2010-10-15T00:01:19 Top 10 players: dmj111(3566), MariusCG(3535), Hazard(3528), felixcoto(3527), rsergio(3516), deepblue(3480), ruilov(3462), davidjliu(3461), savchenko(3459), _iouri_(3452) 2010-10-15T00:01:36 Congrats dmj. 2010-10-15T00:02:05 dmj111: ya! 2010-10-15T00:03:28 amstan_: thanks guys. 2010-10-15T00:03:30 *** AlliedEnvy has joined #aichallenge 2010-10-15T00:03:37 Azrathud: both... 2010-10-15T00:03:38 dmj111: new bot version? 2010-10-15T00:04:17 kind of. it is a little experimental. it even crashes (look at its first loss :/ ) 2010-10-15T00:04:47 any hints on how it works? 2010-10-15T00:05:12 its not too complicated... 2010-10-15T00:05:15 * dmj111 is typing more 2010-10-15T00:05:17 wow! second place guy is romanian 2010-10-15T00:05:35 yes! http://ai-contest.com/visualizer.php?game_id=5881697 2010-10-15T00:05:37 epic battle 2010-10-15T00:06:32 yeah, that one is pretty amusing. I almost feel bad getting a win for it though. 2010-10-15T00:07:11 a1k0n is nowhere to be seen? 2010-10-15T00:07:16 a1k0n: did you give up? 2010-10-15T00:07:16 My bot does little more than only take what it can take, and defend what it can defend. One of its weaknesses is that it is too conservative on the attack. 2010-10-15T00:07:55 So, this version has some code it in to realize that it doesn't need to defend each planet with only the ships at the planet. They can defend each other. 2010-10-15T00:08:19 but, the code is ugly, and it is not clear I got it right. 2010-10-15T00:08:26 lol 2010-10-15T00:08:50 aww, there's no way to get submission id from the website 2010-10-15T00:08:51 i am waiting for a1k0n to unleash a monster at some point. although, I am thankful for minecraft. 2010-10-15T00:08:54 phpmyadmin is too much work 2010-10-15T00:09:06 lol 2010-10-15T00:09:18 dmj111: i have a hunch he gave up 2010-10-15T00:09:24 maybe he does not want more glory 2010-10-15T00:09:49 once you've won one... what more do you need to prove :) 2010-10-15T00:09:58 yeah, pretty much 2010-10-15T00:10:17 he just dominated that final tournament... it was amazing. 2010-10-15T00:10:40 i wish we had contestbot then 2010-10-15T00:10:43 to log everything 2010-10-15T00:11:06 maybe i can digg up my logs from #googleai 2010-10-15T00:11:15 dmj111: did you participate in the other one? 2010-10-15T00:11:19 i never even lurked back then. I wish I would have... listening/talking on irc has been the best part of this. 2010-10-15T00:11:46 I did. I even led, briefly. I think I was one of the earlier voronoi adopters. 2010-10-15T00:11:54 Then, everyone else figured it out better. 2010-10-15T00:11:56 and boom. 2010-10-15T00:13:06 yeah, actually.. last term the channel was 1/3 of the size 2010-10-15T00:13:13 *** amstan_ is now known as amstan 2010-10-15T00:13:54 ohh... tell me you aren't going to do this each term. I have spend too much time on this ... 2010-10-15T00:14:11 it would be cool 2010-10-15T00:15:26 well, take it easy. I haven't slept much the last few weeks... so I better head off. 2010-10-15T00:15:38 good night 2010-10-15T00:20:39 *** davidjliu has quit IRC (Quit: Page closed) 2010-10-15T00:22:57 dmj111: you know what should be done for the next contest? something that will have to get you addicted to minecraft first.. 2010-10-15T00:23:01 like a minecraft AI 2010-10-15T00:23:24 so contestants will have to deal with the addiction to minecraft while making their bot 2010-10-15T00:23:27 that is diabolical 2010-10-15T00:23:37 evening folks 2010-10-15T00:23:49 bhasker: it's just me actually, talking to myself 2010-10-15T00:23:58 lol 2010-10-15T00:23:58 hmm 2010-10-15T00:26:12 o/ 2010-10-15T00:26:25 so.. let's talk! everyone! lol 2010-10-15T00:26:54 i was going to work a lot on my bot today... 2010-10-15T00:27:02 aww 2010-10-15T00:27:07 i'm lonely then 2010-10-15T00:27:17 but then i sat down at the computer and was like...ugh... 2010-10-15T00:27:28 maybe saturday 2010-10-15T00:27:31 :P 2010-10-15T00:28:08 if i don't have a working bot by november it'll be my fault 2010-10-15T00:28:08 Zannick: did you see "the social network"? 2010-10-15T00:28:24 i think i'm going to see it tomorrow 2010-10-15T00:28:39 i've heard some things already 2010-10-15T00:28:39 Zannick: make sure you watch it at the beginning of the day 2010-10-15T00:28:53 after you watch it, you'll be in perfect mood to program 2010-10-15T00:28:56 it's magical 2010-10-15T00:29:00 heh 2010-10-15T00:29:08 you'll be like, zomg! code code! 2010-10-15T00:29:14 program what, is the question 2010-10-15T00:29:21 that's your problem, lol 2010-10-15T00:29:25 * Zannick has so many projects besides this one :P 2010-10-15T00:29:35 like what? 2010-10-15T00:30:05 The current rankings don't show up on Firefox but they do on Chromium. 2010-10-15T00:30:13 Azrathud: fun stuff 2010-10-15T00:30:23 they work in contest bot, so it's all good 2010-10-15T00:30:24 my own irc bot, a roguelike, and some other things 2010-10-15T00:30:36 Zannick: irc bot you say? 2010-10-15T00:31:00 aye 2010-10-15T00:31:06 tell me more 2010-10-15T00:31:08 #moar 2010-10-15T00:31:09 amstan: Error: You haven't asked me a command; perhaps you want to see someone else's more. To do so, call this command with that person's nick. 2010-10-15T00:31:59 it is a pretty basic thing acting as a joke repository 2010-10-15T00:32:31 you should work on bot technologies then 2010-10-15T00:32:36 Zannick: I have 5 units on top of this and my side project 2010-10-15T00:32:49 bot technologies i say! 2010-10-15T00:32:52 i am very much interested in ai 2010-10-15T00:32:53 plus i've been audting post grad measure theory :| 2010-10-15T00:33:19 everyone needs to have a bot! 2010-10-15T00:33:23 * amstan looks at contestbot 2010-10-15T00:33:33 i'm going to implement a minimax bot this weekend and see how it goes 2010-10-15T00:33:49 antimatroid: was the feasible then? 2010-10-15T00:33:49 i need to rewrite a lot of my backend to do it properly though 2010-10-15T00:34:02 was the feasible what? 2010-10-15T00:34:07 minimax 2010-10-15T00:34:20 it's not going to be an exhaustive tree at each layer 2010-10-15T00:34:22 s/the/that/ 2010-10-15T00:34:23 maybe i should implement the simpler idea before the more complicated one... 2010-10-15T00:34:25 but i expect to to do well if set up properly 2010-10-15T00:34:34 too simple and it'll suck 2010-10-15T00:34:47 at least it'll be something that runs! 2010-10-15T00:35:05 copybot doesn't count 2010-10-15T00:35:16 i was thinking about abstracting ships and stuff 2010-10-15T00:35:21 don't think of such low levels 2010-10-15T00:35:37 abstract to things like, waves, attacks 2010-10-15T00:35:49 don't leave planets undefended 2010-10-15T00:35:50 and all that 2010-10-15T00:35:52 nah, you need to handle the micro side of things 2010-10-15T00:35:55 or else you'll get trampled 2010-10-15T00:36:12 i had an idea wherein i would program a bot that knew nothing but would learn from watching games and analyzing strategies 2010-10-15T00:36:24 Zannick: been done, neural networks 2010-10-15T00:36:33 yeah 2010-10-15T00:36:45 but really hard for me since i have no experience in that sort of thing 2010-10-15T00:36:47 topic for it: http://ai-contest.com/forum/viewtopic.php?f=17&t=486 2010-10-15T00:36:54 i was thinking about it 2010-10-15T00:37:10 could be done 2010-10-15T00:37:15 with proper inputs and outputs 2010-10-15T00:37:34 i have never been able to find a decent coverage of neural networks 2010-10-15T00:37:47 inputs could be... major incoming fleets, enemy planet size, closest enemy planet 2010-10-15T00:37:50 do you guys know anywhere on the net that would satisfy the standards expected of a maths student? 2010-10-15T00:38:01 outputs could be: attack biggest enemy planet, defent base being attacked 2010-10-15T00:38:04 defend* 2010-10-15T00:38:20 antimatroid: i saw a guide somewhere 2010-10-15T00:38:22 idk about guides 2010-10-15T00:38:25 hold on 2010-10-15T00:39:13 i don't want a guide, i want to know the maths behind nueral networks 2010-10-15T00:39:31 http://www.student.cs.uwaterloo.ca/~cs486/ 2010-10-15T00:39:35 cheers 2010-10-15T00:39:38 there's a book on it 2010-10-15T00:39:43 maybe you can aquire it 2010-10-15T00:39:53 pretty sure it has neural networks 2010-10-15T00:40:23 *** jmcarthur has quit IRC (Read error: Operation timed out) 2010-10-15T00:40:35 antimatroid: nvm, html version, under CC 2010-10-15T00:41:04 antimatroid: i was reading this the other day: http://www.ai-junkie.com/ann/evolved/nnt1.html 2010-10-15T00:41:06 seems cool 2010-10-15T00:41:16 the mutation part seems the most complicated 2010-10-15T00:41:20 actually making the bot is easy 2010-10-15T00:41:23 the trainer's hard 2010-10-15T00:42:25 Zannick: neural networks link: http://www.ai-junkie.com/ann/evolved/nnt1.html 2010-10-15T00:43:54 cool, thanks 2010-10-15T00:44:59 *** jmcarthur has joined #aichallenge 2010-10-15T00:45:18 i remember some of this from my AI class... 2010-10-15T00:45:37 looks, cool, thanks 2010-10-15T00:48:19 pitty the guy goes easy on the maths, it's very hard to find a coverage of neural networks with the maths, everyone seems to want to do it without it :| 2010-10-15T00:48:31 too, many, commas, too 2010-10-15T01:01:20 Top 10 players: dmj111(3547), MariusCG(3532), Hazard(3501), felixcoto(3496), rsergio(3491), deepblue(3487), ruilov(3449), savchenko(3448), ofrias(3446), _iouri_(3428) 2010-10-15T01:04:09 antimatroid: that maths... there isn't much 2010-10-15T01:04:19 the maths... 2010-10-15T01:04:34 rwa: i still prefer to do things properly 2010-10-15T01:04:47 NN are little more than a function fitting procedure 2010-10-15T01:05:03 the only non-trivial part is the back propagation, no? 2010-10-15T01:05:08 I can't recall it being that hard 2010-10-15T01:05:13 i've never really looked at them 2010-10-15T01:05:16 antimatroid: nn isn't about math, not really 2010-10-15T01:06:25 foucist: that's an odd statement... what do you mean? 2010-10-15T01:06:26 from wikipedia "The perceptron is essentially a linear classifier for classifying data  specified by parameters  and an output function f = w'x + b. Its parameters are adapted with an ad-hoc rule similar to stochastic steepest gradient descent. Because the inner product is a linear operator in the input space" 2010-10-15T01:06:32 along those lines is what i'm looking for 2010-10-15T01:08:31 i think i have a pretty good idea of how to do my game tree anyway, hopefully that will work well 2010-10-15T01:08:31 how about this (I just followed links from wikipedia): http://www.doc.ic.ac.uk/~nd/surprise_96/journal/vol4/cs11/report.html#Appendix%20B%20-%20The%20back-propagation%20Algorithm%20-%20a%20mathematical%20approach 2010-10-15T01:08:42 i'm still going to work out how to try and train the parameters though 2010-10-15T01:09:41 good luck 2010-10-15T01:09:44 I'm not quite ready to try to train parameters yet, but when I am, I know exactly how I'm going to do it 2010-10-15T01:09:51 i am :o at the entire idea 2010-10-15T01:10:05 no idea how to do it without running millions of games 2010-10-15T01:10:22 what's wrong with running millions of games? 2010-10-15T01:10:36 nothing unless you can do the same with thousands of games 2010-10-15T01:10:41 we're intending to do it once we have the tree etc. going, i suspect we'll take parameters as inputs from a file and just update them there, then write a program to just test lots of parameters etc. against each other, the fine details are still very vague though 2010-10-15T01:10:42 sure 2010-10-15T01:10:43 i'm just going to entertain myself watching these minesweeper tanks 2010-10-15T01:11:13 minesweeper is one of my favourite games, it's incredibly underated 2010-10-15T01:11:24 *** Azrathud has quit IRC (Quit: Meh) 2010-10-15T01:11:41 antbot disconnecting on turn 1 mcleopold 2010-10-15T01:11:45 what's wrong? i dunno, i wouldn't really know how well the network worked until the end 2010-10-15T01:12:16 *** delt0r___ has joined #aichallenge 2010-10-15T01:13:01 my current version of codeblocks wont let me run multiple sessions, even when i fiddle with the settings, it's very annoying 2010-10-15T01:13:30 multiple sessions? 2010-10-15T01:13:50 like, multiple codeblocks windows for different projects 2010-10-15T01:13:56 have to have them all in the one window 2010-10-15T01:14:05 ah, i see 2010-10-15T01:14:32 *** delt0r_ has quit IRC (Ping timeout: 276 seconds) 2010-10-15T01:14:35 emacs 4tw 2010-10-15T01:15:36 vim :P 2010-10-15T01:15:47 also nano, pico, and ed 2010-10-15T01:15:54 and echo 2010-10-15T01:15:57 and sed 2010-10-15T01:16:18 and word 2010-10-15T01:16:34 (things i'm sure people program in) 2010-10-15T01:17:26 but emacs 4tw 2010-10-15T01:17:28 it's going to suck once tree bots are up and running for testing on tcp 2010-10-15T01:17:33 rest are just wannabes 2010-10-15T01:17:43 i'm determined to have one going by the end of the weekend 2010-10-15T01:18:04 *** jmcarthur has quit IRC (Ping timeout: 245 seconds) 2010-10-15T01:18:20 tree bots are hard! 2010-10-15T01:18:25 I'm sure everyone is trying 2010-10-15T01:18:39 i've worked out how to do it, just need to implement 2010-10-15T01:18:54 well, a way to do it 2010-10-15T01:18:57 i've already seen *spanning* tree bots... >.> 2010-10-15T01:19:10 are you talking about bartwe? 2010-10-15T01:19:20 spanning tree was a bad idea, imo 2010-10-15T01:19:23 yes 2010-10-15T01:19:26 terrible idea 2010-10-15T01:19:34 it was an interesting idea 2010-10-15T01:19:45 it worked well when no one else was redistributing ships 2010-10-15T01:19:57 someone else mentioned building convex hulls that kept trying to increase their area 2010-10-15T01:20:00 any sort of supply chain is better than none 2010-10-15T01:20:05 bhasker I think 2010-10-15T01:20:07 that was sort of cool, but again i don't think it would work very well 2010-10-15T01:20:23 i was thinking of it 2010-10-15T01:20:26 but didnt do it 2010-10-15T01:20:29 I was right! 