2010-09-16T00:04:38 finally 2010-09-16T00:06:21 *** shades has joined #aichallenge 2010-09-16T00:06:56 *** inverselimit has quit IRC (Quit: Leaving) 2010-09-16T00:08:20 *** ShaedS has quit IRC (Quit: Page closed) 2010-09-16T00:13:37 *** Olathe has quit IRC (Ping timeout: 272 seconds) 2010-09-16T00:13:41 http://www.benzedrine.cx/planetwars/canvas?game_id=2480 2010-09-16T00:13:49 dhartmei goes down 2010-09-16T00:15:12 http://www.benzedrine.cx/planetwars/canvas?game_id=2466 75% tied 2010-09-16T00:18:12 *** other_donkey has joined #aichallenge 2010-09-16T00:19:22 *** smhanov_ has quit IRC (Ping timeout: 240 seconds) 2010-09-16T00:20:23 what exactly do the changes in bold for (2) and (3) do here? http://ai-contest.com/simple_strategy_guide.php 2010-09-16T00:20:23 *** Queue29 has joined #aichallenge 2010-09-16T00:21:58 it makes the score (the good-ness) of a planet depend on the number of ships that are on it as well as the growth rate 2010-09-16T00:22:05 *** other_donkey has quit IRC (Client Quit) 2010-09-16T00:22:26 yup, in the unmodified code only the number of ships was used 2010-09-16T00:23:57 number of ships / (1+p.GrowthRate()) 2010-09-16T00:24:16 what kind of values are we seeing for GrowthRate? 2010-09-16T00:24:26 0,1,3,5 2010-09-16T00:25:22 the 0's were kind of a surprise 2010-09-16T00:25:39 0-5 2010-09-16T00:28:09 amstan_: ? 2010-09-16T00:28:31 amstan_: We have that, its called vpn 2010-09-16T00:28:35 greghaynes: read some about it, forward it to the person which needs to look into it, it seems like your club is more involved 2010-09-16T00:28:44 greghaynes: no... this is for wifi access 2010-09-16T00:29:10 greghaynes: for example.. i can go to europe now, to an univ that also supports this, and login to wpa as amstan@uwaterloo.ca and my normal pass 2010-09-16T00:29:30 Ah, we wouldnt allow that 2010-09-16T00:29:50 *** amstan_ is now known as amstan 2010-09-16T00:30:26 We have a very large pipe so someone wanting to be malicious could take advantege 2010-09-16T00:31:00 greghaynes: it's something very official and organized, and seems very beneficial 2010-09-16T00:32:27 hey guys any tips on how to programattically capture the output of playgame before all the moves are piped out? 2010-09-16T00:32:33 *** smhanov_ has joined #aichallenge 2010-09-16T00:32:46 hornairs: if you're using windows 2010-09-16T00:32:48 ie the stuff that showgame doesn't need, the turns and the winer? 2010-09-16T00:32:53 wrap your command in this: 2010-09-16T00:33:04 *nix sorry 2010-09-16T00:33:17 tee input.log|yourbotcommand|tee output.log 2010-09-16T00:33:46 it will dump everything that your bot receives to one file and dump everything outgoing to another 2010-09-16T00:34:07 amstan sorry not the messages going to the bot, the messages coming from playgame.jar 2010-09-16T00:34:19 doesn 2010-09-16T00:34:27 and i want to grab it and put it in a string as opposed to write to file 2010-09-16T00:34:27 doesn't playgame already do log.txt? 2010-09-16T00:34:36 viewgame nukes log.txt 2010-09-16T00:34:42 yeah but thats really verbose move output 2010-09-16T00:34:44 change the log.txt argument to log2.txt 2010-09-16T00:34:54 i just want the "Turn 1 \n Turn 2\n " ... 2010-09-16T00:35:00 oh.. 2010-09-16T00:35:02 Player 1 Wins is whats important 2010-09-16T00:35:02 umm 2010-09-16T00:35:21 or change the "| java -jar ShowGame.jar" to "| tee output.log | java -jar ShowGame.jar" 2010-09-16T00:35:27 yeah 2010-09-16T00:35:29 that will work 2010-09-16T00:35:38 dump everything going to showgame also to a file 2010-09-16T00:35:58 yes, that will write what playgame outputs in output.log 2010-09-16T00:36:17 is there any command i could do to just write what playgame outputs to stdout and discard the big output at the end? 2010-09-16T00:36:29 no files involved? 2010-09-16T00:36:52 you could just drop the "| java -jar ShowGame.jar" 2010-09-16T00:37:15 oh, you mean you don't want the playback 2010-09-16T00:37:18 yeah 2010-09-16T00:37:23 and right now all im getting is the playback 2010-09-16T00:37:26 none of the stuff i want 2010-09-16T00:37:33 right, so 2010-09-16T00:37:41 the stuff you want is coming from stderr 2010-09-16T00:37:47 really? 2010-09-16T00:38:02 yes, stdout is being piped to the showgame jar 2010-09-16T00:38:07 but std is shown 2010-09-16T00:38:10 that makes a lot of sense 2010-09-16T00:38:15 stdout* is show 2010-09-16T00:38:32 the | will eat stdout and you won't see it 2010-09-16T00:39:09 if you don't want to run showgame, you can drop the | and everything after, and add "> /dev/null" which will toss the playback 2010-09-16T00:39:35 what has to be in the zipfile I submit? I'm assuming my bot file and everything it needs to compile? Do I submit source, or compiled code? Do I have to name it MyBot.java/class or can I retain ErrBot.java/class? 2010-09-16T00:40:18 rogue780: there has to be a MyBot.java and submit all source, class files will be nuked before compile 2010-09-16T00:40:35 Zannick, thanks 2010-09-16T00:41:28 rogue780: uncompiled and all the source needed to compile, main file must be MyBot.[appropriate extension for language] 2010-09-16T00:41:46 i.e. MyBot.java 2010-09-16T00:42:42 languages that compile multiple files into one binary don't have to put the main() function in the MyBot source file, but java is not one of those languages 2010-09-16T00:43:17 Zannick, yeah. lame. I have to rename the file, the class, and all the references. 2010-09-16T00:43:35 rogue780: not really 2010-09-16T00:43:53 make a new class 2010-09-16T00:44:17 whose main calls ErrBot.main 2010-09-16T00:45:21 wow, I must be more tired than I thought 2010-09-16T00:45:26 why didn't I think of that? 2010-09-16T00:45:41 and why did I forget to recompile my source before I began playing it again on the tcp server? 2010-09-16T00:45:59 sorry to whoever errbot is playing. im about to timeout 2010-09-16T00:50:17 *** dmj111 has left #aichallenge ("ERC Version 5.3 (IRC client for Emacs)") 2010-09-16T00:54:15 So, apparently there can't be any compiler warnings either 2010-09-16T00:57:08 *** eburnette_ has quit IRC (Ping timeout: 252 seconds) 2010-09-16T01:02:52 *** mightybyte has joined #aichallenge 2010-09-16T01:03:36 *** mightybyte has left #aichallenge 2010-09-16T01:05:12 *** whiteside has quit IRC (Ping timeout: 252 seconds) 2010-09-16T01:09:38 *** Queue29 has quit IRC (Remote host closed the connection) 2010-09-16T01:10:05 woohoo, scored my first win against deepblue 2010-09-16T01:11:12 rwa: link 2010-09-16T01:11:31 http://www.benzedrine.cx/planetwars/canvas?game_id=2803 2010-09-16T01:11:37 cool 2010-09-16T01:12:17 *** staiii has joined #aichallenge 2010-09-16T01:12:27 the visualizer needs more flashyness 2010-09-16T01:12:29 more glow! 2010-09-16T01:14:21 *** rogue780_ has joined #aichallenge 2010-09-16T01:14:31 Howdy 2010-09-16T01:16:17 only thing i really dislike about being an organizer 2010-09-16T01:16:23 i have no idea what ppl are doing 2010-09-16T01:16:29 i want stories 2010-09-16T01:16:43 how are you making your bot be epic? 2010-09-16T01:16:49 what smart startegy did you make? 2010-09-16T01:17:01 i don't wanna wait till postmortem 2010-09-16T01:17:29 *** hornairs has quit IRC (Quit: hornairs) 2010-09-16T01:18:21 amstan cant you enter your own bot and just have it removed when finals come? 2010-09-16T01:18:35 or play with a bot on tcp server 2010-09-16T01:18:46 we decided it a long time ago that this would not be fair 2010-09-16T01:18:52 tcp server yes 2010-09-16T01:19:01 but like.. i feel out of the loop now 2010-09-16T01:19:50 *** hornairs has joined #aichallenge 2010-09-16T01:27:09 amstan: I'm doing well enough that I don't want to talk strategy yet. When bots run away from mine and it looks like I have no hope of placing top 10, I'll talk strategy. 2010-09-16T01:28:35 amstan: the tcp server actaully have a fair range of users 2010-09-16T01:28:44 and the games are instant 2010-09-16T01:29:04 a fair few of the top players have their bots running on there 2010-09-16T01:29:33 yeah, def come play with us on tcp server if you have an itch to play 2010-09-16T01:29:37 tbh it feels MORE in the 'loop' than the regular server 2010-09-16T01:29:46 because you dont have to wait 9 hours for a game 2010-09-16T01:29:51 I have to say this is more fun than I've had in awhile programming 2010-09-16T01:31:31 is greghaynes here yet? 2010-09-16T01:34:25 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-09-16T01:36:39 *** DirtyKeyboard has quit IRC (Quit: DirtyKeyboard) 2010-09-16T01:38:14 *** kozlovsky has joined #aichallenge 2010-09-16T01:45:29 fun yes, but alot harder then expected :O first time i try to make a bot like this 2010-09-16T01:51:30 just added a sleep to my bots, off to watch david copperfield on agt 2010-09-16T01:52:25 *** syntaxglitch has quit IRC (Ping timeout: 240 seconds) 2010-09-16T01:59:07 *** HakanD has joined #aichallenge 2010-09-16T01:59:52 *** graphitemaster has joined #aichallenge 2010-09-16T02:00:04 amstan, has op hehe 2010-09-16T02:00:06 graphitemaster: so... ask in here questions 2010-09-16T02:00:12 graphitemaster: yes, i'm one of the organizers 2010-09-16T02:00:31 *** hellman has quit IRC (Remote host closed the connection) 2010-09-16T02:00:35 where is your implementation? 2010-09-16T02:00:41 i don't have one.. 2010-09-16T02:00:56 i have a python kit if you want to use it, but i doubt you want python: http://github.com/amstan/planetwars-python-kit 2010-09-16T02:01:05 graphitemaster: there's a c++ starter package 2010-09-16T02:01:12 yeah, already tweaking it :D 2010-09-16T02:01:34 i strongly suggest you use this visualizer for testing though, it's much more advanced: http://github.com/DanielVF/Planet-Wars-Canvas-Visualizer 2010-09-16T02:01:53 it's the one that's used on the main site 2010-09-16T02:01:53 how do I run these4 .jar files :( 2010-09-16T02:02:16 huh? 2010-09-16T02:02:20 what files? 2010-09-16T02:02:49 like the visualiser. 2010-09-16T02:03:07 ShowGame.jar 2010-09-16T02:03:16 sorry, here I don't use Java, so... 2010-09-16T02:03:57 amstan, folder "tools" 2010-09-16T02:04:26 graphitemaster: i just told you not to use that visualizer 2010-09-16T02:04:41 for the game engine, that's pretty stable, just have java there shouldn't be problems 2010-09-16T02:04:50 I'm not using the one you sent, it's python 2010-09-16T02:04:52 if you don't like it.. make your own engine in c++ 2010-09-16T02:05:01 http://github.com/DanielVF/Planet-Wars-Canvas-Visualizer 2010-09-16T02:05:03 that one.. 2010-09-16T02:05:22 webbased, javascript. 2010-09-16T02:05:38 graphitemaster: ok.. you don't like it... you don't have to participate? ok 2010-09-16T02:05:39 I'll stick to "java -jar ShowGame.jar" 2010-09-16T02:05:42 i'm tired already 2010-09-16T02:05:47 *** TapewormGalactos has joined #aichallenge 2010-09-16T02:05:53 anyway.. 2010-09-16T02:05:55 i'm going to bed 2010-09-16T02:06:00 ask greghayes to help you 2010-09-16T02:06:14 contestbot: what repo 2010-09-16T02:06:14 amstan: repo = http://code.google.com/p/ai-contest/ 2010-09-16T02:06:22 that's the repository, you can find moar code there 2010-09-16T02:09:54 *** TapewormGalactos has quit IRC (Client Quit) 2010-09-16T02:18:40 *** otherAntimatroid has joined #aichallenge 2010-09-16T02:19:50 @tcpserver 2010-09-16T02:19:51 delt0r___: tcpserver = http://www.benzedrine.cx/planetwars/ 2010-09-16T02:25:28 *** Naith has quit IRC (Quit: Naith) 2010-09-16T02:38:13 *** phreeza has joined #aichallenge 2010-09-16T02:41:28 *** Yoshi-TS4 has joined #aichallenge 2010-09-16T02:43:12 *** dhaval_ has joined #aichallenge 2010-09-16T02:44:13 *** Palmik has joined #aichallenge 2010-09-16T02:45:58 *** vizier has joined #aichallenge 2010-09-16T02:46:58 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-09-16T02:46:58 *** dhaval_ has quit IRC (Client Quit) 2010-09-16T02:47:02 http://www.benzedrine.cx/planetwars/canvas?game_id=3365 2 of my bots fought 2010-09-16T02:50:53 *** tapwater has quit IRC (Quit: tapwater) 2010-09-16T02:52:27 *** Janzert has left #aichallenge 2010-09-16T03:03:33 *** Utkarsh has joined #aichallenge 2010-09-16T03:07:06 that custom visuliser is impossible to get working 2010-09-16T03:07:22 the command listed on the site dun work 2010-09-16T03:09:45 *** Snowplan has left #aichallenge 2010-09-16T03:12:05 yea it doesnt seem to work so good 2010-09-16T03:12:43 i actually managed to make it start but it just spewed like 1000 lines of numbers into the console 2010-09-16T03:15:26 *** nullkuhl has joined #aichallenge 2010-09-16T03:16:39 what was round 1 about ? i came too late i guess.. are they independent ? can i join in round 2 directly ? 2010-09-16T03:17:08 *** Utkarsh has quit IRC (Ping timeout: 264 seconds) 2010-09-16T03:20:43 damnit, peoples bots are getting better and i have too much damn uni work 2010-09-16T03:21:57 hahaha 2010-09-16T03:22:39 dont use the python visualizer it doesnt work well 2010-09-16T03:25:20 *** Utkarsh has joined #aichallenge 2010-09-16T03:28:06 http://ai-contest.com/forum/viewtopic.php?f=18&t=621 perl starter package 2010-09-16T03:30:56 *** staiii has quit IRC (Ping timeout: 264 seconds) 2010-09-16T03:34:35 *** sigh has quit IRC (Remote host closed the connection) 2010-09-16T03:35:03 *** staiii has joined #aichallenge 2010-09-16T03:36:54 animatroid: you have an intelligent working bot at least 2010-09-16T03:37:49 *** Kastra has joined #aichallenge 2010-09-16T03:38:53 *** delt0r_ has joined #aichallenge 2010-09-16T03:40:54 *** delt0r___ has quit IRC (Ping timeout: 272 seconds) 2010-09-16T03:49:37 *** Palmik has joined #aichallenge 2010-09-16T03:51:15 *** aerique has joined #aichallenge 2010-09-16T03:51:31 this way lies madness 2010-09-16T03:56:17 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2010-09-16T04:02:58 *** Utkarsh has joined #aichallenge 2010-09-16T04:03:37 *** Kastra has quit IRC (Ping timeout: 240 seconds) 2010-09-16T04:03:46 *** jaspervdj has joined #aichallenge 2010-09-16T04:05:23 *** staiii has quit IRC (Ping timeout: 240 seconds) 2010-09-16T04:08:11 *** Utkarsh has quit IRC (Ping timeout: 265 seconds) 2010-09-16T04:10:06 http://www.benzedrine.cx/planetwars/canvas?game_id=2386 hehe 2010-09-16T04:10:51 *** dhartmei has joined #aichallenge 2010-09-16T04:13:11 rofl 2010-09-16T04:13:17 bot failure 2010-09-16T04:14:10 *** amstan has quit IRC (Ping timeout: 245 seconds) 2010-09-16T04:17:59 *** Utkarsh has joined #aichallenge 2010-09-16T04:21:02 *** sinan has joined #aichallenge 2010-09-16T04:21:48 *** boegel has joined #aichallenge 2010-09-16T04:23:02 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-09-16T04:28:54 *** Utkarsh has joined #aichallenge 2010-09-16T04:32:57 *** Naktibalda has joined #aichallenge 2010-09-16T04:34:51 *** sigh has joined #aichallenge 2010-09-16T04:44:46 http://www.benzedrine.cx/planetwars/canvas?game_id=4028 2010-09-16T04:51:03 *** Theobon has quit IRC (Ping timeout: 276 seconds) 2010-09-16T04:52:27 *** s_denchev has joined #aichallenge 2010-09-16T04:57:05 *** smellyhippy has joined #aichallenge 2010-09-16T04:57:10 lo all 2010-09-16T04:57:18 yo yo 2010-09-16T04:58:11 I was wondering if anyone was using c# for thier bot and if they had tried to make a test runner using the Process class to execute the PlayGame.jar ? I'm having aweful trouble getting the result of a match back from the console 2010-09-16T04:59:31 that or if i had the source for PlayGame.jar I could modify it output what i require 2010-09-16T05:00:03 http://ai-contest.com/forum/viewtopic.php?f=18&t=621 perl starter package 2010-09-16T05:01:25 Accoun: create an issue http://code.google.com/p/ai-contest/issues/list 2010-09-16T05:02:37 ahhh hah ... it's using the standard error stream to report the text and the stanrad out to pass the data ... wish i could have noticed that last night ... oh well 2010-09-16T05:11:49 *** boegel has quit IRC (Quit: Leaving) 2010-09-16T05:12:25 rwa, hh, nice game xD a little repetitive :P 2010-09-16T05:12:45 yeah, that one was odd 2010-09-16T05:12:48 how about a C starter package :D 2010-09-16T05:13:17 benzedrine.cx is now ending games on invalid moves (previously they were just ignored) 2010-09-16T05:13:58 yep 2010-09-16T05:14:26 found a bug that way 2010-09-16T05:15:13 dhartmei: tonight your bot made an invalid move in final turn against my bot ::) 2010-09-16T05:15:38 yes, i had a nasty bug that lost me some games i should have own :) 2010-09-16T05:17:54 dhartmei: is it a setting or a patch? can't benzedrine return a reason of timeout? 