2010-11-13T00:01:25 what's one with close starting angle 2010-11-13T00:04:42 201 is pretty close, is that one even legal? 2010-11-13T00:05:31 it is 2010-11-13T00:05:44 ok that's strange, i took out all the defense code and it's still defending?! 2010-11-13T00:06:27 legal requirements at the start are that at least one neutral is claimable by both players without losing their home planet to ragebot 2010-11-13T00:06:38 although i think i'm in favour of increasing the min starting distance between players 2010-11-13T00:08:20 I guess defending is just attacking from a different point of view 2010-11-13T00:08:36 *** iFire has quit IRC (Read error: Connection reset by peer) 2010-11-13T00:10:20 give me some map numbers to try antimatroid, before I have to go 2010-11-13T00:11:12 *** iFire has joined #aichallenge 2010-11-13T00:12:32 i haven't really looked through for cool ones sorry 2010-11-13T00:12:36 i am trying to write a new bot 2010-11-13T00:12:42 i only wrote the generator because no one else was 2010-11-13T00:13:03 and i want things in concrete 2010-11-13T00:14:13 *** delt0r___ has joined #aichallenge 2010-11-13T00:16:12 *** delt0r has quit IRC (Ping timeout: 264 seconds) 2010-11-13T00:17:02 amstan: Error: There was a problem accessing the interface to ai-contest.com 2010-11-13T00:17:10 contestbot: shut up 2010-11-13T00:17:11 amstan: User error -- Replace user. 2010-11-13T00:19:47 I took out that thing in my bot that was relying on a line between the starters running through center, no great loss 2010-11-13T00:19:58 eburnette: :) 2010-11-13T00:20:13 i really don't think it will change rankings, just gives more awesome maps 2010-11-13T00:21:21 hmm, i'm unsure whether i like the idea of having a map generator set in stone at the start of future contests either 2010-11-13T00:21:35 by the sounds of it, a lot of people are making their bots to do well on specific maps rather than general ones 2010-11-13T00:21:45 Ok here's what I think of these maps... 2010-11-13T00:21:59 - they're awesome! I wish we'd had them sooner but my bot runs ok with them and there are 2 weeks left before the contest starts anyway. 2010-11-13T00:22:17 :P, not a single person has said they don't actually like them 2010-11-13T00:22:37 i was worried you were going to be the first for a second ehe 2010-11-13T00:23:17 I don't like 201, but almost all the others I tried were good 2010-11-13T00:23:30 yeah i don't like 201 either 2010-11-13T00:23:42 i originally set their min starting distance to 7, but people in here suggested removing that 2010-11-13T00:24:57 I'm not too fond of the ones that are reeeeeally spaced out like 104 but I guess there has to be a mix 2010-11-13T00:25:09 104 was done with the old generator :P 2010-11-13T00:25:27 lol 2010-11-13T00:25:53 100-200 all old 2010-11-13T00:25:53 201-300 line symmetric 2010-11-13T00:25:53 301-400 combinations 2010-11-13T00:26:13 looks like i'm doing worse on the 100-200 range actually 2010-11-13T00:26:49 there seems to be quite a few maps in there with >23 planets 2010-11-13T00:27:51 234 is a bit close for my tastes 2010-11-13T00:28:12 yeah, i think min starting distance between players needs to be greater 2010-11-13T00:28:28 236 is close but not too bad 2010-11-13T00:28:40 236 is good 2010-11-13T00:28:47 *** burny has joined #aichallenge 2010-11-13T00:28:51 that's an awesome map 2010-11-13T00:28:56 259 looks like a ladybug 2010-11-13T00:29:08 i had a space invader the other day, and a cock and ball 2010-11-13T00:29:09 s 2010-11-13T00:29:13 but they were from other generations 2010-11-13T00:30:07 I can't figure out how to make an AI.. I downloaded the "cpp starter", and it had ZERO c or cpp files in it? and the instructions given to use the test bots, just says "'java' is not recognized as an internal or external command" .. am I doing something wrong? 2010-11-13T00:30:09 277 looks like a catfish 2010-11-13T00:30:14 the players home planets being the eyes 2010-11-13T00:30:46 burny: edit your bot in MyBot.cc 2010-11-13T00:30:55 and planetwars is your game state file 2010-11-13T00:31:21 which i renamed to game state and completely rewrote cause it was rank 2010-11-13T00:31:50 what is a cc file? 2010-11-13T00:32:00 no idea, but it works the same as a cpp one 2010-11-13T00:32:00 well you have my vote, good night (I just started a new bot on tcp - Houndoom) 2010-11-13T00:32:38 night 2010-11-13T00:33:09 so is it just a fairly sloppy contest, that files are named wacky, and I have to have java, and other stuff, in order to do it in c++? 2010-11-13T00:34:00 the contest isn't sloppy, it just lets people use many languages 2010-11-13T00:34:07 and parts of the engine require java to run 2010-11-13T00:34:15 your bot will just be running with c++ 2010-11-13T00:34:43 i have no idea why .cc was used, i'm just a student, i just go with a flow ;) 2010-11-13T00:35:44 *** amriedle has quit IRC (Quit: Lost terminal) 2010-11-13T00:35:44 I never went to school, so setting up stuff like this is real painful for me 2010-11-13T00:35:45 http://stackoverflow.com/questions/1545080/correct-c-code-file-extension-cc-vs-cpp 2010-11-13T00:36:17 only the client engine requires java 2010-11-13T00:36:18 probable the reason the contest uses .cc is because that is the google styleguide way 2010-11-13T00:36:50 janzert: :) 2010-11-13T00:37:01 what are your current thoughts about the new maps? 2010-11-13T00:37:30 as a whole, i think people like them 2010-11-13T00:37:38 some are just worried about change 2010-11-13T00:37:53 some == about half 2010-11-13T00:38:16 I still mildly disapprove but don't mind whatever is decided 2010-11-13T00:38:37 I would really like the decision made in the next day or so though 2010-11-13T00:38:59 my understanding of both the tron contest and this one was that provided maps in way represent final maps 2010-11-13T00:39:05 no way* 2010-11-13T00:39:25 i might send an email through the discussions saying jeff, make a decision 2010-11-13T00:39:29 would that be best do you think? 2010-11-13T00:39:53 in the end he has to make the decision i think 2010-11-13T00:39:55 yeah, point him at the forum thread and ask him to make a final call 2010-11-13T00:39:59 yep 2010-11-13T00:42:22 for the submission, it has to have all the maps tools etc.? or just zip up the single source file? 2010-11-13T00:42:57 it needs any files needed for your bot, so not the tools or maps or anything but all your source files 2010-11-13T01:08:58 sent 2010-11-13T01:09:00 now we wait 2010-11-13T01:17:02 amstan: Error: There was a problem accessing the interface to ai-contest.com 2010-11-13T01:17:45 @rankings 2010-11-13T01:17:47 ... 2010-11-13T01:17:55 amstan: Error: There was a problem accessing the interface to ai-contest.com 2010-11-13T01:18:04 @schedule 2010-11-13T01:18:05 amstan: User error, it's not my fault. 2010-11-13T01:18:11 @scheduler delete rankings 2010-11-13T01:18:12 amstan: Run as fast as you can and don't look back. 2010-11-13T01:18:14 darn 2010-11-13T01:19:17 amstan: what's the time over there? 2010-11-13T01:19:23 @time 2010-11-13T01:19:23 amstan: 01:19 AM, November 13, 2010 2010-11-13T01:19:27 i sent a discussion email thing asking jeff to make decisions 2010-11-13T01:19:39 i guess it'll be a while then :P 2010-11-13T01:23:04 *** flag has joined #aichallenge 2010-11-13T01:23:13 @rankings 2010-11-13T01:23:14 flag: Top 10 players: bocsimacko(4458), FlagCapper(4185), dmj111(4184), Raschi(4149), shangas(4149), asavis(4131), george(4124), bix0r4ever(4114), _Astek_(4087), GreenTea(4065) 2010-11-13T01:24:23 :) 2010-11-13T01:24:38 ;) 2010-11-13T01:24:53 flag: have you taken a look at the new maps? 2010-11-13T01:25:00 no 2010-11-13T01:25:10 Where are they? 2010-11-13T01:25:11 http://ai-contest.com/forum/viewtopic.php?f=19&t=1087&start=0 2010-11-13T01:25:25 it's quite possible line symmetric maps will be used too 2010-11-13T01:25:33 and varying number of planets etc. 2010-11-13T01:25:49 they're more awesome 2010-11-13T01:27:02 interesting 2010-11-13T01:27:22 I hope they update it soon so it's not sprung on us right before the contest ends though 2010-11-13T01:27:29 on the third page there is a link to 300, the first 100 are done with the original generator, the second 100 are line symmetric, and the last 100 is combinations 2010-11-13T01:27:43 hopefully the decision will be made in the next 24 hours 2010-11-13T01:28:12 although, the rules have always stipulated that the example maps aren't representative of all maps that may be used 2010-11-13T01:28:48 besides, the strategy doesn't really change i don't think, they just provide more interesting games and more variety 2010-11-13T01:30:25 I think most people that are already doing well with these ones will be against it 2010-11-13T01:30:50 the official server still only has the 100 maps that were handed out to everyone 2010-11-13T01:30:59 that's the worst possible set to be trying to get rankings off 2010-11-13T01:31:37 I assumed that the ones on the tcp server were from the official one 2010-11-13T01:31:46 there are more on tcp 2010-11-13T01:31:49 but tcp has the offical ones 2010-11-13T01:32:30 the starting maps are only ever provided as examples, i don't think many people caught that 2010-11-13T01:32:52 yes, but it looks like most people think its too late to change them 2010-11-13T01:32:55 *** CryBaby has joined #aichallenge 2010-11-13T01:33:26 When they made that rule they probably intended on updating the maps sooner 2010-11-13T01:36:14 so did i 2010-11-13T01:36:31 but i've been waiting for the changes, and just finished exams, so finally getting around to pushing for the change 2010-11-13T01:36:39 anyone else could have gone ahead and pushed for it 2010-11-13T01:36:51 *** McLeopold has joined #aichallenge 2010-11-13T01:36:52 the maps are better and still random 2010-11-13T01:41:59 flag: you beat bocsimacko :P 2010-11-13T01:43:12 wow 2010-11-13T01:43:24 I'm not in first though 2010-11-13T01:43:27 Which is odd 2010-11-13T01:43:45 not really, you've had only 2 games 2010-11-13T01:44:22 the site is crawling atm 2010-11-13T01:44:41 cause I'm busy making it faster :P 2010-11-13T01:44:54 Well usually if I've won every game I've played my rank is higher than the best person I beat 2010-11-13T01:45:05 But since he's so far ahead of everyone else maybe that's not the case 2010-11-13T01:45:05 that wasn't a complaint, just a comment :) 2010-11-13T01:45:24 yeah, it'll take more than that to get ahead of bocsimacko 2010-11-13T01:45:29 yeah, just thought I'd throw an explanation out for the interuptions 2010-11-13T01:45:52 Anyway, I'm pretty sure his bot is still far better than mine 2010-11-13T01:45:57 a weird unhelpful explanation but... :) 2010-11-13T01:46:04 Just two lucky games so far :) 2010-11-13T01:46:22 i salute you for taking java out of second place though 2010-11-13T01:46:54 I'm actually surprised at how many Java programmers there are near the top 2010-11-13T01:47:01 I thought less of them D: 2010-11-13T01:47:09 so did i :P 2010-11-13T01:47:19 do** 2010-11-13T01:47:25 And then there's lisp 2010-11-13T01:48:05 I wonder if it's worth learning 2010-11-13T01:48:22 meh, i need to learn python and haskell before anything else 2010-11-13T01:48:48 lisp is worth learning 2010-11-13T01:48:58 javascript can be sorta like lisp 2010-11-13T01:49:18 at least some functional language should be learned since it will help you learn a new way to think about programs 2010-11-13T01:49:19 it takes a mental jump to grok it 2010-11-13T01:50:38 i just finished a pure maths major, i hope i could handle it 2010-11-13T01:50:49 @rankings 2010-11-13T01:50:50 flag: Top 10 players: FlagCapper(4528), bocsimacko(4424), Mangara(4181), Raschi(4141), shangas(4128), asavis(4123), bix0r4ever(4116), george(4116), dmj111(4108), _Astek_(4095) 2010-11-13T01:50:56 life goal accomplished! 2010-11-13T01:51:04 haha, congratulations 2010-11-13T01:51:11 Thanks 2010-11-13T01:51:17 I'll drop down soon enough 2010-11-13T01:51:48 congrats 2010-11-13T01:51:52 still, i don't think many people have done what you just did in the last two months without it resulting from bocsimacko resubmitting 2010-11-13T01:52:27 I guess, but not many people were lucky enough to play him in one of their first few games 2010-11-13T01:52:43 not many people have beaten him :P 2010-11-13T01:52:47 I'm gonna review your match and give my commentary 2010-11-13T01:52:59 If it will load... 2010-11-13T01:53:04 yeah, not to damper it but the new resubmission pairings make it much more likely 2010-11-13T01:53:06 McLeopold: that is your job 2010-11-13T01:53:42 janzert: Yeah, I noticed I wasn't having to climb my way back up anymore 2010-11-13T01:54:13 Wow, mine was the second loss he's had 2010-11-13T01:54:18 *** hellman has joined #aichallenge 2010-11-13T01:55:27 McLeopold: what do you think of the line symmetric map idea where all neutrals start in the "center" 2010-11-13T01:55:50 it'd probably be hard to make such a map that didn't end up in a draw actually 2010-11-13T01:56:00 so, 2 starting planets, and a line of planets? 2010-11-13T01:56:07 yep 2010-11-13T01:56:26 it would be very unstable. That type of map is closer to r/p/s than other ones. 2010-11-13T01:56:27 but line symmetric, not point symmetric 2010-11-13T01:56:34 that's my point :P 2010-11-13T01:57:20 I think if a few line maps end up looking like that, that would be fine, but I wouldn't want a whole series of them. 2010-11-13T01:57:51 no, i'm not even considering stuff like that for the contest 2010-11-13T01:58:05 i want that to be random 2010-11-13T01:58:22 and as you might have guessed, as general as possible :P 2010-11-13T02:00:10 flag: impressive win over mega1. You seemed to pick the better targets and make up for a small growth disadvantage. 2010-11-13T02:00:35 my bot should have taken that 68 when it was ahead 2010-11-13T02:01:04 Also, I noticed a 121 fleet at turn 62 that gave you the ability to take the 68 down south. 2010-11-13T02:01:29 I can't even load the game, just a sec 2010-11-13T02:01:30 by mega1, it flew back a forth 2010-11-13T02:02:30 Also, it looks like you concentrated your attacks where as mega1 spammed you line. Spamming is usually less effective. 2010-11-13T02:03:14 I'm at a loss (:-)) how to concentrate attacks 2010-11-13T02:03:37 The planet hopping you did to take the back growth 5 was good too. You didn't hold it, but it gave you a ship advantage. 2010-11-13T02:03:40 * janzert aims a kick at the database. wake up over there and quit complaining 2010-11-13T02:03:46 "(6:03:16 PM) mega1: I'm at a loss" that's a first :) 2010-11-13T02:04:11 um, 1 enemy planet at a time? 2010-11-13T02:04:29 Although, I still can't beat you, so I'm not one to talk. 2010-11-13T02:04:36 I don't have much direct control over that. 2010-11-13T02:04:50 really, why not? 2010-11-13T02:04:56 there a many situation where that would hurt too 2010-11-13T02:05:15 because behaviour comes from the evaluation function 2010-11-13T02:05:41 oh, your minimax 2010-11-13T02:05:56 yes, even if 1 ply 2010-11-13T02:06:22 although I now have a 2 ply version that beats it 2 to 1 but is too slow 2010-11-13T02:06:26 I suppose if it tells you to swap sides, then you can't argue. 2010-11-13T02:06:35 yes 2010-11-13T02:06:50 I may hack it still, but I'm not too keen on it. 2010-11-13T02:06:52 But, none of your targets landed with success. 2010-11-13T02:07:21 I suppose then, it thought the other bot would do something else. 2010-11-13T02:07:58 I'm gonna have to try 1 ply something. I can't see to get better position on you with just heuristics. 2010-11-13T02:08:20 Let me start cranking up the games and see if we play. 2010-11-13T02:08:25 on tcp... 2010-11-13T02:08:39 I'm testing stuff locally, can't play now. 2010-11-13T02:08:45 :( 2010-11-13T02:16:53 Top 10 players: FlagCapper(4534), bocsimacko(4431), Mangara(4187), Raschi(4147), shangas(4135), asavis(4130), bix0r4ever(4122), george(4122), dmj111(4114), _Astek_(4101) 2010-11-13T02:19:30 hmmm, that thing about infinite points on a circly, then extending those out to another cirlce and it not giving a continuous line is rather confusing 2010-11-13T02:30:06 ? 2010-11-13T02:40:18 if i use my generator, migi32 pointed out that the density of planets is going to be high around the centre 2010-11-13T02:40:39 so one should use like sqrt(rand(0,1))*maxRadius for generating a random radius 2010-11-13T02:41:20 but when you think about it, from a single point, you are just as likely to generate a random point on any circle centred around that point, but there is "more" points anyway the further out you go 2010-11-13T02:41:50 i find that confusing anyway 2010-11-13T02:42:10 it's like, the further you go out from a point, the more "space there is" but the same happens in the reverse etc. 2010-11-13T02:42:21 sqr rather than sqrt probably 2010-11-13T02:42:27 *** choas has joined #aichallenge 2010-11-13T02:42:29 sqrt 2010-11-13T02:42:39 0.5^2 = 0.25 2010-11-13T02:42:48 oh, ya.. you're right 2010-11-13T02:42:49 we're tring to have them not clump up around the middle 2010-11-13T02:43:28 you want a radius of 2 to be 4 times more likely to occur than a radius of 1 2010-11-13T02:43:29 oh man 2010-11-13T02:43:33 somehow jeff is still up 2010-11-13T02:44:06 actually, that might be for 3d 2010-11-13T02:44:18 @seen j3camero 2010-11-13T02:44:18 amstan: j3camero was last seen in #aichallenge 1 day, 5 hours, 21 minutes, and 18 seconds ago: The servers should crap out soon. But we're gonna set up the server-kicking script as a cron job soon. 2010-11-13T02:44:43 amstan: i thought he had responded about the map gen for a second 2010-11-13T02:44:56 antimatroid: he's online on gtalk 2010-11-13T02:45:18 the sooner that's settled the better 2010-11-13T02:45:29 amstan: really? not on my gtalk 2010-11-13T02:45:43 he's green for me, lol 2010-11-13T02:45:50 janzert: you have gtalk? 2010-11-13T02:45:52 heh, wierd 2010-11-13T02:46:06 mine is there, but i have never used it 2010-11-13T02:46:15 some people have randomly talked to me on there, and i'm just like wtf 2010-11-13T02:46:30 pidgin actually, but I have gtalk on it 2010-11-13T02:46:52 janzert: of course, that's clearly superior 2010-11-13T02:46:53 i love pidgin, but it's so buggy 2010-11-13T02:46:56 antimatroid: actually you want to use pi 2010-11-13T02:46:56 ahh, now I see him wierd 2010-11-13T02:47:02 *** jem has joined #aichallenge 2010-11-13T02:47:18 McLeopold? 2010-11-13T02:47:30 let's assume you would like to generate a map where all the planets are evenly spread out 2010-11-13T02:47:43 or at least, on average they will be 2010-11-13T02:47:48 *** jem has quit IRC (Client Quit) 2010-11-13T02:48:00 so rand(0,1)^(1/pi)? 2010-11-13T02:48:09 i can't see how the power could be greater than 1 2010-11-13T02:48:25 any idea how the top 10 AI's perform compared to a human player? 2010-11-13T02:48:28 the circumference of radius 1 is 1*pi*2, and radius 2 is 2*pi*2 2010-11-13T02:48:41 or 6.28 and 12.56 2010-11-13T02:48:45 burny: so.. you know the AI from galcon? that can kick your ass in hard? 2010-11-13T02:48:49 burny: the only way you're going to beat a top 50 bot is from ruining it on the first few moves 2010-11-13T02:49:00 burny: it's somewhere in the last few pages 2010-11-13T02:49:04 I would argue that twice as many planets at radius 1 than radius 2 would be even 2010-11-13T02:49:08 whoops 2010-11-13T02:49:15 twice as many at radius 2 2010-11-13T02:49:38 ah, i never tried the AI on galcon 2010-11-13T02:49:46 level10? 