2010-10-15T01:20:47 i wouldn't be surprised if there was quite a few :P 2010-10-15T01:20:58 quite a few what? 2010-10-15T01:20:59 i was not really looking at using a convex hull to increase the area but more as an influence area 2010-10-15T01:21:01 http://72.44.46.68/canvas?game_id=495698 2010-10-15T01:21:06 to identify the edges where i should flow my ships too 2010-10-15T01:21:07 this is sort of a tree bot 2010-10-15T01:21:32 what makes you say that? 2010-10-15T01:21:39 oh wait, nevermind 2010-10-15T01:21:50 so is it working reasonable well? 2010-10-15T01:21:54 eh 2010-10-15T01:22:00 are you using A*? and what about alpha-beta? 2010-10-15T01:22:06 i'm not sure about the second 2010-10-15T01:22:08 it is beating some strong bots, but also losing to weak bots 2010-10-15T01:22:09 intend to use the first 2010-10-15T01:22:25 are you going lots of moves at each layer? that's my intention 2010-10-15T01:23:01 and i'm storing states for extendable nodes, and a whole tree, but not game state info for already processed nodes, despite what people on here advised :P 2010-10-15T01:23:20 I am not using either right now 2010-10-15T01:23:37 is memory an issue too? it wasn't for tron even when storing the tree, i think it is for this though 2010-10-15T01:23:47 I'm not sure how A* is applicable for searching the game state 2010-10-15T01:23:54 no, my tree is not big enough to be a problem with mem 2010-10-15T01:25:59 *** jmcarthur has joined #aichallenge 2010-10-15T01:26:15 that's why I say "sort of" a tree bot. There's a tree there, but it's very incomplete. 2010-10-15T01:30:11 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-10-15T01:31:16 choosing the ways to complete or not complete it sort of leaves you in heuristic land... 2010-10-15T01:34:27 rwa: i'm planning to have a priority queue of nodes to extend 2010-10-15T01:34:32 that way you can just extend for as long as you want 2010-10-15T01:34:54 *** Utkarsh has joined #aichallenge 2010-10-15T01:35:58 curious to see how the result plays 2010-10-15T01:36:25 i have to start getting ready for exams soon, so if i don't get it up and running before the end of the weekend, it's going to take me a while 2010-10-15T01:37:31 exams are only what, 3hrs? plenty of time left for working on your bot 2010-10-15T01:37:40 i have 5 of them :P 2010-10-15T01:38:03 and analysis, pdes, and topics in adv. maths aren't going to be walks in the park 2010-10-15T01:38:24 the other two, yeah probably 2010-10-15T01:38:30 :P 2010-10-15T01:39:07 PDEs is a final year subject? 2010-10-15T01:39:32 yeah, and it was very much more of an applied physics unit, have hated it 2010-10-15T01:39:47 i can't stand applied maths, it's a lot of grunt work for no fun, i'm doing all pure units for honours 2010-10-15T01:39:53 haha 2010-10-15T01:40:16 yet i constantly do the most tedious things for fun while programming stuff, go figure 2010-10-15T01:40:57 well, a computer is closer to pure maths than the physical world 2010-10-15T01:41:03 I'm the opposite. Any article that starts with something like "Consider a compact operator on a Hilbert space" goes directly into the round file. 2010-10-15T01:41:32 i've only just learnt about hilbert spaces this semester, so still a bit shaky 2010-10-15T01:41:53 Don't waste my time with that sh7t, what problem are you trying to solve? I don't care about your pseudo-art. 2010-10-15T01:41:56 i'm constantly amazed at how much von neumann did 2010-10-15T01:42:12 rwa: analysis and algebra were my two favourite undergrad units :P 2010-10-15T01:42:18 Von Neumann was an absolute genius for sure 2010-10-15T01:42:33 he did two degrees simultaneously 2010-10-15T01:43:01 i swear that's impossible, i have done 5 officially this semester, plus lectures and some assignments for measure and ai, and i have not had a spare second 2010-10-15T01:43:07 he did that for a good 3 years 2010-10-15T01:43:43 you still have enough time for this contest... your work load is not very high 2010-10-15T01:43:45 :P 2010-10-15T01:44:01 you should see the quality of some of the assignments i handed in :P 2010-10-15T01:44:13 :) 2010-10-15T01:44:14 here's for high weighted exams saving my life 2010-10-15T01:44:24 how much do the exams count for/ 2010-10-15T01:44:41 70-80% 2010-10-15T01:44:47 yeah, that's not too bad 2010-10-15T01:45:23 i shouldn't even be passing computer graphics, haven't been to a class since first week and my assignments were dreadful, but our school for that here is that terrible 2010-10-15T01:46:50 meh, if you are a math student computer graphics should be easy 2010-10-15T01:47:02 it's just basic linear algebra 2010-10-15T01:47:47 you haven't seen what i handed in as assignments, they were all about using a pre written engine like a black box, the school sucks, they're entirely math averse 2010-10-15T01:48:06 ah :( 2010-10-15T01:48:13 an exam question is likely to be write a function for multiplying two 3x3 matrices, i mean a third year unit, come on 2010-10-15T01:48:24 anyway, I've been informed there's free food nearby... cya 2010-10-15T01:48:26 it's 4 lines 2010-10-15T01:48:30 ha, bye 2010-10-15T01:48:46 5* 2010-10-15T01:56:28 has a decision been made about whether bots get to know the number of turns before a draw in the final tournament? 2010-10-15T02:01:20 Top 10 players: dmj111(3538), MariusCG(3516), Hazard(3510), rsergio(3498), felixcoto(3488), deepblue(3476), ofrias(3437), savchenko(3437), ruilov(3414), _iouri_(3402) 2010-10-15T02:08:31 *** acieroid` has joined #aichallenge 2010-10-15T02:10:09 *** acieroid has quit IRC (Ping timeout: 240 seconds) 2010-10-15T02:15:38 is it quicker to reset values in a vector to 0 or just create a new vector and have the old one destroyed? 2010-10-15T02:17:16 *** McLeopold_ has joined #aichallenge 2010-10-15T02:21:56 *** sigh has quit IRC () 2010-10-15T02:24:23 may depend on how the garbage collector works 2010-10-15T02:25:15 i'd say that usually the former is probably quicker 2010-10-15T02:26:37 http://72.44.46.68/canvas?game_id=496433 new speed record 2010-10-15T02:27:04 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-10-15T02:27:11 and 8 games in a row without a disconnect. yeah 2010-10-15T02:27:38 yeah, don't think i've done single digits yet, well done 2010-10-15T02:27:43 i have done under 20 2010-10-15T02:28:06 So, because of a logging error, I have a lot of losses with games under 10 turns. 2010-10-15T02:28:27 i was referring to legitimate wins :P 2010-10-15T02:28:38 My bot looks very impressive against who I'm playing. 2010-10-15T02:28:43 i don't care for outcomes if messed up games, it's not the point 2010-10-15T02:28:50 Nobody good yet, I'm too far down. 2010-10-15T02:28:58 is this the knapsack like bot? 2010-10-15T02:29:11 It still gets to me, but yeah, me neither. 2010-10-15T02:29:20 Kinda? 2010-10-15T02:29:31 Not really. 2010-10-15T02:29:40 i've just cleaned out a copy of mine for a complete clean then minimax attempt 2010-10-15T02:30:03 The knapsack problem is fun to solve, but not totally applicable to this game. 2010-10-15T02:30:30 yeah, but it's similar in concept to picking planets to send ships too 2010-10-15T02:30:38 if not entirely the same 2010-10-15T02:30:44 to***** 2010-10-15T02:30:44 This bot has growth management and super defense. 2010-10-15T02:30:59 It's meant to barely win. 2010-10-15T02:31:11 yeah, i prioritise defense over anything in my heuristic got, and i have tried to cut back on expanding 2010-10-15T02:31:58 This bot also tries really hard to steal planets. 2010-10-15T02:32:14 *** Utkarsh has joined #aichallenge 2010-10-15T02:32:38 i must admit my very poor hack to try and also cripple their front line with spare ships seemed to work better than it hurt 2010-10-15T02:33:09 I found that to be beneficial as well. That's what my official entry does. 2010-10-15T02:33:31 Especially against dmj and deepblue. 2010-10-15T02:33:34 my official entry atm seems to suck 2010-10-15T02:34:16 if you hammer their front lines, they make more mistakes, or overcommit to defense. 2010-10-15T02:35:20 At first I though abandoning planets was a bad idea, but I'm beginning to think it may be useful. I have to do more calculations. 2010-10-15T02:35:42 *** JCS^ has quit IRC (Quit: \x00) 2010-10-15T02:35:47 And if jmcarthur is listening, I have a really good scenario to solve. 2010-10-15T02:37:52 contestbot later tell jmcarthur I have a scenario that is close to r/p/s. I'll pastebin it later if interested. 2010-10-15T02:37:52 McLeopold_: Yep. 2010-10-15T02:38:26 *** McLeopold_ has quit IRC (Quit: Page closed) 2010-10-15T02:46:33 * antimatroid requests a pastebin 2010-10-15T02:46:39 oh, he's gone, my bad 2010-10-15T02:48:28 arg 2010-10-15T02:48:58 @later tell McLeopold i want that scenario :) 2010-10-15T02:48:58 jmcarthur: Yep. 2010-10-15T03:01:21 Top 10 players: dmj111(3568), MariusCG(3536), Hazard(3514), felixcoto(3509), rsergio(3508), deepblue(3474), savchenko(3455), ofrias(3437), ruilov(3426), _iouri_(3383) 2010-10-15T03:15:34 *** javagamer has quit IRC (Ping timeout: 245 seconds) 2010-10-15T03:17:29 *** perror has joined #aichallenge 2010-10-15T03:23:53 *** aerique has joined #aichallenge 2010-10-15T03:24:29 *** edcba has quit IRC (Read error: Connection reset by peer) 2010-10-15T03:26:29 *** javagamer has joined #aichallenge 2010-10-15T03:30:34 *** espes has quit IRC (Ping timeout: 252 seconds) 2010-10-15T03:31:34 *** sigh has joined #aichallenge 2010-10-15T03:33:14 *** espes has joined #aichallenge 2010-10-15T03:35:58 *** sigh has quit IRC (Remote host closed the connection) 2010-10-15T03:43:19 *** synmnky has joined #aichallenge 2010-10-15T03:45:10 *** Florian95 has joined #aichallenge 2010-10-15T03:45:19 *** jukkerknott has quit IRC (Quit: Leaving) 2010-10-15T03:48:10 *** yasith_ has quit IRC (Ping timeout: 252 seconds) 2010-10-15T03:50:42 hmmm, building redistribution into the game tree is hard 2010-10-15T03:52:46 *** smellyhippy has joined #aichallenge 2010-10-15T04:01:13 *** yasith_ has joined #aichallenge 2010-10-15T04:01:22 Top 10 players: dmj111(3576), MariusCG(3556), felixcoto(3529), rsergio(3515), Hazard(3506), deepblue(3493), savchenko(3462), ruilov(3433), evja08(3417), ofrias(3409) 2010-10-15T04:09:54 *** hellman has quit IRC (Remote host closed the connection) 2010-10-15T04:10:55 *** boegel has joined #aichallenge 2010-10-15T04:18:32 *** hellman has joined #aichallenge 2010-10-15T04:27:38 *** hellman has quit IRC (Remote host closed the connection) 2010-10-15T04:27:59 *** hellman has joined #aichallenge 2010-10-15T04:31:36 *** HakanD has joined #aichallenge 2010-10-15T04:35:58 morning 2010-10-15T04:36:25 evening 2010-10-15T04:40:35 *** hellman has quit IRC (Remote host closed the connection) 2010-10-15T04:49:28 *** Naktibalda has joined #aichallenge 2010-10-15T05:01:23 Top 10 players: dmj111(3559), felixcoto(3523), MariusCG(3517), deepblue(3485), Hazard(3485), rsergio(3477), savchenko(3414), Manwe56(3413), evja08(3411), ruilov(3411) 2010-10-15T05:01:25 *** Olathe has joined #aichallenge 2010-10-15T05:01:39 @rankings 15 2010-10-15T05:01:40 HakanD: Top 15 players: dmj111(3559), felixcoto(3523), MariusCG(3517), deepblue(3485), Hazard(3485), rsergio(3477), savchenko(3414), Manwe56(3413), evja08(3411), ruilov(3411), ofrias(3387), davidjliu(3387), McLeopold(3362), EBraun(3347), _iouri_(3347) 2010-10-15T05:07:40 *** sigh has joined #aichallenge 2010-10-15T05:13:46 i almost have the perfect elo score 2010-10-15T05:14:00 3 twenty 6 is kind of a novalty number between a group for friends 2010-10-15T05:14:03 and i have 3207 2010-10-15T05:14:23 group of friends* 2010-10-15T05:22:03 *** Olathe has quit IRC (Ping timeout: 276 seconds) 2010-10-15T05:22:55 are the maps used on the tcp server available somewhere? 2010-10-15T05:24:21 mega1: yes, http://72.44.46.68/getmaps 2010-10-15T05:24:28 thanks 2010-10-15T05:24:56 *** yasith_ has quit IRC (Ping timeout: 265 seconds) 2010-10-15T05:24:57 there's a download of a tarball with all the maps, I'll see if I canfind it 2010-10-15T05:25:53 http://72.44.46.68/maps.tar.gz 2010-10-15T05:27:31 dhartmei mentioned create tarball of game log too, didn't he? 2010-10-15T05:27:39 *creating tarballs 2010-10-15T05:28:00 last I heard it was too big and he didn't have anywhere to host it 2010-10-15T05:28:50 a pity 2010-10-15T05:40:40 *** yasith_ has joined #aichallenge 2010-10-15T05:43:41 *** Meatkat has joined #aichallenge 2010-10-15T05:46:16 *** hellman has joined #aichallenge 2010-10-15T05:55:58 *** espes has quit IRC (Ping timeout: 240 seconds) 2010-10-15T05:56:31 *** davidd has joined #aichallenge 2010-10-15T05:56:36 hello 2010-10-15T06:00:26 According to http://ai-contest.com/game_stats.php, there are no cloud servers running at the moment. All cloud servers "died" yesterday and never came back up, so there is only a single slow server playing games (only 5-6 games per minute for thousands of bots). 2010-10-15T06:00:37 http://ai-contest.com/forum/viewtopic.php?f=18&t=926 <- any news on this"? 2010-10-15T06:01:23 Top 10 players: dmj111(3525), felixcoto(3497), MariusCG(3494), Hazard(3470), rsergio(3451), deepblue(3448), Manwe56(3388), savchenko(3383), evja08(3377), ofrias(3366) 2010-10-15T06:02:17 why after reuploading your code is your ranking reset to unranked, instead of leaving it around the level of your last upload? 2010-10-15T06:04:29 !seen DanielVF 2010-10-15T06:04:29 HakanD: DanielVF was last seen in #aichallenge 1 week, 3 days, 8 hours, 23 minutes, and 28 seconds ago: * DanielVF out. :( 2010-10-15T06:10:06 *** Appleman1234 has quit IRC (Ping timeout: 255 seconds) 2010-10-15T06:15:38 *** zorane has joined #aichallenge 2010-10-15T06:22:31 davidd: were they allive yesterday? they were dead since Tuesday 2010-10-15T06:24:29 davidd: The way the elo score works it seemed straight forward to put every new entry in the middle and watch it climb or fall. Someone said he might be doing something about it. But with the recent rate of changes I don't think that will happen ^^. 