2010-09-16T05:18:06 it would be helpful :) 2010-09-16T05:18:51 it tells the client why the order is invalid 2010-09-16T05:19:23 or when someone disconnects or times out 2010-09-16T05:19:26 *** botman has joined #aichallenge 2010-09-16T05:20:00 oh, you mean display in the games list if a game was not properly finished? yes, i want to add that. 2010-09-16T05:20:11 that would be great 2010-09-16T05:21:01 hmm, I haven't noticed that tcp display something else than Timeout 2010-09-16T05:25:13 I have yet to score a win against dmj, but I need to go to bed!! :( 2010-09-16T05:31:35 http://www.benzedrine.cx/planetwars/canvas?game_id=4309 2010-09-16T05:32:27 *** jerkface03 has quit IRC (Ping timeout: 276 seconds) 2010-09-16T05:35:45 The server si shutting down :) 2010-09-16T05:36:41 benzedrine? 2010-09-16T05:37:03 yep 2010-09-16T05:37:24 just for a sec or for awhile? 2010-09-16T05:37:25 btw... with how much elo do you start on benzedrine? 2010-09-16T05:37:30 no idea 2010-09-16T05:37:45 rwa, dunno, I!m just forwarding the message server i giving to me :) 2010-09-16T05:37:53 *I'm 2010-09-16T05:38:10 that sucks cause I wanna run my bots overnight 2010-09-16T05:38:42 Palmik: it says that your score is -999999 2010-09-16T05:38:46 if you have infinite loop, it will join 2010-09-16T05:38:57 Naktibalda, yeah, I know... is that really so? 2010-09-16T05:39:33 *** oh_cripes has joined #aichallenge 2010-09-16T05:40:33 *** phreeza has joined #aichallenge 2010-09-16T05:41:55 alright, now showing abnormal game ends (T=timed out, D=disconnected, I=invalid input) on the games list 2010-09-16T05:42:22 only for games played from now on 2010-09-16T05:42:46 dhartmei, nice :) 2010-09-16T05:43:56 dhartmei, btw... for updating the stats, you use cron I guess? 2010-09-16T05:44:28 could you share the command? :) 2010-09-16T05:45:31 are you building another server? 2010-09-16T05:45:39 it's in the current server.tar.gz, rate.sh script: printf "prompt off\nreadpgn results.pgn\nelo\nmm\nadvantage 0\nexactdist\nratings\nx\nx\n" | ./BayesElo/bayeselo >ratings.txt.new 2>&1 2010-09-16T05:46:09 dhartmei, thanks :) 2010-09-16T05:46:58 Naktibalda, yep :) We are using some free virtual server for dueling :) 2010-09-16T05:47:51 moin 2010-09-16T05:48:47 dhartmei: what do you think of scraping the html of benzedrine to get the playback of games 2010-09-16T05:49:24 *** num1 has joined #aichallenge 2010-09-16T05:49:34 you want all games? 2010-09-16T05:49:54 seemed like an obvious thing to do if you want to do a statistical approach 2010-09-16T05:50:10 http://www.benzedrine.cx/planetwars/canvas?game_id=4417 Hmm... why is not the red planet on the left captured? 2010-09-16T05:50:29 i can zip up the files for you, so you only need to download a single compressed tarball... 2010-09-16T05:50:52 Palmik: which turn? 2010-09-16T05:50:56 dhartmei: sounds usefull, you'd post a new one every day or so ? 2010-09-16T05:51:15 yes 2010-09-16T05:51:38 Naktibalda, 281 for example 2010-09-16T05:51:41 there are more 2010-09-16T05:51:42 Palmik: smr: knows it has one, so it goes defensive 2010-09-16T05:51:52 s/one/won 2010-09-16T05:52:11 Palmik: growth rate is 4, 7+4 = 11 11-11 = 0, so you don't capture it 2010-09-16T05:52:17 you need 1 ship more 2010-09-16T05:52:32 ahnd then smr would have sent one more in defensive support 2010-09-16T05:52:38 :) 2010-09-16T05:52:51 so I guess I made some mistake in my calculations :P 2010-09-16T05:52:59 possibly 2010-09-16T05:53:09 no, it's just a n oponent is much smarter 2010-09-16T05:53:16 or you can't win from that possition of smr defends in that style 2010-09-16T05:53:25 he is sending ships out all the time, but keeps a correct number 2010-09-16T05:53:40 bartwe_: 4500 games right now are 60MB .tar.gz.... 2010-09-16T05:54:12 Palmik: try formula num_ships + growth_rate * distance 2010-09-16T05:55:03 if you send more than that, oponent will send more ships to that planet 2010-09-16T05:55:33 Naktibalda, I'm doing along those lines... does not matter :) 2010-09-16T05:56:07 dhartmei: thats alot smaller then i expected 2010-09-16T05:56:39 Palmik: you have sent 19 ships for faraway planets to a planet that can grow this amount in 5 turns 2010-09-16T05:56:55 dhartmei: should make for a great testbed for the testing of futurer prediciting logic 2010-09-16T05:57:01 still, i'd rather upload it to sendspace.com or such, and then people fetch it from there 2010-09-16T05:57:07 Naktibalda, the bot is 50 lines, I!m not even taking distance in account yet :) 2010-09-16T05:57:08 dhartmei: sure 2010-09-16T05:57:25 not sure how much gb/month my slicehost has 2010-09-16T05:57:29 ok, i'll prepare the first upload later today 2010-09-16T05:58:06 guess i could host maybe upto a 50GB of cummulative bandwidth 2010-09-16T05:58:36 dhartmei: you rock 2010-09-16T06:00:02 what was round 1 about ? i came too late i guess.. are they independent ? can i join in round 2 directly ? 2010-09-16T06:01:04 round 1? the contest doesn't have rounds, you can join anytime before the deadline. 2010-09-16T06:01:27 do you mean a first contest? 2010-09-16T06:01:43 *** vikhyat has joined #aichallenge 2010-09-16T06:02:56 *** Utkarsh has quit IRC (Ping timeout: 245 seconds) 2010-09-16T06:03:23 dhartmei, is D really disconnect? Not timeout? 2010-09-16T06:03:57 Oh nice, a real victory against dhartmei: http://www.benzedrine.cx/planetwars/canvas?game_id=4503 2010-09-16T06:04:08 yes, if the client crashes/segfaults, you get a disconnect, for instance 2010-09-16T06:04:24 dhartmei, ok, thanks :) 2010-09-16T06:04:39 the latest tcp.c should show the reason for the death of the child, too (SIGSEGV, exit with status, etc) 2010-09-16T06:05:08 shartmei: i keep getting 'kill: permission denied' at the end of games 2010-09-16T06:05:14 not sure why you put that in there 2010-09-16T06:05:16 dhartmei, ok, I shall recompile then :) 2010-09-16T06:06:10 dhartmei: woops 2010-09-16T06:06:19 my typo rate is horrible 2010-09-16T06:06:30 the child (the bot) should terminate when it get EOF on stdin, some bots don't 2010-09-16T06:06:30 too many different keyboard layouts 2010-09-16T06:06:52 you can edit out the kill() call, if that's no issue with yours 2010-09-16T06:06:58 some people creates many versions of bot every day, it looks like my rate will be weekly :( 2010-09-16T06:07:37 Nak: some people spam with lots of possible parameter tweeaks, and let the bots fight it out 2010-09-16T06:07:48 thats not really representative of versions 2010-09-16T06:08:04 also how much difference do you think there is beterween .77 and .8 of mine ? 2010-09-16T06:08:10 maybe a dozen lines 2010-09-16T06:08:10 but your bots look like versions 2010-09-16T06:08:18 *** Utkarsh has joined #aichallenge 2010-09-16T06:08:20 mine are versions yes 2010-09-16T06:09:59 then your changes are more incremental, and mine is a complete rewrite 2010-09-16T06:10:53 still, that doesn't express either success or investment 2010-09-16T06:10:57 do what works for you 2010-09-16T06:11:08 i might plateau while you keep growing 2010-09-16T06:11:10 dhartmei, hmm, the new version does not output the game progress? 2010-09-16T06:12:16 seems like that... no biggie :) 2010-09-16T06:12:41 yes, duplicate stdin/stdout to stderr in your bot, if you still want to see it 2010-09-16T06:12:54 ok :) 2010-09-16T06:13:54 is there a limit on the number of fleets you can have in the official server? 2010-09-16T06:14:00 will there be? 2010-09-16T06:14:03 hope not 2010-09-16T06:14:19 is it a valid strategy to max out one's fleet count to crash the opponent? 2010-09-16T06:14:27 it shouldn't be an issue :) 2010-09-16T06:14:31 no, it's sgainast the rules :) 2010-09-16T06:14:35 *against 2010-09-16T06:14:39 how much time does parsing the game state take for a something like 1000 fleets? 2010-09-16T06:14:45 since it's not how sportsman would behave :P 2010-09-16T06:15:01 if you own 23 planets, and you send out one ship from each planet to each other planet each turn, how many fleets will you reach? 2010-09-16T06:15:12 > 1000 :) 2010-09-16T06:15:23 1000 fleets are like 10kb of text, should be very fast 2010-09-16T06:15:28 but limiting the abount of fleets would only add more complexity imo and might completely change the way you approach the game 2010-09-16T06:16:19 but if there is officially no limit, then maxing out fleet count (for whatever purpose) must be legal as well, no? 2010-09-16T06:16:49 I think so 2010-09-16T06:16:58 it shouldn't help you to win against top bots :) 2010-09-16T06:17:09 dhartmei, and yould you do that? I mean doing that to crah your opponent? 2010-09-16T06:17:30 i just noticed my bot crashes (or times out) on large fleet inputs, so i was wondering 2010-09-16T06:17:32 Palmik: I think it's a "who will crash first" game :) 2010-09-16T06:18:04 it's not just reading and parsing the fleet list, it's doing "computations" with the fleets :) 2010-09-16T06:18:13 Entries which are deemed to violate the spirit of fair and sportsmanlike competition will be disqualified without any opportunity for appeal. 2010-09-16T06:18:14 hmm, I should get rid of Fleet objects, they are useless 2010-09-16T06:18:17 might apply to that 2010-09-16T06:19:19 Hmm... I got 3 DCs in row on first turn... 2010-09-16T06:19:35 well, there is a physical limit, you can't send more than growth_rate fleets every turn :) 2010-09-16T06:19:38 have to do something about that :( 2010-09-16T06:20:23 dhartmei, what is the tim elimit for the first turn on your server? 2010-09-16T06:20:23 runtime error? 2010-09-16T06:20:25 *time 2010-09-16T06:20:47 DC means that your bot crashed 2010-09-16T06:20:52 Naktibalda, possible, but it's hard to debug 2010-09-16T06:21:29 add a log message (to stderr) when you end a turn, see if you end the first turn :) 2010-09-16T06:21:30 *** danielvf has quit IRC (Read error: Connection reset by peer) 2010-09-16T06:21:35 *** danielvf has joined #aichallenge 2010-09-16T06:21:48 ok 2010-09-16T06:22:14 you can print anything to stderr, it won't get sent to the server, and you'll see it 2010-09-16T06:22:57 startup, begin of turn, end of turn, orders you issue, planets/fleets at start of turn, etc. :) 2010-09-16T06:24:04 it only happens in first turn, and I have different functions for initialising and updating the game state, so at least I know here to look at, thanks :) 2010-09-16T06:24:30 Palmik: the timeout is 5s for every turn, but the network latency is in the order of 1000ms at times. i suggest measuring the time in the bot as well, and printing it at the end of the turn. 2010-09-16T06:26:09 dhartmei: sorry to bother you but do you have tcp.c written in another language? ideall java or c#? 2010-09-16T06:26:27 *** nullkuhl has quit IRC (Ping timeout: 276 seconds) 2010-09-16T06:26:31 smellyhippy, java version id on the forums 2010-09-16T06:26:32 (I'm guess no otherwise it'd be linked on the site ;)) 2010-09-16T06:26:39 *is 2010-09-16T06:26:42 Palmik: ohhh I'll have a look :) 2010-09-16T06:26:46 thanks 2010-09-16T06:26:48 np 2010-09-16T06:27:42 i actually thought the forums were totally broken :P but it just seems to be with IE 2010-09-16T06:27:44 * Naktibalda is running his bot in free shell server, no more timeouts! 2010-09-16T06:29:29 UloPe-04.7 is loosing because of invalid timeout everytime :) 2010-09-16T06:29:34 *** Bobng has joined #aichallenge 2010-09-16T06:29:47 err, invalid input 2010-09-16T06:30:38 yep 2010-09-16T06:32:24 ErrBot3.2a does the same thing faster (good name by the way) :) 2010-09-16T06:32:28 Ah. Is that being checked now? Then I maybe should disable my bot until i fix this 2010-09-16T06:40:09 *** sinan has quit IRC (Ping timeout: 272 seconds) 2010-09-16T06:47:40 *** zaph has joined #aichallenge 2010-09-16T06:47:55 *** dmj111 has joined #aichallenge 2010-09-16T06:48:57 *** zaph has quit IRC (Client Quit) 2010-09-16T06:51:00 Planet-Wars-Canvas-Visualizer: Daniel Von Fange master * r4ed12ad / js/visualizer.js : Linkify player names. From patch from gvsmirnov.CP - http://bit.ly/cVsVsY 2010-09-16T06:51:56 Palmik: oh, your bot won in 54 turns against mine 2010-09-16T06:52:29 :) 2010-09-16T06:53:59 *** ratatata has joined #aichallenge 2010-09-16T06:55:07 *** storrgie has joined #aichallenge 2010-09-16T06:55:10 hello! 2010-09-16T06:56:30 *** vikhyat has quit IRC (Ping timeout: 255 seconds) 2010-09-16T07:00:25 *** jmreardon has joined #aichallenge 2010-09-16T07:01:43 http://www.benzedrine.cx/planetwars/canvas?game_id=4891 2010-09-16T07:01:48 ha, nicely done 2010-09-16T07:01:56 i hope your not overfit to my bot 2010-09-16T07:02:02 but great work anyways 2010-09-16T07:03:25 *** vikhyat has joined #aichallenge 2010-09-16T07:03:40 very good :) 2010-09-16T07:04:00 time for me to implement the neutral capture fitness function :) 2010-09-16T07:04:31 what is a fitness function? 2010-09-16T07:05:36 heuristic that gives a number to choices that indicates their 'fitness' ie. how good the choice is 2010-09-16T07:05:51 ah, right, I have something like that 2010-09-16T07:09:02 smr's bots also send quite a lot of invalid orders :) 2010-09-16T07:09:36 really ? 