2010-11-13T02:49:52 hand on, i'm so confused, at the moment it would calculate 2010-11-13T02:49:52 r = sqrt(rand(0,1) )*maxRadi 2010-11-13T02:49:55 what should it be? 2010-11-13T02:50:03 http://www.galcon.com/flash/ 2010-11-13T02:50:19 space planets along the circumference using a minimum distance (arc) between planets 2010-11-13T02:50:36 antimatroid: i'm thinking... 2010-11-13T02:50:44 i don't think it needs to be perfect 2010-11-13T02:50:57 i think even the sqrt fix would be fine, but i'll play around with things when i clean it up 2010-11-13T02:53:41 I just beat vl10 3 times in a row 2010-11-13T02:55:21 antimatroid: I think you want log(rand(0,1),2)? 2010-11-13T02:56:57 i'll try them both out after and see what the maps generated look like 2010-11-13T02:57:00 i'm out for a while 2010-11-13T02:57:25 sqrt will create more doughnut shaped maps 2010-11-13T02:57:40 too much is taken from the center 2010-11-13T02:57:46 yeah, okay 2010-11-13T02:58:31 one could generate a in (0.5, 1) then use that for the whole map generating r=rand(0,1)^a*maxr 2010-11-13T02:59:52 *** CryBaby has left #aichallenge 2010-11-13T03:01:42 I think you'll do better playing with the log base 2010-11-13T03:02:10 And, I would suggest 2 is the better number 2010-11-13T03:02:19 is it just me or is the main server zzzzing? 2010-11-13T03:02:23 @rankings 2010-11-13T03:02:24 amstan: Top 10 players: bocsimacko(4423), FlagCapper(4322), bix0r4ever(4176), Raschi(4147), shangas(4140), george(4137), dmj111(4136), Mangara(4126), asavis(4126), _Astek_(4099) 2010-11-13T03:05:48 amstan: it's off and on right now, I'm doing some database reorganization 2010-11-13T03:06:01 oh, ok, np 2010-11-13T03:09:05 *** Azrathud has quit IRC (Remote host closed the connection) 2010-11-13T03:15:54 so regarding AI vs human.. I beat the AI on lvl10 about 75% of the time, but that has a large reaction time component to it, that doesn't really apply in the contest 2010-11-13T03:16:54 Top 10 players: bocsimacko(4404), FlagCapper(4191), Raschi(4144), shangas(4137), bix0r4ever(4130), asavis(4129), _Astek_(4126), george(4124), bpellom(4107), oldman(4089) 2010-11-13T03:25:06 *** paulmcq_ has quit IRC (Ping timeout: 265 seconds) 2010-11-13T03:33:35 *** Frontier has joined #aichallenge 2010-11-13T03:44:18 *** Utkarsh has joined #aichallenge 2010-11-13T03:56:24 burny: you can play against your own bot using jBotManager 2010-11-13T03:57:15 ya, I'm just trying to get an idea of how good it has to be, to compete in the top bunch 2010-11-13T03:58:05 they are far from optimal 2010-11-13T03:58:22 it's a surprisingly intricate game 2010-11-13T03:58:33 the slightest tweak of numbers completely changes the best strategy 2010-11-13T03:58:42 compile that into a map with over 20 planets and it's insane 2010-11-13T03:58:43 ya, 2010-11-13T03:59:04 it's highly dependant on luck though(atleast the galcon one is) 2010-11-13T03:59:27 maps in this are always symmetrical 2010-11-13T03:59:30 but in teh last 20games I did against the lvl10 ai, I only lost 1 2010-11-13T03:59:41 ah, so zero luck factor: cool 2010-11-13T04:00:18 there can still be situations where there is not best move in pure strategies 2010-11-13T04:00:22 but imo they are going to be quite rare 2010-11-13T04:01:00 well, the best move then, is just more complicated 2010-11-13T04:01:22 like, if you have a losing situation, and you determine that if other player plays well, they win no matter what.. 2010-11-13T04:01:28 once you're using mixed strategies, results aren't consistent 2010-11-13T04:01:29 you can still win 2010-11-13T04:02:00 as in the game of goe. you don't concede if the other player gets to place 5 stones before your first stone 2010-11-13T04:03:08 oh.. the only reason lvl10 difficulty is even hard, is cause it has more units than you do 2010-11-13T04:03:34 that would explain why i couldn't own it before 2010-11-13T04:03:43 it was going for a neutral near me, i waited and still lost 2010-11-13T04:04:15 ya, 2010-11-13T04:04:20 that's why I just lost another one 2010-11-13T04:04:39 it went for the same thing as me right off the start, and I sent my entire army 2010-11-13T04:04:51 you do want to arrive just after them 2010-11-13T04:04:57 so they waste their ships taking out neutral 2010-11-13T04:05:04 also, i take it you've just started for this? 2010-11-13T04:05:12 make sure you read how neutrals are resolved 2010-11-13T04:05:22 there was a number of ways that could have been done 2010-11-13T04:06:06 ya, just starting now 2010-11-13T04:06:37 and yes, I arrived just after them, it was just on their side ofmiddle, and I sent slightly later 2010-11-13T04:16:54 Top 10 players: bocsimacko(4367), FlagCapper(4222), Raschi(4185), _Astek_(4143), george(4128), shangas(4128), oldman(4117), asavis(4116), dmj111(4110), bix0r4ever(4108) 2010-11-13T04:18:39 *** tapwater has quit IRC (Quit: tapwater) 2010-11-13T04:46:43 *** hellman has quit IRC (Remote host closed the connection) 2010-11-13T04:58:24 *** McLeopold has quit IRC (Ping timeout: 276 seconds) 2010-11-13T05:05:23 *** Rubicon-|-Cross has joined #aichallenge 2010-11-13T05:16:56 Top 10 players: bocsimacko(4379), Raschi(4242), _Astek_(4135), bix0r4ever(4127), george(4124), shangas(4114), asavis(4090), dmj111(4087), oldman(4082), FlagCapper(4054) 2010-11-13T05:55:09 *** Apophis has joined #aichallenge 2010-11-13T05:55:23 *** Apophis has quit IRC (Read error: Connection reset by peer) 2010-11-13T05:59:13 burny: When you snipe a neutral make sure you can hold it until the planet produced twice the ships you lost on capturing. Otherwise you could just have let the enemy take it. 2010-11-13T06:13:01 Frontier: why does it matter? It doesn't cost anything (relatively speaking) to throw ships at the enemy, so any turn you own it and he doesn't is a net gain for you 2010-11-13T06:14:30 lavalamp_: Just keep your belief. The things I make up when I go through the park at night never failed me so far :p 2010-11-13T06:15:17 I can see it might be a bad idea to trade ships in some circumstances, but in my mind that's a separate issue :) 2010-11-13T06:15:23 he said when sniping neutrals 2010-11-13T06:15:27 is that expanding? 2010-11-13T06:15:48 i have always disagreed with the term "sniping", i don't consider it any different to normal offense 2010-11-13T06:15:51 If you're sniping, by definition they took it first 2010-11-13T06:16:20 antimatroid: yeah, my code just looks for optimal time to take a planet, sniping is a side effect 2010-11-13T06:16:56 Top 10 players: bocsimacko(4375), Raschi(4175), bix0r4ever(4156), george(4111), shangas(4103), _Astek_(4098), asavis(4079), dmj111(4059), GreenTea(4027), Accoun(4025) 2010-11-13T06:20:41 *** Bobng has joined #aichallenge 2010-11-13T06:23:07 Ah that is the meaning of sniping. What I meant to say is: "When you take a neutral just before the enemy does." 2010-11-13T06:23:58 ah... in that case... don't do that!! 2010-11-13T06:24:42 I think my code shouldn't take neutrals it doesn't think it can hold indefinitely 2010-11-13T06:25:59 but suddenly your original comment makes sense ^^ 2010-11-13T06:28:58 *** mceier has joined #aichallenge 2010-11-13T06:47:11 *** _xvinyl <_xvinyl!~xvinyl@ANantes-256-1-33-33.w86-199.abo.wanadoo.fr> has quit IRC (Remote host closed the connection) 2010-11-13T07:09:04 *** agravier has joined #aichallenge 2010-11-13T07:10:41 hi all. 2010-11-13T07:12:24 hello 2010-11-13T07:13:03 I'd like to get an official update from an admin (amstan if you are reading this...) on the Clojure SP. It's not so much that I insist to get a posivite response, but just _any_ word from someone at the CS club would do :) 2010-11-13T07:13:43 *** narnach has joined #aichallenge 2010-11-13T07:13:43 *** narnach is now known as narnach_mb 2010-11-13T07:13:46 i expect he's fast asleep atm sorry 2010-11-13T07:13:47 @time 2010-11-13T07:13:47 antimatroid: 07:13 AM, November 13, 2010 2010-11-13T07:13:48 *** yasith has joined #aichallenge 2010-11-13T07:13:54 he was up at like 2am 2010-11-13T07:14:15 Ho I see sorry I have no idea, I'm at GMT +8 2010-11-13T07:14:22 I'm at +10 :) 2010-11-13T07:14:36 well, +11 atm, daylight savings <3 2010-11-13T07:14:53 wait, that'd be +9 wouldn't it? :P 2010-11-13T07:15:19 +11 would be Japan no? 2010-11-13T07:15:25 tasmania 2010-11-13T07:15:28 (australia) 2010-11-13T07:15:29 *+10 I mean 2010-11-13T07:15:30 ok 2010-11-13T07:16:37 Alright, thanks for the info anyway, so, I'll try going on spamming amstan's PM inbox 2010-11-13T07:16:56 Top 10 players: bocsimacko(4345), Raschi(4196), bix0r4ever(4153), george(4109), shangas(4106), _Astek_(4096), dmj111(4091), asavis(4081), Accoun(4039), GreenTea(4031) 2010-11-13T07:17:04 I have to go, have fun all :) 2010-11-13T07:17:07 *** agravier has quit IRC (Quit: leaving) 2010-11-13T07:32:44 *** krokkrok has joined #aichallenge 2010-11-13T07:38:28 *** krokkrok has quit IRC (Quit: Page closed) 2010-11-13T07:46:17 *** fluffluff1 has left #aichallenge 2010-11-13T08:03:43 *** hellman has joined #aichallenge 2010-11-13T08:07:58 Error323: around? 2010-11-13T08:08:35 mega1: am now 2010-11-13T08:08:56 lol, like a few sec later 2010-11-13T08:10:58 Error323: wanna know the results? 2010-11-13T08:11:05 hurt me 2010-11-13T08:11:10 xD 2010-11-13T08:11:27 16/3/224 (win/draw/loss) 2010-11-13T08:11:36 hahaha omg 2010-11-13T08:12:43 not too bad :P 2010-11-13T08:12:58 given it's mega1 and all 2010-11-13T08:14:29 that's about in line with what's they do on tcp 2010-11-13T08:14:34 *what 2010-11-13T08:15:23 *** delt0r has joined #aichallenge 2010-11-13T08:15:56 *** delt0r___ has quit IRC (Ping timeout: 245 seconds) 2010-11-13T08:16:57 Top 10 players: bocsimacko(4355), Raschi(4220), bix0r4ever(4164), dmj111(4130), george(4111), _Astek_(4106), shangas(4096), asavis(4074), Accoun(4057), oldman(4053) 2010-11-13T08:18:04 mega1: was that only on old maps? 2010-11-13T08:18:10 yes 2010-11-13T08:18:27 haven't been tempted to try against it on new ones? :P 2010-11-13T08:18:31 i'd be interested in those results 2010-11-13T08:18:48 it's running already 2010-11-13T08:18:52 :) 2010-11-13T08:19:12 hopefully the final map generator will be pretty much in stone by 24 hours time 2010-11-13T08:19:52 i sent jeff an email earlier saying he needs to make a decision 2010-11-13T08:19:59 you have probably seen it :P 2010-11-13T08:23:25 yes 2010-11-13T08:28:38 5/0/95 2010-11-13T08:29:00 Error323: ^^^: on antimaps 2010-11-13T08:30:04 i'm surprised there was no draws 2010-11-13T08:32:50 sample size is too small 2010-11-13T08:33:14 and the difference is too great to tell much 2010-11-13T09:07:35 *** hellman has quit IRC (Remote host closed the connection) 2010-11-13T09:08:50 lavalamp_: did you go changing your mind again? 2010-11-13T09:09:01 hm? what about? 2010-11-13T09:09:10 the poll 2010-11-13T09:09:18 one person did :P 2010-11-13T09:09:24 haha it wasn't me 2010-11-13T09:09:34 i'm up 16-15 atm :P 2010-11-13T09:09:42 haha 2010-11-13T09:09:47 assuming options 2 and 3 go to line symmetry 2010-11-13T09:11:03 mega1: haha damn 2010-11-13T09:11:15 you created a true monster yes 2010-11-13T09:11:17 *** mk5875 has joined #aichallenge 2010-11-13T09:12:11 what you don't realise is that mega1 is a bot, hence why his bot is so good :) 2010-11-13T09:16:56 Top 10 players: bocsimacko(4368), bix0r4ever(4210), Raschi(4161), dmj111(4135), george(4128), shangas(4104), Accoun(4084), GreenTea(4071), asavis(4065), oldman(4042) 2010-11-13T09:20:23 *** Migi32 has joined #aichallenge 2010-11-13T09:48:01 *** fawek has joined #aichallenge 2010-11-13T09:54:43 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-11-13T09:56:17 *** narnach has joined #aichallenge 2010-11-13T09:56:17 *** narnach is now known as narnach_mb 2010-11-13T10:07:23 *** McLeopold has joined #aichallenge 2010-11-13T10:09:13 *** iNtERrUpT has joined #aichallenge 2010-11-13T10:09:57 *** p4p4 has joined #aichallenge 2010-11-13T10:10:00 *** p4p4 has quit IRC (Client Quit) 2010-11-13T10:11:17 *** p4p4 has joined #aichallenge 2010-11-13T10:14:32 *** tapwater has joined #aichallenge 2010-11-13T10:16:57 Top 10 players: bocsimacko(4380), bix0r4ever(4233), Raschi(4177), dmj111(4136), george(4127), shangas(4101), ClusterTeam(4100), FlagCapper(4086), Accoun(4081), asavis(4067) 2010-11-13T10:22:14 *** iNtERrUpT has quit IRC (Remote host closed the connection) 2010-11-13T10:25:21 *** McLeopold has quit IRC (Ping timeout: 276 seconds) 2010-11-13T10:53:26 *** phreeza has joined #aichallenge 2010-11-13T10:59:55 *** mk5875 has quit IRC (Ping timeout: 265 seconds) 2010-11-13T11:00:27 hey everybody. i haven't been following the contest since I dropped. i'm curious if there have been any breakthroughs or if all improvement have continued to be small and incremental? 2010-11-13T11:00:37 *improvements 2010-11-13T11:13:32 small and incremental as far as i know 2010-11-13T11:16:57 Top 10 players: bocsimacko(4385), bix0r4ever(4244), Raschi(4181), george(4115), dmj111(4111), Accoun(4095), _Astek_(4094), shangas(4092), FlagCapper(4074), asavis(4060) 2010-11-13T11:17:29 well mega1 (bocsimacko) had a major breakthrough but only he knows what it is 2010-11-13T11:17:38 you dropped? 2010-11-13T11:17:42 what does that mean 2010-11-13T11:22:11 *** p4p4 has quit IRC (Ping timeout: 245 seconds) 2010-11-13T11:25:17 i'm not sure he had any major breackthrough, rather he's just doing it better than anyone else and has been a long time 2010-11-13T11:27:00 mega1 doped his bot -.- It must be using multithreading on all 8 server cores or something 2010-11-13T11:27:27 lol 2010-11-13T11:28:17 lisp is an ai language, ok. he might be an ai specialist, ok. but isn't lisp still a script language? 2010-11-13T11:29:04 *** RainCT has joined #aichallenge 2010-11-13T11:29:33 due to the size of the search space, i'm convinced how one picks their moves is much more important than the number of moves you process 2010-11-13T11:29:41 so language really isn't mattering 2010-11-13T11:29:41 *** superflit has joined #aichallenge 2010-11-13T11:30:13 antimatroid: any conclusion reached on the map thing? 2010-11-13T11:30:33 phreeza: not yet, waiting for jeff to make a decision 2010-11-13T11:30:39 it'll be finalised after that 2010-11-13T11:30:46 cool 2010-11-13T11:31:13 ah fuck, jeff told me to make it under my best judgement 2010-11-13T11:32:49 janzert: you about? 2010-11-13T11:32:52 i see there are compilers for lisp. and you are right, using good heuristics is the only way to keep anything below 1 second 2010-11-13T11:33:35 lisp isn't a script language. 2010-11-13T11:34:10 i always mix up script and interpreted :/ 2010-11-13T11:34:33 mega1: have your thoughts changed at all on including new maps? 2010-11-13T11:35:22 i didn't expect to, and don't really want to be the one to make this decision, as i'm so obviously biased 2010-11-13T11:35:44 jeff didn't give a real answer? unfortunately that doesn't surprise me 2010-11-13T11:35:57 his answer was for me to decide 2010-11-13T11:36:30 he meant for everyone?? I assumed you meant he said something like "use your best judgment" 2010-11-13T11:36:44 make an issue on the tracker for people to discuss it? 2010-11-13T11:36:51 antimatroid: no, they haven't 2010-11-13T11:37:09 it's ridiculous; the contest organizers should make a decision and stick with it 2010-11-13T11:37:24 he sort of said yes do it, but not completely 2010-11-13T11:37:33 and they should at least tell us max and min parameters that we need to be able to handle 2010-11-13T11:37:41 I support this change. I think that almost anything would be better than the map generator that we have now. The current map generator is just something that I whipped up on the bus a couple months ago to create some non-trivial maps to test the first game engine. I didn't put much thought into creating maps that had interesting properties, and I always meant to circle back and create a better map generator. 2010-11-13T11:37:41 Can I put you in charge of creating the new map gen? It sounds like you have studied the problem in much greater detail than I have. I will defer to your best judgement on all the decision points that you raised. Try to build some consensus in the community on each of the points, but if you can't reach consensus just make an executive decision. Like you said, getting it done fast is better than getting it done perfectly. 2010-11-13T11:37:49 that's what he has said 2010-11-13T11:38:05 so it's not for me to just pick waht i want, but i pretty much get executive decision if i want to make it 2010-11-13T11:38:12 holy cow 2010-11-13T11:38:14 but i am way too biased for that i think 2010-11-13T11:38:32 yeah, no offense, but they should not be delegating that to contest participants 2010-11-13T11:38:58 I'm going to say no, he has to go through and say yes or no to each point i raised 2010-11-13T11:39:33 haha, amazing. now try... and try hard not to make the map generator spit out maps that your bot runs well on ^^ 2010-11-13T11:39:38 that sounds like the right thing for you to do... 2010-11-13T11:40:04 ideally they should write a new generator 2010-11-13T11:40:08 Frontier: map generator will be publicly viewable and making random maps 2010-11-13T11:40:20 it doesn't matter who writes it when it's open source 2010-11-13T11:40:29 does he not have time at all? It's like half an hour to read through that topic. 2010-11-13T11:40:34 and think it through too. 2010-11-13T11:41:12 not all properties are apperant in the source code 2010-11-13T11:41:31 If your bot favours dense centers and your random maps have a lot of good planets in the center compared to what the current generator spits out that would be problematic. But I trust you to do it well. 2010-11-13T11:41:45 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-11-13T11:41:48 if they were we weren't pulling each other's hair 2010-11-13T11:42:33 Are there any more problems with the map generator than just spawn positions too close for any real expansion? 2010-11-13T11:43:00 with the current one? 2010-11-13T11:43:08 yeah 2010-11-13T11:44:08 maps could be more varied 2010-11-13T11:44:45 *** p4p4 has joined #aichallenge 2010-11-13T11:44:55 i would at least like the possibility of multiple neutrals starting at an equal distance 2010-11-13T11:45:02 hmm, in what sense? less random, more template based -> "large map", "high growth map", "jump map", ...? 2010-11-13T11:45:06 but i really really like the line symmetrical maps 2010-11-13T11:45:25 they over a lot more variety 2010-11-13T11:45:28 offer* 2010-11-13T11:45:35 in my experience the line symmetrical maps tend to create stable fronts 2010-11-13T11:46:03 even more than the point symmetrical ones. 2010-11-13T11:46:06 one could set a minimum distance between players angles 2010-11-13T11:46:18 but decisions need to be made now 2010-11-13T11:46:24 so games are more likely to end in draw with line symmetry than with circular symmetry? 