2010-10-15T06:27:00 *** jesionaj has quit IRC (Read error: Connection reset by peer) 2010-10-15T06:27:07 I bet if they made it so you start at say 30 your entry will still jump a lot, maybe even far above rank 1 for a while until the score becomes more reliable. 2010-10-15T06:27:46 *** Frontier has left #aichallenge 2010-10-15T06:35:12 *** Frontier has joined #aichallenge 2010-10-15T06:45:48 *** andrew-101 has joined #aichallenge 2010-10-15T06:46:52 *** phreeza has joined #aichallenge 2010-10-15T06:48:00 if one starts counting moves from 0, is 199 the last turn or 200? 2010-10-15T06:48:46 199 2010-10-15T06:49:06 should be 199 2010-10-15T06:55:23 *** Bobng has joined #aichallenge 2010-10-15T06:59:05 *** andrew-101 has quit IRC (Quit: Leaving.) 2010-10-15T06:59:54 *** andrew-101 has joined #aichallenge 2010-10-15T07:01:23 Top 10 players: dmj111(3489), felixcoto(3468), MariusCG(3464), deepblue(3443), Hazard(3439), rsergio(3420), Manwe56(3370), ofrias(3366), evja08(3358), savchenko(3351) 2010-10-15T07:14:46 *** Utkarsh has quit IRC (Read error: Connection reset by peer) 2010-10-15T07:22:14 *** Utkarsh has joined #aichallenge 2010-10-15T07:27:05 *** Utkarsh has quit IRC (Ping timeout: 272 seconds) 2010-10-15T07:28:13 *** narnach has joined #aichallenge 2010-10-15T07:28:23 *** Utkarsh has joined #aichallenge 2010-10-15T07:36:43 *** oPless has joined #aichallenge 2010-10-15T07:39:44 *** Meatkat has quit IRC (Ping timeout: 252 seconds) 2010-10-15T07:42:20 *** BtbN has joined #aichallenge 2010-10-15T07:43:14 *** hellman has quit IRC (Remote host closed the connection) 2010-10-15T07:50:09 *** Cyndre has quit IRC (Quit: Leaving) 2010-10-15T08:01:26 Top 10 players: dmj111(3484), MariusCG(3455), felixcoto(3439), deepblue(3420), Hazard(3419), rsergio(3403), evja08(3352), Manwe56(3352), EBraun(3347), ofrias(3335) 2010-10-15T08:05:32 *** krokokrusa has joined #aichallenge 2010-10-15T08:06:02 *** krokokrusa has joined #aichallenge 2010-10-15T08:08:17 *** Appleman1234 has joined #aichallenge 2010-10-15T08:15:15 *** Olathe has joined #aichallenge 2010-10-15T08:15:22 i foresee a lot of debugging in my future 2010-10-15T08:15:32 once i even get to a testing point, it seems so far away 2010-10-15T08:16:14 *** ichti has joined #aichallenge 2010-10-15T08:16:30 @predict 2010-10-15T08:16:30 I predict that davidd will win! 2010-10-15T08:16:35 @predict 2010-10-15T08:16:35 I predict that acieroid` will win! 2010-10-15T08:16:43 pls decide 2010-10-15T08:16:55 @roulette 2010-10-15T08:16:55 HakanD: *click* 2010-10-15T08:18:04 :) 2010-10-15T08:18:09 @roulette 2010-10-15T08:18:09 rabidus: *click* 2010-10-15T08:19:47 not roulette again 2010-10-15T08:20:10 lamer 2010-10-15T08:20:11 :) 2010-10-15T08:20:22 @roulette 2010-10-15T08:20:22 delt0r___: *click* 2010-10-15T08:20:24 i'ma win damnit 2010-10-15T08:20:32 :P 2010-10-15T08:21:43 *** Cyndre has joined #aichallenge 2010-10-15T08:22:07 3/6 2010-10-15T08:22:27 it is 1:2 odds now 2010-10-15T08:22:31 1/3 probability 2010-10-15T08:23:52 @roulette 2010-10-15T08:23:52 *** dmj111 was kicked by contestbot (BANG!) 2010-10-15T08:23:53 * contestbot reloads and spins the chambers. 2010-10-15T08:24:02 :( 2010-10-15T08:24:06 *** dmj111 has joined #aichallenge 2010-10-15T08:24:09 :) 2010-10-15T08:24:19 i've been waiting for a chance to try that :) 2010-10-15T08:24:25 :P 2010-10-15T08:24:32 so i won? 2010-10-15T08:24:35 @roulette 2010-10-15T08:24:35 HakanD: *click* 2010-10-15T08:24:39 o ou 2010-10-15T08:24:41 @roulette 2010-10-15T08:24:41 rabidus: *click* 2010-10-15T08:24:45 win (: 2010-10-15T08:25:23 @roulette 2010-10-15T08:25:23 *** dmj111 was kicked by contestbot (BANG!) 2010-10-15T08:25:24 * contestbot reloads and spins the chambers. 2010-10-15T08:25:32 *** dmj111 has joined #aichallenge 2010-10-15T08:25:36 *** ratatata has joined #aichallenge 2010-10-15T08:25:41 if dmj then BANG 2010-10-15T08:25:53 lol :D 2010-10-15T08:26:01 hehe 2010-10-15T08:26:05 better odds now... 2010-10-15T08:26:05 @roulette 2010-10-15T08:26:05 HakanD: *click* 2010-10-15T08:26:05 @roulette 2010-10-15T08:26:06 dmj111: *click* 2010-10-15T08:26:08 phew 2010-10-15T08:26:09 @roulette 2010-10-15T08:26:09 rabidus: *click* 2010-10-15T08:26:12 *phew* 2010-10-15T08:26:28 @roulette 2010-10-15T08:26:28 davidd: *click* 2010-10-15T08:29:06 *** iris1 has quit IRC (Remote host closed the connection) 2010-10-15T08:33:28 (bang) 2010-10-15T08:33:30 @roulette 2010-10-15T08:33:30 *** oPless was kicked by contestbot (BANG!) 2010-10-15T08:33:30 *** oPless has joined #aichallenge 2010-10-15T08:33:31 * contestbot reloads and spins the chambers. 2010-10-15T08:33:38 @roulette 2010-10-15T08:33:38 *** phreeza was kicked by contestbot (BANG!) 2010-10-15T08:33:39 * contestbot reloads and spins the chambers. 2010-10-15T08:33:47 * oPless grins 2010-10-15T08:33:53 @roulette 2010-10-15T08:33:53 oPless: *click* 2010-10-15T08:34:07 *** phreeza has joined #aichallenge 2010-10-15T08:34:13 @roulette 2010-10-15T08:34:13 oPless: *click* 2010-10-15T08:34:14 lol 2010-10-15T08:34:22 @roulette 2010-10-15T08:34:22 oPless: *click* 2010-10-15T08:34:26 how many bullets are there? 2010-10-15T08:34:34 6 chambers usually 2010-10-15T08:34:40 one bullet 2010-10-15T08:34:41 and 1 bullet? 2010-10-15T08:34:42 ok 2010-10-15T08:34:45 reload on fire 2010-10-15T08:35:05 I feel lucky 2010-10-15T08:35:08 @roulette 2010-10-15T08:35:08 *** oPless was kicked by contestbot (BANG!) 2010-10-15T08:35:08 *** oPless has joined #aichallenge 2010-10-15T08:35:09 * contestbot reloads and spins the chambers. 2010-10-15T08:35:14 or not :) 2010-10-15T08:35:31 and it doesn't spin again after the shot is fired? 2010-10-15T08:35:40 ie the probablility goes up? 2010-10-15T08:35:43 clearly not 2010-10-15T08:35:48 hehe 2010-10-15T08:35:54 @roulette 2010-10-15T08:35:54 phreeza: *click* 2010-10-15T08:36:01 @roulette 2010-10-15T08:36:01 *** phreeza was kicked by contestbot (BANG!) 2010-10-15T08:36:02 * contestbot reloads and spins the chambers. 2010-10-15T08:36:02 which isnt the right way to play imho 2010-10-15T08:36:14 @roulette 2010-10-15T08:36:14 oPless: *click* 2010-10-15T08:36:15 @roulette 2010-10-15T08:36:15 *** oPless was kicked by contestbot (BANG!) 2010-10-15T08:36:15 *** oPless has joined #aichallenge 2010-10-15T08:36:16 * contestbot reloads and spins the chambers. 2010-10-15T08:36:19 @roulette 2010-10-15T08:36:19 oPless: *click* 2010-10-15T08:36:21 *** phreeza has joined #aichallenge 2010-10-15T08:36:29 funny 2010-10-15T08:36:30 try now :) 2010-10-15T08:36:35 @roulette 2010-10-15T08:36:35 phreeza: *click* 2010-10-15T08:36:38 @roulette 2010-10-15T08:36:38 oPless: *click* 2010-10-15T08:36:41 @roulette 2010-10-15T08:36:41 phreeza: *click* 2010-10-15T08:36:43 @roulette 2010-10-15T08:36:43 *** oPless was kicked by contestbot (BANG!) 2010-10-15T08:36:44 *** oPless has joined #aichallenge 2010-10-15T08:36:45 * contestbot reloads and spins the chambers. 2010-10-15T08:36:45 :p 2010-10-15T08:36:48 :D 2010-10-15T08:37:16 russian roulette: the ultimate noise 2010-10-15T08:38:47 Did you know there are 41,841,412,812 possible game states only considering ownership of planets and the fact that neutrals cannot capture planets back? 2010-10-15T08:42:00 aha 2010-10-15T08:42:07 (3^21)*4 2010-10-15T08:43:25 so just enumerating them would take 38 GB o_O 2010-10-15T08:43:57 no wait, more 2010-10-15T08:44:04 damn 2010-10-15T08:44:48 how many does chess have? 2010-10-15T08:45:55 on the order of 10^43... 2010-10-15T08:50:35 i'm making game tree atm 2010-10-15T08:50:52 but i'm setting it up to only pick a certain number of moves at each node 2010-10-15T08:52:14 ass to mouth 2010-10-15T08:52:21 wat 2010-10-15T08:53:22 antimatroid: if that works it would be really cool... the tricky thing is which moves to choose i guess 2010-10-15T08:54:44 i was thinking of doing that and just choosing moves at random but i have a feeling that won't do... 2010-10-15T08:54:53 theres just too many 2010-10-15T08:55:19 *** andrew-101 has left #aichallenge 2010-10-15T08:56:07 yeah, i'm being more strategic in move selection than that 2010-10-15T08:56:16 hopefully it will work 2010-10-15T08:56:44 it's basically trying to do a mix of good moves favouring different things, and mixing up selections of some 2010-10-15T08:59:12 I have the skeleton of a c++ bot with game tree search, but it doesn't really do anything at the moment. and i have my python bot which is a bag of heuristics. maybe i'll work some more on the game tree one, too 2010-10-15T08:59:53 and try to come up with a smarter selection criterion 2010-10-15T09:01:23 Top 10 players: dmj111(3488), MariusCG(3474), felixcoto(3419), Hazard(3384), rsergio(3374), deepblue(3373), evja08(3349), EBraun(3347), ruilov(3291), ofrias(3290) 2010-10-15T09:01:36 yeah, mine is in c++, it's like 3/4 constructed, but i expect to have a huge amount of debugging too 2010-10-15T09:01:46 i can't really incrementally code it other than check that it compiles 2010-10-15T09:01:52 i have watched a few games of deepblue 2010-10-15T09:02:00 i think it is the favorite atm 2010-10-15T09:02:07 at the moment rabidus 2010-10-15T09:02:25 i don't think much can be said yet, i'm pretty sure all bots are still heuristic based 2010-10-15T09:02:48 some people seem to be testing out some ideas finally though 2010-10-15T09:02:53 have you watched the top bots? 2010-10-15T09:03:00 albert is doing game tree search i think but it sucks a lot more than his old heuristics bot 2010-10-15T09:03:04 not recently admitidly 2010-10-15T09:03:15 so am I :( 2010-10-15T09:03:19 if i just could write into file @ server...... 2010-10-15T09:03:22 its not working out yet lalala 2010-10-15T09:03:32 i decided against alpha beta because it assumes you have a small number of possible moves 2010-10-15T09:03:37 Error323: searching the tree? 2010-10-15T09:03:46 phreeza: yeah, alpha beta 2010-10-15T09:04:18 if you are thinking of choosing moves at random you might want to look at the UCT algorithm 2010-10-15T09:04:21 how deep do you get? 2010-10-15T09:04:38 sigh: yea thats what i am planning to do 2010-10-15T09:04:45 well, a more interesting and difficult question is how many actions are there :P 2010-10-15T09:04:53 the branching factor is massive 2010-10-15T09:04:58 phreeza: cool, interesting to see how it turns out 2010-10-15T09:05:07 phreeza: have you used that algorithm before? 2010-10-15T09:05:11 sigh: with some offline knowledge though, i think vanilla will still not work 2010-10-15T09:05:14 but atm depth 2-3 2010-10-15T09:05:22 nope 2010-10-15T09:05:32 however i'm ranked somewhere around 3K lol cuz the timer is set too agressive 2010-10-15T09:05:38 but I'm rethinking some stuff 2010-10-15T09:05:42 Error323: ah thats a shame 2010-10-15T09:05:52 delt0r___: JBotManager replays have become visually very pleasing :) 2010-10-15T09:06:09 so you time out mostly or don't calculate enough? 2010-10-15T09:06:22 i currently timeout every time 2010-10-15T09:06:27 sigh: have you used it before? 2010-10-15T09:06:32 Error323: how many moves are you considering at each node? 2010-10-15T09:06:58 well that is a good question ;) 2010-10-15T09:07:00 a1k0n was getting like 30 layers with tron, albeit less moves still quite deep 2010-10-15T09:07:07 but I cant share everything :P 2010-10-15T09:07:20 sigh: theres an open source Go engine that uses it... was consdering maybe using some of their code 2010-10-15T09:07:52 or at least see what optimizations they have 2010-10-15T09:09:01 *** ratatata has quit IRC (Ping timeout: 265 seconds) 2010-10-15T09:10:50 *** Appleman1234 has quit IRC (Quit: Leaving) 2010-10-15T09:12:16 *** delt0r_ has joined #aichallenge 2010-10-15T09:12:55 *** delt0r___ has quit IRC (Read error: Operation timed out) 2010-10-15T09:14:23 *** wvdschel_ has joined #aichallenge 2010-10-15T09:18:29 morning 2010-10-15T09:18:49 phreeza: no, I have not 2010-10-15T09:19:08 and I've never seen it used outside of Go either, which is why I'm interested 2010-10-15T09:19:26 it's used in Hex and some board games 2010-10-15T09:19:57 i see 2010-10-15T09:20:10 *** shades has quit IRC (Ping timeout: 260 seconds) 2010-10-15T09:28:29 *** Accoun has quit IRC (Read error: Connection reset by peer) 2010-10-15T09:34:23 *** mceier has joined #aichallenge 2010-10-15T09:39:22 *** nann has quit IRC (Read error: Connection reset by peer) 2010-10-15T09:44:45 hah interesting flashing lights game http://72.44.46.68/canvas?game_id=500536 2010-10-15T09:45:16 sigh: I saw it being used for an RTS somewhere 2010-10-15T09:46:30 sigh: http://ijcai.org/papers09/Papers/IJCAI09-018.pdf 2010-10-15T09:50:28 *** Accoun has joined #aichallenge 2010-10-15T09:51:17 phreeza: thanks, I'll have a look at that later 2010-10-15T09:51:23 right now, I must sleep 2010-10-15T09:51:25 aerique: Yea--i was bored.. 2010-10-15T09:51:36 *** sigh has quit IRC (Quit: Leaving...) 2010-10-15T09:57:19 bhasker: what do you mean by flashing lights? 