? 2011-06-23T15:12:38 I think most ants will not end the game until reaching current densities anyway so lowering the food rate simply increases the game length 2011-06-23T15:12:49 s/ants/bots 2011-06-23T15:13:03 it would, but the battles would be more interesting 2011-06-23T15:13:15 i think the food rate should be random (let's say uniformly sprayed). So only the bounds are oppen to discussions. 2011-06-23T15:13:27 stocha: it is 2011-06-23T15:13:43 are we discussion the bounds then ? 2011-06-23T15:13:47 we could use bell curves also. 2011-06-23T15:13:54 lower bound is 8/12 per turn per player, upper is 2/30 per turn per player 2011-06-23T15:14:10 I do think change the bounds to be based on either the total number of players or the size of the map would be a really good thing 2011-06-23T15:14:11 it is curved, but not a bell 2011-06-23T15:14:30 janzert: it is based on number of players already 2011-06-23T15:14:34 McLeopold: so lower bound is higher than upper one ? 2011-06-23T15:14:37 ahh, great 2011-06-23T15:14:58 whoops :) 2011-06-23T15:15:11 okay so the bounds are correct then. 2011-06-23T15:15:26 *** okay is now known as okayzed 2011-06-23T15:15:32 Seems quite a wide range then. 2011-06-23T15:17:09 yeah adding map size into the equation might be worthy. 2011-06-23T15:17:10 *** gobang has joined #aichallenge 2011-06-23T15:17:37 my first idea was to have a probability based on empty square to spawn food. So the bigger the map, the more the food. 2011-06-23T15:17:47 the starting food is based on land area 2011-06-23T15:17:50 it makes sense really. 2011-06-23T15:18:03 stocha: per player base should be fine 2011-06-23T15:18:06 so should the new spawned one i think. 2011-06-23T15:18:32 you should use perplayer And per number of ground square on map. IMHO 2011-06-23T15:19:02 Although the per player part, i'm not that sure in fact. 2011-06-23T15:19:14 what's the justification behind per player ? 2011-06-23T15:19:21 McLeopold: what are your thoughts on the numbers of ants btw? 2011-06-23T15:19:26 well, it can't really be both since they don't correspond 1 to 1 although either will roughly correspond to the other 2011-06-23T15:19:31 number of bots? 2011-06-23T15:19:32 i think 10 is too many for one game without too much randomness 2011-06-23T15:19:36 in a game 2011-06-23T15:19:49 2-26 2011-06-23T15:20:08 those are our bounds 2011-06-23T15:20:26 what are your thoughts on more reasonable parameters? or do you actually want games with 26 bots? :P 2011-06-23T15:20:35 antimatroid: do the 10 player games cause large mu and sigma updates? 2011-06-23T15:20:41 it all depends on map size really. 2011-06-23T15:20:45 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-23T15:20:50 the scoring should make more sense in a 2 player setting with spawn points i think 2011-06-23T15:21:05 McLeopold: I don't know much about trueskill 2011-06-23T15:21:10 i think that the number of player simply make the game different. 2011-06-23T15:21:29 especially because then strength also depends on opponents features. 2011-06-23T15:21:39 while in 1vs1 it's less dependent on players features. 2011-06-23T15:21:39 stocha: by basing food rate on per player you can reason about how fast a player should be able to grow their population 2011-06-23T15:22:02 but, with the close game only, i think you'd get a consistent ranking, whatever the number of player 2011-06-23T15:22:28 antimatroid: the 10 player games I see FlagCapper in he still comes in first 2011-06-23T15:22:30 janzert: but you can steal other player's food. 2011-06-23T15:22:39 stocha: yes, and good for you 2011-06-23T15:22:40 janzert: so i don't get it. 2011-06-23T15:22:55 janzert: flag and sir_macelon don't have much competition atm 2011-06-23T15:23:03 i can't believe my bot is ranked so highly 2011-06-23T15:23:09 janzert: why not have a food speed based on map size then ? (and then you can it so that each player zone receive a fair amount) 2011-06-23T15:23:42 i tend to agree with stocha 2011-06-23T15:23:45 i still thing that there should be at least 10% of games be 1vs1 2011-06-23T15:23:48 *** sigh has joined #aichallenge 2011-06-23T15:24:01 for per player amount to be similar, players bases need to be about the same relative size 2011-06-23T15:24:02 1vs1 games are really fun to watch. 2011-06-23T15:24:18 when 4 free for all, can look like a mess. 2011-06-23T15:24:25 making that not implicit would allow more reasonable variation of players base sizes 2011-06-23T15:24:31 because at least I reason about what food rates are reasonable by how fast I think the player populations should grow (on average, ignoring opponents, etc.) i.e. each player can get a new ant every X turns 2011-06-23T15:24:55 janzert: my reonning would be more about max number of ant per game. 2011-06-23T15:24:59 not by every X turns each 100x100 section of map should get a new food 2011-06-23T15:25:17 stocha: that depends much more on bot strategies than food rate though 2011-06-23T15:25:29 janzert: i don't see the value of this player growth factor. 