2010-11-13T11:46:32 *** Utkarsh has joined #aichallenge 2010-11-13T11:46:39 depending on the map, yeah 2010-11-13T11:46:54 no, not now. A long time ago. 2010-11-13T11:46:57 but there usually isn't draws unless the maps are completely trivial 2010-11-13T11:47:15 or you have near identical bots (different revisions of your own) 2010-11-13T11:49:43 right, i responded 2010-11-13T11:49:57 So you are now discussing advanced features for the map generator, instead of only fixing the bugs. I know that my next bot relies on point symmetry. I would like to know if that changes :o) 2010-11-13T11:50:24 that is what i want to change and mega1 doesn't 2010-11-13T11:50:34 and that is why i'm too biased to make a decision :) 2010-11-13T11:50:50 *** irchs has joined #aichallenge 2010-11-13T11:50:51 *** hellman has joined #aichallenge 2010-11-13T11:51:14 'cause if your bot is ok with line symmetry you have no additional work, but others do. ;-) 2010-11-13T11:51:19 even if i wasn't biased, i don't think i as a contestant should be making the decision 2010-11-13T11:51:39 people whose bots rely on point symmetry made their own assumptions 2010-11-13T11:52:07 it was never stated that maps would meet that criteria 2010-11-13T11:52:37 I made a lot of these assumptions. Like a game ends after turn 200, there are 23 planets max and things along the line ^^ 2010-11-13T11:52:50 yeah, bad idea :P 2010-11-13T11:53:08 fun: dropping symmetry alltogether and playing the game twice instead with swapped starting positions. 2010-11-13T11:53:19 turn 200 is the only thing I still rely on 2010-11-13T11:53:29 i don't think 200 will change 2010-11-13T11:53:32 but i don't rely on it 2010-11-13T11:53:41 the only thing i rely on is 2 player 2010-11-13T11:53:43 brx: NOOO, my algorithm can only calculate enemy moves effectively, because they are symmetrical to mine 2010-11-13T11:54:02 Frontier: the game is usually no longer symmetric after the first move? 2010-11-13T11:54:18 they reserve the right to have non-symmetric maps 2010-11-13T11:54:22 but i don't think that's a good idea 2010-11-13T11:54:44 phew 2010-11-13T11:55:00 antimatroid: the first turn is my most important turn. i'm calculating far ahead there. 2010-11-13T11:55:00 and my bot would probably timeout if planets were above 50 or 60 2010-11-13T11:55:29 *** mk453 has joined #aichallenge 2010-11-13T11:55:53 mega1: don't think you'd do well with nonsymmetric? 2010-11-13T11:55:58 i don't think anyone would 2010-11-13T11:56:14 it would be a drawfest 2010-11-13T11:56:15 it's very hard to come back once you're behind 2010-11-13T11:56:33 because we'd have to play both ways 2010-11-13T11:56:43 yeah, that's true 2010-11-13T11:56:51 terrible terrible idea 2010-11-13T11:58:09 my bot would be fine with a few hundred planets 2010-11-13T11:58:09 right, well i'm going to just make a clean full cut map genertor now 2010-11-13T11:58:12 then i can remove anything after 2010-11-13T11:58:13 one possible way to do "non-symmetric": 2010-11-13T11:58:24 let's increase the map count 2010-11-13T11:58:27 really? i haven't tested 2010-11-13T11:58:31 games of that size would look cool 2010-11-13T11:58:44 just being sarcastic 2010-11-13T11:58:45 i have some pretty nasty o(n^x) loops though 2010-11-13T11:59:01 generate a non-symmetric map. duplicate it and translate so that it would take 200 turns to get from one copy to the other. reverse player starting positions on the second copy. 2010-11-13T11:59:10 mega1: I bet your bot would manage to beat the others on any map *sigh* 2010-11-13T11:59:39 lavalamp_: or just run two games and merge results 2010-11-13T11:59:39 there are map types it doesn't like 2010-11-13T11:59:51 that way, even if one player wins each half, the one who did it more efficiently wins 2010-11-13T11:59:52 like any other bot 2010-11-13T12:00:09 but what type they like differs 2010-11-13T12:00:32 or you could put the halves only 100 moves away so there's some possibility of interacting late in the game 2010-11-13T12:00:32 there is no way to change the map generator without favouring some 2010-11-13T12:00:37 my new bot has only just started executing some moves tonight 2010-11-13T12:00:42 i'm taking my time 2010-11-13T12:01:14 the map generator was always meant to change 2010-11-13T12:01:17 antimatroid: I wish I was at that point. I have like 5 good ideas that cannot be merged to a concept ^^ 2010-11-13T12:01:24 you can't not change it without favouring some either 2010-11-13T12:01:49 Frontier: i spent a couple of months playing around with ideas :) 2010-11-13T12:02:05 i've thrown out most of my elaborate game theory ones 2010-11-13T12:02:07 it was too messy 2010-11-13T12:04:36 either way some are favoured. Why not favour those who actually invested effort? 2010-11-13T12:04:57 ok ok, but we need some clear statement now that there are only 2 weeks to go. :D 2010-11-13T12:05:23 mega1: i invested effort into having a very general strategy 2010-11-13T12:05:40 and i do happen to be on the side of what was always intended 2010-11-13T12:05:47 but i will admit it wasn't posted very clearly 2010-11-13T12:06:07 ok, I'm out of this. We are both broken records. 2010-11-13T12:06:16 i would prefer maps to always be different at the end, i just wish people were told blunter 2010-11-13T12:06:28 hehe 2010-11-13T12:07:41 Right I wish I was told there will be massive changes to the map generator ^^. And we need an email notification of the change to all participants so everyone is informed at the same time. 2010-11-13T12:07:44 imo the fuckup comes in that everybody wasn't clear on how it was going to work, rules are allowed to be general provided they're clear 2010-11-13T12:08:41 yeah, the mistake has already been made and the resolution is really unclear 2010-11-13T12:08:53 mega1: do you rely on things that weren't set in stone from the beginning? map radius, growth, anything? 2010-11-13T12:09:02 he doesn't 2010-11-13T12:09:16 maybe best would be to make a decision and give us an extra couple weeks ^^ 2010-11-13T12:09:19 his argument comes with the late change 2010-11-13T12:09:35 i have always expected it and don't see it as a change, neither of us is really wrong, hence the problem 2010-11-13T12:09:48 Frontier: not to my knowledge, but please read what I and others wrote about optimizing for the current map generator in the forum 2010-11-13T12:10:12 and i second that. 2 weeks ago i told janzer or some other official that by now there should be no more changes to the game. 2010-11-13T12:10:18 mega1: you have to understand the counterargument about not just optimizing on that generator 2010-11-13T12:10:31 mega1: in the strategy section? 2010-11-13T12:11:11 yeah, you're right, there is no point us trying to discuss it 2010-11-13T12:12:07 Frontier: http://ai-contest.com/forum/viewtopic.php?f=19&t=1087&start=50 2010-11-13T12:12:10 six pages 2010-11-13T12:12:25 lol, someone suggested extending deadline to january 2010-11-13T12:12:27 i would actually die 2010-11-13T12:12:53 hahah i wouldn't mind a second round though 2010-11-13T12:12:54 would make sense, but please don't 2010-11-13T12:13:00 I can't survive on coffee and pizza for so long Oo 2010-11-13T12:13:09 until the next game comes 2010-11-13T12:13:12 ai-contest.com is a pretty good domain 2010-11-13T12:13:15 how did you get it? 2010-11-13T12:13:19 it must have been squatted. 2010-11-13T12:13:56 i'm not an organiser :P 2010-11-13T12:14:11 It has a dash, so it's probably not that big of a domain 2010-11-13T12:14:47 i swear there wasn't this much issue with new maps for tron contest 2010-11-13T12:15:02 and that was right towards the end 2010-11-13T12:15:08 wow, who suggested 100 planets? my 32-bit-int-bitmasks only can hold up to 31 planets 2010-11-13T12:15:27 Frontier: i think you're safe, they aren't going to fuck people like you over, there's too many of you 2010-11-13T12:15:42 but imo it was a poor choice 2010-11-13T12:15:49 antimatroid: because for tron there was no map generator so we could not see a better way 2010-11-13T12:15:53 lol, that was what i was hoping for. i know i should feel ashamed. 2010-11-13T12:16:47 so the _huge_ shuffling of ranks and rush at the end was accepted 2010-11-13T12:16:58 Top 10 players: bocsimacko(4409), bix0r4ever(4208), Raschi(4172), _Astek_(4165), george(4109), Accoun(4086), shangas(4078), Error323(4077), dmj111(4077), asavis(4057) 2010-11-13T12:17:06 there will be a huge shuffling of results with this if you even regenerate the maps 2010-11-13T12:17:09 it doesn't mean that it didn't suck 2010-11-13T12:17:12 the main server only has the first 100 still 2010-11-13T12:17:16 comment of glevee: "I see what you're saying is that I should have been using maps of New York city to test my Bot, since no where does it say that they won't be used..." made my day 2010-11-13T12:17:33 antimatroid: where is your current generator at? 2010-11-13T12:17:39 The people who are doing well with the current maps will always reject the change and the people that are doing poorly will always want to mix something up 2010-11-13T12:17:50 janzert: which one? :P 2010-11-13T12:17:52 You can't really agree 2010-11-13T12:18:04 the latest one with the most features :) 2010-11-13T12:18:14 can you wait 30 minutes? 2010-11-13T12:18:18 it should really be cleaned 2010-11-13T12:18:19 sure 2010-11-13T12:18:21 flag: I thought you are in the first group? 2010-11-13T12:18:28 are you going to put some maps up? 2010-11-13T12:18:37 Up to 31 planets and no more is fine for me, as are line symmetrical maps, as long as i can optimize for *some* kind of symmetry. I am biased, too. 2010-11-13T12:18:44 I'm in neither 2010-11-13T12:19:09 ok, between the email and an IM he sent it's obvious that he is want to make the change 2010-11-13T12:19:11 I sympathize with both viewpoints too much 2010-11-13T12:19:27 janzert: is want? 2010-11-13T12:19:42 wanting 2010-11-13T12:19:53 k 2010-11-13T12:20:03 the only thing that's obivous from that email is that he wants to delegate it to someone else 2010-11-13T12:20:19 ...are you guys talking about the same email? 2010-11-13T12:20:25 yes 2010-11-13T12:20:30 if you are I agree with mega1 2010-11-13T12:20:38 I'm in neither, but I don't think it's incredibly fair to change them at the last minute 2010-11-13T12:20:48 in the interest of 'getting it done' if you're ok I'll do the specifics for now and we can change them if he says different or it horribly breaks things 2010-11-13T12:21:12 if I may ask what was the voting for? 2010-11-13T12:21:13 janzert: actually, it might take an hour, there were changes that needed making too 2010-11-13T12:21:14 basically one of the organizers needs to be the badguy and decide one way or the other 2010-11-13T12:21:34 half of the contestants will hate you but there's no way around that at this point... 2010-11-13T12:21:43 mega1: he wanted to add the maps before even talking to people 2010-11-13T12:21:52 amstan added the poll 2010-11-13T12:22:11 and there does seem to be an equal split on opinions 2010-11-13T12:22:12 ok, assume yes/use your suggestion for all the specific question right now 2010-11-13T12:22:13 he was "slightly opposed" but wanted to add the maps, I see 2010-11-13T12:22:44 that was janzert? 2010-11-13T12:22:52 mega1: where is he slightly opposed? that's me 2010-11-13T12:23:39 part of his message to me is "I support this change" 2010-11-13T12:23:40 sorry, he = janzert in what I wrote 2010-11-13T12:23:52 *** nullkuhl has quit IRC (Read error: No route to host) 2010-11-13T12:23:53 your "he" is then jeff 2010-11-13T12:24:10 janzert: yeah i took it as pretty much yes, but i still wasn't making that decision 2010-11-13T12:24:18 i'm possibly the worst person to make it 2010-11-13T12:24:39 as much as i would have liked to just go with my way :P 2010-11-13T12:25:16 don't, it's me making it now. although trying to channel Jeff in the doing so :) 2010-11-13T12:25:35 *** yellowseed has joined #aichallenge 2010-11-13T12:25:37 okay, well i'm working 2010-11-13T12:25:42 log is base 2 in python yes? 2010-11-13T12:26:10 and you agree with making the radius more likely to be further from the center? 2010-11-13T12:26:12 natural log 2010-11-13T12:26:17 holy crap I think I finally have a bot that plays better with more iterations 2010-11-13T12:26:42 use math.log(x, base=2) for log base 2 2010-11-13T12:26:44 just correct the start positions to be far enough from each other and everyone is happy. 2010-11-13T12:26:49 lavalamp_: what's an iteration? 2010-11-13T12:27:22 mega1: I'm playing with monte carlo for selecting best move order 2010-11-13T12:27:34 lavalamp_: cool 2010-11-13T12:27:47 antimatroid: you mean making the planets even density throughout the map? 2010-11-13T12:27:55 yeah that's the idea 2010-11-13T12:28:10 migi32 pointed it out, and mcleopold came up with the log(r(0,1), 2) 2010-11-13T12:28:25 yeah, that seems best 2010-11-13T12:28:48 i'm going to make it so that players home planets are 1 and 2 again as well 2010-11-13T12:28:56 people seemed to want that, and given it can happen i might as well 2010-11-13T12:29:09 so on every radius there is the same density but they can still be very close on the 'logitude' 2010-11-13T12:30:11 I know it's probably easy to do but I wouldn't worry about that too much 2010-11-13T12:33:07 antimatroid, log2(r(0,1))? That gives a random number between -inf and 1. Are you sure this is right? 2010-11-13T12:33:17 -inf and 0 I mean 2010-11-13T12:33:37 Why not just pick x, y like the old generator, and then convert to polar? Or, if you want line symmetry, make it symmetric about the x axis all the time, then rotate the map by a random number of degrees? 2010-11-13T12:33:44 *** dmj111` is now known as dmj111 2010-11-13T12:34:21 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-11-13T12:35:44 yea, like this: do { x = rand(-max_r,max_r); y = rand(-max_r,max_r); r = sqrt(x*x+y*y); } while (r > max_r); theta = atan2(y,x); 2010-11-13T12:36:19 should give you the same result as generating it in polar coordinates, but now you are sure. 2010-11-13T12:37:39 *** McLeopold has joined #aichallenge 2010-11-13T12:40:01 Migi32: and i'm a maths student, whacks myself over the head 2010-11-13T12:40:32 Mcleopold thought sqrt would put too many maps on the outside 2010-11-13T12:40:41 maybe it was log2(random(1,4)) 2010-11-13T12:41:17 that wasn't what he said, hmmm 2010-11-13T12:42:13 howdy 2010-11-13T12:42:29 last year I had to integrate over circles and spheres all the time, it was also with a correction factor, lets see if I can remember this... 2010-11-13T12:43:45 jacobian determinant, something like that? 2010-11-13T12:43:51 been a few years for me : 2010-11-13T12:43:52 yea that :) 2010-11-13T12:43:54 :D 2010-11-13T12:44:32 i'm not particularly keen to switch from generating polar coordinates and coverting 2010-11-13T12:44:38 it's already written and works 2010-11-13T12:44:48 i'm just fixing up the creation of central neutrals 2010-11-13T12:44:59 it doesn't do extra ones for point symmetric maps atm 2010-11-13T12:45:38 wait, i don't think you can ever can you? 2010-11-13T12:47:13 nah yeah that's fine 2010-11-13T12:47:18 you could do a density check before accepting new planets 2010-11-13T12:47:38 nah, it's good to have a mix of densities 2010-11-13T12:47:47 we just don't want it to be disproportionate 2010-11-13T12:48:49 yeah, we don't care about the specific distribution of any specific map too much but do want it to average out over all the maps 2010-11-13T12:51:37 ok so the correction factor for integrating over the area of a circle in polar coordinates (r,theta) is just r 2010-11-13T12:52:07 this means it should be twice as likely to generate the coordinates (1,x) than (0.5,x) 2010-11-13T12:52:34 *** McLeopold has quit IRC (Ping timeout: 265 seconds) 2010-11-13T12:52:58 give me something like r = rand(0,1)^x *maxR 2010-11-13T12:54:10 I'm actually fairly sure that for x=1/2 it gives overall equal density. But yea, maybe maps with x different from 1/2 can be interesting as well 2010-11-13T12:54:41 like a map with no central planet and most planets in a circle shape 2010-11-13T12:54:42 *** hellman has quit IRC (Remote host closed the connection) 2010-11-13T12:54:58 brb 2010-11-13T12:55:56 when you send units from planet to planet, do they all leave+land simultaneously? or is there a buffer like there is in the flash game? 2010-11-13T12:56:00 map1 has pretty much all planets on a cirle 2010-11-13T12:56:41 burny: they start at the start of the round and land at the end of the round 2010-11-13T12:56:51 atm the maps are always going to have a center planet 2010-11-13T12:57:00 although that doesn't mean much in line symmetric maps 2010-11-13T12:57:40 burny: so the planets grow while the ships are still in flight. when you are on tick away from a 0 ships 5 growth planet it will have 5 ships when you arrive 2010-11-13T12:58:14 ok, thanks 2010-11-13T12:59:35 burny: it becomes clearer if you imagine the game to play continuously without the turn steps. during a turn the ships fly a distance of '1' and the planets grow x ships 2010-11-13T13:01:34 ya, but in the flash game, if you send 500 ships in one turn, they get spaced out greatly 2010-11-13T13:02:23 that's all I was wondering about, was if it simulated that sort of delay/que 2010-11-13T13:05:07 no we don't, a fleet is a single entity within the game 2010-11-13T13:05:17 yup: thanks 2010-11-13T13:07:45 if planets A B C are spaced out 2 each(4 between A C) if you fly a->b->c it takes 5turns rather than 4 right? 2010-11-13T13:09:47 burny: why would it take 5 turns rather than 4? 2010-11-13T13:10:17 cause the ships landing at B on second turn, can't be sent until the 3rd turn? 2010-11-13T13:11:37 "landing" doesn't take any time. 2010-11-13T13:12:09 so a distance one 1, means it lands at the end of that turn, and can be sent again the next turn? 2010-11-13T13:12:15 burny: i'm not sure about this, but i think a turn is 1. land, 2. grow, 3. send fleets 2010-11-13T13:12:53 burny: so the planets grow while the ships are still in flight. when you are on tick away from a 0 ships 5 growth planet it will have 5 ships when you arrive 2010-11-13T13:13:04 sounds the opposite from what frontier said 2010-11-13T13:13:31 janzert: http://codepad.org/eMMqZnET 2010-11-13T13:13:45 but I gues.. grow, send, land, move vs grow send move land 2010-11-13T13:13:51 burny: http://ai-contest.com/specification.php#game-state-update might help clarify 2010-11-13T13:14:29 antimatroid: thanks 2010-11-13T13:14:40 it doesn't always generate the number of planets it randomly selected, but i'm pretty sure the range is 15 to 30 inclusive 2010-11-13T13:15:24 it certainly doesn't go above 30, and i'm 99% sure 14 doesn't happen 2010-11-13T13:15:24 thanks jaz.. to me that sugests that a distance of 1, means the ships are available at the new planet, on the consecutive round 2010-11-13T13:15:30 but that isn't really that important 2010-11-13T13:16:07 should i make another forum post explicitly outlining what specifications the maps meet? 2010-11-13T13:16:24 burny: yes that's correct 2010-11-13T13:16:29 anti: if some people are aware of it: then I would say so 2010-11-13T13:16:32 So the new maps are a go? 2010-11-13T13:16:39 otherwise some people have an unfair advantage 2010-11-13T13:16:58 Top 10 players: bocsimacko(4421), bix0r4ever(4210), Raschi(4165), george(4106), Accoun(4082), dmj111(4080), _Astek_(4075), shangas(4063), asavis(4043), GreenTea(4042) 2010-11-13T13:17:42 antimatroid: sure but add a disclaimer that there is still a small possibility of change in the specifics 2010-11-13T13:17:48 yep 2010-11-13T13:17:56 assuming no one is aware of specifications: I'd say it depends on the nature of how you want the AI to be working.. is part of the problem suposed to be determining size/shape/etc. or should symetry be part of the strategy, etc. 2010-11-13T13:18:55 burny: the rules were specified for bots to be general, although it wasn't very clear 2010-11-13T13:19:28 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-11-13T13:19:32 for instance: rather than symetry, the maps could have been played twice(once from each side), and the 'winner' is the average score from each game 2010-11-13T13:20:06 then part of the AI would be not just winning vs losing: but lasting longer if you are losing, and winning faster if you have the upper hand, etc. 2010-11-13T13:20:23 antimatroid: why line 156 isn't inside the if/else conditions directly above and not / 2 for the second case? 2010-11-13T13:21:12 *** narnach has joined #aichallenge 2010-11-13T13:21:12 *** narnach is now known as narnach_mb 2010-11-13T13:22:59 *** narnach_mb has quit IRC (Client Quit) 2010-11-13T13:23:24 nevermind I see 2010-11-13T13:23:26 :) 2010-11-13T13:23:37 saved me finishing that explanation 2010-11-13T13:28:14 anti. if bots are not suposed to take advantage of symmetry, it would be hard to make it be generic 2010-11-13T13:28:45 and most likely, we would want the AI to assume each side starts with equal army size 2010-11-13T13:29:42 *** narnach has joined #aichallenge 2010-11-13T13:29:42 *** narnach is now known as narnach_mb 2010-11-13T13:29:43 I don't like the idea of playing 2 times on asymmetric maps. It's like playing chess 2 times, and each time one of the players plays without his queen. As players/bots approach optimal play, the win percentages will come closer and closer to 100% for the favorite. 