2010-10-15T09:57:39 ah just that it looks like lights flashing when he hits my planets 2010-10-15T09:57:49 its like a display of lights 2010-10-15T09:57:57 when he hits and i take it back in a couple of turns 2010-10-15T09:58:57 ah ok :) 2010-10-15T09:58:59 nice game 2010-10-15T10:01:24 Top 10 players: dmj111(3496), MariusCG(3454), felixcoto(3397), Hazard(3367), rsergio(3343), evja08(3340), deepblue(3340), EBraun(3330), Manwe56(3329), _iouri_(3304) 2010-10-15T10:11:30 *** HakanD has quit IRC (Ping timeout: 240 seconds) 2010-10-15T10:13:21 *** espes has joined #aichallenge 2010-10-15T10:19:03 *** JamesMG has joined #aichallenge 2010-10-15T10:23:12 *** Blkt has joined #aichallenge 2010-10-15T10:32:15 *** rebelxt_ has joined #aichallenge 2010-10-15T10:33:26 *** Blkt has quit IRC (Ping timeout: 265 seconds) 2010-10-15T10:35:11 *** rebelxt_ has quit IRC (Client Quit) 2010-10-15T10:42:10 *** Apophis has joined #aichallenge 2010-10-15T10:54:45 *** area has quit IRC (Quit: Leaving) 2010-10-15T11:01:16 *** boegel has quit IRC (Read error: Operation timed out) 2010-10-15T11:01:25 Top 10 players: dmj111(3504), MariusCG(3460), felixcoto(3369), Hazard(3357), EBraun(3347), deepblue(3341), Manwe56(3327), evja08(3326), _iouri_(3319), rsergio(3305) 2010-10-15T11:02:03 *** Sylph2 has quit IRC (Ping timeout: 264 seconds) 2010-10-15T11:02:17 *** wvdschel_ is now known as yukito 2010-10-15T11:04:00 *** aerique has quit IRC (Quit: ...) 2010-10-15T11:09:44 *** Sylph has joined #aichallenge 2010-10-15T11:14:17 *** hornairs has joined #aichallenge 2010-10-15T11:24:05 *** rebelxt has joined #aichallenge 2010-10-15T11:25:09 the starcraft ai competition results were announced yesterday 2010-10-15T11:25:10 http://eis-blog.ucsc.edu/2010/10/starcraft-ai-competition-results/ 2010-10-15T11:25:17 amstan: hey Alex, you there? 2010-10-15T11:25:29 ? 2010-10-15T11:25:56 say no 2010-10-15T11:27:43 *** fluffluff1 has quit IRC (Quit: Leaving.) 2010-10-15T11:29:52 huh 2010-10-15T11:32:25 *** bhasker has quit IRC (Ping timeout: 264 seconds) 2010-10-15T11:32:53 gah, i'm going to sleep 2010-10-15T11:32:57 will keep slogging later 2010-10-15T11:33:11 i have most of the stuff set uo 2010-10-15T11:34:16 *** hornairs has quit IRC (Remote host closed the connection) 2010-10-15T11:34:23 amstan: thinking if i should re-submit or not.. any chance cloud servers will be up soon? if not, i might hold off since it would take a week to get back to the top.. 2010-10-15T11:35:10 phreeza: that pdf on starcraft AI was a good read 2010-10-15T11:35:35 quote on runner-up bot: "The bot uses primarily states machines to make decisions and it can recognize the strategy of its opponent." 2010-10-15T11:36:03 i have no idea about cloud servers 2010-10-15T11:36:14 rebelxt: pdf you mean the presentation? 2010-10-15T11:36:28 or the UCT paper i posted earlier? 2010-10-15T11:36:35 phreeza: yeah, linked on the page you posted a link to 2010-10-15T11:37:04 aha, well yea the info was a bit sparse, haven't had any time to look for more detailed descriptions 2010-10-15T11:38:02 i think many of the top heuristic bots now also use state machines, perhaps implicitly 2010-10-15T11:38:04 yeah, slide 8 of http://eis.ucsc.edu/sites/default/files/AIIDECompetition.pdf is a great list of AI techniques bots had to use 2010-10-15T11:39:07 I only understand about half of those terms 2010-10-15T11:39:11 *** bhasker has joined #aichallenge 2010-10-15T11:39:30 good starting points for literature research 2010-10-15T11:39:51 scripting and dynamic scripting are interesting techniques :) 2010-10-15T11:39:52 potential fields... sounds interesting 2010-10-15T11:40:29 Naktibalda: what do they mean? fixed action sequences? 2010-10-15T11:40:55 i tried to use potential fields from the very beginning :) 2010-10-15T11:40:56 so sniping for example would be a script in our case? 2010-10-15T11:41:35 phreeza: same here. i'm looking up techniques what i have not heard about before 2010-10-15T11:44:09 *** yukito has quit IRC (Remote host closed the connection) 2010-10-15T12:01:26 Top 10 players: dmj111(3426), MariusCG(3414), felixcoto(3323), EBraun(3312), deepblue(3311), Hazard(3288), Manwe56(3283), ruilov(3271), savchenko(3252), rsergio(3237) 2010-10-15T12:08:32 *** boegel has joined #aichallenge 2010-10-15T12:08:38 *** hellman has joined #aichallenge 2010-10-15T12:08:43 *** McLeopold has joined #aichallenge 2010-10-15T12:09:26 http://72.44.46.68/canvas?game_id=497410 2010-10-15T12:14:43 impressively swift 2010-10-15T12:15:15 http://72.44.46.68/canvas?game_id=496433 this one is faster 2010-10-15T12:17:38 *** bhasker has quit IRC (Quit: bhasker) 2010-10-15T12:18:02 McLeopold: can you start AntBot? or is it still in the works? 2010-10-15T12:18:20 It's running, but only every 30 minutes. 2010-10-15T12:19:03 McLeopold: i want to know what that scenario you came up with was 2010-10-15T12:19:04 ahh, my bots have not played with it yet 2010-10-15T12:20:00 okay, give me a second to make sure I have it right... 2010-10-15T12:21:32 *** Itkovian has quit IRC (Quit: Itkovian) 2010-10-15T12:29:09 thanks 2010-10-15T12:32:13 *** Mathnerd314 has quit IRC (Ping timeout: 265 seconds) 2010-10-15T12:38:23 http://pastebin.com/2CSkuKne 2010-10-15T12:38:26 sorry for the wait 2010-10-15T12:39:02 You'll need to analyze that carefully. Let me know when you've got it. 2010-10-15T12:40:50 That's the most brilliant map ever designed. 2010-10-15T12:41:28 Actually, it's still mostly a draw. But it's close. 2010-10-15T12:41:36 Wait, no it's not... 2010-10-15T12:41:53 ... 2010-10-15T12:42:03 * jmcarthur computes a distance matrix for it 2010-10-15T12:42:23 DEBUG - 0 1 2 3 2010-10-15T12:42:23 DEBUG - 0 0 5 2 7 2010-10-15T12:42:23 DEBUG - 1 0 7 2 2010-10-15T12:42:23 DEBUG - 2 0 9 2010-10-15T12:42:23 DEBUG - 3 0 2010-10-15T12:42:48 sweet thanks 2010-10-15T12:42:50 5 to opponent, 2 back to neutral 2010-10-15T12:46:28 That map is to produce an RPS situation? 2010-10-15T12:46:35 jmcarthur: if it looks good, I'm going to post on the strategy forum. 2010-10-15T12:46:46 phreeze: yes, or close to it 2010-10-15T12:47:18 it's more of an odd/even contest 2010-10-15T12:47:40 with the option to not throw a number 2010-10-15T12:48:34 i'm gonna spend a little time on this 2010-10-15T12:49:03 wow 2010-10-15T12:49:27 I may need to put a neutral in the center 2010-10-15T12:49:29 i'm not familiar with that game exactly 2010-10-15T12:49:49 but i can imagine 2010-10-15T12:49:57 very interesting 2010-10-15T12:49:59 odd/even? just like rps, but you hold out 1 or 2 figures... 2010-10-15T12:50:20 ...and someone calls odd or even at the beginning... 2010-10-15T12:50:33 ...then you total the numbers and see if it is odd or even. 2010-10-15T12:51:58 *fingers* 2010-10-15T12:52:40 does this ELO system work correctly, in rock-paper-scissors situation ? 2010-10-15T12:53:18 1st place should be taken by 3 bots simultaneously, right? 2010-10-15T12:54:46 it was taken by 3 bots yesterday 2010-10-15T12:55:05 *** JCS^ has joined #aichallenge 2010-10-15T12:55:55 *** rebelxt has left #aichallenge 2010-10-15T12:57:56 McLeopold: i believe the dominant strategy is for a draw 2010-10-15T12:58:11 krokokrusa: that's not really how the math works, even for rps 2010-10-15T12:58:20 McLeopold: where both players take their near neutrals 2010-10-15T12:58:53 McLeopold: if only one player does it, that player wins 2010-10-15T12:59:00 jmcarthur: what happens if you take a neutral and I attack immediately? 2010-10-15T12:59:58 I do think I need to add a neutral to make it perfect... 2010-10-15T13:00:03 in the center 2010-10-15T13:00:19 but that means I have to tamper with the growth rates and starting ships... 2010-10-15T13:00:34 ah snap you're right. you can lose by taking your neutral and being attacked 2010-10-15T13:00:39 :) 2010-10-15T13:01:00 attack beats neutral, neutral beats hold, hold sucks... 2010-10-15T13:01:23 That's why I need a neutral, so that holding can beat an attack. 2010-10-15T13:01:26 Top 10 players: dmj111(3417), MariusCG(3391), deepblue(3328), Hazard(3304), EBraun(3289), felixcoto(3272), ruilov(3253), ofrias(3246), Manwe56(3243), savchenko(3228) 2010-10-15T13:01:30 McLeopold: so it looks like the best strategy for both players is to pick randomly between staying and attacking the other's home planet 2010-10-15T13:01:55 Well, attacking guarantees a draw, but not a win. 2010-10-15T13:01:59 McLeopold: and i think all combinations of those result in a draw 2010-10-15T13:02:12 right. there is no win if both play optimally 2010-10-15T13:02:35 If I can get hold to beat attack, I've got it. 2010-10-15T13:02:47 *** Migi32 has joined #aichallenge 2010-10-15T13:03:28 yeah, that would put the optimal strategy at picking randomly between holding, attacking, or taking neutral 2010-10-15T13:03:46 and the winner would be nondeterministic 2010-10-15T13:04:17 Only on some maps. If I can figure this out, I wonder if a map generator could always produce these types. 2010-10-15T13:04:42 it would be a draw 1 out of 3 times 2010-10-15T13:06:36 i don't see how you could make holding beat attacking 2010-10-15T13:07:35 *** hellman has quit IRC (Remote host closed the connection) 2010-10-15T13:10:35 hmm... i guess with a neutral in the middle you could have a sniping strategy of some sort take over if your opponent attacks while you hold 2010-10-15T13:10:37 holding leaves you with 5 free ships 2010-10-15T13:10:53 yes, but your opponent also gains 5 2010-10-15T13:11:09 so you're back at square one 2010-10-15T13:11:18 but his won't be available until turn 5, yours are available now 2010-10-15T13:11:28 unless he's streaming 2010-10-15T13:12:09 which he should do if he sees you holding, i think 2010-10-15T13:12:22 a neutral 4 would still go to the person that holds, it just messes up the other scenarios. 2010-10-15T13:12:49 at one point can the holder take a center neutral if the opponent is streaming? 2010-10-15T13:13:01 *at what point 2010-10-15T13:13:19 *** ichti has quit IRC (Ping timeout: 276 seconds) 2010-10-15T13:13:20 maybe a 4 neutral for both sides, that's far away 2010-10-15T13:13:52 seems like taking that would become the dominant strategy then 2010-10-15T13:14:07 no, it should still lose to an attack 2010-10-15T13:14:30 sorry, 2 neutrals, 1 each side, behind the 7 neutral 2010-10-15T13:14:39 right 2010-10-15T13:14:48 I think that would work! 2010-10-15T13:14:57 what growth and distance, and how many neutral ships? 2010-10-15T13:16:00 no, it won't work 2010-10-15T13:16:10 also, i still don't see when you would get a chance to take that after holding 2010-10-15T13:16:15 if you can hold against an attack and get a neutral, that is always best 2010-10-15T13:16:34 it would have to be 1 in the center 2010-10-15T13:17:15 have to adjust all the distances for that now 2010-10-15T13:18:17 i think it might work though 2010-10-15T13:18:40 *** bhasker has joined #aichallenge 2010-10-15T13:18:44 but the center neutral needs a big growth 2010-10-15T13:19:50 it works because it's *only* worth taking the center if you hold and your opponent either attacks or tries to take the center 2010-10-15T13:19:55 *** Bobng has quit IRC (Remote host closed the connection) 2010-10-15T13:20:05 no, the center neutral would have to have less growth than the others 2010-10-15T13:20:05 otherwise you can just get stung 2010-10-15T13:20:22 *** Bobng has joined #aichallenge 2010-10-15T13:20:25 So, I need everything else to have more growth, then I can add the center neutral 2010-10-15T13:20:38 it would have to be enough to counter an attack though 2010-10-15T13:20:56 ? 2010-10-15T13:21:16 *** amstan has quit IRC (Ping timeout: 276 seconds) 2010-10-15T13:21:28 enough to counter an attack at home 2010-10-15T13:21:46 with a center neutral, there is a forth strategy, which I need to make sure has the same outcome as holding 2010-10-15T13:21:48 especially if your home or neutral have a larger growth 2010-10-15T13:22:42 going for the center has to be less desireable than going for the back neutral, thus, the back neutral must have larger growth 2010-10-15T13:23:01 the center is fairly undesirable anyway due to sniping 2010-10-15T13:23:23 regardless of growth 2010-10-15T13:23:24 right, but I need to make sure that neutral beats hold/center 2010-10-15T13:23:56 it doesn't take all your ships to take your near neutral, so you could still snipe an attempt to take the center 2010-10-15T13:24:16 on turn 1? 2010-10-15T13:24:22 *** boegel has quit IRC (Ping timeout: 245 seconds) 2010-10-15T13:24:31 sniping only works on turn 2 2010-10-15T13:24:46 if the center is two turns from each home then you can snipe on turn 2 after moving to take your near neutral on turn 1 2010-10-15T13:25:00 so there is still no motivation to take the center unless your opponent attacks 2010-10-15T13:25:04 so, your left over ships if you take the back neutral must be less than a snipe attempt 2010-10-15T13:25:16 less? 2010-10-15T13:25:32 no i think the snipe is what makes the center neutral feasible for this 2010-10-15T13:25:51 the center of this map would be 3 from each start 2010-10-15T13:25:54 since it makes the neutral only desireable after you hold and your opponent attacks 2010-10-15T13:26:21 I think it might should be offset up or down enough to keep it out of the way 2010-10-15T13:28:42 you have 2 left after taking neutral, if you opponent attacks, you are dead, no sniping 2010-10-15T13:28:59 if your opponent doesn't attack, you can send 2 for a snipe 2010-10-15T13:29:22 which would beat the opponent going straight for the center neutral 2010-10-15T13:29:25 with 5 2010-10-15T13:29:44 If he went with 6 or more? we need that to be a loss 2010-10-15T13:30:13 so the neutral needs to be far enough away that taking the back neutral is faster 2010-10-15T13:30:44 i'm saying we can make center/hold be a loss, center/attack be a win, center/near be a loss, and center/center of course be a draw 2010-10-15T13:30:48 but not so far that holding and taking it would hurt 2010-10-15T13:30:59 where win/loss/draw is for the first player in those combos 2010-10-15T13:31:58 and if it's that far, it needs a reasonably nice growth to be worth it i think 2010-10-15T13:32:27 no, just 1 more than opponent so you have more ships 2010-10-15T13:33:09 it needs to be not worth it at all, unless you opponent tries an all out attack 2010-10-15T13:33:09 your opponent would have both homes and access to both near neutrals if you can't counterattack after taking center 2010-10-15T13:33:13 *** Mathnerd314 has joined #aichallenge 2010-10-15T13:34:11 we may be trying to solve this different ways. Let's both come up with a new map. 2010-10-15T13:34:22 i'm suspecting that as well 2010-10-15T13:34:28 if we can come up with two cases that's even better :D 2010-10-15T13:34:44 we can run a mini contest! 