2011-06-23T15:25:48 janzert: the upper bound only depend on maximum food spawned during a game. 2011-06-23T15:25:52 how long does a player have to maneuver the current ants before expecting to receive another 2011-06-23T15:26:19 janzert: it depends on the number of ant. 2011-06-23T15:26:42 so if no food is visible at start, it will be high. If 4 food is visible, it will be low. All with same number of food on map. 2011-06-23T15:26:53 yes, but especially evident with the bots that aren't terribly smart is that the game won't naturally end until the lead ants reaches a certain density of ants 2011-06-23T15:27:02 (arguably if food is very dense, it doesn't change a thing but well) 2011-06-23T15:27:03 *** olexs has joined #aichallenge 2011-06-23T15:27:13 it doesn't matter much how slow or fast those ants actually spawn 2011-06-23T15:27:33 janzert: so the density of food is important. 2011-06-23T15:27:40 wich the number of player don't affect anyway. 2011-06-23T15:27:48 and slowing the spawn rate just increases the game length 2011-06-23T15:27:49 why should be food denser when there is more players ? 2011-06-23T15:28:30 janzert: I think game progress depends more on when bots clash than on number of ants 2011-06-23T15:28:35 the question i ask is this : for a fixed map size X, why should food spawn faster when there are 4 players (hence it's more dense) than when there is 2 (hence less dense) 2011-06-23T15:29:08 games usually don't take that long if the bots aren't all defense rather than any offense 2011-06-23T15:29:08 shouldn"t the density be the same, and define the map. 2011-06-23T15:29:08 because there are more players to use those ants instead of the food sitting there 2011-06-23T15:29:25 the food never sits there 2011-06-23T15:29:30 and so far it looks to me like some offense is best 2011-06-23T15:29:31 the growth is exponential 2011-06-23T15:29:33 which is awesome 2011-06-23T15:29:35 I'm refering to during the growth phase of course 2011-06-23T15:29:43 the the number of ant spawned per turn, only depend on food spawn density 2011-06-23T15:30:16 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-23T15:30:26 janzert: the growth phase only take a few turn. Or it waits for a food density (example my random bot waits for a food density) 2011-06-23T15:30:51 the only bot i know which may not follow this rule is the snake bot. 2011-06-23T15:31:08 because it only has one ant active. sort of. 2011-06-23T15:31:31 but that's kind rare. and innefficient. 2011-06-23T15:31:33 stocha: what percentage of land area should spawn each turn? can you give a range? 2011-06-23T15:31:48 McLeopold: no. 2011-06-23T15:31:50 finnish programmers, raise your hands.. NOW! \o/ 2011-06-23T15:32:01 McLeopold: i can only say that variety is good. 2011-06-23T15:32:12 that's why I asked for a range :) 2011-06-23T15:32:25 i still have a little more code to write before i finnish ... 2011-06-23T15:32:32 sorry 2011-06-23T15:32:48 McLeopold: i only think it's a good idea to use this particular metric. 2011-06-23T15:32:51 stocha: can you say how many times faster the fastest rate should be over the slowest? 2011-06-23T15:32:58 so let's say we start with around 1/30 2011-06-23T15:33:14 McLeopold: no. 2011-06-23T15:33:21 McLeopold: the slowest should be 0 2011-06-23T15:33:23 george1914: who are you apologising to? 2011-06-23T15:33:31 so, 1 food for every 30 land squares per turn? 2011-06-23T15:33:35 McLeopold: so the anwser here :there is no response to your question. 2011-06-23T15:33:39 for jokes like that? everyone 2011-06-23T15:33:54 McLeopold: I would be interested in seeing some games with less food at least during beta :) 2011-06-23T15:34:12 McLeopold: so, let's say from 0 (rare) up to 1/30 (rare also) 2011-06-23T15:34:21 one out of every thirty squares every turn? thats seems really, really high 2011-06-23T15:34:26 and then we'll adjust if 1/30 is too much. 2011-06-23T15:34:36 it is really really high :) 2011-06-23T15:34:50 janzert: i made experiments with dense food. It was amusing. You have armies then. 2011-06-23T15:34:59 although i did that on 256x256 maps. 2011-06-23T15:35:22 bot will have to build a severe time management on those map. i know i can do it easilly :) 2011-06-23T15:35:28 bots won't have time to maneuver new ants into good positions 2011-06-23T15:35:42 it also requires more cpu time to process 2011-06-23T15:35:50 finnish.. like Finlandeses 2011-06-23T15:35:54 not finish 2011-06-23T15:35:56 :p 2011-06-23T15:37:42 janzert: you understand this game very much more than i then. 2011-06-23T15:38:09 McLeopold: it only depends on your algorithm. 2011-06-23T15:38:32 *** Accoun has quit IRC () 2011-06-23T15:38:51 One thing i insist on is : there should be a upper bound to maps size. There should be a upper bound to maximum number of ants per game. And there should be a possibility the bots will encounter them. 2011-06-23T15:39:17 The upper bounds are to be known and settled in stone before the contests starts. 