2010-11-13T13:30:02 migi: more like playing chess twice, once as white once as black 2010-11-13T13:30:30 nobodies introducing asymetric maps 2010-11-13T13:31:28 and if non-symetric maps were used, I would sugest the winner to be based on like.. if I won in 58 rounds, and you won in 47rounds, that meant you win the match 2010-11-13T13:31:46 *** nullkuhl has joined #aichallenge 2010-11-13T13:31:59 but as anti said: no one is introducing asymetric maps: I was just hypothesizing 2010-11-13T13:32:34 with regards to wether or not information about the maps should be given out 2010-11-13T13:34:10 ah, with a whole bunch of maps that would be cool (if you still play each of them twice to fully eliminate luck) 2010-11-13T13:34:48 it could be cool: but it would completely change the AI mechanics 2010-11-13T13:36:00 I wouldn't say completely. My bot doesn't rely on the fact that the maps are symmetrical. But I agree that probably some bots do. 2010-11-13T13:36:11 like: you would need to determine if you will win or lose, and if on the losing end, choose a path/pattern that makes the enemy take as long as possible to get you.. with symmetry, that's not remotly a viable strategy 2010-11-13T13:37:14 I imagine most bots, if in a losing situation. will fight to the death rather than run away hoping for the buzzer to save them 2010-11-13T13:37:20 ah ok you meant "rounds" as in number of turns. No, sorry, then I completely disagree. I thought you meant we both play 100 games on 50 maps, the winner is the one who won on most maps. 2010-11-13T13:37:30 completely infeasible to do, but cool :) 2010-11-13T13:37:30 or, rather than hoping the opponent will make a mistake 2010-11-13T13:37:30 janzert: http://ai-contest.com/forum/viewtopic.php?f=19&t=1094 2010-11-13T13:37:53 or making a bad move, to entice the opponent into making a worse move, etc. 2010-11-13T13:38:10 * antimatroid quickly changes an isn't to is 2010-11-13T13:38:28 I do that sort of thing in chess all the time(I'm good but not expereinced) 2010-11-13T13:39:04 if I'm losing, I don't just "make it hard to win".. I go suicidal hoping they make a mistake, due to the unorthodox play style 2010-11-13T13:40:02 the specification page is going to need updating 2010-11-13T13:40:28 antimatroid, you removed 0-growths? Awww. 2010-11-13T13:40:37 0-growths were interesting 2010-11-13T13:41:54 but otherwise it's looking good. Why is this rule there: "there can be anywhere from 0 to 4 additional planets starting at an equal distance from both players"? 2010-11-13T13:42:13 Migi32: they were never there 2010-11-13T13:42:17 only centre planet, i kept that 2010-11-13T13:42:41 only the center planet could have 0 growth and I never noticed that? Omg. 2010-11-13T13:42:53 Migi32: because centre planets add more spice to games 2010-11-13T13:43:07 particularly with the line symmetric maps 2010-11-13T13:43:16 yes but why only 4? 2010-11-13T13:43:31 it seemed like a reasonable number 2010-11-13T13:43:36 thats a max of 5 2010-11-13T13:43:48 how much do you plan to increase the planet number? ^^ 2010-11-13T13:43:56 planets numbers are between 15 and 30 2010-11-13T13:44:44 * Frontier quickly edits source code to const int MAX_PLANET_COUNT = 30; 2010-11-13T13:45:17 you really need to specify what "additional" means here. After the normal generation of 15-30 maps? After the center planet has been placed? 2010-11-13T13:46:15 Migi32: what? 2010-11-13T13:46:42 0 to 4 "additional" planets. Additional compared to what? 2010-11-13T13:47:10 fixed? 2010-11-13T13:47:49 ad·di·tion·al, adjective. Added, extra, or supplementary to what is already present or available. Which planets are already present here? 2010-11-13T13:48:11 the center planet? 2010-11-13T13:48:24 yes 2010-11-13T13:48:48 I think you should rewrite it then as "anywhere from 0 to 5 planets starting at an equal distance" 2010-11-13T13:49:15 hmm, but only the center one can have growth between 0 and 5 2010-11-13T13:49:22 and it is the dead centre of the symmetries 2010-11-13T13:49:58 ok, fair point. 2010-11-13T13:50:07 can i remove that edit? 2010-11-13T13:50:29 yea, that is said on rule 1 2010-11-13T13:50:46 yeah, i misunderstood you at first 2010-11-13T13:52:26 no it was my mistake, I read the whole sentence wrong. I thought "There is always one center neutral" was a separate sentence. 2010-11-13T13:55:06 i would laugh if i now sucked at my new maps 2010-11-13T13:55:23 i'm in the middle of rewriting it cleanly and haven't been able to even test on them with a proper bot 2010-11-13T13:55:44 *** Mjothvitnir has joined #aichallenge 2010-11-13T13:56:07 it should be general enough not to matter though 2010-11-13T13:56:51 I bet the new maps will break a ton of bots :) It will break mine for sure, with >23 planets 2010-11-13T13:57:09 why will yours break? 2010-11-13T13:57:26 it never said anywhere 23 is a maximum 2010-11-13T13:57:41 I know, there's a cont int MAX_NUM_PLANETS like Frontier 2010-11-13T13:57:53 I have fixed-size std::bitsets 2010-11-13T13:58:05 not a lot of them, but still 2010-11-13T13:58:52 changing that const int would fix the problems, but I can't be bothered to reupload 2010-11-13T13:59:09 antimatroid, are there any plans to get these maps on TCP too? 2010-11-13T13:59:31 I have a const set to 23, I just have to change it and upload new version. my code shouldn't need to change otherwise. 2010-11-13T13:59:51 Migi32: no one has seen dhartmei 2010-11-13T13:59:58 i tried emailing him, but it bounced 2010-11-13T14:00:17 *** fawek has quit IRC (Ping timeout: 240 seconds) 2010-11-13T14:01:36 @google dhartmei 2010-11-13T14:01:37 Migi32: Advogato: Personal info for dhartmei: ; benzedrine.cx - Daniel Hartmeier: ; c2k2 by dhartmei: ; Google AI challenge 2010: Planet Wars: ; Amazon.com: Profile for dhartmei: ; (1 more message) 2010-11-13T14:02:41 I promise a lot of crashes ahead with the new maps in place. But that is fine. Anyone who doesn't care about his/her bot in the last 2 weeks is out anyway ^^ 2010-11-13T14:03:12 @google Migi32 2010-11-13T14:03:13 Frontier: haxe IRC logs [March 1 - 2010]: ; Google AI Challenge Forums ג€¢ View topic - Bots vs Humans: ; krishnaashish - Google AI Challenge: ; /srv/ufr/supy/logs.www/freenode/2009/12/24/#ubuntu-ir.log: try dhartmei at freebsd dot org or dhartmei at openbsd dot org. www.benzedrine.cx/dharmei.html is still in google's caches 2010-11-13T14:03:46 if making significant changes: might be worth Emailing everyone(some people might be workign on it, but not checking the site, to re-read the rules) 2010-11-13T14:04:33 the rules aren't actually changing 2010-11-13T14:04:40 dang, can't google you, Frontier :P 2010-11-13T14:04:41 a lot of people made assumptions 2010-11-13T14:05:38 actually the rules are changing, even by your standards. They are going from nebulous (only example maps to go on) to being in wet concrete. :) 2010-11-13T14:05:41 Migi32: try NerdIII 2010-11-13T14:08:24 I agree we should send an email with the change in rules. It would be nice if this mail could go along with a notice that C# got fixed. Did it, by the way? 2010-11-13T14:08:37 you cannot have one code cover maps up to 23 planets and maps up to 200 planets at once for example. - well, maybe mega1 can ^^. even if there were no rules, some assumptions had to be made 2010-11-13T14:08:50 Mjothvitnir: nobody ever said conrete information wouldn't be given 2010-11-13T14:08:56 but it certainly hasn't been up until now 2010-11-13T14:09:30 antimatroid: You are doing a good job, don't you want to administrate the next contest? 2010-11-13T14:09:40 vaccuum. 2010-11-13T14:09:46 no, i want to compete in the next contest 2010-11-13T14:09:54 the only reason i'm doing this is because no one else is 2010-11-13T14:10:06 i would have done it a lot earlier, but i didn't have time and no one else has 2010-11-13T14:10:16 hey, you can write your own map generator, who cares if you administrate and compete in the next contest ^^ 2010-11-13T14:10:29 you can't administrate and compete 2010-11-13T14:10:38 the map generator is open source, it doesn't matter who writes it 2010-11-13T14:10:44 speaking of that, can we get someone who is listed by the forums as a "contest admion" to sign off on your post antimatroid 2010-11-13T14:10:45 rules can change over time 2010-11-13T14:11:15 jaznert: they want someone official posting in my thread 2010-11-13T14:11:31 janzert* 2010-11-13T14:11:38 :P 2010-11-13T14:11:47 haha, i quite like jaznert now 2010-11-13T14:13:11 Maybe add a link to new 'sample maps' in the email. Not everyone has python, but there is a need for some local test maps. 2010-11-13T14:15:03 yeah, I hope civ4 will run the map generator, so I don't have to install python 2010-11-13T14:15:12 is it going to be te final map generator? 2010-11-13T14:15:24 or you are going to tune it to keep things interesting? 2010-11-13T14:15:28 mega1: it looks like it 2010-11-13T14:15:41 i expect some refinements will be allowed if necessary 2010-11-13T14:15:51 no mix of old and new maps then? 2010-11-13T14:16:04 it does point symmetric maps too 2010-11-13T14:16:07 old maps are technically still allowed 2010-11-13T14:16:21 and the rules don't say that the generator can't change 2010-11-13T14:16:21 they're a subset of what the new generator will make 2010-11-13T14:16:30 please 2010-11-13T14:16:33 Migi32: mega1 is well aware 2010-11-13T14:16:50 I like the fact that you underlined "current" :) 2010-11-13T14:16:58 Top 10 players: bocsimacko(4441), bix0r4ever(4166), Raschi(4138), george(4092), Accoun(4083), dmj111(4072), ClusterTeam(4051), shangas(4036), GreenTea(4019), flowerbot(4017) 2010-11-13T14:17:29 I'd like to know what to test on. 2010-11-13T14:17:47 atm, that generator is now the official one 2010-11-13T14:17:51 http://codepad.org/eMMqZnET 2010-11-13T14:17:53 If you're turning knobs here and there that's no good. 2010-11-13T14:18:00 I say we put a deadline on when the map generator / rules can't change anymore 2010-11-13T14:18:14 mega1: that will only change if there are issues with it 2010-11-13T14:18:17 yes, finally 2010-11-13T14:18:59 i'm unsure about the function to select a radius though 2010-11-13T14:19:13 let's ask #math 2010-11-13T14:20:35 antimatroid: so am I understanding it correctly if planetsToGenerate is odd and symmetryType = 1 the planets out will be one less? 2010-11-13T14:21:01 *** yasith has quit IRC (Ping timeout: 265 seconds) 2010-11-13T14:22:31 does that mean 14 can happen? 2010-11-13T14:22:52 yes if my logic is correct above 2010-11-13T14:23:00 My mistake Civ4 has python dll, it is Open Office that has python exe... 2010-11-13T14:23:02 just change the minPlanets variable to 16 2010-11-13T14:23:04 in which case I'll just make it bump up to 16 instead 2010-11-13T14:23:15 yeah sorry, i did say that might be the case ;) 2010-11-13T14:23:20 just wanted someone to double check my logic :) 2010-11-13T14:23:37 so are you saying that it won't be symmetric 2010-11-13T14:24:27 no 2010-11-13T14:24:43 it's confirmed, it was sqrt(random(0,1)) :) 2010-11-13T14:24:49 the map generator picks a number between 15 and 30 for planets to generate, it may not actually generate that 2010-11-13T14:24:49 my intuition was correct :) 2010-11-13T14:24:52 here's where I'm at right now 2010-11-13T14:24:55 http://codepad.org/eytoZCJy 2010-11-13T14:24:57 how is centerplanet + odd number of planets symmetric? 2010-11-13T14:25:23 or am I just misunderstanding what you were saying? 2010-11-13T14:26:03 janzert: what am i looking for? 2010-11-13T14:26:34 how can it generate an even number of planets by these rules? 2010-11-13T14:26:48 ah, ok in line symmetry, I see 2010-11-13T14:27:20 antimatroid: nothing in particular, I've basically just made minor formatting changes is all 2010-11-13T14:27:56 oh sure, i'd never done anything in python before, rip it to pieces if you want ;) 2010-11-13T14:30:02 janzert: is it safe to safe turn limit wont change from 200? 2010-11-13T14:30:51 I would have said it was safe to say the current map_generator will be used a week ago 2010-11-13T14:31:06 how do you know how many turns are remaining? 2010-11-13T14:31:12 i had always expected maps to change :\ 2010-11-13T14:31:13 can make a global variable and count it yourself? 2010-11-13T14:31:28 burny: yes you can 2010-11-13T14:31:34 yes burny that's the only way 2010-11-13T14:31:40 k 2010-11-13T14:31:58 well, doesn't have to be global, but just, a variable tracking it throughout the game 2010-11-13T14:32:17 k, and how do we initialize? put it inside main? 2010-11-13T14:32:25 yep 2010-11-13T14:33:25 k, didn't know you can have more than 1 main in a single app 2010-11-13T14:33:44 there isn't? 2010-11-13T14:34:19 the main doesn't ever return, so how does the other bot get to go? 2010-11-13T14:34:23 oh, is it networked? 2010-11-13T14:36:25 my main function looks like this these days 2010-11-13T14:36:26 http://codepad.org/1a2mZxUU 2010-11-13T14:37:15 I just would have thought the final game, would have an alternate main, and it just calls 'DoTurn' 2010-11-13T14:37:42 no, your bot is playing the whole game 2010-11-13T14:37:49 not a bunch of semi repeated ones 2010-11-13T14:38:04 k 2010-11-13T14:40:03 anyway, it's 6 40am, i should go to bed 2010-11-13T14:40:42 can you post your latest generator first? 2010-11-13T14:40:57 it's linked in the forum 2010-11-13T14:41:34 ok great go sleep ^^ 2010-11-13T14:42:01 when i say i'm going to bed, it'll take a good 30 minutes yet :) 2010-11-13T14:42:14 i'm shocking at actually getting up and doing the deed 2010-11-13T14:43:07 i can get back to working on my bot when i wake up now that the map thing is pretty much sorted 2010-11-13T14:43:32 which code to use? anti or janz? 2010-11-13T14:44:47 they should do exactly the same 2010-11-13T14:44:53 janzerts is closest to what will be used 2010-11-13T14:45:40 yeah, still making a few minor changes to assure number of planets range is always met 2010-11-13T14:47:59 what other problems were there? 2010-11-13T14:48:15 actually my logic before was wrong 2010-11-13T14:48:27 oh yeah 2010-11-13T14:48:30 symmetry 1 can only produce odd planets 2010-11-13T14:48:58 :) 2010-11-13T14:49:30 well it could do even, but not with the centre planet 2010-11-13T14:49:38 but you have to assure that going into the final generation there are an even number of planets left to generate 2010-11-13T14:49:50 ah yeah 2010-11-13T14:50:12 wait, wouldn't the /2 handle that? 2010-11-13T14:51:04 so with radial we have center + even (sym) planets, with bilat we have center + 0-X on line planets + even (sym) planets, yes? 2010-11-13T14:51:36 aahhh what? :P 2010-11-13T14:51:42 i did the generator using polar coordinates 2010-11-13T14:51:57 then converted those for saving maps 2010-11-13T14:52:58 each player has a rotation angle theta1 and theta2, for a new point, pick r and theta, then radial symmetry -> theta1+theta and theta2+theta, bilat -> theta1+theta and theta2-theta 2010-11-13T14:53:05 wrapping around 360 degrees 2010-11-13T14:54:25 symmtryType == 1 is radial right? 2010-11-13T14:54:48 yes 2010-11-13T14:54:57 the difference I was asking about was that bilat can have an even number of total planets because there are an odd number of on "the mirror" line planets to off set the center planet 2010-11-13T14:55:18 (offset in terms of count) 2010-11-13T14:55:33 maps can have up to 5 "central" neutrals 2010-11-13T14:55:54 although it has to be an odd number for radially symmetric maps 2010-11-13T14:56:59 antimatroid: not if you left out the center planet 2010-11-13T14:57:29 the centre planet is always added 2010-11-13T14:57:36 that way players planets are 1 and 2 2010-11-13T14:57:47 people wanted that for some reason, so i gave it to them by always adding centre planet 2010-11-13T14:58:31 *** wh1teside has quit IRC (Ping timeout: 265 seconds) 2010-11-13T14:58:46 so with both types of maps there is the possiblility of more than 1 (1-5) (1 since there is always a "center" planet) non-symmetrically-paired planets. (planets on the "mirror line" as it were) 2010-11-13T14:59:04 http://codepad.org/8JSzFQRA 2010-11-13T14:59:24 unless someone sees a problem with that, that is what I'll use 2010-11-13T15:01:27 btw, why is the center not at 0,0? I figure that would protect the maps from rounding errors that could make maps asymmetrical. 2010-11-13T15:01:53 Frontier: people wanted positive coordinates 2010-11-13T15:02:13 my understanding is the visualizer also requires it 2010-11-13T15:02:33 janzert: you didn't copy the last part? but looks good to me 2010-11-13T15:02:53 ? 2010-11-13T15:03:00 wait, page didn't load, nm 2010-11-13T15:04:46 i generated a few, they still look symmetric 2010-11-13T15:04:54 so i assume nothing got messed up 2010-11-13T15:04:59 ok, I stumbled my way through the python (don't know python) and I see the answer to my questions, so you can ignore them. wait you already did. 2010-11-13T15:05:22 :P sorry, i'm trying to read the code for janzert 2010-11-13T15:06:17 I think the only real change I should have made is assuring the number of planets falls within the correct range 2010-11-13T15:06:57 yeah, i thought i'd make a couple anyway 2010-11-13T15:07:19 looks fine to me modulo some style things 2010-11-13T15:07:32 when i first put in multiple centre planets i didn't do it in pairs for radially symmetric maps, and 100 maps got posted with some not even symmetric 2010-11-13T15:07:39 but that was quickly fixed 2010-11-13T15:07:54 Zannick: i'd never used python, style was the last thing one thought of 2010-11-13T15:08:27 no worries 2010-11-13T15:08:54 anyone have a preference on naming scheme for the new maps? 2010-11-13T15:09:12 newmap.txt? 2010-11-13T15:09:46 or map_newstyle.txt 2010-11-13T15:10:07 if you have to have "new" then "newmap" is better 2010-11-13T15:10:54 yeah, but if you make a next iteration, would you call it newnewmap? 2010-11-13T15:10:57 yeah, newmap is fine 2010-11-13T15:11:53 Zannick: yeah, I'm just not creative enough to think of something else :) 2010-11-13T15:11:58 personally, I will call them "map#.txt" and put them in "newMap" directory. 