2010-09-16T07:10:40 bartwe_: I beat them a few times yesterday by virtue of them issuing bad orders 2010-09-16T07:10:42 also shows the weekness of elo to show rock/paper/scissor bots 2010-09-16T07:13:40 *** RainCT has joined #aichallenge 2010-09-16T07:14:51 bartwe_, 8 times almost in row against me 2010-09-16T07:22:22 i built tcp.c binary with cygwin, if anyone is using windows and doesn't use the tcp server yet: http://ai-contest.com/forum/viewtopic.php?f=18&t=424&start=40#p3786 2010-09-16T07:24:36 *** hellman has joined #aichallenge 2010-09-16T07:26:41 I'm trying to get connected to the test server at benzedrine.cx. I just wanted to check if 213.3.30.106:9999 is still the right IP to use? As a test i connected to it using putty and sent the USER command manually but i didnt get a reply 2010-09-16T07:26:48 dhartmei: ohhh that might be easier ;) 2010-09-16T07:28:54 dhartmei: could you consider temporarily banning bots who routinely timeout? 2010-09-16T07:32:11 *** Olathe has joined #aichallenge 2010-09-16T07:33:27 http://baphomet.dontexist.com/tcp.class for pre compiled tcp.c that works in windows for anyone that wants it 2010-09-16T07:35:05 dhartmei: thanks for compiling that ... that was super useful and I've just ran my run on the server \o/ now for a bit of wireshark to see why my cleitn isnt behaving 2010-09-16T07:40:00 Pallmik: try a random sleep to avoid getting reset against the same 2010-09-16T07:48:37 *** bob___ has joined #aichallenge 2010-09-16T07:50:24 wtf 2010-09-16T07:50:27 passpack isn't loading 2010-09-16T07:52:48 Palmik: what's wrong with your bot? http://www.benzedrine.cx/planetwars/canvas?game_id=5164 issueOrder doesn't work? 2010-09-16T07:53:53 yes, it does the same in all games 2010-09-16T07:59:35 Naktibalda, hmm... possible I broke it :) 2010-09-16T08:00:20 thanks for report 2010-09-16T08:01:36 *** EvGenius has joined #aichallenge 2010-09-16T08:02:36 *** Yoshi-TS4 has quit IRC (Quit: Minus One IRC) 2010-09-16T08:07:14 *** ermau has joined #aichallenge 2010-09-16T08:08:52 *** ermau is now known as ermau|work 2010-09-16T08:09:04 *** s_denchev has quit IRC (Quit: Page closed) 2010-09-16T08:11:29 http://ai-contest.com/visualizer.php?game_id=4501311 2010-09-16T08:11:30 http://ai-contest.com/forum/viewtopic.php?f=18&t=621 perl starter package 2010-09-16T08:11:50 that is what will happen if we make rankings take into account number of ships :p 2010-09-16T08:12:51 lol 2010-09-16T08:13:03 strange 2010-09-16T08:13:10 krokkro had moves left too 2010-09-16T08:14:35 yea 2010-09-16T08:14:42 but that's the basic reason anyways 2010-09-16T08:14:45 well one of em 2010-09-16T08:15:00 if games finish in 50 turns that's good, the faster games finish the more we can run 2010-09-16T08:15:15 if every game ends up taking exactly 200 turns, we won't run as many games 2010-09-16T08:17:12 I had a thought. You limit each turn to 1 second of CPU time 2010-09-16T08:17:25 why not limit each game to 1*max_turns worth of cpu time? 2010-09-16T08:17:42 Bobng: actually the limit is 1 second of wall time 2010-09-16T08:17:59 Bobng: and i wish it was 1* max_turns worth of cpu time 2010-09-16T08:18:05 oh yeah, but that is being changed soon isn't it? 2010-09-16T08:18:07 it would make things so much easier 2010-09-16T08:18:13 I submitted some code that measures the cpu time 2010-09-16T08:18:24 firstly, we could impose the limit by just using limits.conf lolllll 2010-09-16T08:18:43 *** aaa_ has joined #aichallenge 2010-09-16T08:18:52 yeah but then you couldn't limit it per turn could you 2010-09-16T08:19:43 Bobng: thats what i was saying, i wish it was just like speed chess 2010-09-16T08:19:55 but it's not, and that would be a pretty big change 2010-09-16T08:19:57 to the rules 2010-09-16T08:19:58 D: 2010-09-16T08:20:02 that would be much simpler 2010-09-16T08:20:07 it wouldn't be much of a change 2010-09-16T08:20:14 *** sinan has joined #aichallenge 2010-09-16T08:20:25 I think it would work nice personally though 2010-09-16T08:20:28 yes a total amount of time would be nicer 2010-09-16T08:20:53 i'll post about it on the mailing list if people actually like the idea 2010-09-16T08:21:37 would be much simpler to code as well. 2010-09-16T08:21:50 doubt it 2010-09-16T08:21:58 doubt what? 2010-09-16T08:21:58 why? 2010-09-16T08:22:16 tracking the cpu time per turn means you need to track the turn state 2010-09-16T08:22:36 tracking the cpu time in total means you only need to track the time the program has been executing 2010-09-16T08:22:39 i doubt it would be easier to code 2010-09-16T08:22:45 its already coded. 2010-09-16T08:23:12 oh you meant easier for the server ! 2010-09-16T08:23:12 have different divisions depending on how long it takes for the bot to run 2010-09-16T08:23:16 http://code.google.com/p/ai-contest/issues/detail?id=35 server.py patch 2010-09-16T08:23:21 yeah, easier for the server. 2010-09-16T08:23:43 for the client it rewards people who can get stuff done in fewer turns 2010-09-16T08:24:16 if you have 200 seconds of cpu time, and you can win in 50 turns, that's 4 seconds of cpu time per turn 2010-09-16T08:24:36 yeah 2010-09-16T08:24:47 it let's you front load your bot too 2010-09-16T08:25:04 you can use even 10seconds of cpu time for the first turn to run heavy calculations 2010-09-16T08:25:08 yeah 2010-09-16T08:25:14 like map the game out 2010-09-16T08:25:24 plan ahead etc. 2010-09-16T08:25:58 *** hellman has quit IRC (Remote host closed the connection) 2010-09-16T08:26:03 i never actually had to care, but im sure the languages have standard ways of getting the cpu time you've used in the process 2010-09-16T08:26:17 fischer timing with some new draw rule 2010-09-16T08:27:31 dstufft not really 2010-09-16T08:27:51 well at least not with python and I think php and perl 2010-09-16T08:28:15 i hope they leave this up after the competition is finished 2010-09-16T08:28:33 you would have to make it yourself, it wouldn't be hard at all though. 2010-09-16T08:28:54 wouldnt be hard if you know what you are doing :p 2010-09-16T08:29:24 bob___: well ideally the starer bots would be modified to include help in that regards then 2010-09-16T08:30:18 There isnt a website that keeps track of competitions like this is there? 2010-09-16T08:30:27 One that has all the AI competitions and rankings 2010-09-16T08:31:25 not just for this competition but for all of them 2010-09-16T08:32:24 *** nullkuhl has joined #aichallenge 2010-09-16T08:33:08 posted on the mailing list 2010-09-16T08:33:17 regarding using Speed Chess Timing 2010-09-16T08:33:19 *** Appleman1234 has quit IRC (Ping timeout: 265 seconds) 2010-09-16T08:34:26 *** DirtyKeyboard has joined #aichallenge 2010-09-16T08:34:29 *** kuwabara has quit IRC (Read error: Operation timed out) 2010-09-16T08:35:29 *** oh_cripes has quit IRC (Ping timeout: 252 seconds) 2010-09-16T08:42:23 *** bob___ has quit IRC (Quit: Page closed) 2010-09-16T08:45:17 *** kuwabara has joined #aichallenge 2010-09-16T08:50:44 dstufft: what is the mailing list addr? 2010-09-16T08:50:51 I might subscribe and give my 2c 2010-09-16T08:54:15 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-09-16T09:03:53 *** aaa_ has quit IRC (Quit: Page closed) 2010-09-16T09:13:59 *** mightybyte has joined #aichallenge 2010-09-16T09:14:15 *** DirtyKeyboard has quit IRC (Quit: DirtyKeyboard) 2010-09-16T09:14:39 *** otherAntimatroid has left #aichallenge 2010-09-16T09:15:37 http://ai-contest.com/forum/viewtopic.php?f=18&t=621 perl starter package 2010-09-16T09:15:54 I think that it's annoying 2010-09-16T09:23:23 I'm annoyed my compile_anything patch hasn't been commented on :< 2010-09-16T09:27:24 *** braddunbar has joined #aichallenge 2010-09-16T09:29:06 *** Palmik_ has joined #aichallenge 2010-09-16T09:29:44 *** Palmik has quit IRC (Ping timeout: 272 seconds) 2010-09-16T09:30:46 *** ademar_ has joined #aichallenge 2010-09-16T09:31:45 *** Palmik_ is now known as Palmik 2010-09-16T09:32:50 hi guys 2010-09-16T09:32:53 good morning 2010-09-16T09:33:04 hi 2010-09-16T09:33:05 hi 2010-09-16T09:33:07 *** ajhager has quit IRC (Ping timeout: 252 seconds) 2010-09-16T09:33:25 hehe .. firtst time i use irc 2010-09-16T09:34:15 is anybody of you is familiar running the tournament_manager.py ? 2010-09-16T09:34:22 *** EvGenius has left #aichallenge 2010-09-16T09:36:39 never touched python before and i'm wondering if there should be a server_info.py that the 'from server_info import' is referecing .. 2010-09-16T09:40:12 woo, looks nearly occult 2010-09-16T09:40:15 http://www.benzedrine.cx/planetwars/canvas?game_id=6411 2010-09-16T09:40:32 i dub thee, blockhole 2010-09-16T09:41:02 haha 2010-09-16T09:41:03 cool (: 2010-09-16T09:41:42 thousands and thoushands in a continal cycle of destruction 2010-09-16T09:42:21 mirrorbot/copybot wishes he was this cool ;p 2010-09-16T09:42:41 still, fighting over a zero growth neutral, silly bot 2010-09-16T09:43:02 tactical position may count for somethign i guess 2010-09-16T09:43:05 *** phreeza has joined #aichallenge 2010-09-16T09:43:05 *** bduc has quit IRC (Read error: Connection reset by peer) 2010-09-16T09:43:07 ademar_: What are you trying to do? 2010-09-16T09:43:23 bartwe: nice 2010-09-16T09:43:41 wow cool game 2010-09-16T09:43:48 wow .. that battle looks kool 2010-09-16T09:43:52 planet is like: om nom nom nom nom 2010-09-16T09:44:27 danielvf: i'm trying to run the tournament_manager.py in my machine 2010-09-16T09:45:09 where does he pulls the configuration data for the connection strings, etc .. 2010-09-16T09:45:24 http://ai-contest.com/forum/viewtopic.php?f=18&t=645 2010-09-16T09:45:35 https://zeus.ugent.be/wiki/Aichallenge 2010-09-16T09:46:53 thanks, i think that will do it .. 2010-09-16T09:48:51 *** HakanD has quit IRC (Ping timeout: 240 seconds) 2010-09-16T09:50:19 *** hakand has joined #aichallenge 2010-09-16T09:50:55 ademar_: why do you want to run your own serve r? 2010-09-16T09:51:02 *** kozlovsky has quit IRC (Ping timeout: 252 seconds) 2010-09-16T09:51:05 silly bugs -_- 2010-09-16T09:53:12 *** Sean_McEligot has joined #aichallenge 2010-09-16T09:54:18 edcba: just testing support for F# and C# .. 2010-09-16T09:54:47 http://www.ai-contest.com/profile.php?user_id=6149 2010-09-16T09:55:04 someone should launch a new server with a good prizes and take over a challenge :) 2010-09-16T09:55:07 *** nigg_ has joined #aichallenge 2010-09-16T09:55:54 Naktibalda, why? 2010-09-16T09:56:22 for fun 2010-09-16T09:56:23 It's fine as it is as long they speed up a thing a bit... at least non-random pairing :) 2010-09-16T09:56:47 Naktibalda, you mean for prizes right? :P 2010-09-16T09:57:10 uhm, right 2010-09-16T09:57:15 not a bad idea 2010-09-16T09:57:41 *** willyd has joined #aichallenge 2010-09-16T09:58:07 ademar_: C# is supported 2010-09-16T09:58:24 *** Bobng has quit IRC (Ping timeout: 265 seconds) 2010-09-16T09:59:24 yes.. c# is supported .. partially .. folk have been complaining they can not use linq .. 2010-09-16T09:59:54 lol lazy ppl ! 2010-09-16T09:59:57 isn't linq a database thing? 2010-09-16T10:00:18 Its query language support for c shart 2010-09-16T10:00:30 it works as fine vs objects as it does vs databases 2010-09-16T10:00:36 what's a use of it for bots? 2010-09-16T10:00:50 oh 2010-09-16T10:01:18 what version of Mono is in contest server? 2010-09-16T10:01:32 hehe 2010-09-16T10:01:36 c# user here ;p 2010-09-16T10:01:41 Naktibalda: it brings some functional programming capabilities into c# and there are other goodies 2010-09-16T10:01:52 linq is unsuited for this contest 2010-09-16T10:02:03 you can with a few lines ask the linq to give you all planets, belonging to you, which have been targeted by enemy ships, and you will get a nice little collection of whatever it finds 2010-09-16T10:02:03 too many allocations 2010-09-16T10:03:27 bartwe_: your bots are c# ? 2010-09-16T10:03:29 however im not quite sure, but i think linq have a bit of overhead, i have not confirmed this and it might be wrong, but it looks like typical overhead-ish heavy work, however there are quite a few good solutions for c# already so.. ;) 2010-09-16T10:03:35 (about linq that is) 2010-09-16T10:03:43 edcba: yes in c# 2010-09-16T10:04:13 edcba: more modern then java, less explosive then c++ 2010-09-16T10:04:45 *** DirtyKeyboard has joined #aichallenge 2010-09-16T10:05:01 *** zerd has quit IRC (Quit: No Ping reply in 180 seconds.) 2010-09-16T10:05:06 *** zerd has joined #aichallenge 2010-09-16T10:08:02 *** sinan has quit IRC (Quit: WeeChat 0.2.6.3) 2010-09-16T10:08:39 *** iFire has quit IRC (Read error: Connection reset by peer) 2010-09-16T10:08:55 *** iFire has joined #aichallenge 2010-09-16T10:09:33 woohoo 2010-09-16T10:09:44 beat McLeopold.5 2010-09-16T10:09:58 *** DirtyKeyboard has quit IRC (Quit: DirtyKeyboard) 2010-09-16T10:10:46 *** sinan has joined #aichallenge 2010-09-16T10:10:50 *** willyd has quit IRC (Ping timeout: 252 seconds) 2010-09-16T10:11:10 http://www.benzedrine.cx/planetwars/canvas?game_id=6898 utterly pwned 2010-09-16T10:11:40 wiw 2010-09-16T10:11:48 'who the heck is macuyiko? 2010-09-16T10:11:53 impressive victory 2010-09-16T10:12:11 interesting strategy 2010-09-16T10:12:37 only take planets your sure to hold onto for long enough for a payback on investment 2010-09-16T10:12:46 all out on first turn -> vulnerable first planet 2010-09-16T10:12:50 he raped your starting 5-gr planet immediately after you launched ships to take over crappy neutrals 2010-09-16T10:13:03 i think my algo would have survived 2010-09-16T10:13:11 as it would have targetted his planet as well 2010-09-16T10:13:17 what aspect would have saved you ? 2010-09-16T10:13:32 why would you attack his planet ? 2010-09-16T10:13:36 my algo right now chooses planets to attack based heavily on proximity and growth rate 2010-09-16T10:13:56 so his planet would have been highest on my priority queue to attack 2010-09-16T10:14:07 i doubt attacking his home is a zero loss strategy due to loss in your relative flexibility compared to his 2010-09-16T10:14:12 *** vikhyat has quit IRC (Ping timeout: 245 seconds) 2010-09-16T10:14:12 *** Queue29 has joined #aichallenge 2010-09-16T10:14:28 it would have been too attractive to pass up 2010-09-16T10:14:36 now if only i could make an Ev+ function out of that 2010-09-16T10:14:48 blister: even when superiorly defended ? 2010-09-16T10:14:56 no 2010-09-16T10:15:06 but (assuming he launched ships), then i would have had more 2010-09-16T10:15:14 a good protection would be protecting home planet if enemy is close in first turns 2010-09-16T10:15:19 yeah 2010-09-16T10:15:28 thats something i dont take into account (and probably should) 2010-09-16T10:15:58 lots of bots doing this hom 2010-09-16T10:16:01 home planet stealing 2010-09-16T10:16:04 and neutral stealing 2010-09-16T10:16:15 neutral stealing is a good strategy 2010-09-16T10:16:16 *** justin_pdx has joined #aichallenge 2010-09-16T10:16:19 yea 2010-09-16T10:16:20 yep 2010-09-16T10:16:25 home stealing is just plain old planet stealing 2010-09-16T10:16:45 what makes this intresting is the very short trip time from a strong position 2010-09-16T10:17:00 i figure the final winner will be someone who's really really good at doing just enough to beat you every time you launch a fleet 2010-09-16T10:17:16 meh 2010-09-16T10:17:45 nope, you can always send a second fleet at second turn, than things get complicated 2010-09-16T10:17:45 bartwe_: are you using multiple intelligence algorithms? 