2011-06-23T15:39:27 you wont get that last bit 2011-06-23T15:39:40 well, depending on what you mean by the contest starting 2011-06-23T15:39:46 If that is done, then it will simply define the game, whatever rules you arbitrarily choose :) 2011-06-23T15:39:54 antimatroid: i mean the beta end. 2011-06-23T15:40:13 we can aim to have good parameters at launch, but we can't assume we'll have found the best maps/parameters at that stage 2011-06-23T15:40:25 antimatroid: if i don't get it, then there is a big problem. Because people won't be able to make informed decisions. 2011-06-23T15:40:48 stocha: everyong has to make the same decisions 2011-06-23T15:41:03 i agree about the comment of the parameters etc. needing to be as general as they were allowed for 2011-06-23T15:41:10 for example, some people won"t care. And either they will time out and loose, or never time out and win. 2011-06-23T15:41:14 otherwise it becomes a contest of who can make the best assumptions 2011-06-23T15:41:21 and some other will design robust algorithm. And maybe be weaker. 2011-06-23T15:41:47 antimatroid: exactly. So you have to settle parameters before the contest begin. You have the beta for choosing them though. 2011-06-23T15:42:38 I think during the beta some wild thing should be tryed; 1vs1 games, 256x256 games. Very high food rate. 0 food rate. Etc. 2011-06-23T15:42:46 Assymetric maps also. 2011-06-23T15:43:01 i would prefer parameters to be roughly set by end of beta as well... 2011-06-23T15:43:04 all that should be tryed. And eventually then rejected. after we have evidence to discuss. 2011-06-23T15:43:13 mainly though, i would like to get to start of contest ... :) 2011-06-23T15:43:31 george1914: why ? 2011-06-23T15:43:48 george1914: what does starting it early bring against beta testing ? 2011-06-23T15:44:00 george1914: because in fact, you'll be beta testing it. 2011-06-23T15:44:07 because i have ideas for algorithms that might or might not work well, depending on map size, density of ants, etc 2011-06-23T15:44:27 george1914: yeah, but why do you want it to start early ? 2011-06-23T15:44:27 (that is why i like to roughly know parameters) 2011-06-23T15:44:42 stocha: some of us are impatient people :) 2011-06-23T15:44:51 i just want it to start :) 2011-06-23T15:44:52 we think of something and want to do it NOW 2011-06-23T15:44:58 will be fun 2011-06-23T15:45:08 but there is the beta. you can try your ideas. and with the tcp you can compare your bot to others. 2011-06-23T15:45:20 what prevent you from doing it. 2011-06-23T15:45:24 i have my ideas 2011-06-23T15:45:31 maybe the server costs. 2011-06-23T15:45:35 i don't want to implement them without doing it properly 2011-06-23T15:45:45 maybe you can't have the same server for the beta and the contest. So that's an argument. 2011-06-23T15:45:47 randomly throwing stuff into a bot is really messy for development 2011-06-23T15:46:05 stocha: i do not have infinite time for contest - would prefer to only do what turns out to be good, once i know what parameters are 2011-06-23T15:46:42 george1914: everybody do. but if you want to know the parameters, some people need to do some extensive testings. Or you might run into not liking the result. 2011-06-23T15:46:53 *** gobang has quit IRC (Ping timeout: 250 seconds) 2011-06-23T15:47:40 Okay, so we need to settle all the contests code. And then we need to have enougth data to choose a good set of parameters. 2011-06-23T15:48:07 hence we should try as many things as we can, during the time needed for testing the contests code (workers etc) 2011-06-23T15:48:28 then it won't extend the testing period. Only give more information out of it( roughly) 2011-06-23T15:48:45 stocha: i think organisers have a decent setup, i am happy to go with something that seems roughly good enough. trying all you suggest seems like a time-waste to me .. just go with something good enough instead of imagining perfection can be achieved ... 2011-06-23T15:49:09 george1914: okay. you have a point in there i think. 2011-06-23T15:49:31 anyway, you wont know the 'best' parameters until the top bots have been written - which wont happen until parameters are fixed :/ 2011-06-23T15:49:34 george1914: if what i propose make thing more complicated ant time consuming for core team, then it may not be worth it. 2011-06-23T15:49:50 *** Accoun has joined #aichallenge 2011-06-23T15:50:08 george1914: you still could strongly propose a set for end of beta. and test from there. 2011-06-23T15:50:11 stocha: i guess the core team could say on that - i am just interested observer ... :) 2011-06-23T15:50:17 you could do that now in fact. 2011-06-23T15:50:49 george1914: yeah, but it's good that you tell what's on your mind. :) Because you are also a customer of the contest. 2011-06-23T15:51:11 stocha: you also :) 2011-06-23T15:51:15 george1914: there may be 4000 customers or so, after the contest launch. That'll bring a lot of diversity in views. 