2010-11-13T15:12:03 *** Accoun has quit IRC () 2010-11-13T15:12:13 betamap 2010-11-13T15:12:19 omegamap 2010-11-13T15:12:27 antimap ;) 2010-11-13T15:12:48 i hate that i don't have a "shift" key for the greek alphabet 2010-11-13T15:12:48 adding them into a different directory would require changing other code 2010-11-13T15:12:56 and standar maths notation at my fingertips 2010-11-13T15:13:24 or we could go to a versioning style naming map_1_.txt 2010-11-13T15:13:40 that's a decent idea 2010-11-13T15:13:43 do that 2010-11-13T15:13:44 or some variant 2010-11-13T15:13:47 i like that style less, but *shrug* 2010-11-13T15:14:02 File "official_gen.py", line 166, in assert planetsToGenerate % 2 == 0, "Odd number of planets left to add" 2010-11-13T15:14:02 map2..txt 2010-11-13T15:14:04 Zannick: it's reusable 2010-11-13T15:14:12 anyone else get that? 2010-11-13T15:14:16 are these going to replace the old maps or be added to them? 2010-11-13T15:14:36 odd -> even 2010-11-13T15:14:49 jaznert: ^^ that's a problem 2010-11-13T15:15:00 the comment is wrong 2010-11-13T15:15:02 gah, lavalamp it means my checks to enforce the correct number of planets failed 2010-11-13T15:15:05 isn't "Odd ..." the error message? 2010-11-13T15:15:08 no it's right 2010-11-13T15:15:23 the assert throws if there are and odd left 2010-11-13T15:15:23 so yeah, if the assert that it's even fails, the error says it's odd 2010-11-13T15:15:31 ah, see what happens when you don't know the lang 2010-11-13T15:15:33 ok 2010-11-13T15:15:47 well then there's a bug, I get it a lot ^^ 2010-11-13T15:16:20 bah, I can't get it at all :/ 2010-11-13T15:16:26 really? 2010-11-13T15:16:28 hm 2010-11-13T15:16:32 there we go 2010-11-13T15:16:38 we both thought that there was something screwy about that line (I just figured the words were wrong since it seemed to be the correct check) 2010-11-13T15:17:00 Top 10 players: bocsimacko(4413), bix0r4ever(4226), Raschi(4161), george(4117), Accoun(4078), _Astek_(4058), dmj111(4050), shangas(4038), GreenTea(4033), flowerbot(4025) 2010-11-13T15:17:04 i'm confused, is it sorted? 2010-11-13T15:17:07 I have a script to call a map generator N times, I asked for 200 and I get it a lot 2010-11-13T15:17:24 Mjothvitnir: when it throws you get this AssertionError: Odd number of planets left to add 2010-11-13T15:17:35 i haven't got it once 2010-11-13T15:17:36 *** Azrathud has joined #aichallenge 2010-11-13T15:17:45 janzert: line 149 2010-11-13T15:17:53 x = (x+1)%5 2010-11-13T15:17:54 ok, not a lot, like 9 times out of 200 2010-11-13T15:18:18 if x is 4 (even) then x becomes 0 (even) and then (p-x) % 2 is still 1 2010-11-13T15:18:34 Zannick: ahh 2010-11-13T15:19:49 * janzert debates whether to change it to % 4 and have the logic look a bit weird or change if to while 2010-11-13T15:20:14 I guess either way will look weird 2010-11-13T15:20:21 a comment is required :P 2010-11-13T15:20:33 well, you could change centerneutrals to randint(0,3) and the mod to 4 2010-11-13T15:20:46 or you could just drop the mod 2010-11-13T15:21:06 or randint(0,5) and mod 6 2010-11-13T15:21:32 * Zannick shrug 2010-11-13T15:21:36 actually I should probably make that 4 a constant set at the beginning of the file 2010-11-13T15:21:43 and use while 2010-11-13T15:22:30 bah, i really am climbing into bed now, enjoy working that out :) 2010-11-13T15:23:02 jaznert: it might be worth posting the code once you finalise it 2010-11-13T15:23:16 yep :) 2010-11-13T15:23:45 noCenterNeutrals -= (noCenterNeutrals == 0) ? -1 : 1; 2010-11-13T15:23:57 or how ever you say that in python 2010-11-13T15:24:19 *** Accoun has joined #aichallenge 2010-11-13T15:24:24 -= -1 if noCenterNeutrals == 0 else 1 2010-11-13T15:24:27 well that would be easier to read the other way 2010-11-13T15:24:45 noCenterNeutrals += (noCenterNeutrals == 0) ? 1 : -1; 2010-11-13T15:25:01 or += (noCenterNeutrals == 0) and 1 or -1 2010-11-13T15:25:47 that gives a bias towards have 1 NoCenterNeutrals instead of an uniform distribution 2010-11-13T15:26:02 same as using a while loop 2010-11-13T15:26:03 good point 2010-11-13T15:26:44 since you need to switch from even to odd, but there are 3 even numbers mod 5 and 2 odd numbers 2010-11-13T15:27:10 that's why i recommend using (0,3) and mod 4 or (0,5) and mod 6 if you want a more even distribution through that 2010-11-13T15:29:06 awesome, I found a dumb bug and now my launches all sync up :) 2010-11-13T15:30:29 hmm, I think I'll switch to an appropriate randrange call actually 2010-11-13T15:31:20 janzert, gpm from the official server's been awesome lately, if you had anything to do with that, thanks :) 2010-11-13T15:32:44 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-11-13T15:35:32 I just helped Jeff get his instances set up to stay up :) 2010-11-13T15:37:17 well, it rocks now, good job 2010-11-13T15:37:43 http://codepad.org/suGqd4Hk 2010-11-13T15:38:35 I did bump up the maximum number of centrally located planets to 5 2010-11-13T15:39:16 only because it seemed rather nice to have it 1/3 of the total minimum 2010-11-13T15:40:05 You should probably say so in antimatroid's thread 2010-11-13T15:40:22 I can't imagine depending on that but *shrug* 2010-11-13T15:40:25 yeah, I will once we reach agreement here 2010-11-13T15:40:48 I'll post a link to the final generator there as well 2010-11-13T15:40:50 what does the ",2)" do in the rand? 2010-11-13T15:41:00 step size 2010-11-13T15:41:25 so if minCentral is even give even, if odd give odd 2010-11-13T15:41:34 ok that one made 200 maps without error 2010-11-13T15:41:37 yep basically 2010-11-13T15:41:42 cool 2010-11-13T15:42:09 minCentral = planetsToGenerate % 2 2010-11-13T15:42:25 Zannick: ahh, yeah :) 2010-11-13T15:43:34 *** iFire` has joined #aichallenge 2010-11-13T15:46:17 *** iFire has quit IRC (Ping timeout: 250 seconds) 2010-11-13T15:47:08 would this be a c 2010-11-13T15:47:17 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-11-13T15:47:46 would this be a correct definition of the "central neutral planets" 2010-11-13T15:47:57 planets located equidistant from both players 2010-11-13T15:48:14 yeah 2010-11-13T15:48:35 since they will always be on a specific line centrally located between the starting planets 2010-11-13T15:52:06 *** narnach has joined #aichallenge 2010-11-13T15:52:06 *** narnach is now known as narnach_mb 2010-11-13T15:53:39 *** narnach_mb has quit IRC (Read error: Connection reset by peer) 2010-11-13T15:53:42 *** narnach has joined #aichallenge 2010-11-13T15:53:42 *** narnach is now known as narnach_mb 2010-11-13T15:56:41 http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/map_generator.py 2010-11-13T15:56:58 I'll post that in the forum shortly 2010-11-13T16:05:04 so how many central neutrals do you expect the code to produce on purpose (we'll ignore the ones it can create randomly as normal planets) 2010-11-13T16:05:31 1 in the very center and 0-5 along the central line. 2010-11-13T16:07:39 what does this line produce: (0,2,4,6;1,3,5) or??: noCenterNeutrals = random.randrange(minCentral, maxCentral+1, 2) 2010-11-13T16:08:22 random.randrange(0,5+1,2) 2010-11-13T16:08:48 http://docs.python.org/library/random.html#random.randrange 2010-11-13T16:09:15 the stop will never be generated 2010-11-13T16:09:42 ok 2010-11-13T16:10:21 so 0,2,4 or 1,3,5 2010-11-13T16:10:49 *** iFire` is now known as iFire 2010-11-13T16:14:40 *** delt0r___ has joined #aichallenge 2010-11-13T16:15:56 *** delt0r has quit IRC (Ping timeout: 245 seconds) 2010-11-13T16:16:53 isn't centre spelled center? 2010-11-13T16:16:59 Top 10 players: bocsimacko(4392), bix0r4ever(4297), Raschi(4106), Accoun(4085), george(4077), _Astek_(4062), GreenTea(4042), flowerbot(4025), hutabhu(4011), dmj111(3992) 2010-11-13T16:17:15 depends one if you spell english or not 2010-11-13T16:17:16 depends on if you still feel british :) 2010-11-13T16:17:31 It is so hard for a foreigner to the english language to get this stuff into their trunk 2010-11-13T16:17:52 its like a boot kick sometimes 2010-11-13T16:18:02 Yes, the British and the Americans are seperated by a common language 2010-11-13T16:19:00 When I learnt English we were presented with the new way to spell "Hello!": "Hallo!" I have never seen it written like this since. 2010-11-13T16:20:41 Lol, when my sister was asked "What is the simple present?" she was half asleep and ansered in German something like "A simple gift." The teacher though she was mobbing him. He was kinda paranoid anyway. 2010-11-13T16:21:11 hmm, the only "hallo" I know is german 2010-11-13T16:23:30 We had a british student there and he was *allowed* to use hello instead of hallo. the rest of the class had to learn it the other way. anyway the internet was the best teacher so far ^^ 2010-11-13T16:24:06 *** irchs is now known as janstaunton 2010-11-13T16:24:11 * janstaunton and lavalamp_ be battlin; 2010-11-13T16:24:16 batllin' even 2010-11-13T16:24:45 god damn it 2010-11-13T16:24:47 battlin' 2010-11-13T16:28:31 hehe 2010-11-13T16:28:40 *** hellman has joined #aichallenge 2010-11-13T16:29:04 *** eburnette has quit IRC (Ping timeout: 265 seconds) 2010-11-13T16:29:11 looks like janstaunton is winning most of those :/ 2010-11-13T16:29:31 BRING IT LAMPY 2010-11-13T16:29:37 oh, I guess my official submission can still beat you once in a while 2010-11-13T16:29:41 haha 2010-11-13T16:29:43 hehe 2010-11-13T16:30:10 @seen j3camero 2010-11-13T16:30:10 amstan: j3camero was last seen in #aichallenge 1 day, 19 hours, 7 minutes, and 9 seconds ago: The servers should crap out soon. But we're gonna set up the server-kicking script as a cron job soon. 2010-11-13T16:30:34 yeah we had a long series against each other there... 2010-11-13T16:31:08 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-11-13T16:33:06 amstan: are you happy with the latest generator? 2010-11-13T16:33:07 just wait until I get my engine and my move generator on speaking terms... 2010-11-13T16:33:14 http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/map_generator.py 2010-11-13T16:33:14 hehe 2010-11-13T16:33:20 I am also in the middle of a reqrite 2010-11-13T16:33:24 *** p4p4 has joined #aichallenge 2010-11-13T16:33:24 janzert: in what way? i don't like the code, i do like the map types 2010-11-13T16:33:25 our rewrites will clash 2010-11-13T16:33:30 May the best bot win 2010-11-13T16:33:36 bring it ^^ 2010-11-13T16:33:46 :D 2010-11-13T16:33:59 as in fine with it being used the way it is 2010-11-13T16:34:21 I decided not to clean up a lot of the style 2010-11-13T16:34:32 janzert: sure... 2010-11-13T16:34:42 janzert: hold on, is the number of planets variable? 2010-11-13T16:34:47 yes 2010-11-13T16:34:50 good 2010-11-13T16:34:55 15-30 2010-11-13T16:35:25 I think I explain all the settable constants at the top 2010-11-13T16:35:49 although probably not well :) 2010-11-13T16:35:49 # maximum number of planets located equidistant from both players 2010-11-13T16:36:06 is technically not true, since it can generate them "accidentally" 2010-11-13T16:36:29 nit-pick, nit-pick. :) 2010-11-13T16:36:43 *** Mathnerd314 has joined #aichallenge 2010-11-13T16:36:54 maybe it should be "maximum specifically generated to be equidistant..."? 2010-11-13T16:37:19 wow, you're better at that than me, I couldn't think of a good way to say it. 2010-11-13T16:37:37 #topic remove 4 2010-11-13T16:37:42 @topic remove 4 2010-11-13T16:37:43 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask." 2010-11-13T16:38:45 @topic add New MapGen is being used: http://tinyurl.com/aimapgen 2010-11-13T16:38:45 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask. || New MapGen is being used: http://tinyurl.com/aimapgen" 2010-11-13T16:38:55 now I'm gonna wonder what topic 4 used to be all day 2010-11-13T16:39:15 lavalamp_: devs wanted 2010-11-13T16:39:24 oh right 2010-11-13T16:40:08 Are we gonna see these new map types on the tcp server? 2010-11-13T16:40:19 idk, tcp is not really official 2010-11-13T16:40:24 i haven't seen the admin in a while 2010-11-13T16:40:57 *** virdo has joined #aichallenge 2010-11-13T16:45:26 amstan: I'm thinking of closing the thread on the proposed map change now that the decision has been made, what do you think? 2010-11-13T16:45:36 janzert: sure 2010-11-13T16:51:11 are you going to post a couple of hundred sample maps? 2010-11-13T16:51:50 I have a couple hundred, you want them? 2010-11-13T16:52:52 lavalamp_: yes. Are they generated by the latest version? 2010-11-13T16:53:15 Yeah, I just made them 2010-11-13T16:53:30 http://ai-contest.com/forum/viewtopic.php?f=19&t=1094&p=7233#p7233 2010-11-13T16:53:35 I posted them 2010-11-13T16:54:27 thank you 2010-11-13T16:54:39 what was the max distance previously? 2010-11-13T16:54:44 no problem, I can make more if you want 2010-11-13T16:54:52 I don't know but it was larger 2010-11-13T16:55:19 *** Sylph has quit IRC (Read error: Connection reset by peer) 2010-11-13T16:56:23 I believe it was a square with sides of length 24 2010-11-13T16:56:31 now it is a circle with radius 24 2010-11-13T16:56:40 err, diameter 24 2010-11-13T16:57:01 so you just chopped off the corners, so to speak. 2010-11-13T16:57:29 yes, if you is the collective you :) 2010-11-13T16:57:30 and tuning just starts over 2010-11-13T16:57:45 I'm just looking at the old and new generators 2010-11-13T16:58:22 true, antimatroid wrote it 2010-11-13T16:58:32 i believe 2010-11-13T16:58:33 it's a little confusing because the old generator also called it's variable a radius 2010-11-13T16:58:43 but from a quick glance that doesn't seem true 2010-11-13T16:58:51 the square bounds the circle 2010-11-13T16:59:07 yes 2010-11-13T16:59:36 and it would have planets outside of the max radius defined 2010-11-13T16:59:58 outside the square or outside the circle 2010-11-13T17:00:24 outside the circle, that's what a radius defines right? 2010-11-13T17:01:02 yes. radius is a silly name for the length of a side of a square that bounds a circle 2010-11-13T17:01:02 strictly I think of the radius of a square as rather nonsensical 2010-11-13T17:02:09 ok, taking a short break and then I'll start generating and getting the new maps loaded 2010-11-13T17:02:37 I'm planning on only generating 30 initially to see how this will work for the final contest as well 2010-11-13T17:02:54 the current maps will still be there but will be disabled 2010-11-13T17:02:56 so all server fights will be on 30 maps? 2010-11-13T17:02:58 I guess it will soon be time for me to update my submission, then 2010-11-13T17:03:42 'cause these maps crash my current one all over the place 2010-11-13T17:03:43 *** hellman has quit IRC (Remote host closed the connection) 2010-11-13T17:03:44 Mjothvitnir: yes, at least until there is a pairing that has to repeat maps then I'll increase it 2010-11-13T17:04:24 *** Sylph has joined #aichallenge 2010-11-13T17:04:37 * janzert afk 2010-11-13T17:10:32 I ran 1000 maps and they seemed to come out well distributed in terms of # of planets (~3 times as many odds as evens. which implies that radial and bilateral are ~equal in number) 2010-11-13T17:11:06 new maps in svn here http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/#planet_wars/maps 2010-11-13T17:11:37 well they should be anyway :P 2010-11-13T17:11:40 that is my only question (about "30") is will they be well distrubuted? 2010-11-13T17:12:37 as soon googlecode refreshes its cache I suppose 2010-11-13T17:13:38 until then I guess you can see them here http://code.google.com/p/ai-contest/source/detail?r=529 2010-11-13T17:13:56 or do a svn update on any checkout you have 2010-11-13T17:17:00 Top 10 players: bocsimacko(4397), bix0r4ever(4351), GreenTea(4160), Raschi(4115), george(4074), Accoun(4057), flowerbot(4026), _Astek_(4001), hutabhu(3998), dmj111(3998) 2010-11-13T17:17:06 http://dl.dropbox.com/u/9140666/newmaps.zip 2010-11-13T17:18:37 what about the tcp server? 2010-11-13T17:18:46 Mjothvitnir: yeah, there is a tension between more maps to make sure there is a good distribution and less to get people to play the same maps 2010-11-13T17:19:09 mega1: that is up to dhartmei 2010-11-13T17:19:25 and I don't think he has been around for the last month or more 2010-11-13T17:19:38 bummer 2010-11-13T17:19:46 there goes the main testing venue 2010-11-13T17:19:51 @seen dhartmei 2010-11-13T17:19:51 janzert: dhartmei was last seen in #aichallenge 6 weeks, 3 days, 5 hours, 35 minutes, and 13 seconds ago: project selection problem? 2010-11-13T17:22:31 one could fake it by generating 15 maps radial & 15 bilateral (harder to do density since the code only has min distance between planets) 2010-11-13T17:23:47 same idea with planet numbers 2010-11-13T17:24:28 just adding a couple of loops to the map generator to set what are no consts and poof! 2010-11-13T17:32:12 *** McLeopold has joined #aichallenge 2010-11-13T17:32:32 is dhartmei's server code availible? 2010-11-13T17:32:41 yes 2010-11-13T17:33:14 is linked to on front page 2010-11-13T17:33:19 http://72.44.46.68/ 2010-11-13T17:33:29 hm, if no one can reach him and i can get my bot working maybe I will try to set up an alt server :/ 2010-11-13T17:35:51 lavalamp_: I think you need good bandwidth 2010-11-13T17:36:00 and storage too. 2010-11-13T17:36:25 he mentioned his bandwidth wasn't too bad 2010-11-13T17:36:41 if I understand correctly he is running the current server from Ec2 2010-11-13T17:36:51 And, I don't think he keeps games over 24 hours, so storage shouldn't be too bad. 2010-11-13T17:36:54 not sure what instance type though 2010-11-13T17:38:11 Mjothvitnir: did you happen to take a look at the distribution of the new maps actually generated for the server? 2010-11-13T17:40:18 to few can't infer, so I need to write a little script 2010-11-13T17:40:59 *** Bobng has quit IRC (Quit: Leaving) 2010-11-13T17:41:13 :) need to learn python :) 2010-11-13T17:41:30 ahh I see I thought you already had such a script 2010-11-13T17:44:37 well, sort of :) wc|sed|grep|sort|uniq|sort|gawk 2010-11-13T17:44:52 :) 2010-11-13T17:45:49 to actually determine symmetry takes a little more omph than the last gawk bit, but gawk will still do the job 2010-11-13T17:46:21 lavalamp_: that's only 199 maps 2010-11-13T17:46:44 * mega1 wonders what language is at fault 2010-11-13T17:46:52 I bet on shell script 2010-11-13T17:48:36 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-11-13T17:49:32 *** p4p4 has joined #aichallenge 2010-11-13T17:51:00 mega1: got it in 1 :) 2010-11-13T17:51:47 bash hash the most insane numeric support 2010-11-13T17:51:51 more likely user error, actually, since it takes command line params XP 2010-11-13T17:51:51 *has 2010-11-13T17:52:07 oh, then I was too kind :-) 2010-11-13T17:52:31 *** janstaunton has quit IRC (Quit: janstaunton) 2010-11-13T17:52:32 haha 2010-11-13T17:52:36 *checks* 2010-11-13T17:52:42 Yeah, user error, doh 2010-11-13T17:54:41 Error323: update on new maps: 17/0/182 2010-11-13T17:56:19 who is bix0r4ever? 2010-11-13T17:59:33 These new maps seem to be really drawish 2010-11-13T18:00:34 awesome, my bot gives me hope: http://72.44.46.68/canvas?game_id=1007720 2010-11-13T18:00:54 flag, my bots don't draw on them locally... 