2010-09-16T10:17:56 *** eburnette has quit IRC (Quit: Page closed) 2010-09-16T10:18:04 doing different things at different stages of the game? 2010-09-16T10:18:10 blister: the design allows that, but it would probably suck 2010-09-16T10:18:22 yeah, i think thats where i've been going wrong 2010-09-16T10:18:33 cobalt-1.0 was pretty good (won about 50%) 2010-09-16T10:18:41 then i started adding things to try to be smarter 2010-09-16T10:18:44 i think in this case i incorrectly value the defense force, and also neutrals 2010-09-16T10:18:53 and cobalt-2.0 - 7.3 were all pretty awful 2010-09-16T10:19:00 kept tweaking things and losing even harder 2010-09-16T10:19:01 hah 2010-09-16T10:19:02 been thinking about rooting those in real theory 2010-09-16T10:20:13 1. determine 'stay home' defensive forces 2. determine attacks 3. determine grouping 2010-09-16T10:21:17 except that 1 should be recalculated after 2, whch should be recalculated again, ad inifintum 2010-09-16T10:21:37 delaying actions is also something that would be great 2010-09-16T10:22:06 no there is no delaying you should just conquer the faster the safer 2010-09-16T10:22:35 do you mean reinforcing at the moment of attack? 2010-09-16T10:22:45 instead of 5 turns earlier 2010-09-16T10:23:52 i wish i could figure out how to bolster attacks 2010-09-16T10:23:59 my algo does a great job of picking out targets 2010-09-16T10:24:10 but once i launch an attack, it no longer cares about that target 2010-09-16T10:24:32 it'd be nice if i could wait a turn or two and then send a handful of ships to the same target to help out in case of counter-attack 2010-09-16T10:24:41 lol 2010-09-16T10:25:44 i maintain my prediction of top 5 c++ winners :) 2010-09-16T10:25:44 ok, i need some help 2010-09-16T10:25:53 apparently im sending some kind of invalid command to the server 2010-09-16T10:26:01 and its causing my bot to lose matches that im clearly winning 2010-09-16T10:26:06 http://www.benzedrine.cx/planetwars/canvas?game_id=7050 2010-09-16T10:26:16 which is listed as a lossI for me 2010-09-16T10:26:18 edcba: so you are predicting that the winners will all be search algorithms? 2010-09-16T10:26:27 maybe you are sending more ships than available 2010-09-16T10:26:30 but the server isn't reporting back what invalid input im sending 2010-09-16T10:26:43 edcba: i have code in place to trap that 2010-09-16T10:26:44 jmcarthur: yes 2010-09-16T10:26:44 Palmik: server isn't doing random pairings anymore 2010-09-16T10:26:47 :/ 2010-09-16T10:27:24 edcba: interesting prediction. i'm not going to say either way yet, but that's still interesting 2010-09-16T10:27:36 *** vikhyat has joined #aichallenge 2010-09-16T10:27:57 jmcarthur: you dont think its just a min max problem? 2010-09-16T10:28:00 i think i have an interesting search algorithm in mind, but if i ever reveal it people are going to think it's doomed to failure 2010-09-16T10:28:10 Cyndre: certainly not. not even tron was a minimax problem 2010-09-16T10:28:23 and this is even harder to make minimax 2010-09-16T10:28:30 definitly harder 2010-09-16T10:29:02 *** sinan has quit IRC (Ping timeout: 240 seconds) 2010-09-16T10:29:41 *** sigh has quit IRC (Remote host closed the connection) 2010-09-16T10:29:41 jmcarthur: hmmm, I thought tron was definatly minmax turn ranking 2010-09-16T10:30:00 Cyndre: it was simultaneous play. that alone makes minimax incorrect 2010-09-16T10:30:10 Cyndre: the top bots all used it, but that doesn't mean it was the best fit 2010-09-16T10:30:56 jmcarthur: yea it was - simulate turns, assume the other bot is going to maximize your your returns and minimize yours 2010-09-16T10:31:10 *** retybok_ has quit IRC (Ping timeout: 272 seconds) 2010-09-16T10:31:36 *** retybok has joined #aichallenge 2010-09-16T10:31:49 Cyndre: i never did it, but it is theoretically possible for a bot to exploit minimax bots 2010-09-16T10:32:03 Cyndre: because the move chosen is not always optimal, even with perfect information 2010-09-16T10:32:26 because tron is not turn-based like go or chess 2010-09-16T10:33:02 jmcarthur: yea it was - the more turns ahead I was able to look the higher my ranking went 2010-09-16T10:33:41 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-09-16T10:34:00 Cyndre: just read this http://en.wikipedia.org/wiki/Minimax_theorem#Example 2010-09-16T10:34:31 Cyndre: it says right there that simultaneously play games can lead to unstable strategies with minimax 2010-09-16T10:35:03 where does it mention unstable 2010-09-16T10:35:24 Cyndre: "Then, the minimax choice for A is A2 since the worst possible result is then having to pay 1, while the simple minimax choice for B is B2 since the worst possible result is then no payment. However, this solution is not stable, since if B believes A will choose A2 then B will choose B1 to gain 1; then if A believes B will choose B1 then A will choose A1 to gain 3; and then B will choose 2010-09-16T10:35:26 B2; and eventually both players will realize the difficulty of making a choice." 2010-09-16T10:35:44 *** Naith has joined #aichallenge 2010-09-16T10:36:04 Cyndre: a mixed strategy is the only stable strategy in that game 2010-09-16T10:36:28 tron had a lot of situations like that. a few were mentioned in the forums at the time 2010-09-16T10:37:16 jmcarthur: how else would you decide which way to move in tron with out doing a minmax eval? 2010-09-16T10:37:48 Cyndre: just make a game matrix for each simultaneous turn and find a mixed strategy from it. there are lots of algorithms for that 2010-09-16T10:38:08 i was going to make my bot do that, but work took over and i was unable to finish my bot 2010-09-16T10:39:00 jmcarthur: but then you would still run minmax on your new position to determine if it was good 2010-09-16T10:39:02 Cyndre: the most common algorithm is to treat it as a linear programming problem 2010-09-16T10:39:05 no 2010-09-16T10:39:30 minimax was straight up *wrong* for tron. it happened to work out pretty well, but it was not optimal 2010-09-16T10:39:35 unless you and the enemy cant see each other, then simple flood fill 2010-09-16T10:39:42 sure 2010-09-16T10:39:47 but there are no real choices, then 2010-09-16T10:40:11 if by "see" you really mean "smell" 2010-09-16T10:40:19 minimax worked good from what I seen and I felt dumb at the end of the contest for not doing it like that from the start 2010-09-16T10:41:26 and you can use what ever algortihms you want to decide the outcomes of the individual moves, but without simulating 10 turns of every possible decision... 2010-09-16T10:41:42 *** seanmceligot has joined #aichallenge 2010-09-16T10:42:01 danielvf, nice 2010-09-16T10:42:03 :) 2010-09-16T10:43:22 *** Sean_McEligot has quit IRC (Ping timeout: 245 seconds) 2010-09-16T10:43:32 *** Theobon has joined #aichallenge 2010-09-16T10:43:34 *** seanmceligot is now known as Sean_McEligot 2010-09-16T10:44:02 *** aerique has quit IRC (Quit: ...) 2010-09-16T10:46:02 *** Kastra has joined #aichallenge 2010-09-16T10:46:19 *** nigg_ has quit IRC (Quit: Leaving) 2010-09-16T10:46:19 Cyndre: minimax worked out pretty well. i'm just saying that minimax doesn't fit the model 2010-09-16T10:46:31 Cyndre: in this case, minimax was just a close approximation 2010-09-16T10:46:39 jmcarthur: I tested whether minimax was really hurting my bot by playing a variant of it against itself. That variant knew that the original thinks in term of minimax (that is, my turn, your turn) so ti could have in theory taken advantage of that. 2010-09-16T10:46:58 Found the bug, fine... crashing only on maps with 0-growth planets was great indication :P 2010-09-16T10:47:05 mega1: no dice? 2010-09-16T10:47:11 nothing 2010-09-16T10:47:26 mega1: i suspect that to truly take advantage of it the evaluation functions would have to be pretty dang good anyway 2010-09-16T10:47:32 although I got to read about game theory ... 2010-09-16T10:47:54 *** vikhyat has quit IRC (Remote host closed the connection) 2010-09-16T10:48:34 maybe the point here is that my-turn _then_ your-turn minimax is pessimistic 2010-09-16T10:48:49 and your-turn then my-turn minimax is optimistic. 2010-09-16T10:49:05 maximin? 2010-09-16T10:49:21 or maybe i'm mixing things up now 2010-09-16T10:49:27 i've never written anything to use maximin 2010-09-16T10:49:48 ah they are the same 2010-09-16T10:49:54 nevermind me :) 2010-09-16T10:50:34 mega1: thanks for reporting on that though. i was always curious if anybody tried i 2010-09-16T10:50:36 t 2010-09-16T10:51:14 no probs, I like reading about failures too. 2010-09-16T10:53:25 *** sinan has joined #aichallenge 2010-09-16T10:55:37 *** hellman has joined #aichallenge 2010-09-16T10:55:38 *** oh_cripes has joined #aichallenge 2010-09-16T10:56:40 *** fucawb has joined #aichallenge 2010-09-16T11:00:30 ahh 2010-09-16T11:00:40 found out why i was losing matches where i was doing awesome on 2010-09-16T11:00:47 i was running past the turn time limit 2010-09-16T11:00:57 so it was causing me to lose 2010-09-16T11:01:08 shootSelfInHead = false; 2010-09-16T11:01:16 *** kozlovsky has joined #aichallenge 2010-09-16T11:01:32 nice problem to have i guess... losing because you have so many planets and ships that its taking too long to decide what all it wants to do 2010-09-16T11:01:42 yet i wish the server would at least give me a heads up 2010-09-16T11:01:48 instead of saying i lost because of an invalid command 2010-09-16T11:02:06 shouldn't it be a Timeout error? 2010-09-16T11:02:28 you'd think 2010-09-16T11:02:38 well, crap... my timer code i just added didn't fire 2010-09-16T11:02:40 wtf is happening here? http://www.benzedrine.cx/planetwars/canvas?game_id=7289 it seems at turn 22, vortext's fleets make are hard left-turn out of nowhere 2010-09-16T11:02:41 and it still lost 2010-09-16T11:03:54 *** artyom_shegeda has joined #aichallenge 2010-09-16T11:04:11 rogue780, you're talking about the fleets in between the 68's? 2010-09-16T11:04:29 yes 2010-09-16T11:04:38 rogue780: isn't here a small planet in a center? 2010-09-16T11:04:41 http://ai-contest.com/forum/viewtopic.php?f=18&t=621 perl starter package 2010-09-16T11:04:43 there is 2010-09-16T11:04:52 in turn 21 it looks like >< 2010-09-16T11:04:55 *** zerd has quit IRC (Read error: Operation timed out) 2010-09-16T11:04:58 *** zerd has joined #aichallenge 2010-09-16T11:05:09 oh sh#t. I didn't notice that. I thought it was a rendering bug with the canvas visualizer 2010-09-16T11:05:33 space anomalies :) 2010-09-16T11:05:36 fooled me too 2010-09-16T11:05:40 there's a planet behind 2010-09-16T11:05:54 there is always a planet in the middle 2010-09-16T11:05:56 oh, right, >< is a planet 2010-09-16T11:06:25 it's captured in turn 17 2010-09-16T11:06:42 Naktibalda: yup 2010-09-16T11:07:04 *** hoare has joined #aichallenge 2010-09-16T11:10:23 hi guys. I've just played galcon. that's fantastic. I will participate the contest! 2010-09-16T11:10:38 hmm, galcon sucks 2010-09-16T11:10:50 you have to code here, not to play 2010-09-16T11:11:08 lol 2010-09-16T11:11:54 http://www.youtube.com/watch?v=Eq3CuMDXaPs 2010-09-16T11:12:57 *** hakand has quit IRC (Ping timeout: 245 seconds) 2010-09-16T11:14:37 *** Yoshi-TS4 has joined #aichallenge 2010-09-16T11:25:13 http://www.youtube.com/watch?v=d85p7JZXNy8 2010-09-16T11:26:04 *** seanmceligot has joined #aichallenge 2010-09-16T11:26:57 hrrm. 2010-09-16T11:27:04 i guess im really sending an invalid command of some type 2010-09-16T11:27:14 crap. my bot was not doing this last night 2010-09-16T11:27:21 server code change since yesterday? 2010-09-16T11:28:15 *** Sean_McEligot has quit IRC (Ping timeout: 252 seconds) 2010-09-16T11:28:27 *** seanmceligot is now known as Sean_McEligot 2010-09-16T11:30:51 *** antimatroid1 has joined #aichallenge 2010-09-16T11:30:52 *** antimatroid has quit IRC (Read error: Connection reset by peer) 2010-09-16T11:31:58 *** sinan has quit IRC (Ping timeout: 272 seconds) 2010-09-16T11:32:13 server ignored invalid commands yesterday 2010-09-16T11:34:18 has anyone had any experience compiling the tcp server? I'm getting an error from canvas.cpp about dprintf() 2010-09-16T11:34:51 conflicting types with previous declaration in stdio.h 2010-09-16T11:35:06 sorry canvas.c not .cpp 2010-09-16T11:35:34 hoare: Have fun! 2010-09-16T11:36:30 ahh 2010-09-16T11:36:33 yep, that was it. 2010-09-16T11:36:50 in rare conditions in which i am not winning, but not really losing all that bad 2010-09-16T11:37:03 a planet would try to send 1 ship too many 2010-09-16T11:37:18 didn't happen yesterday because my algorithm was much less good at being competetive 2010-09-16T11:37:21 :/ 2010-09-16T11:37:52 *** delt0r_ has quit IRC (Read error: Operation timed out) 2010-09-16T11:38:21 coder songs are better than admin songs 2010-09-16T11:38:52 *** delt0r_ has joined #aichallenge 2010-09-16T11:40:50 danielvf: congrats, great ranking. 2010-09-16T11:41:08 sigh, if only there were a perl starter package ... 2010-09-16T11:41:15 mega1: there is 2010-09-16T11:41:16 or is it too early for that? 2010-09-16T11:41:16 is it possible to edit contest account fields? 2010-09-16T11:41:20 Just not on server yet 2010-09-16T11:41:27 mega1: look in the forum 2010-09-16T11:41:43 Sorry, I'm waiting for the regular announcement. 2010-09-16T11:41:55 hoare: Thanks. I had to stop competing though, once I took over keeping the server happy 2010-09-16T11:42:04 mega1: Okay. Hopefuly this week. 2010-09-16T11:42:40 I'm just wondering, isn't it possible for a bot to send a lot of fleets containing 0 or 1 ship and timeout the oppenent? 