2011-06-23T15:51:41 best to fixmost things before they arrive, else it will never start :) 2011-06-23T15:51:41 george1914: true. But i'm a loud one. always saying my opinion about everything. 2011-06-23T15:52:20 i would like to have many more beta tester. but then there is the problem of computing power needed to play all the games :p 2011-06-23T15:52:24 *** gobang has joined #aichallenge 2011-06-23T15:52:47 btw - anyone had much luck with writing combat code (alpha-beta tree or whatever) ? 2011-06-23T15:52:49 if we have only the final server for 3 month that's a problem. 2011-06-23T15:53:09 george1914: i didn't try. The top bot seems to do some combat maneuvres though. 2011-06-23T15:53:21 there are not likely to explain what they do though. 2011-06-23T15:53:48 ... i am unsure on how to pick my enemies' moves ... with it not being zero-sum 2011-06-23T15:53:55 george1914: haven't programmed anything like that, and i worry about the ability to use game trees 2011-06-23T15:54:07 am pondering adding an 'enemy evaluation' function 2011-06-23T15:54:24 i have some more testing about non combat stuff to do first. Then i might consider implementing something for combat. i would take a few more week i guess. 2011-06-23T15:54:25 i wanted to partition "battle ants" up into sets of ants that could affect each other then use game trees to select those moves, but SOOO many states to churn through :\ 2011-06-23T15:54:38 ah, well i have clever tricks for that :) only takes 10s a move atm :/ 2011-06-23T15:54:46 only? :P 2011-06-23T15:54:50 :) 2011-06-23T15:55:02 10s a move is pretty good. if your bot is very strong. 2011-06-23T15:55:09 george1914: so you're partitioning the ants and running game trees in 10 seconds? 2011-06-23T15:55:09 george1914: what language are you using ? 2011-06-23T15:55:15 that's a lot better than i would have expected 2011-06-23T15:55:20 c++ 2011-06-23T15:55:40 yes, well, i only search to 1move deep atm :/ 2011-06-23T15:55:49 yes, i don't expect to go further 2011-06-23T15:55:59 george1914: still with pure optimization, you can reach 1s. and then by doing a bit less, you can reach the 0,5 pretty good. If it make you very strong. 2011-06-23T15:56:01 have you got a cap on the maximum size of a partition? 2011-06-23T15:56:16 no - but i will have to add one 2011-06-23T15:56:22 george1914: it seems promising though. 2011-06-23T15:56:26 george1914: what's your evaluation function like? 2011-06-23T15:56:34 you don't have zero sum anymore? 2011-06-23T15:56:36 er, bad 2011-06-23T15:56:48 atm (i just finished first draft) 2011-06-23T15:56:55 +1 for each ant of mine 2011-06-23T15:57:00 -1 for each enemy ant 2011-06-23T15:57:17 and enemy always moves after me, and goes for worst move for me 2011-06-23T15:57:34 so it makes for a cautious/timid player 2011-06-23T15:57:50 how are you evaluating the state? you're not evaluating the whole board i hope? you should be able to iteratively work out the result for each ant in the partition 2011-06-23T15:57:52 that misses lots of good moves because enemy 2011-06-23T15:58:01 no, just the local area 2011-06-23T15:58:09 george1914: yeah, that's how i set my game trees up 2011-06-23T15:58:41 anyway, the main thing is that i cannot reliably predict enemy moves in this non-zero-sum game 2011-06-23T15:58:58 no, my intention is to do backward induction slightly differently 2011-06-23T15:59:29 rather than bringing up a players worst outcome from the best alternatives of the moving player, i'm going to bring up the worst outcome from any alternative to anyone but the moving player 2011-06-23T15:59:37 so you don't assume opponents are rational/optimal 2011-06-23T15:59:58 that is what i am doing, i think 2011-06-23T16:00:13 but i miss many chances to get good kill ratio 2011-06-23T16:00:23 you could get really complicated there and use a tit for tat like strategy as to whether you want to cooperate with an enemy, or "data" on how good bots are as to whether you think they'd move optimally, but i think i've decided against both of those 2011-06-23T16:00:43 because the enemy (my own older bots atm) do not try 100% to damage me - they play for themselves 2011-06-23T16:01:41 how do you implement your game trees? I usually have 2 structs, Tree and Node, then I have a queue of extendable nodes that I iteratively generate the tree from for as long as I allocate time to it 2011-06-23T16:02:35 also, i refuse to use alpha beta while generating my tree :P 2011-06-23T16:02:37 er, recursive functions 2011-06-23T16:02:48 like for alpha-beta in chess 2011-06-23T16:04:41 so, maybe i write MI6 function to report eg - 'enemy indentified as stocha snakebot - deploy mongoose tactic' ... :) 2011-06-23T16:05:27 i like my strategy to be as general as possible 2011-06-23T16:06:28 well, if i play only against strong(ish) opponents, i can assume a fair degree of optimality in their play, maybe 2011-06-23T16:08:01 i'd be more inclined to just track how many kills vs deaths the ant has 2011-06-23T16:08:24 yes, i was thinking of that - 2011-06-23T16:08:24 and judge the opponent bots abilities off that information rather than trying to identify the bot individually 2011-06-23T16:08:57 it is not entirely trivial to keep track of it, is it? - i dont see where enemy moved to, just the dead ants at the end of the turn ... 