2010-11-13T18:00:57 lavalamp_: can you help test my tcp server 2010-11-13T18:01:06 sure, where is it? 2010-11-13T18:01:12 give me 2 minutes 2010-11-13T18:01:19 k 2010-11-13T18:01:44 lavalamp_: Everytime I spawn close to my enemy (lets say within distance 12 or so) I get a draw 2010-11-13T18:02:04 flag: against whom? 2010-11-13T18:02:12 Often you can only take the 1 planet that's garenteed so you don't die to rageBot and that's it 2010-11-13T18:02:19 Different versions of myself 2010-11-13T18:03:22 Not every game is like that, but there's still a fair number of them 2010-11-13T18:03:57 flag: I don't get significantly higher number of draws so far 2010-11-13T18:03:58 that wouldn't be a problem for me since currently my bot doesn't care if it loses to ragey 2010-11-13T18:05:14 mega1: I suppose I'm not changing the versions enough then 2010-11-13T18:05:18 *** RainCT has quit IRC (Remote host closed the connection) 2010-11-13T18:05:36 I compared two version on the old and the new maps 2010-11-13T18:05:51 lavalamp_: zeroviz.us:9999 2010-11-13T18:06:00 18/7/45 on the old 2010-11-13T18:06:03 I'm hoping my firewall is correct 2010-11-13T18:06:18 17/6/32 on the new 2010-11-13T18:06:25 I know, different number of games too 2010-11-13T18:06:47 but the relative strengths are different 2010-11-13T18:07:06 and anyone else that wants to try the new maps on a tcp server 2010-11-13T18:07:32 McLeopold: yes 2010-11-13T18:07:51 zeroviz.us:9999 2010-11-13T18:08:06 I don't have the web site running yet. I look into that. 2010-11-13T18:08:29 A very old version of my bot is waiting to be clobbered by the first customer. 2010-11-13T18:08:34 I'm getting network unreachable 2010-11-13T18:08:42 crap 2010-11-13T18:09:27 how about http://zeroviz.us do you see the paintball logo? 2010-11-13T18:09:28 Command is "./tcp zerovis.us 9999 flag -p test ../flagBot" 2010-11-13T18:09:29 me too 2010-11-13T18:09:46 No logo 2010-11-13T18:09:49 yes, I see it 2010-11-13T18:10:00 hmm, okay, I'll check iptables again 2010-11-13T18:10:08 Now I do, sorry 2010-11-13T18:11:13 ** server can't find zerovis.us: NXDOMAIN 2010-11-13T18:11:29 ah, typo 2010-11-13T18:11:35 I used your IP and I connected 2010-11-13T18:11:42 oh 2010-11-13T18:11:45 Playing you now 2010-11-13T18:11:56 who's shampoo? 2010-11-13T18:11:57 why are all the newmap planet zero at 14,14 ?? 2010-11-13T18:12:01 That's I 2010-11-13T18:12:15 Oh man, I *hate* your bot! 2010-11-13T18:12:22 lol 2010-11-13T18:12:25 :) 2010-11-13T18:12:32 :) 2010-11-13T18:12:39 yeah, me too 2010-11-13T18:13:02 *** Bobng has joined #aichallenge 2010-11-13T18:13:04 Everyone hates me apparently 2010-11-13T18:13:07 D: 2010-11-13T18:13:17 Is there some way I can view the games? 2010-11-13T18:13:19 Mjothvitnir: because the max radius is 14 and the new generator just adds the radius basically 2010-11-13T18:13:24 Or do you still have to set that up? 2010-11-13T18:13:27 I'm working on that... 2010-11-13T18:13:30 Ok 2010-11-13T18:13:47 ok 2010-11-13T18:13:52 http://zeroviz.us:8080/ 2010-11-13T18:13:57 Anyone want to trade binaries? I'm getting tired of practicing against myself 2010-11-13T18:14:45 McLeopold: Not loading 2010-11-13T18:14:48 Mjothvitnir: are these http://dl.dropbox.com/u/9140666/newmaps.zip in there? 2010-11-13T18:15:23 McLeopold: the page works for me 2010-11-13T18:15:33 yep ratings and maplist works here 2010-11-13T18:15:57 visualizer doesnt seem to work tho 2010-11-13T18:16:22 *** p4p4 has quit IRC (Read error: Connection timed out) 2010-11-13T18:17:01 Top 10 players: bocsimacko(4401), bix0r4ever(4317), GreenTea(4167), Raschi(4139), Accoun(4078), george(4073), flowerbot(4037), dmj111(4034), hutabhu(4005), _Astek_(4000) 2010-11-13T18:17:19 the only problem is the visualizer doesn't load. :( 2010-11-13T18:17:33 McLeopold: do I need to use the ip? 2010-11-13T18:17:37 i guess 2010-11-13T18:18:23 I'll put a bot named "server with new maps on zerovis.us" on the benzendrine leaderboard 2010-11-13T18:18:35 zeroviz, damn it 2010-11-13T18:18:55 :8080 2010-11-13T18:18:57 mega1, hahahahaha 2010-11-13T18:18:59 um, well, this is my comcast connection 2010-11-13T18:19:06 let's keep this on the down low 2010-11-13T18:19:10 Do you have a static IP? 2010-11-13T18:19:16 kinda 2010-11-13T18:19:23 Dynamic DNS? 2010-11-13T18:19:23 uh-oh 2010-11-13T18:19:37 hmmmmm.... did I break my bot or is this server doing state differently? 2010-11-13T18:19:57 It's the same code on benzendrine 2010-11-13T18:20:01 lavalamp_: Did you take 30 planets into account? 2010-11-13T18:20:23 Yes 2010-11-13T18:20:39 I'm using the new maps locally just fine 2010-11-13T18:21:06 and have you played on line symmetrical maps yet? 2010-11-13T18:21:30 yes 2010-11-13T18:21:57 mega1: this is the only leaderboard when I will be ahead of you. :) 2010-11-13T18:21:58 I may have an unrelated bug that is getting triggered for some reason 2010-11-13T18:22:26 all maps in the new set (30maps) are line symmetric 2010-11-13T18:22:49 if only we could see these games... 2010-11-13T18:23:26 i think still working on it 2010-11-13T18:24:27 bbl 2010-11-13T18:24:40 if anyone is really good at reading c code, then maybe you could help get the visualizer going 2010-11-13T18:25:41 Error0: fopen: (null): Too many levels of symbolic links 2010-11-13T18:26:25 lavalamp_: I watching a successful game from you. 2010-11-13T18:26:29 oops 2010-11-13T18:27:09 McLeopold: Isn't the visualizer a canvas one? 2010-11-13T18:27:12 Frontier: that error is usually triggered by recursive links 2010-11-13T18:27:28 lol, i talk to my self. it is late here ^^ 2010-11-13T18:27:30 hehe 2010-11-13T18:27:36 being recursive too 2010-11-13T18:27:37 flag: yes, it is a problem with the getgame.c code. When it reads the pgn file, it crashes. 2010-11-13T18:28:01 Frontier: and a recursive conversation at that. 2010-11-13T18:28:58 So, my thought is I'm building the pgn file wrong. 2010-11-13T18:29:15 but what about the symbolic link issue? 2010-11-13T18:29:25 Is that my issue? 2010-11-13T18:29:29 YES 2010-11-13T18:29:29 Where did you see that? 2010-11-13T18:29:34 view source ;) 2010-11-13T18:29:40 of html ;) 2010-11-13T18:29:44 doh! 2010-11-13T18:30:47 hang on to your keyboards, I'm about to break the server for a few seconds... 2010-11-13T18:32:46 itermission - Linux users, use alsaequal if you need an equalizer that sits directly on top of alsa. It really helps with crappy laptop speakers. 2010-11-13T18:33:50 it WORKS! 2010-11-13T18:33:59 thx McLeopold 2010-11-13T18:34:30 okay, problem 2 is the http server times out alot when viewing the page for the second time 2010-11-13T18:34:56 I should probably switch from thttp to apache 2010-11-13T18:34:58 So you guys are seeing something when you go to http://zeroviz.us:8080/, right? 2010-11-13T18:35:17 2 tiny little links 2010-11-13T18:36:40 i have no timeouts so far 2010-11-13T18:37:26 oh forget what i just said 2010-11-13T18:37:28 really? when I click on a name in ratings list, it doesn't load 2010-11-13T18:37:31 k 2010-11-13T18:38:14 only lavalamp.oblina.5 is problematic, the other links work 2010-11-13T18:38:34 *** Rubicon-|-Cross has joined #aichallenge 2010-11-13T18:38:49 aside from the fact that we all don't like the decision made ;) what does everyone think about throwing the switch and going to the new maps on the main server? I'm ready to do it now. 2010-11-13T18:38:50 ...and a few others 2010-11-13T18:39:27 omg my old bot will get killed and my new bot isnt ready ^^ 2010-11-13T18:39:58 was there a reason why the map size was shrunk? 2010-11-13T18:40:02 just do it, ill fix the errors tomorrow, and delegate the decision to.. McLeopold 2010-11-13T18:40:07 janzert: the sooner the better, probably 2010-11-13T18:40:13 * janzert idly wonders how much wailing and gnashing of teeth Jeff is willing to allow before reversing the decision 2010-11-13T18:40:28 I'm expecting great volumes of it 2010-11-13T18:40:43 Yeah, I can already tell my bot isn't terribly good with these new maps 2010-11-13T18:40:44 do it 2010-11-13T18:40:47 was there a reason why the map size was shrunk? 2010-11-13T18:40:48 reversal is on the cards too? 2010-11-13T18:41:06 I'm going to bed. 2010-11-13T18:41:09 mega1: who knows, I didn't think change was in the cards 2010-11-13T18:41:56 everyone is scared of change 2010-11-13T18:41:57 ok my old bot is too primitive, it will accept the new maps 2010-11-13T18:43:28 pairing bots with slightly lower ranked ones after resubmission would help against gaming the leaderboard 2010-11-13T18:43:37 the visualizer needs to be adapted, this game is really ugly: http://zeroviz.us:8080/canvas?game_id=1289690094|McLeopold.AntBot.py|lavalamp.pwwax.12 2010-11-13T18:43:51 mega1: yeah, I've thought about doing that 2010-11-13T18:44:32 janzert: is there a link I can use for the latest visualizer? 2010-11-13T18:44:56 "http://brains.s3.amazonaws.com/posts/2010/vis1.2/js/visualizer.js?v=1 2010-11-13T18:45:27 @repo 2010-11-13T18:45:28 janzert: repo = http://code.google.com/p/ai-contest/ 2010-11-13T18:45:50 http://code.google.com/p/ai-contest/source/browse/#svn/trunk/planet_wars/www/visualizer/js 2010-11-13T18:45:57 is the one in use on the main server 2010-11-13T18:46:46 *** Azrathud has quit IRC (Ping timeout: 272 seconds) 2010-11-13T18:51:40 hmm... broke the visualizer 2010-11-13T18:52:53 *** Frontier has quit IRC (Ping timeout: 260 seconds) 2010-11-13T18:56:08 okay, that's worse 2010-11-13T18:56:14 code must be different from dhartmei's latest, right -- the number of turns a game takes isn't output, for example 2010-11-13T18:56:39 or am i misremembering that 2010-11-13T18:57:24 ok, 5-10 minutes and the main server will be playing on the new maps 2010-11-13T18:58:13 * janzert thinks is the last 5 minutes of peace he will likely have :} 2010-11-13T18:59:14 janzert: which maps become the official example maps? lavalamp's 200? 2010-11-13T18:59:47 anything generated with the new generator is good 2010-11-13T18:59:57 I'll post the actual maps in use on the server in a bit 2010-11-13T19:00:38 or you can grab them from here now http://dl.dropbox.com/u/9140666/newmaps.zip 2010-11-13T19:01:06 thanks! 2010-11-13T19:01:37 Who has the Zoo bots on McLeopold's server? 2010-11-13T19:03:55 16 radial, 14 bilateral, planet# 15-30 (mean:23.5 mode:25) 2010-11-13T19:05:12 games starting at game_id http://ai-contest.com/visualizer.php?game_id=7219268 are on the new maps 2010-11-13T19:05:28 forgot median=25 2010-11-13T19:05:59 so we ended up on the high side 2010-11-13T19:06:19 okay, working again 2010-11-13T19:06:35 janzert: I couldn't get the newer one to work, although I didn't try very hard. 2010-11-13T19:07:10 well the first few games didn't result in crashes at least 2010-11-13T19:07:20 Time to write a asymmetrical map generator. :) 2010-11-13T19:07:23 for purposes of visualizer, that game you've pasted here has 3 planets on top of each other (impossible to see troops hiding behind the planet) 2010-11-13T19:07:24 slightly, since 23 was the old number 2010-11-13T19:07:53 and average now should be 22.5 right? 2010-11-13T19:08:06 *** choas has quit IRC (Quit: leaving) 2010-11-13T19:08:09 23.5 for those 30 maps 2010-11-13T19:08:31 yes, but overall distribution of the new generator 2010-11-13T19:08:37 sorry, yes 2010-11-13T19:09:32 what a game! http://zeroviz.us:8080/canvas?game_id=1289691003|McLeopold.AntBot.py|lavalamp.pwwax.12 2010-11-13T19:10:02 where's the graph? 2010-11-13T19:10:17 um, 221 turns? 2010-11-13T19:10:38 flag: me 2010-11-13T19:10:53 flag: expandtesting and ragebot 2010-11-13T19:11:15 yeah, no graph and 221 turns?? 2010-11-13T19:11:22 well no crashes in the first 20 games, that is a good sign 2010-11-13T19:11:38 man 2010-11-13T19:11:40 these maps own 2010-11-13T19:11:49 turn number is the last thing I depend on :/ 2010-11-13T19:11:50 and the overall errors per minute didn't seem to shoot way up 2010-11-13T19:12:07 indeed... I dont have any real attacking or defending and yet I manage to keep positive elo :| 2010-11-13T19:13:12 *** Utkarsh has quit IRC (Ping timeout: 264 seconds) 2010-11-13T19:13:24 http://zeroviz.us:8080/canvas?game_id=1289692711|McLeopold.AntBot.py|ZooExpand1 2010-11-13T19:13:34 I'm sure I raised the turn limit to 250 for my server. 2010-11-13T19:13:39 I just can'r remember where 2010-11-13T19:13:56 ok time to upload a new submission, I guess... 2010-11-13T19:17:01 Top 10 players: bocsimacko(4421), bix0r4ever(4340), GreenTea(4200), Accoun(4087), DarthMasta(4083), dmj111(4061), george(4050), flowerbot(4040), _Astek_(4021), hutabhu(4015) 2010-11-13T19:17:27 I don't think the above was refering to this, but turn limit on the main server is still 200 2010-11-13T19:17:37 http://ai-contest.com/visualizer.php?game_id=7219744 mega1's first game on the new maps 2010-11-13T19:17:38 yes, my server must be 300 2010-11-13T19:18:00 against a 116 2010-11-13T19:18:07 yeah :) 2010-11-13T19:19:22 aw, my first match on a new map and i lost :P 2010-11-13T19:20:31 OH BOY, Accoun crashed on his 2010-11-13T19:20:32 *** Utkarsh has joined #aichallenge 2010-11-13T19:20:38 janzert: can I get the maps generated for official? 2010-11-13T19:20:44 oh, well, my bot seems to have forgotten how to defend 2010-11-13T19:21:07 McLeopold: http://dl.dropbox.com/u/9140666/newmaps.zip 2010-11-13T19:21:40 ew, the map is shifted a little too much to the right 2010-11-13T19:22:11 yeah, the new maps are a little off center in the visualizer 2010-11-13T19:22:50 I noticed that too 2010-11-13T19:22:55 thx, there running on my server now 2010-11-13T19:23:17 I think the old visualizer always put the min(x) and min(y) at 0 2010-11-13T19:23:23 err generator that is 2010-11-13T19:23:31 yeah, it did 2010-11-13T19:24:34 *** amriedle has joined #aichallenge 2010-11-13T19:25:33 my bot does for i has quit IRC (Quit: Leaving) 2010-11-13T19:31:02 hmm, I don't see in the visualizer why the center is offset 2010-11-13T19:32:18 Come on, server, play me a game! >:| 2010-11-13T19:36:36 ok, switched back to an older version of my bot that kicks newer ones' butt on the new maps 2010-11-13T19:36:42 how I love local tournaments ;-) 2010-11-13T19:40:50 *** Azrathud has joined #aichallenge 2010-11-13T19:41:02 TODO: 1. Fix close starting positions 2010-11-13T19:41:16 2. Fix dealing with equidistant planets 2010-11-13T19:42:56 um, there were already equidistant planets (planet zero + any on the line perpendicular to the line between planets 1&2) 2010-11-13T19:43:10 goingthrough zero 2010-11-13T19:43:14 Yeah, well I dealt with one just fine ^^ 2010-11-13T19:43:33 but that's in relation to my new code, not the new maps... 2010-11-13T19:44:14 new code? you starting from scratch 2010-11-13T19:44:33 actually, now would be a good time if one was planning to do that 2010-11-13T19:44:54 haha no just making a new move generator 2010-11-13T19:45:01 *** mk453 has quit IRC (Quit: Page closed) 2010-11-13T19:49:14 *** Bobng has joined #aichallenge 2010-11-13T19:51:18 *** jmpespxoreax has quit IRC (Changing host) 2010-11-13T19:51:18 *** jmpespxoreax has joined #aichallenge 2010-11-13T19:53:25 gah, 2010-11-13T19:53:59 the maps are offset because the new generator does translatest the planets by radius + 2 instead of radius 2010-11-13T19:54:27 s/does translatest/translates 2010-11-13T19:57:22 the visualizer should just look for max x and y 2010-11-13T19:57:45 *** _xvinyl <_xvinyl!~xvinyl@ANantes-256-1-154-37.w90-25.abo.wanadoo.fr> has joined #aichallenge 2010-11-13T19:59:00 lavalamp_: having fun yet? 2010-11-13T20:00:28 McLeopold: I'll probably add that but the generator should probably be fixed as well 2010-11-13T20:00:45 although I'm debating that 2010-11-13T20:00:56 visualizer is easier 2010-11-13T20:01:15 well, to deploy yes 2010-11-13T20:01:48 McLeopold: always! except when my bot sucks... as it does atm 2010-11-13T20:02:02 *** narnach has joined #aichallenge 2010-11-13T20:02:03 ooh, close one right now 2010-11-13T20:02:43 on your server? 2010-11-13T20:03:19 yeah 2010-11-13T20:03:35 ah I should try my real bot there again 2010-11-13T20:03:41 you're missing an attacking opportunity right now 2010-11-13T20:03:48 pwwax will miss a lot 2010-11-13T20:03:51 hehe 2010-11-13T20:04:02 oh, wait 2010-11-13T20:04:05 I'm playing a 2010-11-13T20:05:10 *** narnach_mb has quit IRC (Ping timeout: 240 seconds) 2010-11-13T20:05:53 <_xvinyl> hi everyone 2010-11-13T20:06:09 <_xvinyl> i've been away from this contest for quite some time now 2010-11-13T20:06:23 <_xvinyl> so what's the general state of things ? 2010-11-13T20:07:25 bocsimacko is kicking everyone's collective ass 2010-11-13T20:07:52 <_xvinyl> saw that :) 2010-11-13T20:08:12 <_xvinyl> doesn't he play on dhartme's server ? 2010-11-13T20:08:32 <_xvinyl> also, the forums looks horribly outdated with respect to TCP servers 2010-11-13T20:08:35 is there a sort of high end limit to the elo score? 2010-11-13T20:09:20 burny, depends on how many players there are and the strength distribution 2010-11-13T20:09:32 like, if a bot wins 100% of games.. would it be caped at like 200ish above the second highest? 2010-11-13T20:10:12 burny, I don't think there's a cap built in, but the math probably works out such that you can only be so much higher than everyone else 2010-11-13T20:10:47 the online visualizer doesn't seem to work for me, need a certain browser? 2010-11-13T20:13:17 shouldn't although IE is stuck using the java visualizer which isn't nearly as nice 2010-11-13T20:13:44 <_xvinyl> burny: very slow in Firefox but ok in Chrome 2010-11-13T20:13:56 <_xvinyl> burny: not tried Opera for a long time though 2010-11-13T20:14:01 I'm using IE 2010-11-13T20:15:33 hmm, it's working here with IE8 2010-11-13T20:15:46 you do have to accept a java security warning though 2010-11-13T20:16:12 and it's no where near as nice as the js visualizer other browsers get 2010-11-13T20:16:26 so I would recommend using a different one 2010-11-13T20:16:35 *** narnach has quit IRC (Quit: Leaving.) 2010-11-13T20:16:59 ah 2010-11-13T20:17:02 Top 10 players: bocsimacko(4440), bix0r4ever(4335), Raschi(4161), GreenTea(4127), Accoun(4078), dmj111(4076), george(4063), DarthMasta(4062), shangas(4020), _Astek_(4015) 2010-11-13T20:17:02 think I have ie4 2010-11-13T20:17:14 janzert: is the official server playing line symmetry maps now? 2010-11-13T20:17:19 yes 2010-11-13T20:17:25 cool 2010-11-13T20:17:53 since game_id 7219268 it is using the new maps 2010-11-13T20:18:14 ah ok :D 2010-11-13T20:18:54 Error323 the 30 new maps are: 16 radial, 14 bilateral; with planet# 15-30 (mean:23.5 median=25 mode:25) 2010-11-13T20:19:18 <_xvinyl> mmh so are there trends in the strategies people use now ? 2010-11-13T20:19:48 <_xvinyl> like Minimax used to appear after some time during the Tron contest ? 2010-11-13T20:20:14 think the trend is they are getting better 2010-11-13T20:21:00 <_xvinyl> yea I guess so... 2010-11-13T20:21:24 <_xvinyl> just wondering if I can do something decent if I restart now 2010-11-13T20:21:48 I'm only starting yesterday.. lots of time left 2010-11-13T20:22:28 antimatroid: testmap9 is very drawish 2010-11-13T20:22:40 *** Migi32 has quit IRC (Remote host closed the connection) 2010-11-13T20:22:52 janzert: is it using both types? 2010-11-13T20:24:12 Mjothvitnir: ah ok, how many in total? 2010-11-13T20:24:37 Error323: play on my tcp server zeroviz.us:9999 2010-11-13T20:24:42 30 maps 2010-11-13T20:24:42 http://zeroviz.us:8080 2010-11-13T20:25:00 McLeopold: yes 2010-11-13T20:25:13 McLeopold: not @ home atm 2010-11-13T20:26:12 janzert: so the old maps are running as well? 2010-11-13T20:26:24 <_xvinyl> McLeopold: is there a place where all this kind of information is stored ? 2010-11-13T20:26:37 no 2010-11-13T20:27:00 <_xvinyl> McLeopold: I mean, who runs TCP servers, what are the resources available to all players (game engines, visualizers, etc.) 2010-11-13T20:27:09 oh 2010-11-13T20:27:10 McLeopold: no 2010-11-13T20:27:42 the main web site is good, and the forums 2010-11-13T20:27:47 <_xvinyl> I tried to look at the forums but it seems nobody uses thel anymore 2010-11-13T20:27:50 <_xvinyl> ok 2010-11-13T20:28:01 what do you need to know? 2010-11-13T20:28:12 hm, something weird is going on... testmap16 works fine when I play locally but gives a state violation when I get it on McLeopold's server 2010-11-13T20:28:38 unintialized variable? 