2010-09-16T11:42:53 0 ships must be an error 2010-09-16T11:42:58 Yoshi-TS4, yea 2010-09-16T11:42:59 0's are removed the real game server 2010-09-16T11:43:02 yes, it's possible 2010-09-16T11:43:13 But a server can merge fleets 2010-09-16T11:43:16 And there's a patch in to combine fleets 2010-09-16T11:43:22 my game server does 2010-09-16T11:43:27 send to different directions 2010-09-16T11:43:28 the tcp one does too IIRC 2010-09-16T11:43:33 so... the official server doesn't? 2010-09-16T11:43:57 you shouldn't relly on that, it isn't a viable tactic 2010-09-16T11:44:43 Yoshi-TS4, you can merge them in your client too 2010-09-16T11:45:07 its a n log n worst case thing 2010-09-16T11:45:57 to get so many ships your bot must have a superior tactic, it's pointless trying to crash your oopponent in such position 2010-09-16T11:46:37 Hmm, that's good 2010-09-16T11:46:38 http://ai-contest.com/forum/viewtopic.php?f=18&t=621 perl starter package 2010-09-16T11:46:59 *** Baphomet has quit IRC (Ping timeout: 240 seconds) 2010-09-16T11:47:26 *** whiteside has joined #aichallenge 2010-09-16T11:48:23 *** Queue29 has quit IRC (Remote host closed the connection) 2010-09-16T11:50:44 *** mega1 has quit IRC (Read error: Connection reset by peer) 2010-09-16T11:51:06 *** kcm1700 has joined #aichallenge 2010-09-16T11:51:12 hello~ am a noob. 2010-09-16T11:51:13 *** mega1 has joined #aichallenge 2010-09-16T11:52:03 we know that 2010-09-16T11:52:13 wrong channel 2010-09-16T11:52:30 Naktibalda, kinda worked anyway 2010-09-16T11:52:35 :) 2010-09-16T11:53:09 *** RainCT has quit IRC (Remote host closed the connection) 2010-09-16T11:53:23 *** Baphomet has joined #aichallenge 2010-09-16T11:54:27 kcm1700: welcome :) 2010-09-16T11:58:29 wow, for the last week, my bot has been absolutely AWFUL 2010-09-16T11:58:36 i figured i had no chance in heck of winning 2010-09-16T11:58:59 then when they stopped ignoring errors, i discovered i had a pretty major flaw that was leaving me weak all the time 2010-09-16T11:59:13 i fixed that bug with one line of code, and all of the sudden my bot is kicking ass 2010-09-16T11:59:17 yay 2010-09-16T11:59:22 *** ratatata has quit IRC (Ping timeout: 252 seconds) 2010-09-16T12:00:25 http://www.benzedrine.cx/planetwars/getplayer?player=cobalt-8.2 2010-09-16T12:00:34 you can quite clearly see the moment when i fixed the bug 2010-09-16T12:00:44 went from almost all lossses to a long string of wins 2010-09-16T12:00:49 heh. invalid input, nice 2010-09-16T12:01:08 dhartmei fixed the little superscript notation thingies. awesome. 2010-09-16T12:01:35 yeah 2010-09-16T12:01:37 its nice 2010-09-16T12:01:48 the invalid input error i was having was small and stupid 2010-09-16T12:01:59 but it was messing with the ai logic that i pictured in my head 2010-09-16T12:02:06 so i couldn't figure out why i kept losing 2010-09-16T12:02:16 wow the game with r7 was a close one 2010-09-16T12:02:26 yeah... 2010-09-16T12:02:30 i watched it live 2010-09-16T12:02:37 i've got my bot logging all my moves to the screen 2010-09-16T12:02:56 IE doesn't properly understand http://ai-contest.com/forums/index.php ... it seems that the reason is a comment which includes some tags inside it. 2010-09-16T12:04:05 who uses ie? 2010-09-16T12:04:06 use another browser :) 2010-09-16T12:04:08 kcm1700: ahha! can you post a bug on http://code.google.com/p/ai-contest/issues/list ? 2010-09-16T12:04:17 IE sucks lol. 2010-09-16T12:04:45 danielvf will probably fix it post-haste 2010-09-16T12:04:58 thanks a lot :-) 2010-09-16T12:06:28 *** McLeopold has joined #aichallenge 2010-09-16T12:07:01 Is it possible to unsubmit a bot from the contest? 2010-09-16T12:07:22 McLeopold: upload a blank zip file? 2010-09-16T12:07:22 I don't see phpBB settings there(respository), is this issue related to ai-contest project code? 2010-09-16T12:07:33 McLeopold: going into stealth mode? 2010-09-16T12:07:47 No, I uploaded a bot that forfeits on turn one every match. :( 2010-09-16T12:07:56 i'ma busy with uni assignments, i'm claiming first spot on the weekend though 2010-09-16T12:08:09 hey Naktibalda, that was a good game 2010-09-16T12:08:13 I have no idea why, either. It would be nice to know if I timed out or send invalid orders. 2010-09-16T12:08:31 kcm1700: oh, yeah i dunno. just post a bug and point to the tag whatever you get from view source 2010-09-16T12:08:34 run it on tcp and debug there? 2010-09-16T12:08:42 http://www.benzedrine.cx/planetwars/canvas?game_id=7825 2010-09-16T12:08:45 I run fine on the tcp. 2010-09-16T12:08:51 Same code. 2010-09-16T12:09:02 *** deepblue has joined #aichallenge 2010-09-16T12:09:18 thanks again for your help ; ) 2010-09-16T12:09:55 wait for my next version :) 2010-09-16T12:10:51 kcm1700: i use Maxthon browser and i created the filter to replace --!> with -->. All works nice ;) 2010-09-16T12:11:04 wow 2010-09-16T12:11:14 i think im about to win a match that i had no business winning 2010-09-16T12:12:16 artyom_shegeda: oh, interesting.. lol I didn't noticed that. then it wasn't f[a-z]+ing IE problem 2010-09-16T12:13:37 i'm searching for info about html comments. is it possible to close comment with --!>? 2010-09-16T12:13:56 you guys HAVE to watch this one 2010-09-16T12:13:58 its hilarious 2010-09-16T12:14:00 http://www.benzedrine.cx/planetwars/canvas?game_id=7840 2010-09-16T12:14:06 i won because he timed out 2010-09-16T12:14:09 but i was GOING to win 2010-09-16T12:14:20 i have no idea how this happened or what i should have done differently 2010-09-16T12:14:29 but this is the craziest game i've ever seen 2010-09-16T12:15:06 html comments must be formed as following : and comment : all proper strings except consecutive -'s 2010-09-16T12:15:41 there are bots spaming lots of 1 unit fleets 2010-09-16T12:15:53 Made some great enhancements to the Java TCP client: http://ai-contest.com/forum/viewtopic.php?p=3836#p3836 2010-09-16T12:15:59 kcm1700: where did you find it? 2010-09-16T12:16:09 kills the web game viewer 2010-09-16T12:16:27 from my old old textbox (...) i'm searching for the recent docs.. 2010-09-16T12:16:29 oh, doesn't kill mine 2010-09-16T12:16:34 but i have a faster computer 2010-09-16T12:16:52 blister, are you talking to me? 2010-09-16T12:16:53 get a better browser 2010-09-16T12:16:58 *** kozlovsky has quit IRC (Ping timeout: 252 seconds) 2010-09-16T12:17:04 right.. 2010-09-16T12:17:17 yes 2010-09-16T12:17:17 :D 2010-09-16T12:17:23 google chrome for the win 2010-09-16T12:17:27 White space is not permitted between the markup declaration open delimiter(""). A common error is to include a string of hyphens ("---") within a comment. Authors should avoid putting two or more adjacent hyphens inside comments. 2010-09-16T12:17:38 well tbh i get widely different performance with the web previewer 2010-09-16T12:17:45 blister, I am on linux 2010-09-16T12:17:50 ahh 2010-09-16T12:18:05 out of the 30 computers here at work there are 0 windows machines 2010-09-16T12:18:26 and i don't get to choose what we have installed either 2010-09-16T12:18:29 must be nice 2010-09-16T12:18:30 *** kozlovsky has joined #aichallenge 2010-09-16T12:18:32 artyom_shegeda: yes,, that's what my textbook says... gj :) 2010-09-16T12:18:32 <3 linux 2010-09-16T12:18:35 I am on FF 3.X something 2010-09-16T12:18:41 i use linux at home and on all my servers 2010-09-16T12:18:47 but we have windows boxes here at work 2010-09-16T12:18:55 I only use linux at home too 2010-09-16T12:19:06 what distro you run? 2010-09-16T12:19:10 but thats slackware 2010-09-16T12:19:13 i run ubuntu 2010-09-16T12:19:15 ahh 2010-09-16T12:19:16 slack is awesome 2010-09-16T12:19:16 you should be able to install firefox/chrome in your home directory 2010-09-16T12:19:21 i haven't touched slackware in years 2010-09-16T12:19:26 deepblue, yea --true 2010-09-16T12:19:28 honestly i didn't even know it was still a project 2010-09-16T12:19:35 but honestly its not really a big deal 2010-09-16T12:19:47 *** atmz443 has joined #aichallenge 2010-09-16T12:19:54 I was commenting more on the fact that bots are spamming single size fleets 2010-09-16T12:20:00 yeah 2010-09-16T12:20:10 the offical gameplayer should merge fleets 2010-09-16T12:20:13 my algo depends heavily on constant cooperation amongst planets 2010-09-16T12:20:21 it should... 2010-09-16T12:20:41 but most of those spammed fleets are sent on consecutive turns 2010-09-16T12:20:47 and have a different arrival time 2010-09-16T12:20:57 in this case its sending 10 or more fleets of size 1 from a the same source to the same desitnation in the same turn 2010-09-16T12:21:36 blister, I don't consider than spamming... the fleets *can't* be merged and have the same game results 2010-09-16T12:21:54 why? 2010-09-16T12:21:54 *** Janzert has joined #aichallenge 2010-09-16T12:22:05 Naktibalda, > 2010-09-16T12:22:07 ? 2010-09-16T12:22:09 ops 2010-09-16T12:22:30 if you send 10 ships from a to b, it must be possible to merge them 2010-09-16T12:23:01 If you send 5 at the start of the game and 5 at the end, it's not possible to merge them. 2010-09-16T12:23:03 Naktibalda, yes--but not in blisters cases since they are not all sent on the *same* turn 2010-09-16T12:23:45 grr. I keep timing out / sending invalid command on the first move. Very frustrating 2010-09-16T12:24:00 check your first command :) 2010-09-16T12:24:06 i 2010-09-16T12:24:13 it doesn't happen all the time though 2010-09-16T12:24:22 just most of the time 2010-09-16T12:24:29 tonight I have debugged my code for 40 minutes to find out that return is missing in a new version of Distance() :) 2010-09-16T12:24:38 I should go to sleep earlier :( 2010-09-16T12:24:42 I've noticed that the server drops the connection without even an error message if you send fleets from planet A to planet A. 2010-09-16T12:25:18 Olathe, yes --thats in the rules/game description 2010-09-16T12:25:32 Yeah, but it should at least say that you lose. 2010-09-16T12:28:18 true 2010-09-16T12:33:24 *** justin_pdx has joined #aichallenge 2010-09-16T12:33:40 *** sinan has joined #aichallenge 2010-09-16T12:33:57 *** HakanD has joined #aichallenge 2010-09-16T12:35:14 i posted --!> problem at the forum. hope it will get fixed soon. 2010-09-16T12:36:47 it shouldn't take more than a 3 days 2010-09-16T12:36:57 if you post a patch 2010-09-16T12:37:24 does anyone know what version of python the official servers are running? 2010-09-16T12:37:39 *** HakanD has quit IRC (Read error: Connection reset by peer) 2010-09-16T12:37:47 *** HakanD has joined #aichallenge 2010-09-16T12:38:08 2.5 right now 2010-09-16T12:38:34 is it going to change? 2010-09-16T12:38:45 Not sure 2010-09-16T12:39:49 how are the coordinates structured on the maps? is (0,0) the center planet, or the topleft corner? 2010-09-16T12:39:51 *** hoare has quit IRC (Quit: Page closed) 2010-09-16T12:40:55 it can breake many things, so probably not :) 2010-09-16T12:40:57 Python3 is where it's at 2010-09-16T12:40:58 break 2010-09-16T12:42:01 I want to get information about the c++ compiler version and compile&linking options on server. where can i get such information? 2010-09-16T12:42:53 kcm1700, I'm *assuming* it's gcc4.x on bsd 2010-09-16T12:43:00 or g++ 2010-09-16T12:43:03 rather 2010-09-16T12:43:51 4.2 iirc 2010-09-16T12:44:50 *** vizier has quit IRC (Ping timeout: 252 seconds) 2010-09-16T12:44:59 I'm wondering of that is why my bot is failing on the contest server... 2010-09-16T12:45:01 rogue780: It's the top left corner. 2010-09-16T12:47:30 Is it okay to submit a code with public libraries or something that the submitter didn't programmed? 2010-09-16T12:48:40 *** danielvf has quit IRC (Ping timeout: 276 seconds) 2010-09-16T12:50:01 shouldn't be an issue if it compiles 2010-09-16T12:50:35 *** braddunbar has quit IRC (Ping timeout: 252 seconds) 2010-09-16T12:50:50 Naktibalda: thank you. 2010-09-16T12:51:33 kcm1700: keep in mind that they don't use any submitted makefiles, they just scan for all of your source code and compile it with their automagic script.. 2010-09-16T12:53:41 ok :-) I was just checking for legitimate things... 2010-09-16T12:54:59 have you a library that may be usefull in contest? 2010-09-16T12:55:43 *** dhartmei has quit IRC (Quit: brb) 2010-09-16T12:56:14 *** dhartmei has joined #aichallenge 2010-09-16T12:56:34 c++ library for some matrix, vectors, linear algebraic things... and I'm planning to use some graph related algorithm codes. 2010-09-16T12:59:36 *** amstan has joined #aichallenge 2010-09-16T12:59:36 *** ChanServ sets mode: +o amstan 2010-09-16T13:00:07 hello 2010-09-16T13:00:43 Arrg! Python 2.5 doesn't support the string's format function! 2010-09-16T13:00:48 morning 2010-09-16T13:01:00 amstan: how's the server going under this load? 2010-09-16T13:01:04 there's well over 1000 bots now 2010-09-16T13:01:06 McLeopold: why do you need that? 2010-09-16T13:01:16 antimatroid1: it's just playing less games 2010-09-16T13:01:22 I'm used to python3. 2010-09-16T13:01:42 For issuing orders to the server. 2010-09-16T13:01:43 i am actually quite happy with the number of games its been giving me 2010-09-16T13:01:52 more bots!=more load 2010-09-16T13:01:54 tcp if one wants to really test 2010-09-16T13:02:13 you still have lots more people on there all the time, uploading etc. 2010-09-16T13:02:26 that does increase server load 2010-09-16T13:02:47 probably not really that many people though really aha 2010-09-16T13:03:32 *** retybok_ has joined #aichallenge 2010-09-16T13:05:31 Come on, Naktibalda, you used to beat me and now I'm beating you. what happened? 2010-09-16T13:05:57 *** amstan has quit IRC (Read error: Connection reset by peer) 2010-09-16T13:06:04 you updated your bot 2010-09-16T13:06:31 and my bot is the same since Sunday 2010-09-16T13:06:42 *** amstan has joined #aichallenge 2010-09-16T13:06:42 *** ChanServ sets mode: +o amstan 2010-09-16T13:06:47 I will beat you on Saturday 2010-09-16T13:06:56 ugh 2010-09-16T13:06:56 interenet is horrible here 2010-09-16T13:07:03 can anyone help me navigate the site? 2010-09-16T13:08:08 what do you mean? 2010-09-16T13:09:51 which page do you want? 2010-09-16T13:11:44 *** ajhager has joined #aichallenge 2010-09-16T13:12:01 So, the .format fix didn't work either. 2010-09-16T13:12:04 *** amstan has quit IRC (Ping timeout: 276 seconds) 2010-09-16T13:12:41 I know the rules say stderr will be absorbed, but it sure would be nice to get contestants 1000 characters at least. 2010-09-16T13:12:56 I have know idea why my bot is failing on the contest server. 2010-09-16T13:12:58 *** td123 has left #aichallenge ("WeeChat 0.3.3") 2010-09-16T13:13:26 My only option is to keep uploading possibly bad code and waste server time until it works. 2010-09-16T13:13:33 well, there's a risk of hacking server if that is permitted. 2010-09-16T13:13:40 get python 2.7 :) 2010-09-16T13:13:49 The server needs python 2.7 2010-09-16T13:13:55 maybe. 2010-09-16T13:14:19 It wouldn't be hacking if they captured the first 1000 characters of stderr and displayed it in the match results. 2010-09-16T13:14:30 Heck, even 100 characters. Something. 2010-09-16T13:15:01 Even only if the bot timed out. 2010-09-16T13:15:11 there's a patch to add a reason for failure in issues 2010-09-16T13:15:20 Is it a high priority? 2010-09-16T13:15:26 I mean that you should use python 2 in dev environment 2010-09-16T13:15:37 unlikelly :) 2010-09-16T13:15:38 eg "timed out" or "invalid order: ..." 