2010-10-15T13:35:07 okay, thinking... 2010-10-15T13:36:12 *** oPless has quit IRC (Quit: This computer has gone to sleep) 2010-10-15T13:45:59 the center neutral makes things a lot more complicated... 2010-10-15T13:51:12 *** narnach1 has joined #aichallenge 2010-10-15T13:51:28 *** narnach has quit IRC (Quit: Leaving.) 2010-10-15T13:51:48 *** jkl__ has joined #aichallenge 2010-10-15T13:52:46 *** narnach1 is now known as Narnach 2010-10-15T13:52:58 *** rebelxt has joined #aichallenge 2010-10-15T13:53:07 my idea won't work. if i make the center growth too little then it's never worth taking and if i make it worth taking then it dominates everything except for holding :\ 2010-10-15T13:53:23 and adjusting the number of neutrals there doesn't seem to make any difference 2010-10-15T13:53:39 i should really write a program to come up with this for me or something 2010-10-15T13:53:43 kind of tedious like this 2010-10-15T13:54:12 writing the program would be just as tedious i suppose 2010-10-15T13:55:13 *** boegel has joined #aichallenge 2010-10-15T13:57:21 how the heck I'm I back up to #10 on the official site? And what happened to all the 3900+ elos? 2010-10-15T13:57:21 *** jkl__ has quit IRC (Quit: Page closed) 2010-10-15T13:57:42 McLeopold: AntBot still running? i wanna start mine and get a few games in with yours 2010-10-15T13:57:56 yeah, I can step it up if you want. 2010-10-15T13:58:15 yeah please do. started mine 2010-10-15T13:58:37 rebelxt113 2010-10-15T13:59:02 how come you dropped to 36 on the official? 2010-10-15T13:59:07 McLeopold: ur bot has super powers to work its way back into top10 every time it falls out of it 2010-10-15T13:59:38 McLeopold: makes no sense to me.. i was #23 in the morning, then won 3 games and fell down, lol? 2010-10-15T14:00:09 did you notice the huge drop in everyone elo? 2010-10-15T14:00:17 The must have removed a lot of starter packs. 2010-10-15T14:00:35 my old bot just got antbotted on the TCP server 2010-10-15T14:00:36 wow, yeah you must be right 2010-10-15T14:00:42 playing JamesMG.old 2010-10-15T14:01:04 or is this ELO drop a side effect of very low game count? should not be the case 2010-10-15T14:01:10 My next bot name...AntLion 2010-10-15T14:01:23 Man, your bot is mean with those early game planet takeovers 2010-10-15T14:01:25 no, it has to be less contestants. 2010-10-15T14:01:28 Top 10 players: dmj111(3433), MariusCG(3362), deepblue(3333), Hazard(3280), EBraun(3264), Manwe56(3252), felixcoto(3230), ofrias(3207), ruilov(3202), McLeopold(3200) 2010-10-15T14:01:30 :D 2010-10-15T14:02:09 woot! back on the list! 2010-10-15T14:02:47 if they did prune inactive bots, that's awesome.. no sense in having dual bots play rage bots over and over ;-) 2010-10-15T14:02:49 JamesMG: that was a shellacking 2010-10-15T14:02:55 http://72.44.46.68/canvas?game_id=502994 2010-10-15T14:03:07 Yeah, more games for real bots. 2010-10-15T14:03:34 I wish they would allow you to remove you bot, if you know you don't want it playing. 2010-10-15T14:03:48 You kind of have the perfect counter to my bot there, with the aggressive attacks 2010-10-15T14:04:10 upload source that won't compile, it'll remove your bot 2010-10-15T14:04:18 or just try and upload C# that will compile :) 2010-10-15T14:04:39 Yeah, but upload something that always fails on turn 1... 2010-10-15T14:05:16 hehe, IssueOrder(1,1,1); 2010-10-15T14:05:23 better yet, 1,10 2010-10-15T14:05:27 err, 1,1,0 2010-10-15T14:05:51 0,0,0 2010-10-15T14:06:00 -1,-1,-1 2010-10-15T14:06:04 I like the way you think 2010-10-15T14:06:09 http://72.44.46.68/canvas?game_id=502908 2010-10-15T14:06:12 IssueOrder(false) 2010-10-15T14:06:13 :( 2010-10-15T14:06:14 http://72.44.46.68/canvas?game_id=503019 2010-10-15T14:06:40 well, just output something like "pass" or "fail" or "help i'm trapped in a planet wars bot" 2010-10-15T14:06:51 people are too lazy to go and remove their bot when they give up.. they could prune all bots not updated within last 3 weeks as long as their ranking is in bottom X%. don't want McLeopold to get purged! ;-) 2010-10-15T14:07:23 rebel watch those games , lets see how this variation of my bot does overall 2010-10-15T14:07:32 yeah, my bot is ancient 2010-10-15T14:07:45 bhasker: malazon0.1 isn't static? 2010-10-15T14:07:54 malazan even 2010-10-15T14:07:55 not really i am lazy to version 2010-10-15T14:08:00 sometimes i tweak 2010-10-15T14:08:05 haha, that explains a bit 2010-10-15T14:08:05 without bumping versions 2010-10-15T14:08:20 bhasker: very nice! i like what i'm seeing 2010-10-15T14:08:38 yea i changed some logic of when i launch attacks 2010-10-15T14:08:44 its a bit more aggresive and opportunistic 2010-10-15T14:08:51 earlier it was too conservative 2010-10-15T14:09:25 rebelxt: my elo might be too low for use to get paired 2010-10-15T14:10:41 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-10-15T14:10:57 It should be going up pretty quick shouldn't it? 2010-10-15T14:11:02 McLeopold: doh well i'll keep it running since your bot is climbing up 2010-10-15T14:12:01 whats new about antbot leopold? 2010-10-15T14:14:06 lots 2010-10-15T14:14:35 bhasker, i think your new bot will kick mine's butt.. you seem to have better timing on initial expansion it seems.. hope we get paired up 2010-10-15T14:14:42 lets see, 2010-10-15T14:14:47 i just beat CCCF again 2010-10-15T14:14:50 3rd time in a row 2010-10-15T14:14:53 its my lucky day 2010-10-15T14:14:54 ! 2010-10-15T14:15:23 i still have trouble with some maps were i am late in taking neutrals 2010-10-15T14:15:28 or if my neutral get sniped 2010-10-15T14:15:28 no kidding? nice! my best is 1 in a row, lol 2010-10-15T14:15:38 :D 2010-10-15T14:16:02 http://72.44.46.68/getplayer?player=malazan0.1 2010-10-15T14:16:12 3 consecutive games with CCCF :D 2010-10-15T14:16:27 *** boegel has quit IRC (Ping timeout: 245 seconds) 2010-10-15T14:17:01 its depressing though 2010-10-15T14:17:08 i lose to bots under 100 elo on some maps 2010-10-15T14:17:09 :( 2010-10-15T14:17:31 ahh.. too opportunistic? how do they manage to beat you? 2010-10-15T14:17:36 *** oPless has joined #aichallenge 2010-10-15T14:17:53 mostly neutral sniping at start 2010-10-15T14:17:55 is how i lose 2010-10-15T14:18:00 i miscalculate at start 2010-10-15T14:18:11 or send too many ships on their way too early 2010-10-15T14:18:16 so most of my ships are in flight 2010-10-15T14:18:23 while he happily snipes my neutrals 2010-10-15T14:18:27 then it just goes downhill 2010-10-15T14:19:20 and my bot doesn't do too well on split maps where planets are really far apart 2010-10-15T14:19:29 it sends too many ships back and forth 2010-10-15T14:21:01 *** Narnach has quit IRC (Quit: Leaving.) 2010-10-15T14:21:24 http://72.44.46.68/canvas?game_id=503150 whoops 2010-10-15T14:24:23 rebelxt: I'm taking AntBot offline. My elo is too low. 2010-10-15T14:24:35 hmm... or I could give it a new name... 2010-10-15T14:24:49 http://72.44.46.68/canvas?game_id=503239 :( lucky run over 2010-10-15T14:24:51 but as i said 2010-10-15T14:24:53 neutral snipe at start 2010-10-15T14:24:56 screws me over 2010-10-15T14:28:14 okay, running as AntLion with high elo... 2010-10-15T14:30:38 http://72.44.46.68/canvas?game_id=503299 I always crush a1k0n with sniping... 2010-10-15T14:35:00 bhasker: i think you got relatively small issues to figure out (not exposing urself to sniping, and first turn action), but your main algorithm is solid imho.. wish i could say this about mine at this point - it's a wreck ;) 2010-10-15T14:35:23 *** Tobu has quit IRC (Read error: Operation timed out) 2010-10-15T14:36:05 how to do first turn: http://72.44.46.68/canvas?game_id=503352 2010-10-15T14:36:26 heh 2010-10-15T14:36:34 hard to argue with that 2010-10-15T14:36:46 :D sometimes it doesn't do that good 2010-10-15T14:36:49 That throw to the center is so nice when the map calls for it 2010-10-15T14:37:04 yeah, I'm loving it 2010-10-15T14:37:14 McLeopold: yep ideal first move for this particular map 2010-10-15T14:37:19 it's like playing paintball 2010-10-15T14:37:40 *** perror has quit IRC (Quit: Bye all !) 2010-10-15T14:38:40 scoring function taking neigbour planets into account - it's so hard to get this right!... 2010-10-15T14:38:49 well what do you know 4 wins and losses to CCCF and counting 2010-10-15T14:39:09 http://72.44.46.68/canvas?game_id=503398 2010-10-15T14:39:32 ooh, guess who's #1 on tcp... 2010-10-15T14:39:59 it will drop 2010-10-15T14:40:02 ragebot? 2010-10-15T14:40:08 McRagebot 2010-10-15T14:40:28 McLeopold: still trying to catch a game with you... 2010-10-15T14:40:32 I broke my code to the point it usually draws with RageBot now 2010-10-15T14:40:40 thus why I am starting fresh 2010-10-15T14:40:52 nice work mc 2010-10-15T14:41:14 there we go, down to #6 2010-10-15T14:41:38 *** oPless has quit IRC (Quit: This computer has gone to sleep) 2010-10-15T14:42:04 *** oPless has joined #aichallenge 2010-10-15T14:42:10 is your bot not sniping in that game McLeopold? 2010-10-15T14:42:47 *** amstan has joined #aichallenge 2010-10-15T14:42:47 *** ChanServ sets mode: +o amstan 2010-10-15T14:42:54 which game? 2010-10-15T14:43:06 Nevermind, I didn't watch all of it 2010-10-15T14:43:16 the one you lost to drop you out of 1st 2010-10-15T14:43:20 *** bhasker_ has joined #aichallenge 2010-10-15T14:46:43 *** bhasker has quit IRC (Ping timeout: 240 seconds) 2010-10-15T14:47:41 *** acieroid` is now known as acieroid 2010-10-15T14:48:36 McLeopold: i just re-submitted.. going to take forever to get back up :( you gonna submit this new bot soon? 2010-10-15T14:49:14 No, too many bugs. 2010-10-15T14:49:21 *** bhasker_ has quit IRC (Ping timeout: 252 seconds) 2010-10-15T14:49:32 I had promised everyone I'd be #1 on official this week. :( 2010-10-15T14:49:43 But then CrowBot happened... 2010-10-15T14:49:47 take your time! :) 2010-10-15T14:49:54 It changed my whole outlook. 2010-10-15T14:49:55 hey, bugs did not stop me from submitting. hehe 2010-10-15T14:50:04 McLeopold, what's CrowBot? 2010-10-15T14:50:08 almost no games are played currently on off. server anyway 2010-10-15T14:50:16 makes no sense to submit now.. till servers are fixed... 2010-10-15T14:50:51 yeah, submitting is only for bragging rights right now. Or if you want to play someone not on tcp 2010-10-15T14:51:06 When I stop losing on tcp, then I'll submit. 2010-10-15T14:51:22 Ouch, loss to CCCF 2010-10-15T14:52:28 McLeopold, in that last game sending 100 to the small neutral with 3 ships wasn't the best opening move I think 2010-10-15T14:53:46 but what's "a" CrowBot? :P 2010-10-15T14:54:06 Migi32: I think it still was. It was not holding the center that was bad. 2010-10-15T14:54:10 the webcomic monster commute has crowbots 2010-10-15T14:54:27 I had no idea... 2010-10-15T14:54:38 CrowBot was build to steal planets. 2010-10-15T14:54:48 ah ok :P 2010-10-15T14:55:00 Some crows/blackbirds are known as thieves. 2010-10-15T14:55:42 should've called it ravens. Those are pesky thieves as well, but they're pretty smart 2010-10-15T14:56:26 http://www.youtube.com/watch?v=1P8Nwl7FAJk 2010-10-15T14:57:41 *** krokokrusa has left #aichallenge ("going to sleep") 2010-10-15T14:58:23 no wait in that video it's a crow :P 2010-10-15T14:59:03 *** boegel has joined #aichallenge 2010-10-15T15:00:34 *** Florian95 has quit IRC (Remote host closed the connection) 2010-10-15T15:01:27 Top 10 players: dmj111(3443), MariusCG(3372), deepblue(3339), EBraun(3292), Hazard(3274), Manwe56(3264), felixcoto(3238), ruilov(3214), savchenko(3208), ofrias(3197) 2010-10-15T15:04:13 *** Naktibalda has joined #aichallenge 2010-10-15T15:07:02 scores are dropping because there are only one server running? 2010-10-15T15:07:56 *** oPless has quit IRC (Quit: This computer has gone to sleep) 2010-10-15T15:09:46 *** phreeza_ has joined #aichallenge 2010-10-15T15:10:14 how could that be? is there score decay? 2010-10-15T15:10:26 *** shades has joined #aichallenge 2010-10-15T15:10:37 *** Itkovian has joined #aichallenge 2010-10-15T15:11:00 or perhaps just the ratio of average # games played / resubmission is dropping 2010-10-15T15:11:14 *** Snowplan has joined #aichallenge 2010-10-15T15:11:47 wasn't there a link to check the log on a TCP match 2010-10-15T15:12:08 I was searching the forums thought this was more direct 2010-10-15T15:12:42 *** phreeza has quit IRC (Ping timeout: 240 seconds) 2010-10-15T15:12:58 McLeopold: finally! wish we played on a different map though 2010-10-15T15:14:47 dmj111: can you start ur powerhouse? 2010-10-15T15:15:02 *** oPless has joined #aichallenge 2010-10-15T15:16:24 Snowplan: what do you mean? you can extract game log from the visualizer html page. but since you run the bot locally, you should have the log on your machine too 2010-10-15T15:16:55 I was just reading that thanks 2010-10-15T15:17:43 *** 13WAAKK2J <13WAAKK2J!~quassel@168.Red-83-36-227.dynamicIP.rima-tde.net> has joined #aichallenge 2010-10-15T15:17:54 rebelxt: http://72.44.46.68/canvas?game_id=503836 2010-10-15T15:18:05 *** 13WAAKK2J is now known as jorge 2010-10-15T15:19:22 *** boegel has quit IRC (Ping timeout: 245 seconds) 2010-10-15T15:19:37 your bot usually beats that one pretty quick 2010-10-15T15:20:02 JamesMG: nice - you beat me on expansion front 2010-10-15T15:21:53 mine should've attacked way earlier though 2010-10-15T15:23:40 better safe than sorry, mine counter-attacks and tries to swap positions 2010-10-15T15:24:42 *** yasith_ has quit IRC (Ping timeout: 240 seconds) 2010-10-15T15:25:05 surprised it just absorbed the hit when mine did move finally then 2010-10-15T15:26:45 yeah it seems i introduced a bug in this version... it should not have tried defending in this case 2010-10-15T15:28:54 *** RainCT has joined #aichallenge 2010-10-15T15:29:32 *** Accoun has quit IRC () 2010-10-15T15:31:58 504K games played on the TCP server = cool 2010-10-15T15:33:19 *** Tobu has joined #aichallenge 2010-10-15T15:33:28 at current rate TCP server is doing more games per minute than official server. 