2011-06-23T16:08:58 * antimatroid has a lot of trouble using bot/ant in the right place 2011-06-23T16:09:12 *** gobang has quit IRC (Ping timeout: 255 seconds) 2011-06-23T16:09:23 george1914: you can search around the dead ant information to see which enemies contributed to it's death? 2011-06-23T16:09:26 attributing kills accurately is not quite possible, i suspect 2011-06-23T16:09:27 including other dead ants 2011-06-23T16:09:39 you might not get perfect information on the edge of vision, but it's better than no information 2011-06-23T16:09:56 hm, forensic battlefield examination 2011-06-23T16:10:07 hehe :) 2011-06-23T16:10:11 yes, you are right, thanks 2011-06-23T16:10:30 i should add that 2011-06-23T16:11:29 well, having said all that i'm not sure if i am going to make use of the information 2011-06-23T16:11:50 so i tend to just ignore the info atm, but really it's just a function you need to call after reading the state info 2011-06-23T16:12:15 yes - time-cost i minimal 2011-06-23T16:12:24 sorry, 'is minimal' 2011-06-23T16:12:40 i would be interested to know kill ratios 2011-06-23T16:12:56 i am not sure whether to evaluate based on the scoring 2011-06-23T16:12:59 or just raw ant numbers 2011-06-23T16:13:18 i might do the former i think 2011-06-23T16:13:56 i reckon: damn the scoring - power/survival is the main thing - the points will follow 2011-06-23T16:14:10 unless the games are made really short 2011-06-23T16:14:29 but the scoring should be able to encapsulate that anyway 2011-06-23T16:14:40 while also making sure you optimise for any micro intricacies it has 2011-06-23T16:15:17 well, if you could predict final score from your moves, then yes 2011-06-23T16:15:21 *** gobang has joined #aichallenge 2011-06-23T16:15:33 but in fact, all you know is change in score at this/next move 2011-06-23T16:15:52 yes, but you're picking the best move in comparison to other moves 2011-06-23T16:15:56 for incrementing your score now 2011-06-23T16:16:12 i say that #ants you have +production capacity is probably what you should optimise 2011-06-23T16:16:28 you will be a very defensive bot if you do that :P 2011-06-23T16:16:50 whereas you are saying 'hey i'm winning this marathon race because i am at the front at mile 2' 2011-06-23T16:17:15 postion before the finish is not everything 2011-06-23T16:17:30 no - i will expand to increase production 2011-06-23T16:17:31 no it's not, but it can be hard to come back with ai 2011-06-23T16:17:47 ah well, variation of strategies is good :) 2011-06-23T16:18:04 it sounds like we have a lot of the same ideas otherwise 2011-06-23T16:18:19 if at move 100 we have same number of ants, but you have higher score because you killed lots of enemies, we are still ~level 2011-06-23T16:18:25 i'm afraid george1914 is quite right. 2011-06-23T16:18:37 stocha: thanks 2011-06-23T16:18:56 but dont tell everyone :) 2011-06-23T16:19:25 george1914: not necessarily 2011-06-23T16:19:36 the amount of territory you have also affects the amount of food you have first access to 2011-06-23T16:19:36 well not really afraid. it's just an expression. maybe from the return of the jedi. The emperor says that :"I'm afraid your the shield will be on, when your friend arrive" or something like that. he is afraid of something. 2011-06-23T16:19:42 taking out enemies increases your territory 2011-06-23T16:20:26 ok, so measure territory/production in payoff function, but dont worry about score too much 2011-06-23T16:20:44 you can't know the score can you ? 2011-06-23T16:20:53 i mean the engine doesn't send it to you ? 2011-06-23T16:20:58 you can know your own ... 2011-06-23T16:20:59 stocha: you can know the relative change in score 2011-06-23T16:21:12 (well even if it does, i don't certainly don't what to do with it but well) 2011-06-23T16:21:21 antimatroid: you receive that ? 2011-06-23T16:21:29 stocha: you have dead ant information 2011-06-23T16:21:41 antimatroid: yeah ? what to do with it ? 2011-06-23T16:21:48 also, when you're making your moves, you implement the scoring system yourself 2011-06-23T16:21:50 if you know where the bodies are buried ... 2011-06-23T16:21:58 so you end up with aproximating a useless value. 2011-06-23T16:22:06 and then ? 2011-06-23T16:22:07 just like you have to implement battle resolution and ant spawning etc. 2011-06-23T16:22:19 it's not necessarily useless 2011-06-23T16:22:30 you can use the information when selecting moves from a game tree 2011-06-23T16:22:40 the score go higher when your ants dies, than when they do not. 2011-06-23T16:22:47 doesn't it ? 2011-06-23T16:22:56 or is it only food and kill that gives score. 2011-06-23T16:22:59 maybe it is. mm 2011-06-23T16:23:12 still i don' think it's usefull for decision making. 2011-06-23T16:23:15 you can do something like yourScore-enemiesScores or whatever you want to evaluate with 2011-06-23T16:23:28 antimatroid: what's the point ? 