2010-09-16T13:15:41 I did. It was build with python 2.6.4 2010-09-16T13:15:57 it doesn't seem high priority, it's been sitting there for maybe a week now 2010-09-16T13:16:18 It would probably reduce server loaded if they did that. 2010-09-16T13:16:42 Then contestants wouldn't keep uploaded failing bots. 2010-09-16T13:16:51 (like me) 2010-09-16T13:18:15 it's issue 96 2010-09-16T13:18:16 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-09-16T13:19:15 you could gently remind one or more admins about it... (it requires a new column in a table) 2010-09-16T13:28:06 *** smellyhippy has quit IRC (Quit: BUT WAIT! What are those penguins up to?!) 2010-09-16T13:30:15 whoever owns PBot3.1 needs to fix their shit 2010-09-16T13:30:22 disconnects immediately when the game starts 2010-09-16T13:30:43 *** retybok_ has quit IRC (Ping timeout: 255 seconds) 2010-09-16T13:33:59 lol, errbot lost due to invalid input. way to go with that name :D 2010-09-16T13:38:49 *** tapwater has joined #aichallenge 2010-09-16T13:39:18 *** ratatata has joined #aichallenge 2010-09-16T13:40:06 IRC disconnected without my knowledge and here I was wondering why the channel was so silent :) 2010-09-16T13:40:49 *** rogue780_ has quit IRC (Quit: Bye) 2010-09-16T13:41:00 deepblue: what happened to your ranking? why did you drop? 2010-09-16T13:41:29 in the official I just uploaded some new code 2010-09-16T13:41:39 so it hasn't risen yet? 2010-09-16T13:41:47 and on the tcp was testing lots of different stuff and had to bugfix for hours 2010-09-16T13:41:56 yeah 2010-09-16T13:42:01 only three games so far 2010-09-16T13:42:06 rank 250 or so 2010-09-16T13:42:14 http://ai-contest.com/profile.php?user_id=5898 2010-09-16T13:42:19 well I can't even get 1 game to work on the official server. :( 2010-09-16T13:42:46 the python problem? 2010-09-16T13:42:49 yea 2010-09-16T13:42:56 that's annoying :/ 2010-09-16T13:43:10 i'm glad they added -O3 to the official compile command :) 2010-09-16T13:43:21 Well, I know I have a good bot. I just need to be more patient until the give us more match status info. 2010-09-16T13:43:31 yeah 2010-09-16T13:43:49 i would love some feedback from the official server 2010-09-16T13:44:38 does c++ give a noticeable advantage in `stuff you can do before timeout' over the other languages? 2010-09-16T13:44:58 If you have lots o stuff to do, then maybe. 2010-09-16T13:45:29 Like if you were doing planet wars bitboards? 2010-09-16T13:46:04 if you were iterating over every fleet for every planet for every fleet or something equally ridiculous, maybe 2010-09-16T13:46:22 *** craigferguson has joined #aichallenge 2010-09-16T13:46:38 c and c++ will probably be the fastest 2010-09-16T13:47:03 *** sinan has quit IRC (Ping timeout: 240 seconds) 2010-09-16T13:48:09 I've had some intermittent timeout issues in java while doing fairly simple stuff, it's probably some bug I haven't found yet though 2010-09-16T13:48:51 unless you have logging, timeout might also mean silent crash 2010-09-16T13:48:53 so...would it technically be possible to create a network connection with my bot on the main server? 2010-09-16T13:49:23 rogue780: i think there are iptables rules to prevent that 2010-09-16T13:50:01 Oh, and logging makes my bot timeout significantly more 2010-09-16T13:50:04 which is annoying 2010-09-16T13:50:16 ouch 2010-09-16T13:50:28 atmz443, i had that problem too 2010-09-16T13:50:31 what language are you using? 2010-09-16T13:50:34 Java 2010-09-16T13:51:17 so in my Log() method, I had it open and close the file each time it was called. this was causing me to time out. It then struck me that I should just open it in main and close it in main. This made the timeouts go away 2010-09-16T13:51:25 not sure if you have a similar situation or not 2010-09-16T13:51:48 I've been using the standard library's logging thingy 2010-09-16T13:52:02 But am planning to roll my own at some point, so thanks for the tip 2010-09-16T13:52:20 no problem 2010-09-16T13:52:38 omg omg omg omg omg omg omg omg 2010-09-16T13:52:41 FIVE DAYS 2010-09-16T13:53:13 5 days what ? 2010-09-16T13:55:53 *** HarryD|cs has joined #aichallenge 2010-09-16T13:56:28 until the 21st, of course 2010-09-16T13:56:38 *** jukkerknott has joined #aichallenge 2010-09-16T13:57:07 which is? 2010-09-16T13:57:11 wow 2010-09-16T13:57:17 ... 2010-09-16T13:57:21 competition on the official servers is WAAAAAY easier than on the real-time servers 2010-09-16T13:57:28 here's cobalt-8.2: 2010-09-16T13:57:33 http://ai-contest.com/profile.php?user_id=7430 2010-09-16T13:57:36 ranked 89 2010-09-16T13:58:33 blister: it is also way irrelevant 2010-09-16T13:59:10 the official is crowded with half broken modified sample bots 2010-09-16T13:59:13 *** seanmceligot has joined #aichallenge 2010-09-16T13:59:16 FIVE DAYS UNTIL CIV V!!! 2010-09-16T13:59:24 i know 2010-09-16T13:59:25 hehe 2010-09-16T13:59:27 still funny though 2010-09-16T13:59:27 :D 2010-09-16T14:00:47 5 days? oh no it's terrible timing 2010-09-16T14:01:04 I'm hoping it'll take out some of the better players on planet wars 2010-09-16T14:01:12 hahaha 2010-09-16T14:01:17 *** Sean_McEligot has quit IRC (Ping timeout: 245 seconds) 2010-09-16T14:01:21 *** seanmceligot is now known as Sean_McEligot 2010-09-16T14:01:27 only if they're not done 2010-09-16T14:07:54 *** Naktibalda has joined #aichallenge 2010-09-16T14:08:43 *** McLeopold has quit IRC (Quit: Leaving) 2010-09-16T14:08:46 are disconnections from the server caused by my bot crashing or other issues? 2010-09-16T14:09:41 *** jukkerknott has quit IRC (Quit: Page closed) 2010-09-16T14:11:45 *** odinsbane has joined #aichallenge 2010-09-16T14:13:14 *** HakanD has quit IRC (Read error: Connection reset by peer) 2010-09-16T14:14:09 * rogue780 is preloading Civ V now 2010-09-16T14:14:36 *** HakanD has joined #aichallenge 2010-09-16T14:15:31 *** HakanD has quit IRC (Client Quit) 2010-09-16T14:17:41 What would the take be on smoke screening, say sending out massive numbers of fleets so if somebody is trying to evaluate all of them they'll time out. 2010-09-16T14:18:04 odinsbane: seems to be fair game 2010-09-16T14:18:19 this should be a faq question 2010-09-16T14:18:31 but it's not necessairly always going to be a very smart move 2010-09-16T14:18:36 odinsbane: if it works - you win, if it doesn't - you loose 2010-09-16T14:18:59 and by the time it is, you've probably won anyway 2010-09-16T14:19:01 Well I was sending out to many fleets so I'd time out. 2010-09-16T14:19:09 :D 2010-09-16T14:19:23 do you want to repeat it? 2010-09-16T14:19:33 personally i've set mine up so it doesn't really hurt me and i'm not intending to spam others 2010-09-16T14:19:56 if other people time out they get disqualified too 2010-09-16T14:20:01 that's going to hurt 2010-09-16T14:20:13 they better have their timers set up very well 2010-09-16T14:20:21 It seems like if you sent enough out, it could time out on the parsing, which doesn't seem sportsman like. 2010-09-16T14:20:24 ...wtf? I beat ademar_ !? http://www.benzedrine.cx/planetwars/canvas?game_id=8752 2010-09-16T14:20:26 *** Naith has quit IRC (Quit: Naith) 2010-09-16T14:20:32 *** HarryD|cs has quit IRC (Remote host closed the connection) 2010-09-16T14:20:39 *** Vi0 has joined #aichallenge 2010-09-16T14:20:56 *** retybok_ has joined #aichallenge 2010-09-16T14:20:58 odinsbane: i agree and told them to change it many times, but alas this is the decision they seem to have made, so fair game it is 2010-09-16T14:20:59 parsing itself should be very fast 2010-09-16T14:22:57 *** ajhager has quit IRC (Ping timeout: 245 seconds) 2010-09-16T14:23:39 for fleet analyzis you can merge fleets in a bot code after parsing 2010-09-16T14:24:47 rogue780: hehe 2010-09-16T14:25:14 ademar_, for me, beating you once and losing seven times is still a win ;) 2010-09-16T14:26:02 ah my bot is getting kicked all around 2010-09-16T14:26:10 I think there is some optimization that can go into the planetwars.java to, regarding chages and sorting. 2010-09-16T14:27:05 are you still using a default class? 2010-09-16T14:27:36 Naktibalda: yes, everythign except the planet. 2010-09-16T14:28:10 dstufft: are you around? how do go about getting f# on the server ... i have a patch for compile_anything.py .. but need the path to the f# installer .. 2010-09-16T14:28:33 *** kuwabara has quit IRC (Ping timeout: 240 seconds) 2010-09-16T14:30:01 *** kuwabara has joined #aichallenge 2010-09-16T14:30:58 I'm still trying to get an idea on how much I can do in 1 second before I start optimizing. 2010-09-16T14:31:09 Naktibalda: I merge as a pass :D 2010-09-16T14:31:47 and i keep old fleets stored rather than recreating the entire vector every move 2010-09-16T14:31:51 i have so many ideas, but i don't have the time to implement them all :( 2010-09-16T14:32:09 antimatroid1: how do you discover new fleets? 2010-09-16T14:32:13 jmcarthur: i know the feeling, haven't done anything in over 24 hours to my bot, will on the weekend though 2010-09-16T14:32:34 How about the planet distances? I was considering storing all of the planet distances in a map. 2010-09-16T14:32:35 as i pass game info, if number of moves remaining is 1 less than trip length then it's a new fleet 2010-09-16T14:32:49 and you need to update planets every move in case of planets distanced 1 unit apart 2010-09-16T14:33:04 yeah, stored planet distances in a 2d vector 2010-09-16T14:33:19 2d vector ... c? 2010-09-16T14:33:19 * Naktibalda stored distances in triangle matrix 2010-09-16T14:33:24 c++ 2010-09-16T14:33:39 std::vector > v 2010-09-16T14:33:45 * rogue780 doesn't store distances and computes them on the fly 2010-09-16T14:34:01 i dislike nesting vectors 2010-09-16T14:34:02 Do you think that'd its effectiveness as the number of planets increase? 2010-09-16T14:34:05 i did triangle, but it takes no time to make it, so i just did the full one 2010-09-16T14:34:27 Zannick: i love it, i'll nest and nest all day 2010-09-16T14:34:29 *** amstan has joined #aichallenge 2010-09-16T14:34:29 *** ChanServ sets mode: +o amstan 2010-09-16T14:34:40 i'd rather manually allocate 2d vectors in c 2010-09-16T14:34:59 which i am in the process of doing, in fact 2010-09-16T14:35:01 *** retybok_ has quit IRC (Ping timeout: 272 seconds) 2010-09-16T14:35:02 *** supermoose has joined #aichallenge 2010-09-16T14:35:02 in haskell i just use the polymorphic index arrays for that 2010-09-16T14:35:07 you have fun with that :P 2010-09-16T14:35:08 so i can just index on a tuple 2010-09-16T14:35:14 i am 2010-09-16T14:35:26 c++ <3 2010-09-16T14:35:29 You save on matches 2010-09-16T14:35:40 ...when uses cached distances 2010-09-16T14:36:03 i don't think the distance calculation is a limiting factor 2010-09-16T14:36:11 not a big deal, no 2010-09-16T14:36:21 and if it is then you are doing something horribly wrong 2010-09-16T14:36:27 deepblue: no, but i prefer having them in something i can just index easily anyway 2010-09-16T14:36:32 in fact, so much fun that an idea i had yesterday turned out to eliminate the need to store planets or fleets separately 2010-09-16T14:36:41 still, if you use distance a lot in your algorithm it's still probably faster to avoid recalculating 2010-09-16T14:37:00 unless there are enough planets that you kill your cache with it... 2010-09-16T14:37:02 doubtful 2010-09-16T14:37:09 Zannick: err, wasn't that what we were just talking about? 2010-09-16T14:37:14 merging fleets? 2010-09-16T14:37:15 like distance(i,j) 2010-09-16T14:37:21 simple as that as a function :) 2010-09-16T14:37:24 *** Appleman1234 has joined #aichallenge 2010-09-16T14:37:28 did you profile your application? :) 2010-09-16T14:37:37 i have that as a function anyway (the cache is in the closure) ;) 2010-09-16T14:37:44 antimatroid1: no, like, i do not need planet objects/structs or fleet objects/structs 2010-09-16T14:37:47 sory, i'm a bit late to the conversation. what do you mean by merging fleets? 2010-09-16T14:38:01 you still have to store the information :P 2010-09-16T14:38:03 distance calculation takes less time than sorting and multidimensional planets processing 2010-09-16T14:38:08 rogue780: two fleets with the same source/destination/time. 2010-09-16T14:38:24 yes, i am storing it, but in a different way 2010-09-16T14:38:40 i threw fleet struct out the window immediately 2010-09-16T14:38:43 odinsbane, why would you have more than one fleet with the same source/destination/time? 2010-09-16T14:38:49 but i have added quite a few different structs etc. all over the place 2010-09-16T14:39:03 rogue780: well you might merge if they just have the same destination and time. 2010-09-16T14:39:09 rogue780: why would you have more than one fleet with the say destination/time? 2010-09-16T14:39:15 *** craigferguson has quit IRC (Ping timeout: 252 seconds) 2010-09-16T14:39:17 damn beaten :P 2010-09-16T14:39:28 and ruined same* 2010-09-16T14:39:38 oh, i probably still need to store growth 2010-09-16T14:39:58 I can see same destination and time, but if they have the same source, destination and time, they should be sent with a single order, no? 2010-09-16T14:40:00 i don't even store fleets as a special Fleet type 2010-09-16T14:40:18 rogue780: some stupid bots send multiple fleets 2010-09-16T14:40:29 * Naktibalda is going to discard Fleet class soon 2010-09-16T14:40:32 Hey this haskell-DualBot sure improves the record huh. 2010-09-16T14:40:34 well that's just silly 2010-09-16T14:40:59 probably they are trying to force your bot to timeout :) 2010-09-16T14:40:59 i'm just representing planets as timelines and they just get more ships when fleets "arrive" 2010-09-16T14:41:12 and orders simply modify the timelines 2010-09-16T14:42:18 that's approximately what i came up with 2010-09-16T14:42:21 an order just removes ships from a planet at one point in time and adds them to another planet at a later point in time (and i lazily compute states farther in the future using accumulators) 2010-09-16T14:42:32 jmcarthur: I do most of that calculation, but I am not storing anything. 2010-09-16T14:43:48 odinsbane: haskell-DualBot? 2010-09-16T14:44:13 Running on Benzedrine, I don't know whos it is , but I don't think it has worked once. 2010-09-16T14:44:15 * rogue780 is glad he is logging this channel 2010-09-16T14:44:16 ah 2010-09-16T14:44:23 *** oh_cripes has quit IRC (Ping timeout: 252 seconds) 2010-09-16T14:44:30 it won a few games it looks like 2010-09-16T14:45:26 *** braddunbar has joined #aichallenge 2010-09-16T14:45:46 why doesn't the realtime server have a 200 turn limit? 2010-09-16T14:45:59 *** amstan has quit IRC (Ping timeout: 240 seconds) 2010-09-16T14:46:08 * jmcarthur shrugs 2010-09-16T14:46:16 They must have changed it, everytime I went against it, the game ended at turn 1. 2010-09-16T14:46:31 odinsbane: most games screw up like that, but a few didn't 2010-09-16T14:48:57 I wish it were allowed to change the destination of a fleet that's already been launched. You can do this with Galcon. 2010-09-16T14:49:32 i like that sending a fleet is a commmitment, personally 2010-09-16T14:49:43 it makes your decisions more critical 2010-09-16T14:49:59 * rogue780 is scare of commitment 2010-09-16T14:50:08 and defence easier 2010-09-16T14:50:20 defense 2010-09-16T14:50:37 hrr, it's defence :) 2010-09-16T14:50:39 well, otherwise an ai could have fleets "circling" a planet 2010-09-16T14:52:23 *** retybok_ has joined #aichallenge 2010-09-16T14:53:25 Redirected fleets would almost eliminate 'geographic' analysis of a map because you could just sit a fleet outside of a enemy planet just DARING it to send some ships somwhere and then *glomp* 2010-09-16T14:54:01 Maybe if they had a time-limit they could sit in space. 