2010-11-13T20:29:12 <_xvinyl> McLeopold: well, nothing in particular but I'd hate to discover someone already implemented a game engine in Haskell after havong wrote it 2010-11-13T20:29:14 no no... as in my bot resolved something differently than the server 2010-11-13T20:29:30 talk to jmcarthur 2010-11-13T20:29:34 ? 2010-11-13T20:29:35 <_xvinyl> McLeopold: for example 2010-11-13T20:29:43 that was fast 2010-11-13T20:29:49 <_xvinyl> :P 2010-11-13T20:29:57 is somebody working on a haskell game engine? 2010-11-13T20:30:08 you are 2010-11-13T20:30:23 err, well, umm, not publicly :) 2010-11-13T20:30:32 <_xvinyl> heh 2010-11-13T20:31:05 do you mean for planet wars i guess? 2010-11-13T20:31:13 that is not what i meant 2010-11-13T20:31:38 there were a couple people thinking about doing it, but i don't think they ever did 2010-11-13T20:32:06 <_xvinyl> jmcarthur: ok thanks for the update 2010-11-13T20:32:17 is there any rule about what the growth rate will be on your starting planet? 2010-11-13T20:32:22 mightybyte, Mathnerd314, and somebody else i think all independently considered it, iirc, but i could be mixing some people up 2010-11-13T20:33:06 lavalamp_: something is really wrong with my server 2010-11-13T20:33:16 _xvinyl: for my bot i did have basically an entire rewrite of the engine in haskell, but it wouldn't generalize to other bots very well, let alone to a standalone engine 2010-11-13T20:33:19 look at my games against zoorage 2010-11-13T20:34:27 <_xvinyl> jmcarthur: do you still compete ? 2010-11-13T20:34:42 <_xvinyl> jmcarthur: can't see your bot on the official server 2010-11-13T20:34:48 McLeopold: yes, it assigned a planet I took to player 3 2010-11-13T20:34:49 i've dropped out 2010-11-13T20:34:53 on time 2010-11-13T20:34:55 *no time 2010-11-13T20:35:10 that is what is making me barf 2010-11-13T20:35:46 well, I'll leave it running, but I can't fix it now 2010-11-13T20:35:48 gtg 2010-11-13T20:35:59 *** McLeopold has quit IRC (Quit: Leaving) 2010-11-13T21:02:09 *** denis has joined #aichallenge 2010-11-13T21:17:01 Top 10 players: bocsimacko(4440), bix0r4ever(4291), GreenTea(4167), dmj111(4083), Accoun(4076), george(4072), shangas(4032), DarthMasta(4023), flowerbot(3999), asavis(3996) 2010-11-13T21:20:00 highlight of my current bot's carerr: http://72.44.46.68/canvas?game_id=1010260 2010-11-13T21:22:13 *** amriedle has quit IRC (Quit: Lost terminal) 2010-11-13T21:22:21 *** krokkrok has joined #aichallenge 2010-11-13T21:23:22 *** pc has joined #aichallenge 2010-11-13T21:24:33 *** _xvinyl <_xvinyl!~xvinyl@ANantes-256-1-154-37.w90-25.abo.wanadoo.fr> has quit IRC (Ping timeout: 260 seconds) 2010-11-13T21:26:53 piong 2010-11-13T21:32:31 *** pc has quit IRC (Ping timeout: 250 seconds) 2010-11-13T21:33:40 *** pc has joined #aichallenge 2010-11-13T21:34:11 *** wh1teside has joined #aichallenge 2010-11-13T21:35:46 i'm curious what people would want next time? 2010-11-13T21:36:08 i really like the very general rules that were handed out about maps for tron and this, i do wish more people realised it was like that though 2010-11-13T21:38:27 where is the limit if no guildline was given? 2010-11-13T21:39:35 I think alot of people assumed that the example maps were mostly like the final maps and you seem to be saying that they should have assume completely nothing 2010-11-13T21:40:18 Mjothvitnir: my understanding is built from competing in tron and seeing what the original organised interpretations are of their own rules on numerous ocassions 2010-11-13T21:40:37 i have no problem with very general rules, provided they are clear, which i will admit they aren't 2010-11-13T21:41:03 upper bounds should be given on sizes of maps, but I do like having very general rules on what the maps will look like at the end 2010-11-13T21:41:05 There is no problem with the rules; what happened is what I expected to happen. 2010-11-13T21:41:07 ah, see then you have everyone else at a disadvantage. You knew that there was no organization in the Organization ;) 2010-11-13T21:41:13 so people are writing a bot to play the game, not do well on specific types of maps 2010-11-13T21:41:22 krokkrok: me too 2010-11-13T21:41:28 *** Bobng has quit IRC (Quit: Leaving) 2010-11-13T21:41:48 Mjothvitnir: they were never told these are the maps being used, the specification page does say to make your own 2010-11-13T21:42:31 *** xvinyl has joined #aichallenge 2010-11-13T21:42:34 but it doesn't say make them anyway you please. it says here is our example generator make your own 2010-11-13T21:42:35 my concern is how the final tournament is going to be run. Just using the regular ELO algorithm is not very good. 2010-11-13T21:43:02 TCP cerver map set updated? 2010-11-13T21:43:06 TCP server map set updated? 2010-11-13T21:43:12 *** pc has quit IRC (Ping timeout: 264 seconds) 2010-11-13T21:43:14 I agree, I have been concerned with that since the begining, I expected that the maps would change some (but before now) 2010-11-13T21:44:10 krokkrok: the regular elo algorithm works very well compared to any other alternative i've known of 2010-11-13T21:44:24 well, how about, getting the right answer? 2010-11-13T21:44:32 there is no right answer 2010-11-13T21:44:36 yes, there is 2010-11-13T21:44:38 some maps would result in a cyclic ordering 2010-11-13T21:44:46 for some bots 2010-11-13T21:44:49 nope 2010-11-13T21:44:52 because 2010-11-13T21:45:05 you are thinking about ordering wrong, the way ELO does it 2010-11-13T21:45:07 which is wrong 2010-11-13T21:45:34 ELO is assuming a>b>c is transitive, and so, it does all sorts of goofy stuff like running bots near each other to fine to the rank 2010-11-13T21:46:20 if all combinations of games were run, the winner would be the one with the highest win/loss ratio 2010-11-13T21:46:34 you have n bots b1, ..., bn that you want to define a strict preference relation < on, it's not possible to do that without doing it for a given set of maps and a huge round robbin, which is an absurdly inefficient use of resources 2010-11-13T21:47:03 you said"there is no right answer" 2010-11-13T21:47:06 but anything else is pure luck 2010-11-13T21:47:12 but there is, and we should tryto be as close to it as possible 2010-11-13T21:47:16 round robbin isn't right, it's too inefficient 2010-11-13T21:47:38 the current system is meant to be a best approximation to something like that as far as i know 2010-11-13T21:47:40 what is right is, all bots play all bots, and count the wins 2010-11-13T21:47:43 i'm not an official 2010-11-13T21:47:55 krokkrok: that was done for tron with random pairings 2010-11-13T21:47:56 well, the ELO doesn't come close to getting the 'right answer' 2010-11-13T21:47:58 it worked very poorly 2010-11-13T21:48:13 it did better than this for ensuring bots got games against bots of a similar rank 2010-11-13T21:48:20 when running all bots against all bots, there is no random..... 2010-11-13T21:48:35 you can't run bots against all other bots, it's too inefficient 2010-11-13T21:48:50 there is no point running the top 10 bots agains the botoom 3000 on every map 2010-11-13T21:48:53 it's a huge waste of time 2010-11-13T21:48:57 no its not 2010-11-13T21:49:11 it is, you end up with top bots playing each other a couple of times 2010-11-13T21:49:14 because, you dont actually know the top ten bots, without having the answer a priori 2010-11-13T21:49:15 that isn't sufficient to rank them 2010-11-13T21:49:50 http://csclub.uwaterloo.ca/contest/rankings.php 2010-11-13T21:50:00 right now, rage bot can beat many of the top bot sometimes. however, you dont see that in the ELO because they never face it, because they outrank it too much 2010-11-13T21:50:06 please take a look at some of those users pages at the top, too many of the games were against shit bots 2010-11-13T21:50:21 for testing, ELO is fine 2010-11-13T21:50:25 for judgeing, its not 2010-11-13T21:50:50 The best bot is winning isn't it? 2010-11-13T21:50:53 i was one of the people almost demanding a round robin last time, but look at the number of games top bots played against other top ranked bots last time, it sucked 2010-11-13T21:50:58 yes it is, but you're obviously unwilling to be convinced 2010-11-13T21:51:37 each bot plays every bot other bot the same number of times. this is the correct answer. 2010-11-13T21:51:42 *** denis has quit IRC (Quit: Leaving) 2010-11-13T21:52:05 The top two bots might play each other only a few times then and the winner will just end up being a coinflip 2010-11-13T21:52:06 krokkrok: i've be surprised if you even played everyone once 2010-11-13T21:52:19 there's over 4000 bots 2010-11-13T21:52:29 Your method is fair, but only once you've played a million games 2010-11-13T21:52:35 right, but, the goal is to get the best answer 2010-11-13T21:52:38 bots that don't work should be chucked 2010-11-13T21:52:38 which is impossible 2010-11-13T21:52:59 yes the best answer with the resources available is certainly not obtained by round robin 2010-11-13T21:53:04 i would be in favour of a top x round robbin on c maps at the end 2010-11-13T21:53:14 well, here is what is possible: use ELO to get an guess (ONLY A GUESS) at the top 100 or 200, then run all bots in the final round 2010-11-13T21:53:50 round robin is about as inefficient as you can get no matter what 2010-11-13T21:54:02 well using a reasonable system at least 2010-11-13T21:54:06 no, its not reasonable 2010-11-13T21:54:12 its math is unsound 2010-11-13T21:54:21 so is the math for round robin :P 2010-11-13T21:54:28 there is no perfect solution, like i originally said 2010-11-13T21:54:29 efficienct but wrong, this is bad practice 2010-11-13T21:54:38 yes there is a perfect solution 2010-11-13T21:54:51 we seem to have to competeing principles. "correct answer" and "good enough answer given resources" 2010-11-13T21:54:51 look at how rankings are done for sporting events, the rankings for this are a million times better 2010-11-13T21:54:53 all bots vs all bots, win/loss ratio 2010-11-13T21:55:04 ranking is not an easy thing to do right, particularly with 4000 competitors 2010-11-13T21:55:04 efficiency as defined by number of games required to come up with the correct answer 2010-11-13T21:55:06 sporting events are corrupt 2010-11-13T21:55:13 who CARES its a computer 2010-11-13T21:55:22 run ten thousand games 2010-11-13T21:55:25 nothing gives a correct answer in a time fram that is "correct" 2010-11-13T21:55:36 why..not 2010-11-13T21:55:48 krokkrok: it would take too long 2010-11-13T21:56:12 how long would it take to run the top 100 bots against each other top 100 bot 2010-11-13T21:56:19 if you player 4000 bots 1000 times, that's 1000*2000 games or something 2010-11-13T21:56:48 depends on how long players are taking, an upper bound would be 50*200*1second 2010-11-13T21:56:50 i already agreed, you can prune off the bottom by using ELO, and assume ELO 2010-11-13T21:56:52 for one round 2010-11-13T21:57:07 krokkrok: while i don't care about that, i think lower users that put time in wont 2010-11-13T21:57:28 i'm still hoping to make top 10 2010-11-13T21:57:48 its easy to make top ten. just resubmit everytime you lose, you will get top ten eventually 2010-11-13T21:57:56 no, at the end of the contest 2010-11-13T21:58:00 ... 2010-11-13T21:58:11 i don't even have a bot uploaded atm, i don't care for current rankings, its on the original 100 maps 2010-11-13T21:58:41 we should fix either the maps or the visualizer to better center the map 2010-11-13T21:59:27 the visualiser needs to enforece it renders square 2010-11-13T21:59:40 the maps can't be "fixed", they're symmetric :P 2010-11-13T21:59:45 really, messing around with the maps is a huge waste of time. getting the ranking system to approximate the correct answer matters more. otherwise, this is just a lottery 2010-11-13T21:59:46 the website one? 2010-11-13T21:59:59 Zannick: I did try refreshing 2010-11-13T22:00:16 antimatroid: have you watched any games? the maps are offcenter, badly 2010-11-13T22:00:21 krokkrok: the ranking system is pretty damn good at the moment 2010-11-13T22:00:25 oh 2010-11-13T22:00:27 huh 2010-11-13T22:00:28 Zannick: i have 2010-11-13T22:00:28 prove it 2010-11-13T22:00:29 janzert: great 2010-11-13T22:00:33 you cant prove it 2010-11-13T22:00:39 no, you never can 2010-11-13T22:00:40 i *can* prove it doesnt work 2010-11-13T22:00:43 but i was around for tron 2010-11-13T22:00:51 and compared to that, it's a million times better 2010-11-13T22:01:18 and i was sustaining first rank for that quite often towards the end, so if anything i'd have been biased towards the old method 2010-11-13T22:01:32 what was tron? 2010-11-13T22:01:38 http://csclub.uwaterloo.ca/contest/rankings.php 2010-11-13T22:01:40 last ai contest 2010-11-13T22:01:45 in february 2010-11-13T22:01:52 sorry, I mean what ranking system 2010-11-13T22:01:55 this make no sense: you are saying calculating the correct answer is worse that calculating some random ELO 2010-11-13T22:01:57 i got disqualified in the final tournamnet for timing out :( 2010-11-13T22:02:14 krokkrok: we've bene over this, you can't calculate the correct answer 2010-11-13T22:02:26 there is a correct answer. do you agree? 2010-11-13T22:02:37 no 2010-11-13T22:02:44 not if you include time constraints 2010-11-13T22:02:56 huh? everyone has the same contraints 2010-11-13T22:02:56 and it's being unrealistic not to 2010-11-13T22:03:13 time constraints for obtaining your ranking over bots 2010-11-13T22:03:16 I think antim means get the answer in X days 2010-11-13T22:03:40 i would like a longer contest, at least a week, but i don't think that method would be feasible even then 2010-11-13T22:04:41 *** _xvinyl <_xvinyl!~xvinyl@ANantes-256-1-49-75.w86-199.abo.wanadoo.fr> has joined #aichallenge 2010-11-13T22:04:46 the server ran 124000 games in the last 24 hours 2010-11-13T22:05:01 how many games does it take to run the top 100 vs each other... 2010-11-13T22:05:53 *** xvinyl has quit IRC (Ping timeout: 245 seconds) 2010-11-13T22:05:57 is it 50x50 pairings? 2010-11-13T22:06:05 no wait, gah 2010-11-13T22:06:07 100bots x 99 bots * 30 maps * 100 turns * 1 sec/turn = 343 days 2010-11-13T22:06:12 i suck at counting 2010-11-13T22:06:28 Mjothvitnir: *30 maps? lol 2010-11-13T22:06:30 no way 2010-11-13T22:06:32 1 map max 2010-11-13T22:06:42 divide by 2, you're double counting the pairs 2010-11-13T22:06:44 amstan: the point is that it's not feasible 2010-11-13T22:06:49 so that's 10 days then 2010-11-13T22:07:00 the limit is 200 turns too 2010-11-13T22:07:00 its not 10 days 2010-11-13T22:07:17 sorry I menat 34 days 2010-11-13T22:07:24 what we did last time: we played top 25 for a while alone, without anyone else 2010-11-13T22:07:27 not all bots take 1 second per turn, as we can see, it can run 124000 games a day 2010-11-13T22:07:32 I picked 100 since not all go to 200 2010-11-13T22:07:38 since the pairings are random, it's all good 2010-11-13T22:07:49 the pairings are no longer random 2010-11-13T22:07:56 not completely 2010-11-13T22:08:02 but completely random sucked 2010-11-13T22:08:08 well.. w/e 2010-11-13T22:08:15 any randomness in the ranking makes it a lottery 2010-11-13T22:08:17 it'll be luck based on what games you get 2010-11-13T22:08:21 you can do it with zero randomness 2010-11-13T22:08:25 I was counting KrokKrok method everybot plays every bot on every map 2010-11-13T22:08:31 krokkrok: you can't if you wanna play everyone vs everyone 2010-11-13T22:08:56 its top 100 ELO vs top 100 ELO only, all bots play all bots 2010-11-13T22:09:07 that's why there's elo.. so we can get away with not playing a lot of the bots, while still having the same rankings 2010-11-13T22:09:11 but yes as someone said I am counting double 2010-11-13T22:09:15 s/bot/pairs 2010-11-13T22:09:38 amstan: next time, i do like having very general rules, but i think it's important people are told more explicitly that the maps will change, and if it's said that they will, one can't then leave them :) 2010-11-13T22:09:44 ELO is an approximation, the method being used now creats local minima when it tries to fit bots into a rank 2010-11-13T22:10:10 so approximate the top 100, then 100 vs 100, get the right answer, ezpz 2010-11-13T22:10:33 i do admit, i would be in favour of something like that, maybe not top 100, 50 or 25 maybe? 2010-11-13T22:10:42 actually antimatroid we were told that they WOULD change, just not HOW they would change 2010-11-13T22:10:52 krokkrok: but then, the top 100 rankings are made up only of games played with other top 100 bots 2010-11-13T22:11:01 so if they suck against ragebot as you said, that can never happen 2010-11-13T22:11:14 indeed, it is still top perfect, but..its much better 2010-11-13T22:11:51 different symmetries were used for the tron map generator when that was finally made 2010-11-13T22:11:55 and that was way later in the contest 2010-11-13T22:12:33 well good luck guys, im off to..life <<.<< >>.>> 2010-11-13T22:12:39 *** krokkrok has quit IRC (Quit: Page closed) 2010-11-13T22:12:42 not everyone played in the tron contest, so that is not a valid "item of knowledge" 2010-11-13T22:12:51 *** pc has joined #aichallenge 2010-11-13T22:13:05 *** pc is now known as xvinyl 2010-11-13T22:13:13 i'm giving you an example of what happened last time 2010-11-13T22:13:18 and saying this is no different 2010-11-13T22:13:25 not that one should have expected it because of that 2010-11-13T22:13:44 although one did 2010-11-13T22:13:46 ok (but that is what it sounds like when you say it) 2010-11-13T22:13:48 *** _xvinyl <_xvinyl!~xvinyl@ANantes-256-1-49-75.w86-199.abo.wanadoo.fr> has quit IRC (Ping timeout: 264 seconds) 2010-11-13T22:14:33 besides all this kicking and fussing, does anyone honestly expect to do any different on these new maps than new maps generated with the old generator? 2010-11-13T22:14:37 i'd be quite surprised 2010-11-13T22:15:02 my bot did a nice center planet snipe to win one of those games 2010-11-13T22:15:11 while i had done my own testing on made up maps, i'd never done line symmetric maps before making that generator a few days ago 2010-11-13T22:15:12 well I saw in the forum somebady said their bot did worse 2010-11-13T22:15:13 not quite center, i mean 2010-11-13T22:15:15 i expect nobody ha tested on them 2010-11-13T22:15:41 Mjothvitnir: for all we know, they've only tested on the first 100 maps 2010-11-13T22:15:56 if one had done that, they probably could expect their bot to suck on any others 2010-11-13T22:16:53 true all we know is what they said I think they claimed 1000 maps of "old" type and I don't remember a number for "new" maps probably how many ever were in the first bundle 2010-11-13T22:16:54 Accoun seems to crash rather frequently on them for some reason 2010-11-13T22:17:02 Top 10 players: bocsimacko(4377), bix0r4ever(4225), GreenTea(4171), DarthMasta(4061), dmj111(4054), george(4037), Accoun(4006), asavis(4001), shangas(3981), oldman(3969) 2010-11-13T22:17:05 no idea if it's the symmetry or what though 2010-11-13T22:17:17 prolly, planet # 2010-11-13T22:17:20 janzert: count the plants, yeah 2010-11-13T22:17:26 my bot has an issue with lots of equidistant planets atm, but it's because it has a sniping problem 2010-11-13T22:17:37 oh, heh, right, the planet count is different 2010-11-13T22:17:47 except it's not on first turn 2010-11-13T22:17:58 my bot is fine up until 65 planets.. 