2011-06-23T16:23:37 then you pick whatever move does the best change in score between you and your enemies at each decision, and they do the same 2011-06-23T16:23:50 stocha: this is how most people do ai :P 2011-06-23T16:23:56 by score you mean value of the evaluation function then. 2011-06-23T16:23:57 in games* 2011-06-23T16:24:05 ok, i think i go now. nice talking to yous, as always, antimatroid, stocha :) bye 2011-06-23T16:24:20 george1914: bye 2011-06-23T16:24:21 george1914: i think you more sensible than us. 2011-06-23T16:24:42 we just can't stop talking it seems. 2011-06-23T16:25:31 i should go too. have a family etc. 2011-06-23T16:25:41 *** stocha has quit IRC (Quit: Page closed) 2011-06-23T16:27:02 *** chl has joined #aichallenge 2011-06-23T16:32:40 *** Naktibalda_ has quit IRC (Remote host closed the connection) 2011-06-23T16:47:31 *** sigh has joined #aichallenge 2011-06-23T16:50:08 *** gobang has quit IRC (Ping timeout: 250 seconds) 2011-06-23T16:59:03 *** gobang has joined #aichallenge 2011-06-23T17:04:43 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-23T17:09:42 *** Palmik has quit IRC (Remote host closed the connection) 2011-06-23T17:14:59 *** racko has quit IRC (Quit: Page closed) 2011-06-23T17:26:14 *** sigh has joined #aichallenge 2011-06-23T17:30:44 *** nux67 has quit IRC (Ping timeout: 252 seconds) 2011-06-23T17:38:43 @rainbow ranbows rock 2011-06-23T17:38:43 mcstar: ranbows rock 2011-06-23T17:38:48 *** mcstar has quit IRC (Quit: WeeChat 0.3.5) 2011-06-23T17:41:18 *** Zaphus has joined #aichallenge 2011-06-23T17:41:28 morning all 2011-06-23T17:44:11 I've been watching lots of games and having some thoughts... and I was wondering: The food system currently favors being well spread out over the map, i.e. diffuse. If the food were to stop arriving at some point then the ants could turn to attacking more readily, but right now it's barely worth their while to bunch up 2011-06-23T17:44:13 *** okayzed is now known as okay 2011-06-23T17:46:26 *** Zaphus has quit IRC (Quit: Page closed) 2011-06-23T17:59:23 *** olexs has left #aichallenge 2011-06-23T18:00:54 *** foRei has quit IRC (Quit: Bye) 2011-06-23T18:04:28 *** davidjliu has quit IRC (Ping timeout: 252 seconds) 2011-06-23T18:04:55 aichallenge: janzert epsilon * rcc180ee / website/sql.php : Average worker stats over 15 minutes - http://bit.ly/mbzytd 2011-06-23T18:17:06 *** ltriant has joined #aichallenge 2011-06-23T18:17:15 *** Zaphus has joined #aichallenge 2011-06-23T18:20:50 *** amstan has quit IRC (Ping timeout: 260 seconds) 2011-06-23T18:31:32 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-23T18:37:47 *** sigh has joined #aichallenge 2011-06-23T18:46:22 *** amstan has joined #aichallenge 2011-06-23T18:46:22 *** ChanServ sets mode: +o amstan 2011-06-23T19:09:00 contestbot_: seen demerzel 2011-06-23T19:09:00 amstan: demerzel was last seen in #aichallenge 3 weeks, 6 days, 8 hours, 41 minutes, and 16 seconds ago: haha 2011-06-23T19:09:24 contestbot_: later tell janzert we have another month of the beta server 2011-06-23T19:09:24 amstan: I come to serve. 2011-06-23T19:11:47 *** onensora has quit IRC (Ping timeout: 258 seconds) 2011-06-23T19:13:16 *** okay is now known as okayzed 2011-06-23T19:13:44 *** onensora has joined #aichallenge 2011-06-23T19:21:46 *** okayzed is now known as okay 2011-06-23T19:33:50 *** onensora has quit IRC () 2011-06-23T19:33:53 amstan: sounds good 2011-06-23T19:36:46 contestbot_: later tell McLeopold I checked out submission 1223 and it compiles within a few seconds on the worker vm locally 2011-06-23T19:36:46 janzert: Yes master! 2011-06-23T19:48:47 ok 2011-06-23T20:15:41 *** Zaphus has quit IRC (Quit: Bye) 2011-06-23T20:35:46 *** FireFly has quit IRC (Remote host closed the connection) 2011-06-23T20:46:29 McLeopold: go ahead and change game parameters whenever you want 2011-06-23T20:47:10 I'll stop looking at the game cutoff change difference at 35000 2011-06-23T20:49:05 so the big difference overall with the cutoff is it moves 10% of games that would have gone for a full 2000 turns and ends them in less than 200 2011-06-23T20:56:34 what's the average cutoff? 2011-06-23T20:57:32 average length? I didn't look at cutoff lengths separately at this point 2011-06-23T20:58:00 I guess, average number of turns prevented, assuming cutoff games go the full length 2011-06-23T20:58:37 with cutoff enabled the average game length is 702 without it was 869 2011-06-23T20:58:40 I'm also thinking of dropping the game length to 1500. Have you seen winning or ranking turn above that? 2011-06-23T20:59:40 +-------------------+-------------------+-------------------+-------------------+ 2011-06-23T20:59:40 | max(winning_turn) | max(ranking_turn) | avg(winning_turn) | avg(ranking_turn) | 2011-06-23T20:59:41 +-------------------+-------------------+-------------------+-------------------+ 2011-06-23T20:59:41 | 1988 | 1998 | 221.3401 | 380.2404 | 2011-06-23T20:59:41 +-------------------+-------------------+-------------------+-------------------+ 2011-06-23T20:59:41 right now about 14% of games go longer than 1500 turns 2011-06-23T21:00:08 oh, which game when 1988? 