2010-10-15T15:33:53 *** oPless has quit IRC (Quit: This computer has gone to sleep) 2010-10-15T15:34:13 which is quite... hmm.. 2010-10-15T15:34:16 funny 2010-10-15T15:34:32 to be fair, TCP server doesn't have to run the bots themselves 2010-10-15T15:35:25 JamesMG: yeah no upload, compilation, deployment, sandbox.. 2010-10-15T15:35:32 *** ichti has joined #aichallenge 2010-10-15T15:36:07 just some TCP'ing :P 2010-10-15T15:36:25 amstan: does anyone know when Daniel will be back? 2010-10-15T15:36:44 !seen Daniel 2010-10-15T15:36:44 Migi32: I have not seen Daniel. 2010-10-15T15:36:57 !seen danielvf 2010-10-15T15:36:57 a1k0n: danielvf was last seen in #aichallenge 1 week, 3 days, 17 hours, 55 minutes, and 55 seconds ago: * DanielVF out. :( 2010-10-15T15:37:10 rebelxt: i don't think he'll be back by the end 2010-10-15T15:37:20 rebelxt: jeff is pretty much the main admin 2010-10-15T15:37:34 *** yasith_ has joined #aichallenge 2010-10-15T15:37:36 was he an admin as well? 2010-10-15T15:38:02 amstan: got it. any chance i could help out with cloud servers? i run a bunch of servers on EC2 including a TCP server 2010-10-15T15:38:34 !seen j3camero 2010-10-15T15:38:34 rebelxt: j3camero was last seen in #aichallenge 1 week, 3 days, 17 hours, 18 minutes, and 48 seconds ago: amstan: yes, but no names, haha. 2010-10-15T15:38:55 i have no idea how they work 2010-10-15T15:39:00 they left at a suspiciously similar time 2010-10-15T15:39:02 or how to ssh in there 2010-10-15T15:39:04 or so on 2010-10-15T15:41:29 *** Tobu has quit IRC (Ping timeout: 276 seconds) 2010-10-15T15:43:32 *** Accoun has joined #aichallenge 2010-10-15T15:47:51 amstan: i understand they could not just post login credentials etc on the forums ;-) oh well, we'll keep waiting for j3camero then... people who don't use tcp server are probably giving up on the contest. maybe it's a good thing? ;-) 2010-10-15T15:48:58 huh? 2010-10-15T15:49:24 amstan: some frustration on the forums is obvious now - check technical issues threads 2010-10-15T15:49:33 there's not much i can do 2010-10-15T15:49:39 i haven't see jeff on in forever 2010-10-15T15:50:11 i don't know enough about the servers and ec2 to install/fix stuff 2010-10-15T15:50:43 don't the admins see each other in real life at the university of waterloo? 2010-10-15T15:51:44 amstan: of course, i understand.. topic says - contant j3camero, which i tried doing a number of times. won't give up 2010-10-15T15:52:25 jeff graduated and got a job and doesn't seem to have time to run a contest, and then decided to kick it off anyway 2010-10-15T15:53:41 if danielvf hadn't turned up and sorted a lot of things out the website would be forever fucked by now i think 2010-10-15T15:54:25 a1k0n: this explains the situation. hope he is trying to find people willing to keep it running 2010-10-15T15:55:03 well he is when he's around. but if you want to volunteer to run the servers you are disqualified from the contest as you have access to everyone else's submissions 2010-10-15T15:56:12 a1k0n: oh, crap, i did not think of this side effect obviously 2010-10-15T15:57:12 *** bhasker has joined #aichallenge 2010-10-15T15:57:14 http://72.44.46.68/canvas?game_id=504105 2010-10-15T15:57:19 antlion goes down! 2010-10-15T15:57:48 a1k0n: well, i could sponsor a server or two on EC2 if cost is in fact an issue. we just don't know if that's the root cause 2010-10-15T15:58:18 yeah no idea. there's some server status page which shows how many servers are running how many games per minute 2010-10-15T15:58:41 rebel http://72.44.46.68/canvas?game_id=504013 2010-10-15T15:58:50 ai-contest.com/gamestats.php 2010-10-15T15:59:02 yeah - http://ai-contest.com/game_stats.php 2010-10-15T15:59:33 ah so the two extra nodes are down; maybe the account ran out of money or something 2010-10-15T16:00:16 actually the number of games per minute is quite ok. I get a couple games per day, which is fine. 2010-10-15T16:01:23 bhasker: nice. my bot attacked wrong planet too soon and let you steal that game 2010-10-15T16:01:28 Top 10 players: dmj111(3447), MariusCG(3374), deepblue(3341), EBraun(3295), Hazard(3278), Manwe56(3265), felixcoto(3241), ruilov(3218), ofrias(3215), savchenko(3213) 2010-10-15T16:01:33 yup 2010-10-15T16:02:02 but perhaps if you could identify and de-prioritize starter packages (or just old and poor-performing bots) and prioritize >2000ELO people, the "active" people will get more games per minute 2010-10-15T16:03:27 it's mighty misleading right now, my bot is in 300th something just because the servers haven't had time to smack it down with other highly ranked bots 2010-10-15T16:03:34 *** bhasker_ has joined #aichallenge 2010-10-15T16:03:46 Migi32: well, i re-submitted, and now it's going to take days to get back up.. i'm mostly interested in seeing how bot performs against top people who are not on tcp 2010-10-15T16:04:32 starter package submissions are evil 2010-10-15T16:05:22 *** bhasker__ has joined #aichallenge 2010-10-15T16:06:41 interesting - most of my games with McLeopold's AntLion go to 200 turns, and neither bot is willing to attack after expansion phase is over... so first few turns decide the game pretty much 2010-10-15T16:07:08 *** bhasker has quit IRC (Ping timeout: 260 seconds) 2010-10-15T16:07:49 *** Appleman1234 has joined #aichallenge 2010-10-15T16:08:20 rebelxt: thats probably true for any sufficiently advanced bot, right? once one of them has an enge it will be hard for the oponent to catch up 2010-10-15T16:08:26 *** bhasker_ has quit IRC (Ping timeout: 264 seconds) 2010-10-15T16:08:30 edge* 2010-10-15T16:10:27 phreeza_: true. first few turns decide games on "simple" maps where you don't have to split forces over 2-3 different regions 2010-10-15T16:11:44 maybe i should build an opening library ;) 2010-10-15T16:11:44 *** mceier has quit IRC (Quit: leaving) 2010-10-15T16:14:33 phreeza_: exactly in 99% cases it takes me 15 seconds to come up with opening much better than what my bot does... on virtually every map 2010-10-15T16:15:55 once i have my tree search working, maybe i can build it using that, even if it doesn't work as a real bot due to time constraints 2010-10-15T16:16:06 but theres also the 2MB upload limit... 2010-10-15T16:17:47 phreeza_: if you get tree search to work, it's going to be awesome 2010-10-15T16:17:54 only 2MB? I was going to use War and Peace as text to seed my bot too 2010-10-15T16:17:54 but slow ;) 2010-10-15T16:18:17 JamesMG: hahah 2010-10-15T16:19:44 haha War and Peace - compress it! i actually read the whole thing /lol 2010-10-15T16:22:19 *** boegel has joined #aichallenge 2010-10-15T16:22:44 Leo Tolstoy, War and PeacePublished from 1865 to 1869. Original text has some 460,000 Russian and French words. English translation contains over 560,000 words and over 3.1 million characters; typically over 1400 pages as a paperback. 2010-10-15T16:22:51 is anyone running a (probably very limited depth) minimax bot yet? 2010-10-15T16:24:29 albert has it, but its quite crude still he says 2010-10-15T16:24:30 http://ai-contest.com/profile.php?user_id=3897 2010-10-15T16:25:47 actually he says its A* but i'm not sure how that is supposed to work in the context of a tree search 2010-10-15T16:26:13 i tried - it was too slow and too buggy. begging for another attempt 2010-10-15T16:26:45 its fun to watch actually... very erratic in the openings but sometimes it actually works 2010-10-15T16:28:15 Migi32: i bet that's it. scores dropped because games/resubmission dropped 2010-10-15T16:32:14 *** jorge has quit IRC (Remote host closed the connection) 2010-10-15T16:33:07 *** amstan has quit IRC (Ping timeout: 240 seconds) 2010-10-15T16:34:03 phreeza_: wow, albert's bot has got some serious swarm behaviour going on, interesting 2010-10-15T16:34:35 *** BtbN has quit IRC (Quit: Verlassend) 2010-10-15T16:35:42 i expect my search bot to do that to. i do not consider it a good thing 2010-10-15T16:36:30 many of the first bots looked like that, but now they seem to be dominated by the more strategic ones 2010-10-15T16:37:02 in the olden days :) last month 2010-10-15T16:37:36 *** phreeza_ is now known as phreeza 2010-10-15T16:52:30 *** wh1teside has quit IRC (Quit: leaving) 2010-10-15T16:52:59 *** nann has joined #aichallenge 2010-10-15T16:53:03 *** Snowplan has left #aichallenge 2010-10-15T16:56:42 *** narnach has joined #aichallenge 2010-10-15T16:56:56 *** Apophis has quit IRC (Ping timeout: 276 seconds) 2010-10-15T16:57:43 *** choas has joined #aichallenge 2010-10-15T16:59:07 *** dmj111 has quit IRC (Ping timeout: 252 seconds) 2010-10-15T17:01:29 Top 10 players: dmj111(3452), MariusCG(3379), deepblue(3347), EBraun(3299), Hazard(3283), Manwe56(3276), felixcoto(3246), ruilov(3224), ofrias(3221), savchenko(3219) 2010-10-15T17:03:29 bhasker: http://72.44.46.68/canvas?game_id=504689 2010-10-15T17:04:12 *** gat3way has joined #aichallenge 2010-10-15T17:04:15 The AntLion gambit losses... 2010-10-15T17:08:23 yup 2010-10-15T17:08:59 that map was a build fast map for sure. I need to detect that. 2010-10-15T17:10:11 hmm i don't do any such detection, i have to add some map intelligence into the bot 2010-10-15T17:10:18 so that it alters strategy based on map 2010-10-15T17:10:59 I think, if the middle sucks, build fast. Otherwise, throw to the center. 2010-10-15T17:11:17 I have this rule: take what you can hold 2010-10-15T17:11:34 McLeopold, you seem to be sending your first 100 ships to 1 planet a lot, is that intentional? 2010-10-15T17:11:55 Migi32: yes, because I want to react to the enemy and steal stuff. 2010-10-15T17:11:58 migi it makes sense when you have a strong centre 2010-10-15T17:12:05 and to protect your most valuable planet 2010-10-15T17:12:06 it's pretty effective when the planet he throws to is cheap enough and central enough to snipe enemy expansion 2010-10-15T17:12:18 my bot does that too 2010-10-15T17:12:30 though not as aggresive as mcleopolds 2010-10-15T17:12:41 mine had that at one point, but it misused it so I took it away 2010-10-15T17:12:47 *** delt0r___ has joined #aichallenge 2010-10-15T17:12:57 the basic idea is if X turns from now the planet is going to be long to you , you might as well send your ships there 2010-10-15T17:12:59 in advance 2010-10-15T17:13:04 rather than waiting for it to convert 2010-10-15T17:13:07 and then sending 2010-10-15T17:13:21 s/long/belong 2010-10-15T17:13:34 plus, if most of your strength is in the center, the enemy can't expand there without being sniped 2010-10-15T17:13:46 right, but only if that's one of the planets you are going to be funneling ships to 2010-10-15T17:14:19 well the way i pick that planet is by an average of distance to enemy planets and mine 2010-10-15T17:14:29 so its usually well located 2010-10-15T17:14:30 *** delt0r_ has quit IRC (Ping timeout: 265 seconds) 2010-10-15T17:14:36 sometimes it can get messed up 2010-10-15T17:14:41 if i end up losing that planet and i stop sending ships there 2010-10-15T17:14:46 and now i have a skewed setup 2010-10-15T17:15:10 yeah you might want a more complex function. did you try center of mass calculation? 2010-10-15T17:15:19 hmm should try that 2010-10-15T17:15:23 good idea 2010-10-15T17:15:24 thanks rebel 2010-10-15T17:15:31 didn't think of center of mass 2010-10-15T17:15:40 rebelxt, center of mass is always the center of the map 2010-10-15T17:15:42 i replaced some of the other average stuff with an exponential decay function 2010-10-15T17:15:52 well center of mass of contrlled planets 2010-10-15T17:15:54 which can change 2010-10-15T17:15:58 the maps are symmetric, center of mass is always planet 0 2010-10-15T17:16:04 not all planets 2010-10-15T17:16:19 Migi32, no, center of mass is the point where enemy has the most forces 2010-10-15T17:16:29 so it depends on how the ships are allocated 2010-10-15T17:16:30 ah hmm 2010-10-15T17:16:36 ah, the enemy's center of mass 2010-10-15T17:16:41 mass being ships 2010-10-15T17:16:48 yep 2010-10-15T17:17:00 http://en.wikipedia.org/wiki/Center_of_mass 2010-10-15T17:17:08 like cultural influence in civ 2010-10-15T17:17:12 I think you can get skewed results when the enemy has two "groups" of planets 2010-10-15T17:17:31 any statistic can get skewed 2010-10-15T17:17:39 the questions is how resistant it is to skewing 2010-10-15T17:17:46 average is very susceptible 2010-10-15T17:18:03 Migi32, correct, but you can detect clusters of planets. it does not have to be a single center of masss 2010-10-15T17:18:39 rebelxt, I don't think cluster recognition is as easy as in Tron 2010-10-15T17:18:56 it's not hard 2010-10-15T17:19:01 should be easy to do 2010-10-15T17:19:06 test on map5 2010-10-15T17:19:07 just start from a planet 2010-10-15T17:19:08 Migi32, it's doable 2010-10-15T17:19:09 and grow it 2010-10-15T17:19:17 tillthe next planet is > dist 2010-10-15T17:19:24 then do the same from another random planet not in a cluster 2010-10-15T17:19:54 oh. like that. 2010-10-15T17:20:54 my bot can't seem to break the 300 elo rating barrier:( 2010-10-15T17:20:56 :(( 2010-10-15T17:21:07 keeps going upto 285 then back down to 270 2010-10-15T17:21:44 perhaps because your true ELO is only ~280? No offense, I don't really know your bot very well. 2010-10-15T17:22:11 it probably is 280 :) 2010-10-15T17:22:17 though i was hoping it would do better:( 2010-10-15T17:22:29 will have to analyze more games and fix bugs^ 2010-10-15T17:22:37 the trick is to be good against all players, not just the ones ranked on top :) 2010-10-15T17:22:42 :p 2010-10-15T17:22:53 yea need to fix the bugs for neutral sniping 2010-10-15T17:22:57 mine has the opposite problem i guess. 