2010-09-16T14:54:21 that would just get complicated 2010-09-16T14:54:28 Yes. 2010-09-16T14:54:36 Thats sorta tough, the whole 'glomp' idea. Plus they could send their ships back. 2010-09-16T14:54:49 *** braddunbar has quit IRC (Ping timeout: 272 seconds) 2010-09-16T14:55:03 of course the other bot could counter that by just departing and then immediately returning to glomp the glomper 2010-09-16T14:55:42 there would cease to be as much strategy, in any case 2010-09-16T14:55:52 just a game of chicken 2010-09-16T14:55:54 i would much rather like a system where you don't see the enemy fleets 2010-09-16T14:55:54 we could see a lot of forward-backward movement :) 2010-09-16T14:56:08 deepblue: *that* would be interesting 2010-09-16T14:56:08 deepblue: at least not until they get within X turns of you :) 2010-09-16T14:56:17 deepblue: maybe even have a fog of war on the planets 2010-09-16T14:56:25 that would be cool 2010-09-16T14:56:27 sounds fun to me :) 2010-09-16T14:56:27 did you see how #1 play? it's fantastic. this bot catches galaxy in 50 turns 2010-09-16T14:57:48 which server? what's a name of bot? 2010-09-16T14:57:55 jmcarthur: are you correctly accounting for the fact that you never see a fleet send from the enemy to a planet 1 distance from its source? 2010-09-16T14:58:04 sent* 2010-09-16T14:58:12 official :) http://ai-contest.com/profile.php?user_id=7233 2010-09-16T14:59:31 antimatroid1: in my timeline abstraction? yeah 2010-09-16T15:00:09 i just update the initial value of each planet for each turn 2010-09-16T15:00:57 *** hakand has joined #aichallenge 2010-09-16T15:01:08 Okay yeah that Locutus bot is cool. 2010-09-16T15:01:39 It has this big army it sends out capping small planets on its way to the enemies biggest jumps. 2010-09-16T15:01:41 look: it doesn't send more fleets that it needs 2010-09-16T15:01:51 very methodical 2010-09-16T15:02:08 I'd like to see it play itself. 2010-09-16T15:04:07 *** Naith has joined #aichallenge 2010-09-16T15:04:39 *** rafsoaken has joined #aichallenge 2010-09-16T15:05:26 hmm...the more I think about it, the more I really like that timeline idea. 2010-09-16T15:05:36 It looks like it sends fleets to intermedia planets. 2010-09-16T15:06:13 i have the timeline implemented and a really shitty, slow bot to go with it 2010-09-16T15:06:18 heh 2010-09-16T15:06:45 i'm still just taking forever 2010-09-16T15:06:45 *** ajhager has joined #aichallenge 2010-09-16T15:07:10 He has some nice algorithm to decide where to go. 2010-09-16T15:13:42 what does he do in turn 24 of http://ai-contest.com/visualizer.php?game_id=4511579, react to the attack? 2010-09-16T15:14:48 he deals with it. like a boss. 2010-09-16T15:15:23 ^ this 2010-09-16T15:16:16 dhartmei: I didn't see him do anything, he didn't have any new fleets. 2010-09-16T15:16:46 after turn 25 he sends 3 fleets to planet under attack 2010-09-16T15:17:05 yeah seems a little slow with reaction ;) 2010-09-16T15:18:09 I think he waited till the closest of the fleets would arrive just in time to prevent the enemy from actually gaining much income. 2010-09-16T15:18:27 so, is my bot uncommon that it targets more than one planet at a time with no master target list? 2010-09-16T15:18:36 Till his attack could overcome the income, rather. 2010-09-16T15:19:11 you're right 2010-09-16T15:19:25 ^ 2010-09-16T15:19:36 but he can send fleets from 4 other planets one turn earlier 2010-09-16T15:20:14 He also didn't retaliate when the planet under dures wasn't going to be taken over. 2010-09-16T15:20:42 I wish there was a way to unlock preloaded games on steam early 2010-09-16T15:21:19 more time for you to work on your bot before civ5 takes up all your time 2010-09-16T15:21:20 :) 2010-09-16T15:21:47 I think he doesn't send an attack to recapture until after he captures the planet below the enemy home planet, ensuring that if they rush to defend, he can easily glomp their home. 2010-09-16T15:22:45 Which is the strategy he uses to win, when the home tries to send out a new fleet, he strikes and defends simultaneously. 2010-09-16T15:22:50 oops. he lost the game 2010-09-16T15:23:31 *** xvinyl has quit IRC (Remote host closed the connection) 2010-09-16T15:23:31 *** Bob___ has joined #aichallenge 2010-09-16T15:23:34 *** amstan has joined #aichallenge 2010-09-16T15:23:34 *** ChanServ sets mode: +o amstan 2010-09-16T15:23:40 amstan, !!! 2010-09-16T15:23:47 Does anyone have any tips for debugging? the game engine seems to block all output 2010-09-16T15:23:57 graphitemaster: wtf? 2010-09-16T15:24:08 Bob___, compile with -gdbg? then run it using gdb? 2010-09-16T15:24:09 write to a file 2010-09-16T15:24:10 He lost against dungeon_keeper cause dungeon defends on turn 13. 2010-09-16T15:24:13 alright thanks 2010-09-16T15:24:18 Bob___: language? 2010-09-16T15:24:22 java 2010-09-16T15:24:33 ohh, nvm, mine was a gcc thing 2010-09-16T15:24:35 i think the new tournament selector is working out pretty well 2010-09-16T15:24:50 :-( 2010-09-16T15:25:00 Bob___: write to a file 2010-09-16T15:25:01 amstan, we need a C starter package. 2010-09-16T15:25:06 Haha! My new bot version took me from being ranked at 509 down to 274 2010-09-16T15:25:11 damn... i guess ill have to 2010-09-16T15:25:13 Bob___: log to a file or use the 1.2 version from the forums and write to stderr 2010-09-16T15:25:14 Bob___: just make sure not to include that code in the version you upload to the server 2010-09-16T15:25:20 amstan: the C++ package is done all wrong. 2010-09-16T15:25:32 i set 300000 milliseconds timeout and attach the debugger. but i use C# 2010-09-16T15:25:32 *** kcm1700 has quit IRC (Quit: ¡¬(^o^)£¯) 2010-09-16T15:25:34 ok thanks guys ill give it a try 2010-09-16T15:25:45 amstan: I hate working with string, tokenizer. 2010-09-16T15:25:49 the java version seems to be coded up quite nice 2010-09-16T15:26:13 I'm sorry but, using strings to control the bots is silly.. 2010-09-16T15:26:29 *** retybok__ has joined #aichallenge 2010-09-16T15:26:41 what controls require strings? 2010-09-16T15:26:46 or what methods rather 2010-09-16T15:26:58 What does PlayGame.jar do to the bot process when the game is over? Does it kill it? 2010-09-16T15:27:03 graphitemaster: you mean as a form of IPC? what would you prefer? 2010-09-16T15:27:04 sorry i havent played with it much 2010-09-16T15:27:06 im just wondering 2010-09-16T15:27:15 graphitemaster: .... huh? 2010-09-16T15:27:17 http://www.ai-contest.com/profile.php?user_id=6149 2010-09-16T15:27:18 (I want to profile my bot, but it seems gprof doesn't work if the process is killed) 2010-09-16T15:27:22 graphitemaster: so... list 2010-09-16T15:27:29 like* you never have to touch the starter package cod3e 2010-09-16T15:27:34 here is my amazing parser: http://pastebin.org/889943 2010-09-16T15:27:39 everything can be put in one function, stop complaining 2010-09-16T15:27:44 there's your starter package 2010-09-16T15:27:54 lol! 2010-09-16T15:28:06 a1k0n, that's a hell of a lot nicer then that silly string tokenization crap. 2010-09-16T15:28:08 it's probably 100 times faster than the one in the C++ starter pack 2010-09-16T15:28:33 a1k0n: do you think it makes a real difference? 2010-09-16T15:28:38 nope 2010-09-16T15:28:41 lol 2010-09-16T15:28:45 a1k0n, I love that "// no time for error checking!" 2010-09-16T15:28:46 a1k0n: is that going to try to read the 'P' twice? 2010-09-16T15:28:46 :-D 2010-09-16T15:28:54 that should be the C starter package sorry, thats to great 2010-09-16T15:29:01 yeah i went through the trouble of using bytestrings rather than [Char] in my haskell bot and i doubt it's a huge difference either 2010-09-16T15:29:20 but it's also easy, so might as well just do it 2010-09-16T15:29:21 Zannick: why would it? it doesn't that i've noticed 2010-09-16T15:29:22 *** Bob___ has quit IRC (Quit: Page closed) 2010-09-16T15:29:46 oh, wait, you're reading into a buffer and using sscanf 2010-09-16T15:29:51 *** phreeza has quit IRC (Quit: kthxbai) 2010-09-16T15:29:54 right. 2010-09-16T15:30:03 * rogue780 wishes he knew C 2010-09-16T15:30:10 i'm using pure scanf :P 2010-09-16T15:30:19 ha, how? 2010-09-16T15:30:36 scanf(" %c%c", &c, &d); 2010-09-16T15:30:39 switch on c 2010-09-16T15:30:51 ah. 2010-09-16T15:31:12 that gets either "P ", "F ", or "go" 2010-09-16T15:31:20 right, makes sense 2010-09-16T15:31:35 graphitemaster: make a c starter package, polish it, and we'll add it 2010-09-16T15:31:35 what about the newline after go? 2010-09-16T15:31:45 newline is whitespace 2010-09-16T15:31:47 nice code :) 2010-09-16T15:32:01 *** amstan has quit IRC (Remote host closed the connection) 2010-09-16T15:32:03 so the " " in the scanf format will skip past it 2010-09-16T15:32:03 yeah but won't the next call get you "\nP"? 2010-09-16T15:32:10 * Janzert is sure he saw a C starter package thread in the forums the other day, might want to check that before doing any work 2010-09-16T15:32:11 oh, i see. clever. 2010-09-16T15:32:17 i haven't completed it yet 2010-09-16T15:32:24 but that's what is specced to happen 2010-09-16T15:32:36 so i will check to make sure, first 2010-09-16T15:33:59 doesn't it require whitespace before the initial P you get though? 2010-09-16T15:34:32 why are there private ssh keys in the public svn of the project? 2010-09-16T15:34:33 and don't forget to handle comments :) 2010-09-16T15:34:34 oh i guess not 2010-09-16T15:34:38 very clever. 2010-09-16T15:34:49 seems odd :) 2010-09-16T15:34:55 my parser is unfortunately slightly more involved than a1k0n's simple scanf version, but it also does some conversions and error handling 2010-09-16T15:35:02 yes, mine handles comments quite well: anything not starting with 'g', 'P', or 'F' is a comment 2010-09-16T15:35:41 a1k0n, i like the way you think 2010-09-16T15:36:38 *** RainCT has joined #aichallenge 2010-09-16T15:36:43 dangit now i just want to golf my parser 2010-09-16T15:36:47 * jmcarthur resists 2010-09-16T15:37:14 that'd be like the people who spend the entire ICFP contest optimizing their VMs 2010-09-16T15:37:19 lol 2010-09-16T15:37:46 it's not like mine is even that bad 2010-09-16T15:37:59 does a float always hold 10 digits precision? should it be double? 2010-09-16T15:38:14 *** deepblue has quit IRC (Quit: Page closed) 2010-09-16T15:38:28 subtle bug lurking there :) 2010-09-16T15:38:38 http://jake.devio.us/starter/PlanetWars.html#file-format 2010-09-16T15:39:13 *** hellman has quit IRC (Ping timeout: 245 seconds) 2010-09-16T15:39:34 yeah i guess a 24-bit float only gets 7.2 digits of precision 2010-09-16T15:40:31 does that mean you're going to upload a map with planets 25 turns apart but look like they're 24 turns apart if you don't use the last three digits? 2010-09-16T15:40:43 I just checked all maps... non have 2 planets closer than 2... someone here said some of them do have planets as close as 1 2010-09-16T15:43:01 there was a map where a central planet is covered by 2 planets that are next to it 2010-09-16T15:43:19 Naktibalda: Yes... but thats the graphical representation. 2010-09-16T15:43:30 turns out PlayGame.jar kills the bot process with signal 15. If you want to use gprof, you have to trap signal 15 and call exit 2010-09-16T15:43:34 how far was it? 2010-09-16T15:43:40 all planets are point like as far as the game mechanic is concered 2010-09-16T15:43:44 2 2010-09-16T15:43:50 nothing is smaller than 2 2010-09-16T15:44:15 this means a true distance of 1+eta or bigger 2010-09-16T15:45:31 *** odinsbane has quit IRC (Quit: out) 2010-09-16T15:45:56 it's good :) 2010-09-16T15:46:09 Made a much nicer Java version of the benzadrine.cx client: http://ai-contest.com/forum/viewtopic.php?p=3836#p3836 2010-09-16T15:46:46 Olathe: I have too 2010-09-16T15:47:20 Cool :) 2010-09-16T15:47:51 if by "nicer" you mean "lower scheduling priority," my system might be able to run it ;) 2010-09-16T15:48:15 whiteside: ? 2010-09-16T15:48:20 Haha 2010-09-16T15:48:57 I mean it's all multithreaded and can show the traffic and so on. 2010-09-16T15:49:15 I have graphical live view --and replay 2010-09-16T15:49:22 :p 2010-09-16T15:49:29 Oh, that's cool. Was it the Qt one? 2010-09-16T15:49:33 no 2010-09-16T15:49:38 I have done a java one 2010-09-16T15:49:48 I am going to finish it... 2010-09-16T15:49:48 Java's been killing my weak system specs, so when Olathe said "nicer Java" the only way I could keep my brain from exploding was to interpret that as "nice" as in "the *nix command" 2010-09-16T15:49:58 Oh, OK. Do you have it for download? 2010-09-16T15:50:04 its a game player as well 2010-09-16T15:50:19 *** tobiassjosten has joined #aichallenge 2010-09-16T15:50:40 and you can have the bots not exit so you don't have jvm setup and teardown 2010-09-16T15:50:57 or you can use the javabot interface and run in the same jvm 2010-09-16T15:51:14 and its permitiing batching with replay etc 2010-09-16T15:51:29 Finally you can play as a human 2010-09-16T15:51:34 but that needs work 2010-09-16T15:51:40 Ahh, that will be interesting. 2010-09-16T15:51:44 I plan to release it in the weekend 2010-09-16T15:51:51 Oh, OK :) 2010-09-16T15:51:51 code is a bit hacky however 2010-09-16T15:52:09 I should be writing a bot.... not yet anyother game engine 2010-09-16T15:52:26 but you know how it is 2010-09-16T15:52:51 Yes :) I've been putting effort into a more-optimized Haskell backend. 