2010-11-13T22:18:06 *** mceier has quit IRC (Quit: leaving) 2010-11-13T22:18:10 and he's now resubmitted so I can't pull them up easily 2010-11-13T22:18:25 turns out I tried to save calculations in my first-turn nap sack by assuming there were only going to be at max 15 plnets equadistant or less . . . OOPS 2010-11-13T22:18:42 we could ask 2010-11-13T22:19:01 i'm sure he'll tell us if he needs to 2010-11-13T22:20:22 I think best would have been if they'd told us max and min parameters, and then unleashed a brand new map gen (that stayed within those parameters) on the last day 2010-11-13T22:20:37 yeah 2010-11-13T22:20:43 lavalamp_: that's what i mean by general but clear rules 2010-11-13T22:20:47 i want that next time 2010-11-13T22:20:59 we will do that next time, no doubt 2010-11-13T22:21:07 *** iFire has quit IRC (Read error: Connection reset by peer) 2010-11-13T22:21:07 yeah... they have to tell us what we can depend on 2010-11-13T22:21:07 we will do a lot of things next time :P 2010-11-13T22:21:08 Zannick: i said that after tron 2010-11-13T22:21:33 antimatroid: and did you do that? :P 2010-11-13T22:21:36 can you imagine the outrage if even 10% of the bots ended up crashing in the final tournament with the new generator though 2010-11-13T22:21:37 imo, the most important change was made 2010-11-13T22:21:38 OK, note to self :) come back to contest in 5 years to see if they have ironed out all the bugs. ;) 2010-11-13T22:21:42 you no longer get disqualified for timing out 2010-11-13T22:21:51 I'd rather not completely tick off 400 people 2010-11-13T22:22:11 janzert: that's why even generalness needs to be very clear 2010-11-13T22:22:14 that's why i'm glad some things stayed constant 2010-11-13T22:22:22 even if it was explicitly stated everywhere people are *not* going to like it 2010-11-13T22:22:37 I made no assumptions on the maps. But, I also was slightly afraid of crashing at the end (i.e., the 500 planet stress test). Since the maps hadn't changed for so long, maybe people assumed that the wouldn't. During the tron contest, the maps started changing fairly quickly into the contest, so we all realized that the maps could change. 2010-11-13T22:22:53 Also, during tron, the maps were pretty identifiable. So, everyone could tell they were changing. 2010-11-13T22:23:10 you'd have trouble missing the change with a lot of these 2010-11-13T22:23:14 what we really need to do is have more than one map generator 2010-11-13T22:23:19 for next time, i mean 2010-11-13T22:23:24 why? 2010-11-13T22:23:32 one map generator is fine, a less general one 2010-11-13T22:23:41 more general*** 2010-11-13T22:23:46 *** iFire has joined #aichallenge 2010-11-13T22:23:54 janzert does"New submissions" on the Server Stat page mean NEW bots or anyone's submittals 2010-11-13T22:24:06 anyone's submissions 2010-11-13T22:24:06 i mean, like the new one 2010-11-13T22:24:12 janzert: if too many people crash, maybe try restricting maps to 23 planets again? 2010-11-13T22:24:13 it's essentially two in one 2010-11-13T22:24:18 i don't think line symmetry should cause any crashes 2010-11-13T22:24:29 or very few, and if people assumed that, they really did fuck themselves over 2010-11-13T22:24:31 antimatroid: yeah, but I think we are fine 2010-11-13T22:24:32 but the point is to have sufficiently different maps 2010-11-13T22:24:55 line symmetry won't crash copybot, in fact, but it may do some weird things 2010-11-13T22:25:00 can you ask the DB how many people have made submission in say the last week (we talk about 4000+ players but I doubt all are still active) 2010-11-13T22:25:20 yes I know somebody may not be uploading "weekly" but still 2010-11-13T22:25:31 the big difference is in the tron contest there were a few participants that made map generators and the contest was able to pull from that pool of ideas 2010-11-13T22:26:02 here no one really released any with big differences that the community really got into 2010-11-13T22:26:19 well, till now :) 2010-11-13T22:26:22 nobody really released any with big differences did they? 2010-11-13T22:26:36 the rps maps don't really count 2010-11-13T22:26:42 people had talked about these maps but no one ever did them and i only just finished exams from an overloaded semester last week 2010-11-13T22:29:22 *** _xvinyl <_xvinyl!~xvinyl@ANantes-256-1-77-49.w86-210.abo.wanadoo.fr> has joined #aichallenge 2010-11-13T22:30:09 *** xvinyl has quit IRC (Ping timeout: 250 seconds) 2010-11-13T22:30:56 antimatroid> u is evil kid 2010-11-13T22:31:50 Accoun: this has always been expected by quite a few 2010-11-13T22:32:35 minority rules (hey, how said this wasn't a dictatorship) 2010-11-13T22:32:39 who 2010-11-13T22:32:45 where was the minority? 2010-11-13T22:33:01 the poll was in my favour, the rules were in my favour, jeff was in my favour 2010-11-13T22:33:15 it's more a dictatorfleet than a dictatorship... ;) 2010-11-13T22:33:44 how was the poll in your favor? 2010-11-13T22:33:58 *** _xvinyl <_xvinyl!~xvinyl@ANantes-256-1-77-49.w86-210.abo.wanadoo.fr> has quit IRC (Remote host closed the connection) 2010-11-13T22:33:58 go back and look 2010-11-13T22:34:01 15 vs 7 vs 12 2010-11-13T22:34:13 7 wasnt line symmetric too 2010-11-13T22:34:32 huh 2010-11-13T22:34:38 wanted* 2010-11-13T22:34:49 so 19 people wanted line symmetry, 15 didn't 2010-11-13T22:35:11 but 7 didn't want planet # change 2010-11-13T22:35:25 i'm completely indifferent to that, feel free to rip that out 2010-11-13T22:35:28 i think it makes no difference 2010-11-13T22:35:35 still 34 people vs 4000+ isn't a minority in your book 2010-11-13T22:35:36 antimatroid> Canadian creepy 2010-11-13T22:35:37 i just wanted more general maps as promised 2010-11-13T22:35:56 where'd you pull 4000+ opposition from? 2010-11-13T22:36:15 34 people voted vrs 4000+ who didn't 2010-11-13T22:36:15 antimatroid> bad boy 2010-11-13T22:36:22 4000 abstentions 2010-11-13T22:36:26 Accoun: :) 2010-11-13T22:37:06 they weren't abtentions, the vast majority never even knew there was a "vote" 2010-11-13T22:37:11 it was always intended that a map generator would be made like this, and i would consider it fair for them to just use maps like that in the final contest 2010-11-13T22:37:36 "like that"?? 2010-11-13T22:37:46 i was always warned in both contests the rules are meant to be like that, but like i've said, i do wish it was outlined more clearly 2010-11-13T22:37:58 brand new types of maps 2010-11-13T22:38:24 * lavalamp_ puts marshmellows on stick, holds in front of monitor 2010-11-13T22:38:35 :P 2010-11-13T22:38:51 last yar this canadia organizers make the same evil things 2010-11-13T22:38:54 so you are cool if a drastic (for arguement sake) change is made at 12:01 on the 28th. 2010-11-13T22:39:05 Accoun: that's because it was intended both times 2010-11-13T22:39:09 this http://openpaste.org/en/19644/ writen 2010-11-13T22:39:31 Mjothvitnir: this isn't a drastic change as far as i'm concerned, i've always written my bot to work under such general conditions 2010-11-13T22:40:18 the reason they SHOULD be changed is because otherwise people who assumed their own rules got a 2 month advantage on a restricted domain of maps for their bot to work on 2010-11-13T22:40:56 so is mine (with a few oopses). we agree that nothing was in stone, we agree that there should have been more guidelines, you disagree on stuff because of personal knowledge (tron contest) 2010-11-13T22:41:05 and were caught on the selective removal of the results of games 2010-11-13T22:41:20 selective removal? 2010-11-13T22:41:24 yes 2010-11-13T22:41:36 what do you mean? 2010-11-13T22:42:12 "This rewards good AI algorithms, and punishes strategies that are designed with particular maps in mind." That is exactly the point btw 2010-11-13T22:42:20 it is an ai contest afterall 2010-11-13T22:42:50 final games be maked hidden in process and in publised resilt amount of games plaed by some top bots be radicale less than ozer 2010-11-13T22:43:29 yeah, i'm pretty sure rankings weren't reset when the final contest ran last time 2010-11-13T22:43:45 i have made sure people know not to let that happen again 2010-11-13T22:44:07 and you can no longer get disqualified for timing out 2010-11-13T22:44:13 it canadian 2010-11-13T22:44:18 i had a top 10 quality bot last time and got dq'd for that 2010-11-13T22:44:20 if there was a prize to this contest then I might suspect that "we'll create maps rules and game rules, and contest method at the last minute" means that they will stack the deck 2010-11-13T22:44:22 caontru of stupednes 2010-11-13T22:45:17 last year final games be maked hidden in process and in publised resilt amount of games plaed by some top bots be radicale less than ozer 2010-11-13T22:45:33 and this http://openpaste.org/en/19644/ writen actions 2010-11-13T22:46:31 Accoun: wht do you mean by made hidden in process? 2010-11-13T22:46:45 yes game amounts weren't equal last time, i didn't like that and I don't think it will happen this time 2010-11-13T22:46:49 although i'm not positive 2010-11-13T22:47:19 nobody knows anything about what will/won't happen 2010-11-13T22:47:22 http://csclub.uwaterloo.ca/contest/profile_games.php?user_id=2788 that was my game record leading up to being disqualified 2010-11-13T22:47:27 antimatroid> plaed on private server 2010-11-13T22:47:50 Accoun: are you referring to the leaderboard not being viewable? 2010-11-13T22:47:52 antimatroid> ont seen by public in process, omly final result published 2010-11-13T22:48:04 i have tried to ensure that wont happen again, but i can't state it as fact 2010-11-13T22:48:12 with selective game deleted 2010-11-13T22:48:42 ok, we all get it you hate this contest and the organizers please give it a rest 2010-11-13T22:48:48 idiscusion about it be on deleted aw forum 2010-11-13T22:48:50 discusion about it be on deleted aw forum 2010-11-13T22:48:55 discusion about it be on deleted naw forum 2010-11-13T22:50:25 "ananonsment" forem deleted from this http://ai-contest.com/forum/viewforum.php?f=7 2010-11-13T22:50:39 janzert: what are your current thoughts on the leaderboard being live? i think that's another decent point, people are much less likely to complain about games being "removed" if they can see their games throughout the contest 2010-11-13T22:51:44 I think a live leaderboard helps keep the community alive, but I can also see wanting to not spoil the surprise of the winner 2010-11-13T22:51:58 overall I'd prefer keeping it live 2010-11-13T22:52:12 good, it was so painful waiting around not knowing 2010-11-13T22:52:22 i think hiding the games spoils the idea that this contest is open 2010-11-13T22:52:40 yeah, watching the leaderboard evolve and watching the top games will be exciting 2010-11-13T22:52:42 it seems like a poor idea to me 2010-11-13T22:53:59 last yaer result comlety chiting 2010-11-13T22:54:22 then why in the world do you come back again? 2010-11-13T22:55:00 hoping it would be better, which it is, but there are still things that need ironing out 2010-11-13T22:55:08 but it's a new contest, these things are going to happen 2010-11-13T22:55:22 nah, not you Accoun 2010-11-13T22:55:46 he seems to believe in malicious action on the part of the organizers 2010-11-13T22:56:02 I can't understand why you'd come back in that case 2010-11-13T22:56:21 well, if trying to steal everyone's free time is malicious... ;) 2010-11-13T22:56:39 hehe 2010-11-13T22:56:55 trying? they're pretty damn good at it... 2010-11-13T22:57:11 they don't just steal my free time 2010-11-13T22:57:17 not me, i'm free! 2010-11-13T22:57:30 yeah, i finished undergrad under a week ago :) 2010-11-13T22:57:34 feels good still 2010-11-13T22:59:27 congrats :) 2010-11-13T23:00:39 Accoun: were you actually relying on point symmetry? 2010-11-13T23:04:40 very evil canadian! 2010-11-13T23:05:26 in word war 3 cnadian be bornaed 1st! 2010-11-13T23:05:46 or the planet size being 23 2010-11-13T23:06:09 where's the fancy visualizer? I want to recover a game from the tcp server 2010-11-13T23:06:10 Account: i'm australian, you can hate a whole new nationality now too 2010-11-13T23:06:36 aren't you worried about getting bombed second? 2010-11-13T23:06:37 Zannick: planet size change is easy, relying on point symmetry would be ingrained in the strategy 2010-11-13T23:07:01 but it's not something someone did without just flat out assuming it 2010-11-13T23:07:01 I considered relying on point symmetry for my first move 2010-11-13T23:07:22 but it's not something I'd gotten to yet 2010-11-13T23:07:30 and i argue that it's unfair people got to work for 2 months under that restriction 2010-11-13T23:07:40 *** bet has joined #aichallenge 2010-11-13T23:07:51 i dunno about unfair 2010-11-13T23:08:02 probably should have been done sooner 2010-11-13T23:08:07 *** bet has quit IRC (Client Quit) 2010-11-13T23:08:25 a driveby! 2010-11-13T23:08:31 :P 2010-11-13T23:08:41 other people were welcome to write a generator :P 2010-11-13T23:08:46 i did not have time previously 2010-11-13T23:09:26 it could be interesting to do psuedo symetrical.. 2010-11-13T23:09:34 antimatroid> i dont cnow well australian, but canadian very stuped an evil! 2010-11-13T23:09:48 like, it's symetrical from each base.. but not symetrical from every other point 2010-11-13T23:10:04 since distances are rounded up to integer 2010-11-13T23:10:09 wait what 2010-11-13T23:10:13 aren't maps like that now? 2010-11-13T23:10:19 at least line symmetric ones 2010-11-13T23:10:21 i don't think i understand 2010-11-13T23:10:37 the maps are symmetric around the origin, then translated 2010-11-13T23:10:48 and symmetric for players starting positions 2010-11-13T23:11:06 if you have planets ABC, opposite from planets DEF.. the distance ebtween AD and CF doesn't need to be equal 2010-11-13T23:11:37 can you use paint? :P 2010-11-13T23:11:40 you have to be able to apply a transformation to the map to switch the players 2010-11-13T23:12:11 burny: no... because then you can get nonsymmetry in distances from B and E and then the map isn't symmetrical 2010-11-13T23:12:44 thus the 'psuedo' symetrical :P 2010-11-13T23:13:01 there isn't a pseudo to it 2010-11-13T23:13:20 it could be interesting to do psuedo symetrical 2010-11-13T23:13:30 it's either symmetrical or not 2010-11-13T23:13:56 if B and E are the starters, then you have to have distances the same from them to mirrored planets 2010-11-13T23:14:54 the distances from B and E to all other planets can be mirror'd, without the distances between those planets being 100% mirror'd 2010-11-13T23:15:15 if A and F are a mirrored pair and C and D the other, then distances AD and CF must be the same 2010-11-13T23:15:19 as otherwise 2010-11-13T23:15:45 B to A then to D is not the same time as E to F to C 2010-11-13T23:17:03 Top 10 players: bocsimacko(4361), Raschi(4154), bix0r4ever(4154), GreenTea(4123), dmj111(4028), george(4014), asavis(3977), shangas(3970), DarthMasta(3962), Accoun(3950) 2010-11-13T23:17:10 i think you'd also get noneuclidean things when you look at the triangles BAD and EFC, as BA=EF and BD=EC, but AD is not equal to FC 2010-11-13T23:17:27 eg. instead of distances 2.4, 3.5, 2.4 it could be A 2.4, B 3.5, C 2.6 D.. distances are still 3,4,3.. but the distance from BD not= AC 2010-11-13T23:18:13 oh 2010-11-13T23:18:51 well, if you manage that without changing the integer values used in-game for distance, that's not really different 2010-11-13T23:19:17 thatmakes AC=6 and BD=7 2010-11-13T23:19:54 then it's not symmetrical 2010-11-13T23:19:59 obviously 2010-11-13T23:20:57 *** McLeopold has joined #aichallenge 2010-11-13T23:21:08 and then one player has an advantage 2010-11-13T23:21:20 *** yasith has joined #aichallenge 2010-11-13T23:21:29 if A and D are starters, A can reach B and C before the other player 2010-11-13T23:21:35 both players have different advantages 2010-11-13T23:21:55 one can attack more easily.. 2010-11-13T23:21:55 *** yasith_ has joined #aichallenge 2010-11-13T23:21:59 the other can defend more easily 2010-11-13T23:22:01 last year chiting: http://ai-contest.com/forum/viewtopic.php?f=3&t=366&start=30#p2151 2010-11-13T23:22:06 אשדעג רו 2010-11-13T23:22:12 found it 2010-11-13T23:22:41 "That could explain a difference in number of games played per contestant of 50 or 100, but not a difference of over 500 as actually observed." 2010-11-13T23:22:42 yes, and that inequality is something we have to avoid 2010-11-13T23:24:43 Accoun: have you considered any messups were just that, and not blatent tampering? 2010-11-13T23:25:07 and if i'm correct and rankings weren't reset when the final contest ran 2010-11-13T23:25:18 a lot of crap bots would have been quite old, and have had a lot of games in their history 2010-11-13T23:25:37 that would explain crap bots having more games 2010-11-13T23:26:11 antimatroid> if it not chiting why ginal games process be hidden? 2010-11-13T23:26:19 antimatroid> if it not chiting why final games process be hidden? 2010-11-13T23:26:56 Accoun: the contest is open source this time around 2010-11-13T23:27:01 all code can be viewed 2010-11-13T23:27:39 antimatroid> and some games result be deleted 2010-11-13T23:27:57 antimatroid> manyaly by hands 2010-11-13T23:28:20 one thing that might have happened last time, they were removing maps that resulted in lots of draws and i have no idea how they did that 2010-11-13T23:28:35 if they removed those results, people may have lost win results 2010-11-13T23:28:49 it wont happen like that again, i'm pretty sure janzert is on top of such problems 2010-11-13T23:28:52 janzert <3 2010-11-13T23:29:05 janzert is on top 2010-11-13T23:29:10 8-) 2010-11-13T23:29:24 canadian overlord 2010-11-13T23:29:43 Accoun: besides the map generator changing, what do you think of the new maps? 2010-11-13T23:30:07 http://ai-contest.com/visualizer.php?game_id=7244585 2010-11-13T23:30:15 raschi made a really stupid move there 2010-11-13T23:31:36 antimatroid> by it bag based on ficsed max number of planets 2010-11-13T23:31:53 antimatroid> my be it bug based on ficsed max number of planets 2010-11-13T23:32:36 *** yellowseed has quit IRC (Ping timeout: 245 seconds) 2010-11-13T23:33:15 antimatroid> new maps not so bad, bad idea to make changes in last week 2010-11-13T23:33:52 there's still at least 2 weeks left 2010-11-13T23:33:55 antimatroid> canadian evil actions in action 2010-11-13T23:34:03 actually, 2 weeks exactly pretty much 2010-11-13T23:34:18 antimatroid> 2 weeks is nothong 2010-11-13T23:34:53 antimatroid> this yar problem more complex than last year 2010-11-13T23:35:29 last contest was this year :P but i always screw that up 2010-11-13T23:35:40 all bots are meant to be able to handle maps like this already 2010-11-13T23:36:07 i am trolling profiles like nothing else, i can't find any significant different in peoples win/loss ratios as yet 2010-11-13T23:36:13 but i'm only looking in the top 20 2010-11-13T23:38:30 actually, flow bot seems to be doing worse 2010-11-13T23:38:33 http://ai-contest.com/profile_games.php?user_id=10487 2010-11-13T23:38:35 flowe* 2010-11-13T23:45:44 *** Ermalot has joined #aichallenge 2010-11-13T23:46:49 *** Ermalot has left #aichallenge 2010-11-13T23:49:40 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-11-13T23:56:46 http://ai-contest.com/forum/viewtopic.php?f=17&t=801&start=30 2010-11-13T23:56:51 that smiley commentary was great 2010-11-13T23:59:03 oldman seems to be having issues too