2011-06-23T21:00:35 games involving a top ten player less than 5% of games go beyond 1500 though 2011-06-23T21:00:52 oops, just over 5% 2011-06-23T21:01:10 5.444444...% 2011-06-23T21:01:39 http://aichallengebeta.hypertriangle.com/visualizer.php?game=34763 2011-06-23T21:01:46 not a great game, but a late win 2011-06-23T21:08:26 janzert: the average game that goes 2000 turns stops change the rank at about 1000 turns 2011-06-23T21:10:19 out of 451 games, 12 would be different due to a cutoff at 1500 2011-06-23T21:12:09 the blocking in this is pretty cool http://aichallengebeta.hypertriangle.com/visualizer.php?game=35148 2011-06-23T21:13:08 http://imgur.com/a/v4Pu5#jBNhp 2011-06-23T21:13:12 aichallenge: McLeopold epsilon * r9ddb6b3 / (4 files): update to game list - http://bit.ly/knGlb8 2011-06-23T21:13:12 aichallenge: McLeopold epsilon * rc5c0e8b / website/sql.php : Merge branch 'epsilon' of github.com:aichallenge/aichallenge into epsilon - http://bit.ly/khoOas 2011-06-23T21:13:12 aichallenge: McLeopold epsilon * ra3d63da / (5 files): update to game lists and next game calc - http://bit.ly/lUvfxb 2011-06-23T21:13:41 *** jmpespxoreax has quit IRC (Ping timeout: 240 seconds) 2011-06-23T21:14:48 I think I'll keep the insane presicion for next game calcs. :D 2011-06-23T21:15:16 there are the histograms for cutoff and no cutoff games both all games played and games involving a top 10 player only 2011-06-23T21:15:21 hehe 2011-06-23T21:15:50 including -0.10... minutes eh ;) 2011-06-23T21:15:56 what's the y axis? 2011-06-23T21:16:33 some people need validation that they *are* overdue for a game :) 2011-06-23T21:16:45 first bucket is less than 100 turns, second less than 200 and so on 2011-06-23T21:17:13 blue bars show the percentage that fall in that bucket red bars are the cumulative percentage 2011-06-23T21:17:23 oic 2011-06-23T21:18:02 *** jmpespxoreax has joined #aichallenge 2011-06-23T21:18:41 okay, I'll make the parameter switch now 2011-06-23T21:18:41 I think the changes between no cutoff and cutoff show that it's working really well 2011-06-23T21:19:08 it looks like most of the change is in games that would have otherwise gone to the turn limit anyway 2011-06-23T21:19:22 sounds good 2011-06-23T21:19:50 should I try lowering the cutoff? 2011-06-23T21:20:01 85%? 2011-06-23T21:20:20 I want to error on the side of more games just to test 2011-06-23T21:20:27 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-23T21:20:47 I wouldn't mind testing, but I don't like the idea of mixing it in with other parameter changes 2011-06-23T21:21:11 okay, I'll wait for that 2011-06-23T21:21:19 game length is now 1500 2011-06-23T21:21:36 let's see if the gmp goes up 2011-06-23T21:21:41 gpm 2011-06-23T21:21:42 I'd also like to add a per map maximum game length 2011-06-23T21:21:53 that is dynamically adjusted 2011-06-23T21:22:34 <_flag> Your next game should be in about 6.146689498383561643000000000000 minutes. 2011-06-23T21:22:38 <_flag> Nice to know :P 2011-06-23T21:22:48 :D 2011-06-23T21:22:51 <_flag> "About" 2011-06-23T21:23:33 *** eggs_ has joined #aichallenge 2011-06-23T21:24:46 _flag: I'm getting way more games than you 2011-06-23T21:24:58 <_flag> McLeopold: Yeah I noticed that 2011-06-23T21:25:07 <_flag> How are you distributing the match making? 2011-06-23T21:25:51 <_flag> Could it possibly be because nobody wants to seed me into their games? D: 2011-06-23T21:25:56 pick the person to have a game the longest time ago 2011-06-23T21:26:05 Then pick good matches for that player 2011-06-23T21:26:22 <_flag> I'm probably rarely a good match 2011-06-23T21:26:30 so, my bot is probably getting picked more often than yours for playing against the seed 2011-06-23T21:27:20 hmm, seeding by least recently played should be helping compensate some but maybe not fully for that 2011-06-23T21:27:34 and apparently it isn't 2011-06-23T21:27:40 if he is never the best match, he will have the slowest rate 2011-06-23T21:27:43 it means he'd be picked more often as the player to pick others from yeah? 2011-06-23T21:28:16 <_flag> Uhhh... "Your next game should be in about -1.367493112024861690000000000000 minutes." 2011-06-23T21:28:18 yes, but it may still not be enough 2011-06-23T21:28:34 doesn't it pick the bot with the least number of games? 2011-06-23T21:28:37 I need to put in a few more "no replay" exclusions 2011-06-23T21:28:55 especially if matchups aren't spreading out the pairings enough 2011-06-23T21:29:00 A match is usually considered better when the sigma is low 2011-06-23T21:29:42 what happened to don't add anyone that has played with any of the current players in their last 5 games? 