2010-10-15T17:23:01 http://72.44.46.68/canvas?game_id=505180 2010-10-15T17:23:16 just the opening is interesting 2010-10-15T17:23:26 if you beat McLeopold I'm pretty sure you can beat anyone with ELO < 3000 2010-10-15T17:23:29 Accoun seems to have my number. 2010-10-15T17:23:36 he did 2010-10-15T17:23:44 twice 2010-10-15T17:24:06 McLeopold: nice one.. 2010-10-15T17:24:31 *** dmj111 has joined #aichallenge 2010-10-15T17:24:33 OK I can't figure out one thing 2010-10-15T17:24:38 so accoun won by 13 ships? 2010-10-15T17:24:41 are the games scheduled on a regular basis? 2010-10-15T17:24:55 I mean...once I reupload my code, I have a game almost instantly 2010-10-15T17:25:01 then I wait for hours and no new game 2010-10-15T17:25:07 sorry for the dumb question 2010-10-15T17:25:14 it's biased such that bots with no games get the front of the queue 2010-10-15T17:25:33 aaaaah ok thanks 2010-10-15T17:25:34 otherwise, yeah, everyone gets a game every x hours 2010-10-15T17:25:35 a1k0n: yes, slightly better build schedule 2010-10-15T17:25:38 first come, first serve 2010-10-15T17:26:01 actually, I can tell you why mine didn't build faster, and I think it's a bug... 2010-10-15T17:26:06 gat3way: there are usually more servers online but there's only one now for some unknown reason 2010-10-15T17:26:23 *** fluffluff has joined #aichallenge 2010-10-15T17:26:28 we just saw scheduling in class today... but scheduling CPU processes is nothing like scheduling games for a contest. 2010-10-15T17:26:29 how do you check the server stats? 2010-10-15T17:26:45 http://ai-contest.com/game_stats.php 2010-10-15T17:26:53 !stats 2010-10-15T17:26:53 Migi32: I have 1 registered users with 0 registered hostmasks; 1 owner and 0 admins. 2010-10-15T17:27:04 thanks 2010-10-15T17:27:14 !gamestats 2010-10-15T17:27:14 Migi32: User error, it's not my fault. 2010-10-15T17:27:23 !learn gamestats http://ai-contest.com/game_stats.php 2010-10-15T17:27:23 Migi32: (learn [] as ) -- Associates with . is only necessary if the message isn't sent on the channel itself. The word 'as' is necessary to separate the key from the value. It can be changed to another word via the learnSeparator registry value. 2010-10-15T17:27:32 !learn gamestats as http://ai-contest.com/game_stats.php 2010-10-15T17:27:32 Migi32: I think that worked... 2010-10-15T17:28:51 what's the "EPM" of a server? For server #0 it's 0.0 2010-10-15T17:29:36 *** jukkerknott has joined #aichallenge 2010-10-15T17:29:47 rebelxt: just saw your note. I started one up... but I have been having tcp problems lately, so I am not sure how long it will run. 2010-10-15T17:31:04 allright, definitely recursion is bad 2010-10-15T17:31:33 if only they could raise the limit to say 5s/turn 2010-10-15T17:31:40 *** jukkerknott has quit IRC (Client Quit) 2010-10-15T17:31:55 then we'd get a game a day, but the top games would be interesting 2010-10-15T17:32:16 well....more servers, it's google after all :) 2010-10-15T17:32:28 i'm using less than a tenth of a second ... 2010-10-15T17:33:20 gateway must be doing some tree search 2010-10-15T17:33:23 or something rather expensive 2010-10-15T17:33:26 to need 5 seconds 2010-10-15T17:33:54 what happens first in a turn? 2010-10-15T17:34:17 ships crashing into the planet, or growth being added? 2010-10-15T17:34:21 growth 2010-10-15T17:34:25 hmm 2010-10-15T17:34:31 I dropped the idea 2010-10-15T17:34:33 that what I'd expect 2010-10-15T17:34:34 i think its on the FAQ page 2010-10-15T17:34:38 now I use fairly idiotic algorithm 2010-10-15T17:34:49 http://ai-contest.com/specification.php 2010-10-15T17:35:08 I think I found a bug in the Bot Manager 2000 then (JBotManager.jar) 2010-10-15T17:35:17 I will get it improved once I gather enough impressions of that 2010-10-15T17:35:18 first is departure actually 2010-10-15T17:35:20 then growth 2010-10-15T17:35:22 then arrival 2010-10-15T17:35:22 that seems to crash ships into planets first, and then add growth 2010-10-15T17:35:25 I think 2010-10-15T17:35:42 btw, the map scale in the java starter kit (simulation) is different than the game server one 2010-10-15T17:35:43 http://ai-contest.com/specification.php#game-state-update 2010-10-15T17:35:48 that can be a problem 2010-10-15T17:36:03 !learn specs as http://ai-contest.com/specification.php 2010-10-15T17:36:03 bhasker__: Job's done. 2010-10-15T17:36:11 @contestbot specs 2010-10-15T17:36:12 bhasker__: No! 2010-10-15T17:36:14 @specs 2010-10-15T17:36:15 bhasker__: specs = http://ai-contest.com/specification.php 2010-10-15T17:36:19 there! 2010-10-15T17:37:34 depart, travel, grow, arrive 2010-10-15T17:37:43 anyone know who the author of JBotManager is? 2010-10-15T17:39:10 delt0r: ping? 2010-10-15T17:39:58 http://ai-contest.com/game_stats.php 2010-10-15T17:40:08 oops 2010-10-15T17:40:11 srry 2010-10-15T17:43:53 *** davidjliu has joined #aichallenge 2010-10-15T17:44:34 hmm, code im jar file says differently 2010-10-15T17:44:37 wtf 2010-10-15T17:49:09 heh, weird bug 2010-10-15T17:49:54 if my last planet (growth 5), has 95 ships, and in the next turn an enemy fleet of 100 ships hits, what should happen? 2010-10-15T17:50:10 boegel: 0, you own 2010-10-15T17:50:25 McLeopold: yeah, I figured as much 2010-10-15T17:50:39 McLeopold: doesn't matter if it's my last planet, right? 2010-10-15T17:50:43 no 2010-10-15T17:50:52 k, thx 2010-10-15T17:51:09 so, that's a bug somewhere in this game engine, beuh 2010-10-15T17:51:26 unless you sent an invalid order? 2010-10-15T17:51:41 if it was your last planet i would look for a divide by zero 2010-10-15T17:52:06 usually when my bot crashes at last turn its usually because i was dividing by something without checking 2010-10-15T17:52:41 Yeah, I was plagued with divide by zero errors for awhile. 2010-10-15T17:53:03 hmm, right 2010-10-15T17:53:24 I should check and make sure 2010-10-15T17:53:27 thx 2010-10-15T17:53:37 dmj111, tnx. i had to shut mine down - laptop battery ran out. we got one game though 2010-10-15T17:53:57 *** mimidoux has joined #aichallenge 2010-10-15T17:54:26 *** rebelxt has left #aichallenge 2010-10-15T17:54:47 *** yasith_ has quit IRC (Ping timeout: 276 seconds) 2010-10-15T17:55:01 *** rebelxt has joined #aichallenge 2010-10-15T17:55:52 I'm pretty sure it's the game engine, my planet isn't sending out any fleets anymore 2010-10-15T17:55:53 *** rebelxt has left #aichallenge 2010-10-15T17:55:56 *** rebelxt has joined #aichallenge 2010-10-15T17:56:03 it basically gave up (I need to fix that ;)) 2010-10-15T17:56:25 *** rebelxt has left #aichallenge 2010-10-15T17:57:57 *** narnach has quit IRC (Quit: Leaving.) 2010-10-15T17:59:18 *** McLeopold has quit IRC (Quit: Leaving) 2010-10-15T18:01:29 Top 10 players: dmj111(3447), MariusCG(3426), deepblue(3380), Hazard(3305), EBraun(3296), Manwe56(3283), ruilov(3274), felixcoto(3267), ofrias(3260), savchenko(3243) 2010-10-15T18:02:13 *** gat3way has left #aichallenge ("No matter how dark the night, somehow the Sun rises once again") 2010-10-15T18:07:30 *** yasith_ has joined #aichallenge 2010-10-15T18:17:31 hmm 2010-10-15T18:17:43 when is a game *lost* 2010-10-15T18:17:51 if you have zero ships at sone point? 2010-10-15T18:18:14 or if you have no more planets left (or the max turn count is reached, of course)? 2010-10-15T18:18:24 what if I have one (empty) planet 2010-10-15T18:18:32 at some turn 2010-10-15T18:18:36 do I lose instantly? 2010-10-15T18:19:56 "Provided that neither player performs an invalid action, the player with the most ships at the end of the game wins. The game may also end earlier if one of the players loses all his ships, in which case the player that has ships remaining wins instantly." 2010-10-15T18:20:28 ah 2010-10-15T18:20:34 so yes, only empty planets = lost 2010-10-15T18:20:35 so losing all ships is enough to lose 2010-10-15T18:20:58 Migi32: which is kind of sad, because your planets will regain ships in the next turn... 2010-10-15T18:21:16 anyway, glad that's cleared out 2010-10-15T18:21:22 boegel, if you'd win after that it'd be the most insane comeback ever 2010-10-15T18:21:22 Migi32: how's your bot doing? 2010-10-15T18:21:33 Migi32: it would be cool though 2010-10-15T18:21:45 Migi32: also, it depends on the game state I guess 2010-10-15T18:22:15 boegel, turns out the kind of things I wanted to use for my bot are far too cpu-expensive 2010-10-15T18:22:37 1 second is not much 2010-10-15T18:22:52 *** mager has joined #aichallenge 2010-10-15T18:23:09 Migi32: what did you have in mind? 2010-10-15T18:23:18 my AI crashes everytime :/ 2010-10-15T18:23:21 hey, my bot runs in like 100ms each turn now :) 2010-10-15T18:25:31 i think the AI's should run until the battles are over instead of stopping at 200 turns :P 2010-10-15T18:26:04 would be forever in some cases 2010-10-15T18:26:35 Migi32, boegel: actually, the spec as i wrote it is wrong on that point 2010-10-15T18:26:56 the game isn't necessarily over just because you have no ships left. possession of a planet keeps the game going 2010-10-15T18:27:17 YAY!! my AI beat the prospector! :P 2010-10-15T18:27:30 congraturlations mager! :D 2010-10-15T18:27:31 now if only it would quit crashing against everything else.... 2010-10-15T18:27:42 thanks Migi :) 2010-10-15T18:27:54 but it won after 642 turns :/ 2010-10-15T18:28:06 mager, some games would never end 2010-10-15T18:28:35 and we only have 1 server that can do about 1000 turns per minute 2010-10-15T18:28:36 yeah, but 200 turns isn't long enough for the tables to turn :P 2010-10-15T18:28:58 *** Itkovian has quit IRC (Ping timeout: 245 seconds) 2010-10-15T18:29:00 i think 300 or 500 would be a more accurate comparison of the AI's 2010-10-15T18:29:03 free elo 4tw lol broke 300! yay 2010-10-15T18:29:17 haha nice bhasker__ :P 2010-10-15T18:29:21 :D 2010-10-15T18:29:35 cheating 4tw 2010-10-15T18:29:41 cheating? :o 2010-10-15T18:29:48 just changed your name? 2010-10-15T18:30:26 i know this is a bit off topic - but anyone know a good school to study computer science focusing on C++ and game programming? 2010-10-15T18:30:38 yea 2010-10-15T18:30:40 :p 2010-10-15T18:30:47 actually fixed a bug too 2010-10-15T18:31:07 *** kersny has joined #aichallenge 2010-10-15T18:31:25 haha 338 elo 2010-10-15T18:31:40 lets see how high it can go 2010-10-15T18:31:44 before it drops to 280 again 2010-10-15T18:31:47 jmcarthur: you should correct that ASAP then 2010-10-15T18:31:52 woah 387 2010-10-15T18:31:59 boegel: there are many things that need corrections :\ 2010-10-15T18:32:03 rank 15 now !! 2010-10-15T18:32:06 jmcarthur: I can imagine 2010-10-15T18:32:18 *** JamesMG has quit IRC (Quit: Leaving.) 2010-10-15T18:32:22 grats bhasker - im rank 1964 last i checked :P 2010-10-15T18:32:25 jmcarthur: are you guys going to be able to fix everthing before the official end? 2010-10-15T18:33:09 oh this is on tcp 2010-10-15T18:33:12 not on the official server 2010-10-15T18:33:16 mager: if you go to a good school for computer science you will likely learn more than just game programming 2010-10-15T18:33:18 my bot on the official server is at rank 100 odd 2010-10-15T18:33:19 stuck there 2010-10-15T18:34:43 Zannick - Most colleges have moved into Java programming and in the field I intend to head into, i don't really care for it :/ I'm having a lot of trouble finding a school that still teaches C++ 2010-10-15T18:35:01 many don't 2010-10-15T18:35:22 i went to cmu, they did java, c, and sml while i was there 2010-10-15T18:35:31 i dont see why...they stopped teaching C++ to make it easier, but computer science is supposed to be a challenge :/ 2010-10-15T18:35:35 they have added python to the intro courses 2010-10-15T18:35:59 but there are classes that use c++, like the graphics course 2010-10-15T18:36:02 I think we need to start a movement to go back to C++! :) 2010-10-15T18:36:07 plus you can always teach yourself 2010-10-15T18:36:12 k i think i should treat myself to coffee 2010-10-15T18:36:14 rank 5 now 2010-10-15T18:36:15 !! 2010-10-15T18:36:15 bhasker__: No! 2010-10-15T18:37:08 but i'd rather have someone help me. Im 16 and live in the most rural area possible, i practically run my schools technology district. I've taught myself a lot - but somethings you just need someone with experience, you know? 2010-10-15T18:37:49 i still recommend CMU, even if you only want C++ 2010-10-15T18:38:15 whats CMU? 2010-10-15T18:38:26 Carnegie Mellon, Pittsburgh 2010-10-15T18:39:07 you'll see it pop up if you go looking for good cs schools 2010-10-15T18:39:35 i'll check them out - i kinda like full sail university, anyone know anything about them? 2010-10-15T18:39:56 btw - 688 turns takes forever to load the viewr :/ 2010-10-15T18:40:08 time for bed 2010-10-15T18:40:14 while the bot is playing :) 2010-10-15T18:40:17 ttyl y'all! 2010-10-15T18:40:26 *** boegel has quit IRC (Quit: *poof!*) 2010-10-15T18:41:01 it really depends on what you want, cs at cmu is not a lot of game programming but it's very good for programming in general and there's even a club dedicated to making computer games in your free time 2010-10-15T18:41:38 i don't know anything about full sail but i think they're a little more specific in studies 2010-10-15T18:42:27 i'd like to get a game programming degree specifically - thats generally focused directly on C++ unless its called Web game programming. 2010-10-15T18:42:35 *** zorane has quit IRC (Quit: Page closed) 2010-10-15T18:42:39 but I will look into CMU : ) 2010-10-15T18:43:11 cool and good luck :) 2010-10-15T18:43:17 why would a battle be over in 57 turns if both still have planets? 