2011-06-23T21:30:05 *** sigh has joined #aichallenge 2011-06-23T21:30:20 the server seems to be taxed 2011-06-23T21:30:28 at least I thought you mentioned that was in there at one point 2011-06-23T21:31:03 loadaverage isn't too bad 2011-06-23T21:31:08 *** sigh has quit IRC (Client Quit) 2011-06-23T21:33:12 clearly we have code issues whenever the page render is above 1000ms 2011-06-23T21:33:35 with 97 active players and a minimum of 4 players per game how can someone go 51 games without getting a game? 2011-06-23T21:34:23 I've noticed the "player_ahead" often drop by 2 2011-06-23T21:34:47 meaning only 2 distinct users that were behind just played 2011-06-23T21:35:03 also meaning that 2 or more players got 2 games in since your last one 2011-06-23T21:35:29 http://aichallengebeta.hypertriangle.com/visualizer.php?game=35001&user=126 2011-06-23T21:35:31 There should be a certain amount of that, but I'm not sure what the correct amount is 2011-06-23T21:35:34 how to finish a game in 132 turns 2011-06-23T21:35:48 If it was on a strict rotation, you'd always play the same bots 2011-06-23T21:36:28 there we go, cpu to 100 percent for 10 seconds 2011-06-23T21:36:28 server network issues again? 2011-06-23T21:36:58 oh, maybe that was me trying to view flagcapper's profile page 2011-06-23T21:37:45 I think it's both network and some sql stuff going on 2011-06-23T21:38:13 hmm, yeah my ssh session seems flaky as well 2011-06-23T21:38:18 the next_game_in was running in < 0.03 seconds most of the time 2011-06-23T21:38:36 <_flag> Is the matchmaking done on the server? The players ahead of me just went down as it loaded 2011-06-23T21:38:43 I can't think of any change that should have affected it this much 2011-06-23T21:39:01 _flag: yes and what? 2011-06-23T21:39:16 ping is showing quite a bit of packet loss at the moment 2011-06-23T21:39:25 <_flag> Maybe that could explain why it seems to slow down periodically and then speed back up again 2011-06-23T21:39:33 janzert: so, not my code? 2011-06-23T21:39:48 is slow queries on and did it trigger? :) 2011-06-23T21:40:02 as soon as I get control of ssh again, I can check 2011-06-23T21:40:16 but I think it's a good chance most of the slow page loads are and such are network related right now 2011-06-23T21:40:35 ooh, host unreachable now 2011-06-23T21:40:36 okay, that means slow queries wouldn't be triggered then 2011-06-23T21:40:47 right if it's truly the network 2011-06-23T21:41:01 <_flag> Now it's instantaneous for me 2011-06-23T21:41:15 so, when do we get the real vps? :) 2011-06-23T21:41:25 <_flag> So I think it finished doing whatever it was doing 2011-06-23T21:41:35 <_flag> Or maybe not... 2011-06-23T21:41:54 _flag: more likely the router decided to let some traffic through :( 2011-06-23T21:42:19 generate_leaderboard took 20 seconds 8) 2011-06-23T21:42:29 page rendered in 128 milliseconds 2011-06-23T21:42:34 ouch 2011-06-23T21:42:39 <_flag> ^^ 2011-06-23T21:42:53 or wait is that the manager update? 2011-06-23T21:43:02 not so bad if so 2011-06-23T21:43:22 yeah, something is up with the leaderboard stuff 2011-06-23T21:44:08 yeah, it does look like mysql is sitting at 100% cpu an inordinate amount 2011-06-23T21:44:53 heh, 121 packets transmitted, 53 received, +2 errors, 56% packet loss, time 342086ms 2011-06-23T21:45:00 not a great ping run there 2011-06-23T21:46:28 janzert: I flushed the data and we are back to 1 second 2011-06-23T21:46:58 hmm, 2.5 on a clean table 2011-06-23T21:47:08 I'll see if I can make that faster later. 2011-06-23T21:47:17 the next game calculation is rather weird, 17 players ahead, current rate 4.333 players per minute and next game in -1.52 minutes 2011-06-23T21:47:23 Perhaps I need to add archiving to the manager :) 2011-06-23T21:47:44 for who? 2011-06-23T21:48:04 that is from flagcapper 2011-06-23T21:48:49 perhaps the workers went down 2011-06-23T21:48:55 I would expect the next game calculation to be players ahead / game rate though? 2011-06-23T21:49:16 maybe with another constant division for jumping in line 2011-06-23T21:49:24 minus current average time elapsed from running games 2011-06-23T21:49:52 he was in a game 2011-06-23T21:50:02 I suppose I could check for that? 2011-06-23T21:50:47 what I used to do is if they had a game scheduled that wasn't too far down the queue just sage the next game should be soon 2011-06-23T21:50:59 s/sage/say 2011-06-23T21:51:16 I think I'll just check if they are in the matchup table 2011-06-23T21:51:29 since it is generated on the fly 2011-06-23T21:51:34 yeah, that should be fine now since matches aren't scheduled so far ahead anymore 2011-06-23T21:51:52 I could even give a list of players they are in the match with 2011-06-23T21:52:42 *** okay is now known as okayzed 2011-06-23T21:53:49 *** McLeopold has left #aichallenge 2011-06-23T22:18:56 *** sigh has joined #aichallenge 2011-06-23T22:40:35 *** antimatroid-uni has joined #aichallenge 2011-06-23T22:57:12 contestbot_: later tell mcleopold Your next game should be in about 0.050000000000000000000000000000 minutes. 2011-06-23T22:57:12 amstan: Yes master! 2011-06-23T23:40:07 *** ShintoTuna has joined #aichallenge 2011-06-23T23:41:31 *** ShintoTuna has left #aichallenge