2010-10-15T18:44:08 crash 2010-10-15T18:44:23 *** yasith_ has quit IRC (Ping timeout: 245 seconds) 2010-10-15T18:44:39 how do you stop a crash? 2010-10-15T18:45:00 remove the condition that causes it? 2010-10-15T18:45:20 http://a1k0n.net/temp/galcon_histogram.png 2010-10-15T18:45:46 i think that's pretty conclusive: a 200-turn limit is probably about right 2010-10-15T18:46:15 *** jesionaj has joined #aichallenge 2010-10-15T18:46:52 hmmm, nice 2010-10-15T18:46:55 12962 games sampled 2010-10-15T18:50:23 ...my new AI got owned, in its first match : ( 2010-10-15T18:50:40 but the AI it fought against looked pretty good, so i guess it still has a chance 2010-10-15T18:54:08 *** godar has joined #aichallenge 2010-10-15T18:55:18 *** amstan has joined #aichallenge 2010-10-15T18:55:30 yo 2010-10-15T18:57:11 *** yasith_ has joined #aichallenge 2010-10-15T18:59:13 hum is it just me or is the tcp server timing out a lot? 2010-10-15T19:00:10 *** godar has quit IRC (Ping timeout: 265 seconds) 2010-10-15T19:00:21 hola amstan 2010-10-15T19:01:28 ALRIGHT!! 2010-10-15T19:01:30 Top 10 players: dmj111(3449), MariusCG(3419), deepblue(3388), Hazard(3310), Manwe56(3290), ruilov(3281), felixcoto(3280), EBraun(3260), ofrias(3253), savchenko(3251) 2010-10-15T19:01:36 from 1964 to 1342 : D 2010-10-15T19:01:42 *** mimidoux has quit IRC (Quit: Page closed) 2010-10-15T19:05:05 ah no my bad... invalid input, not timeout :) 2010-10-15T19:12:27 my AI is pretty good IMO :) im proud of myself 2010-10-15T19:13:39 *** jukkerknott has joined #aichallenge 2010-10-15T19:14:58 *** yasith_ has quit IRC (Ping timeout: 265 seconds) 2010-10-15T19:22:53 anyone else think it would be interesting to play against these bots? 2010-10-15T19:26:39 *** RainCT has quit IRC (Remote host closed the connection) 2010-10-15T19:27:55 *** yasith_ has joined #aichallenge 2010-10-15T19:28:10 *** rebelxt has joined #aichallenge 2010-10-15T19:28:49 heh looks like fixing a bug and switching names does wonders to elo rating 2010-10-15T19:30:19 yeah, you want to use new name for each version. otherwise ELO calculation takes all old games into account, and your rating takes very long time to climb 2010-10-15T19:30:41 yea i was doing that 2010-10-15T19:30:48 with all my bhaskerv20.X versions 2010-10-15T19:30:58 but was lazy once i switched the names 2010-10-15T19:31:01 *** BtbN has joined #aichallenge 2010-10-15T19:32:01 *** skandy_ has joined #aichallenge 2010-10-15T19:32:22 i'm paranoid - i commit to git even if i change one line of code.. since it's so ez to cause a regression by a one line change 2010-10-15T19:32:55 ah i am committing 2010-10-15T19:33:06 i am just not updating the name on tcp server 2010-10-15T19:33:38 bhasker__: i really wanna play your new bot, but battery died.. will run it from home in a few hours. 2010-10-15T19:33:53 i don't think it will do much better than my previous version 2010-10-15T19:34:01 its just having a string of lucky wins 2010-10-15T19:34:08 i think it will settle in the low 300's after sometime 2010-10-15T19:34:16 already at 439 2010-10-15T19:34:21 it seems much stronger to me - watched a few games 2010-10-15T19:34:34 hmm 2010-10-15T19:34:39 oh well time will tell :) 2010-10-15T19:34:59 a couple more hours of running and i should know where it will settle down 2010-10-15T19:35:03 *** amstan has quit IRC (Ping timeout: 276 seconds) 2010-10-15T19:37:07 that's the problem with ELO, it never really settles down... 2010-10-15T19:37:11 ELO does not mean much though - it really depends on who and how many people are running their bots at the moment, so only relate ELO matter 2010-10-15T19:37:31 bhasker__, what's your bot's name on TCP? 2010-10-15T19:37:39 malazan 2010-10-15T19:37:44 malazan-0.2 is the new one 2010-10-15T19:37:49 i.e. once it settles in a couple of hours, how does its ELO compare to other usual suspects you compare ELO rating with 2010-10-15T19:37:58 ah yes nice ELO indeed :P 2010-10-15T19:38:04 haha http://72.44.46.68/canvas?game_id=506942 2010-10-15T19:38:06 dropping rapidly lol 2010-10-15T19:38:07 :( 2010-10-15T19:38:14 yea yea just saw that:( 2010-10-15T19:38:32 expanded too rapidly 2010-10-15T19:38:51 *** McLeopold has joined #aichallenge 2010-10-15T19:38:56 not sure why it didnt snipe your planet right next to the corner one 2010-10-15T19:39:00 will have to look at the code again 2010-10-15T19:39:08 it should have taken the 3 as soon as you took that one 2010-10-15T19:40:18 haha boegel.20101015.6d292b249a699207cd5336dac41b13e3 2010-10-15T19:40:40 someone is signing their code 2010-10-15T19:40:46 md5? 2010-10-15T19:41:02 possibly. i wonder how much more obnoxious that could get 2010-10-15T19:42:32 people are using git or mercurial commit ID 2010-10-15T19:43:01 (most likely) 2010-10-15T19:43:01 *** oPless has joined #aichallenge 2010-10-15T19:43:46 man tcpserver gets a looooot of games per hour 2010-10-15T19:45:06 *** amstan has joined #aichallenge 2010-10-15T19:46:54 *** Error323 is now known as E323|ZzzZz 2010-10-15T19:48:10 *** choas has quit IRC (Quit: leaving) 2010-10-15T19:48:21 *** amstan has quit IRC (Read error: Operation timed out) 2010-10-15T19:48:43 *** Migi32 has quit IRC (Quit: http://boinc.berkeley.edu) 2010-10-15T19:51:30 yup 2010-10-15T19:53:12 does the tcp server do any kind of swiss tournament or just first come first served matching? 2010-10-15T19:53:52 first come, first server... but there are often more than 2 people trying to connect at the same time, so it does match by elo 2010-10-15T19:54:11 but the choice is usually very limited 2010-10-15T19:56:09 ok 2010-10-15T19:56:52 right now, if I connect, it says 33 con, 16 games 2010-10-15T19:57:03 16*2 is 32 connection in games, only me left 2010-10-15T19:57:12 yea so just one choice probably 2010-10-15T19:57:13 So, I probably get matched with the next person 2010-10-15T19:57:40 I've looked at the code, and it does elo matching, but I don't think it waits for more than one choice 2010-10-15T19:57:42 but it does seem to have some kind of mechanism to stop you from just being matched up with the same person forever? 2010-10-15T19:57:48 nope 2010-10-15T19:57:52 just random draw 2010-10-15T19:58:02 ah 2010-10-15T19:58:05 my bot went up against the same guy almost 9 in a row 2010-10-15T19:58:07 best bet is to run a loop and randomize your wait timer 2010-10-15T19:58:25 that's because you both finished, the both get in the queue right away 2010-10-15T19:58:30 it does actually try to match you up with someon enear you in ELO 2010-10-15T19:58:30 looking at tcp server source, it just picks first user from the FIFO queue, then finds another user in the queue that has the closest ELO, then starts the game 2010-10-15T19:58:39 but it doesn't wait in the "lobby" long enough for such a match to usually occur 2010-10-15T19:59:00 yeah, i wish it would wait 60 seconds or so 2010-10-15T19:59:20 If dmj111 and I ever get a python server going, we'll add that. 2010-10-15T19:59:52 CrowBot's back up. 2010-10-15T20:00:04 I'm fixing bugs in AntBot 2010-10-15T20:00:14 lets see how my bot does against crowbot 2010-10-15T20:00:35 It should win. Unless you're weak on close starts. 2010-10-15T20:00:41 Or unless we draw map1 2010-10-15T20:00:41 i am 2010-10-15T20:00:49 i need to fix close maps 2010-10-15T20:00:53 *** yasith_ has quit IRC (Ping timeout: 265 seconds) 2010-10-15T20:01:00 the new stuff i added kind of does stupid things on close maps 2010-10-15T20:01:12 *** rebelxt has left #aichallenge 2010-10-15T20:01:26 num_ships - start_distance * growth(5) = first turn ships 2010-10-15T20:01:30 Top 10 players: dmj111(3463), MariusCG(3433), deepblue(3403), Hazard(3324), Manwe56(3308), ruilov(3295), felixcoto(3294), EBraun(3275), ofrias(3268), savchenko(3266) 2010-10-15T20:01:55 yea i do that 2010-10-15T20:02:04 do you do it on turn 2? 2010-10-15T20:02:09 need to do that 2010-10-15T20:02:16 :) 2010-10-15T20:02:28 actually i added it at turn 2 2010-10-15T20:02:29 It's fun to watch dhartmei's bot break down on turn 10. 2010-10-15T20:02:32 then removed it don't remember 2010-10-15T20:02:37 why i did that 6 2010-10-15T20:03:53 actually my bot will break down in certain close situations i think i am going to change it so that it keeps doing that till both of us have a couple of planets 2010-10-15T20:03:59 then switch to a different heuristic 2010-10-15T20:04:00 or something 2010-10-15T20:04:16 rather than do it based on turn numbers 2010-10-15T20:04:31 yeah, any code based on turn number is bad 2010-10-15T20:04:46 any magic number is bad too 2010-10-15T20:05:49 *** ratatata has joined #aichallenge 2010-10-15T20:06:17 yea been trying to get rid of magic numbers 2010-10-15T20:06:25 but sometimes till i get a better idea magic number is all i have ^ 2010-10-15T20:06:40 my official bot uses the golden ratio 2010-10-15T20:06:58 I've gotten rid of that, though. 2010-10-15T20:07:42 hi all 2010-10-15T20:10:53 oh well crowbot just crushed my bot:(( 2010-10-15T20:11:43 *** kersny has quit IRC (Quit: I'm Out) 2010-10-15T20:12:05 *** zerd has quit IRC (Remote host closed the connection) 2010-10-15T20:12:13 *** zerd has joined #aichallenge 2010-10-15T20:12:56 too slow to expand 2010-10-15T20:16:22 haha interesting game 2010-10-15T20:16:22 http://72.44.46.68/canvas?game_id=507441 2010-10-15T20:17:15 *** yasith has joined #aichallenge 2010-10-15T20:18:24 *** hornairs has joined #aichallenge 2010-10-15T20:21:41 alright, back to work 2010-10-15T20:31:37 *** McLeopold has quit IRC (Quit: Leaving) 2010-10-15T20:32:19 whats crowbot? where do you play it? 2010-10-15T20:36:54 *** yellowseed has joined #aichallenge 2010-10-15T20:36:56 *** mimidoux has joined #aichallenge 2010-10-15T20:44:03 i'm trying to profile my bot using gprof, has anyone done that? 2010-10-15T20:49:13 *** bhasker__ has quit IRC (Ping timeout: 240 seconds) 2010-10-15T20:49:35 *** yellowseed has quit IRC (Quit: Leaving.) 2010-10-15T20:52:42 *** mager has quit IRC () 2010-10-15T20:53:12 *** BtbN has quit IRC (Quit: Verlassend) 2010-10-15T20:56:22 *** Apophis has joined #aichallenge 2010-10-15T20:59:21 *** nann has quit IRC (Ping timeout: 255 seconds) 2010-10-15T21:01:31 Top 10 players: dmj111(3482), MariusCG(3442), deepblue(3374), Hazard(3336), Manwe56(3316), felixcoto(3315), EBraun(3289), ruilov(3289), fglider(3279), ofrias(3278) 2010-10-15T21:03:52 *** bhasker has joined #aichallenge 2010-10-15T21:07:16 *** yasith has quit IRC (Ping timeout: 264 seconds) 2010-10-15T21:11:28 *** amstan has joined #aichallenge 2010-10-15T21:11:29 *** ChanServ sets mode: +o amstan 2010-10-15T21:13:08 *** jukkerknott has quit IRC (Quit: Page closed) 2010-10-15T21:16:57 oh boy, debugging sucks 2010-10-15T21:23:30 *** yasith has joined #aichallenge 2010-10-15T21:32:22 wtf... gprof doesn't work on OS X 2010-10-15T21:34:11 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-10-15T21:41:40 haha 2010-10-15T21:41:41 http://72.44.46.68/canvas?game_id=508415 2010-10-15T21:44:39 rofl.. good times 2010-10-15T21:45:44 game frequency seems to dropped on official server, 19 games in 52+ hours :/ 2010-10-15T21:46:09 *** phreeza has quit IRC (Ping timeout: 255 seconds) 2010-10-15T21:46:25 *** phreeza has joined #aichallenge 2010-10-15T21:50:05 *** ichti has quit IRC () 2010-10-15T21:51:33 *** Olathe has quit IRC (Ping timeout: 276 seconds) 2010-10-15T21:53:13 *** Frontier has quit IRC (Ping timeout: 240 seconds) 2010-10-15T22:01:31 Top 10 players: dmj111(3538), MariusCG(3424), deepblue(3366), Hazard(3337), ruilov(3300), EBraun(3294), felixcoto(3289), fglider(3283), savchenko(3280), FlameN(3279) 2010-10-15T22:02:19 *** hornairs has quit IRC (Remote host closed the connection) 2010-10-15T22:02:45 *** Blkt has joined #aichallenge 2010-10-15T22:04:57 *** ratatata has quit IRC (Quit: Page closed) 2010-10-15T22:20:52 *** phreeza_ has joined #aichallenge 2010-10-15T22:23:32 *** phreeza has quit IRC (Ping timeout: 240 seconds) 2010-10-15T22:28:31 *** jesionaj` has joined #aichallenge 2010-10-15T22:30:08 *** Blkt has quit IRC (Quit: Error: do not makunbound t please!) 2010-10-15T22:33:05 *** lovebot_ has joined #aichallenge 2010-10-15T22:33:46 *** lovebot has quit IRC (Ping timeout: 240 seconds) 2010-10-15T22:33:46 *** pgpaskar_ has quit IRC (Ping timeout: 240 seconds) 2010-10-15T22:33:47 *** delt0r has quit IRC (Ping timeout: 240 seconds) 2010-10-15T22:33:47 *** jesionaj has quit IRC (Ping timeout: 240 seconds) 2010-10-15T22:33:48 *** pgpaskar_ has joined #aichallenge 2010-10-15T22:33:48 *** pgpaskar_ has joined #aichallenge 2010-10-15T22:34:03 *** delt0r has joined #aichallenge 2010-10-15T22:50:05 *** yellowseed has joined #aichallenge 2010-10-15T23:01:31 Top 10 players: dmj111(3532), MariusCG(3413), deepblue(3364), Hazard(3335), ofrias(3298), ruilov(3298), EBraun(3289), felixcoto(3288), fglider(3285), FlameN(3280) 2010-10-15T23:03:34 *** yasith has quit IRC (Ping timeout: 276 seconds) 2010-10-15T23:06:58 *** phreeza_ has quit IRC (Quit: Computer has gone to sleep) 2010-10-15T23:07:48 *** jukkerknott has joined #aichallenge 2010-10-15T23:19:36 *** andrew-101 has joined #aichallenge 2010-10-15T23:19:45 hi 2010-10-15T23:20:06 what is guy + drunk + irc 2010-10-15T23:20:18 its /me 2010-10-15T23:20:35 at 6am 2010-10-15T23:23:46 *** tapwater has quit IRC (Ping timeout: 240 seconds) 2010-10-15T23:26:57 *** zerd has quit IRC (Read error: Connection reset by peer) 2010-10-15T23:27:02 *** zerd has joined #aichallenge 2010-10-15T23:27:10 *** tapwater has joined #aichallenge 2010-10-15T23:29:20 *** yasith has joined #aichallenge 2010-10-15T23:48:12 *** hellman has joined #aichallenge 2010-10-15T23:53:04 *** Epona has joined #aichallenge 2010-10-15T23:58:09 *** krokkrok has joined #aichallenge 2010-10-15T23:58:49 hoody hoody hoo