2010-09-16T15:53:14 been a while since i did any haskell 2010-09-16T15:54:11 whiteside: Java is not really slow per say... the badly writen java we are seeing is very slow... 2010-09-16T15:54:30 but still the bots are going to take 1000% more cpu than the game client 2010-09-16T15:54:41 and on the tcp side its all lag 2010-09-16T15:55:30 I would be surprised that your having problems with that unless you are one some 100MHz PI or something 2010-09-16T15:55:42 or only have 100Meg ram 2010-09-16T15:56:27 yeah, I have 256 megs of ram 2010-09-16T15:56:42 ah. yeah, java needs all of that to run hello world 2010-09-16T15:57:16 it's a celeron, too, so even my bank account has more cache 2010-09-16T15:57:45 well physical ram will be very low... it has a lot of address space --not physical space 2010-09-16T15:58:11 ie I have not incress my Heap+stack beyone the defaul 64M 2010-09-16T15:58:15 deafult 2010-09-16T15:58:39 *** Accoun has quit IRC () 2010-09-16T15:58:39 celeron...itsn't that something you eat 2010-09-16T15:58:48 isn't 2010-09-16T15:58:55 whiteside: look at the bright side. your specs are closer to the server's than mine :) 2010-09-16T15:59:13 nice 2010-09-16T15:59:22 or at least what it will appear to be like considering load, etc. 2010-09-16T15:59:33 *** rafsoaken has quit IRC (Ping timeout: 252 seconds) 2010-09-16T15:59:47 I don't think computational power is such a big deal for this contest 2010-09-16T15:59:55 i do 2010-09-16T16:00:01 not as big as the last contest 2010-09-16T16:00:09 tell that to my neu... oh, right, I don't have enough horsepower for one 2010-09-16T16:00:11 but a think many bots will still use a lot of computation 2010-09-16T16:00:14 of course it's nice but the key is to find the good heuristics, once you find them, your bot doesn't need to compute that much 2010-09-16T16:00:26 jmcarthur: I think so to 2010-09-16T16:00:29 that's one way 2010-09-16T16:00:35 retybok__: we'll see if you're still saying that a month from now 2010-09-16T16:00:36 I just think they don't really *need* to 2010-09-16T16:00:43 i think there could still be some good search algorithms 2010-09-16T16:00:52 a1k0n: I'm already getting timeouts, don't worry :-D 2010-09-16T16:01:08 while there have a huge branching factor compared to tron, we also can make some other assumptions about planet wars that we couldn't about tron... 2010-09-16T16:01:15 *while we have 2010-09-16T16:01:19 a good search with a good heuristic will be better than just a good heuristic 2010-09-16T16:01:33 and also better than just a good search 2010-09-16T16:01:51 * Janzert will be disappointed if there aren't scalable algorithms dominating the contest by the end 2010-09-16T16:01:56 sure, but you can't search much anyway :-D 2010-09-16T16:01:56 although perhaps "good search" already implies "good heuristic" 2010-09-16T16:02:06 retybok__: sure you can 2010-09-16T16:02:11 just not exhaustivelyu 2010-09-16T16:02:13 -u 2010-09-16T16:02:23 by much I mean much of the search space :) 2010-09-16T16:02:37 you can search options generated by heuristics 2010-09-16T16:02:41 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-09-16T16:02:52 I agree 2010-09-16T16:03:01 I think that's what we all do to some extent 2010-09-16T16:03:07 retybok__: same thing for chees --but you can prune the tree enough that you see the part of the search space that matters 2010-09-16T16:03:26 in chess the tree is much, much less branchy than for this 2010-09-16T16:03:34 true 2010-09-16T16:03:47 that just means pruning is more important here 2010-09-16T16:04:20 pruning the right parts of the space too 2010-09-16T16:04:26 so important that it's probably better to generate very few options in the first place rather than simply pruning from the entire set of options 2010-09-16T16:04:42 I agree 2010-09-16T16:04:50 *** RainCT has quit IRC (Remote host closed the connection) 2010-09-16T16:05:14 exactly my point: good pruning heuristics are actually more important than the search itself 2010-09-16T16:05:40 i'd say you don't have much of a search without pruning in *most* games, so i don't think your statement is saying anything all that shocking 2010-09-16T16:05:41 wtf? I just won another game, and it made my rank decrease in placement by 2 points!? http://www.ai-contest.com/profile.php?user_id=6149 2010-09-16T16:06:06 rogue780: read up on bayeselo 2010-09-16T16:06:37 it's still BS. I beat someone who is still higher ranked than me and my ranking went down 2010-09-16T16:06:44 rogue780: also, keep in mind that the ranking script doesn't update the rankings after every single game. you could be looking at an old rank, in actuality 2010-09-16T16:06:49 no, it's not bs 2010-09-16T16:07:04 rogue780: it might be completely unrelated to that game 2010-09-16T16:07:23 it might be unrelated to you, actually 2010-09-16T16:07:23 well, nevermind. my rank just updated 2010-09-16T16:07:25 rogue780: it just means that your ranking was overestimated to begin with. past opponents playing new games can also retroactively change your score 2010-09-16T16:07:33 jumped from 284 to 186 2010-09-16T16:09:05 rogue780: i proposed a system that would result in bots almost always climbing rather than jumping around, but it would mean that resubmission would put you near the bottom and then take many hours to get back where it was. that's the price you way for making early guesses worth less :\ 2010-09-16T16:09:31 *** HarryD|cb has joined #aichallenge 2010-09-16T16:09:50 there is also a reasonable in-between we could try, but i don't think it's been tested yet 2010-09-16T16:11:55 *price you pay 2010-09-16T16:12:00 I'm really just pleased that this version of the bot is doing so well when my last one was in the 500's 2010-09-16T16:12:39 i still have no idea why my throwaway bot is so good; my more serious attempts all suck in comparison 2010-09-16T16:13:33 a1k0n: same problem here 2010-09-16T16:13:45 my "simplest thing that could work" bot was surprisingly strong 2010-09-16T16:13:52 and it's very hard to climb from there 2010-09-16T16:13:58 *** Accoun has joined #aichallenge 2010-09-16T16:14:50 although it's not as good as yours :-D 2010-09-16T16:16:13 well...off to calculus. see ya'll later 2010-09-16T16:16:41 rogue780: maybe that will give you good ideas, who knows? :) 2010-09-16T16:17:07 one could hope 2010-09-16T16:18:44 http://www.galcon.com/forums/14/17/3629/?cur=20 2010-09-16T16:19:12 is that so, the floor() in distance means you travel faster by hopping? i didn't think of that! 2010-09-16T16:19:47 awesome to know! 2010-09-16T16:20:04 I do hopping anyway, but now it's an extra bonus 2010-09-16T16:20:23 oh dear... using -O3 hurts my program's logic 2010-09-16T16:20:31 this is a problem 2010-09-16T16:20:33 dhartmei: but it's supposed to be ceiling... 2010-09-16T16:20:37 *** Pitel has joined #aichallenge 2010-09-16T16:20:38 oh that's galcon 2010-09-16T16:20:39 hopping? 2010-09-16T16:20:42 n/m me :) 2010-09-16T16:20:53 or what does the article mean? they are talking about this contest 2010-09-16T16:20:57 planet wars uses ceiling 2010-09-16T16:20:59 oh? 2010-09-16T16:21:08 it's supposed to be ceiling, not floor 2010-09-16T16:21:10 so hopping is slower 2010-09-16T16:21:49 right, ceil(). then what does he mean. is he just wrong? :) 2010-09-16T16:22:07 i guess 2010-09-16T16:23:52 *** amstan has joined #aichallenge 2010-09-16T16:23:52 *** ChanServ sets mode: +o amstan 2010-09-16T16:23:53 *** amstan has quit IRC (Read error: Connection reset by peer) 2010-09-16T16:26:10 why is your game hating my bot? it works on my machine, beating randombot (:D) byt whe I pack and upload it, it seems to still ends after 1 turn :( http://ai-contest.com/profile.php?user_id=7372 2010-09-16T16:26:38 server has an old version of Python 2010-09-16T16:26:55 you have to adapt your bot to work with it 2010-09-16T16:27:34 or... may I wait until server python will be upgraded? 2010-09-16T16:27:50 no 2010-09-16T16:28:37 and what is server python version? 2010-09-16T16:29:55 2.6.something 2010-09-16T16:29:59 bartwe: you're killing me... :( 2010-09-16T16:30:04 if I remember correctly 2010-09-16T16:30:28 2.5.x 2010-09-16T16:31:03 oh, actually, the forgotten debug "print" made it fail. it works now \o/ 2010-09-16T16:31:38 *** bob___ has joined #aichallenge 2010-09-16T16:32:00 i know this may be a bit of a stupid question but can I alter any class in the main folder? 2010-09-16T16:32:13 or will it get overwritten 2010-09-16T16:32:15 when i upload it? 2010-09-16T16:32:25 you can change it 2010-09-16T16:32:26 no it wont 2010-09-16T16:32:31 sweet 2010-09-16T16:33:33 thanks 2010-09-16T16:33:49 They use ceiling because, if you can only get a certain distance per turn, going a certain distance plus a bit more on the last turn is not quite right :) 2010-09-16T16:35:11 *** jukkerknott has joined #aichallenge 2010-09-16T16:35:28 what is the movement rate of the units? say i want to divide the growth rate of a planet by the movement speed of my fleet so that I know how much extra units to add on so that they can take the planet when they get there 2010-09-16T16:35:29 *** Vi0 has quit IRC (Ping timeout: 276 seconds) 2010-09-16T16:37:10 jesus i slept forever today 2010-09-16T16:39:09 bob___: 1 unit per turn 2010-09-16T16:39:26 *** boegel has joined #aichallenge 2010-09-16T16:39:37 1 unit of distance 2010-09-16T16:40:02 ok thanks 2010-09-16T16:40:21 MassAttackBot vs RageBot 87/100 Avg Turns: 75.0 My new version is much faster :) Older took 200 turns in average 2010-09-16T16:41:11 *** MorkBork has left #aichallenge ("Ex-Chat") 2010-09-16T16:41:26 which premade bot is the best? 2010-09-16T16:41:43 *** jukkerknott has quit IRC (Ping timeout: 252 seconds) 2010-09-16T16:41:50 I thought I read somewhere that dualbot was but I haven't checked it. 2010-09-16T16:42:04 sweet, got 62 turns on duel bot :p 2010-09-16T16:42:12 got lucky though 2010-09-16T16:42:22 *** bob___ has quit IRC (Quit: Page closed) 2010-09-16T16:42:37 *** ermau|work has quit IRC (Quit: ermau|work) 2010-09-16T16:44:49 RageBot wins more agains my bot 2010-09-16T16:47:54 *** pjreddie has joined #aichallenge 2010-09-16T16:48:51 *** zaph has joined #aichallenge 2010-09-16T16:52:09 *** sigh has joined #aichallenge 2010-09-16T16:59:09 *** Mathnerd314 has quit IRC (Ping timeout: 265 seconds) 2010-09-16T17:00:19 *** amstan has joined #aichallenge 2010-09-16T17:00:19 *** ChanServ sets mode: +o amstan 2010-09-16T17:00:41 yo 2010-09-16T17:00:48 so... what's up guys? 2010-09-16T17:00:59 hey 2010-09-16T17:01:05 someone hire me to do something cool 2010-09-16T17:01:07 imo 2010-09-16T17:01:20 im tired of coding for idiot clients 2010-09-16T17:01:31 lol 2010-09-16T17:01:35 haha 2010-09-16T17:01:50 livin in a dream world if you think there are such things as non idiot clients 2010-09-16T17:02:03 dstufft: put this contest on your resume, apply to google 2010-09-16T17:02:25 Zannick: i could put helping admin it i guess, didn't think about that 2010-09-16T17:02:31 dstufft: can you create a bot for me? 2010-09-16T17:02:37 yeah, it's good stuff 2010-09-16T17:02:40 Naktibalda: create a bot? 2010-09-16T17:02:48 for contest :) 2010-09-16T17:02:48 I hate living in Philly 2010-09-16T17:02:51 but i can't leave D: 2010-09-16T17:03:13 not until my wife's daughter is old enough to tell her shitbag father to go fuck himself 2010-09-16T17:03:21 So... is there any way we can get some sort of carry over for our ranking on re 2010-09-16T17:03:26 *submissions 2010-09-16T17:03:31 dstufft: :/ 2010-09-16T17:03:47 i get tired of submitting updates and dropping back to like 500 2010-09-16T17:04:00 *** Nubick has joined #aichallenge 2010-09-16T17:04:39 pjreddie: i don't know bayeselo that well, but i think the reason that is done is because of that, but don't quote me on it 2010-09-16T17:04:43 pjreddie: there are ways, but in tests so far they have negative consequences 2010-09-16T17:05:03 yeah, and bayeselo doesn't support that kind of thing 2010-09-16T17:05:21 can anybody explain how http://ai-contest.com/profile.php?user_id=8123 grows so quickly? this bot fighted against bots are weaker then it. 2010-09-16T17:05:27 we tested some things other than bayeselo and they have pretty bad results 2010-09-16T17:05:39 *** rwa has quit IRC (Quit: rwa) 2010-09-16T17:05:45 *** tobiassjosten has quit IRC (Ping timeout: 272 seconds) 2010-09-16T17:06:05 artyom_shegeda: the rank is a best guess, not concrete 2010-09-16T17:06:42 ok. will first lose show the real position? :) 2010-09-16T17:06:45 artyom_shegeda: the rank is an estimate of what that bot's ELO score would be, the system produces a "range" of where that bot's ELO score would probably fall 2010-09-16T17:06:56 artyom_shegeda: there is no "real" position 2010-09-16T17:07:05 artyom_shegeda: it converges to the correct value after many games 2010-09-16T17:07:06 the more games you play, the more confident the system is of it's estimate 2010-09-16T17:07:14 ok, system is optimistic :) 2010-09-16T17:07:23 it is neither optimistic nor pessimistic 2010-09-16T17:07:27 if i recall we use the mean of the range 2010-09-16T17:07:28 i wish it was pessimistic 2010-09-16T17:07:31 of possible values 2010-09-16T17:07:32 yeah we use the mean 2010-09-16T17:07:37 *** boegel has quit IRC (Quit: Leaving) 2010-09-16T17:07:47 we are still experimenting with using the lower bound instead 2010-09-16T17:07:49 * artyom_shegeda run away 2010-09-16T17:07:58 one problem with ranking is that a "rank" doesn't really exist 2010-09-16T17:08:14 Its not true that a>b and b>c then a>c is also true 2010-09-16T17:08:30 in practice 2010-09-16T17:08:41 the 95% lower confidence bound has appeared to be unfavorable so far, but maybe we can go with something like 50% lower bound or something... 2010-09-16T17:08:44 yeah 2010-09-16T17:08:54 actually 50 is silly 2010-09-16T17:08:56 :P 2010-09-16T17:09:15 75% is what i meant 2010-09-16T17:09:16 *** Kastra has quit IRC (Ping timeout: 240 seconds) 2010-09-16T17:09:27 yea you can't say that bots would win in order of the ranking 2010-09-16T17:09:37 but there's no real perfect system 2010-09-16T17:09:53 atleast this is better then something like double elimination lol 2010-09-16T17:10:10 we could have 100-elimination :P 2010-09-16T17:10:23 doesn't condorcet voting give an ordering with good properties? 2010-09-16T17:10:26 well the ranks are only for --well fun. How is the final competition done? stil with elo? 2010-09-16T17:10:38 delt0r_: currently that's the plan 2010-09-16T17:10:42 oh 2010-09-16T17:10:45 delt0r_: it's not set in stone though 2010-09-16T17:10:49 delt0r_: it's still elo. basically just attempt to give enough games to the top end to make the confidence intervals small 2010-09-16T17:10:49 right.. 2010-09-16T17:11:34 If I understand correctly, the condorcet system could be use to rank people at the end, not to give quick estimates 2010-09-16T17:11:37 well if you really have a case of a>b b>c a then getting small intervals could be impossible 2010-09-16T17:12:00 yeah 2010-09-16T17:12:14 delt0r_: a counter-argument would be that we can't declare a real winner, though 2010-09-16T17:12:21 and since we want to do that, we have to sacrifice something 2010-09-16T17:12:21 true 2010-09-16T17:12:30 Oh I am not complaining... 2010-09-16T17:12:51 delt0r_: no discussion is good, if something better is found im sure there will be no issues in switching to it 2010-09-16T17:12:55 I think awarding points for wins and losses is good enough 2010-09-16T17:13:15 but its not like it doesn't suffer from the same problem 2010-09-16T17:13:24 we experimented with trueskill already. i didn't see the results of the experiment myself, but i was told that it was kind of crappy looking 2010-09-16T17:13:39 delt0r_: the issue with that is, it doesn't differntiate between winning against a good bot, and a bad bot 2010-09-16T17:13:48 yea 2010-09-16T17:13:48 and for most of tron we had a simple points based system, and that *really* sucked 2010-09-16T17:13:53 Oh i get Elo 2010-09-16T17:14:08 yea, and bayeslo isn't true elo 2010-09-16T17:14:14 it's used to get an estimate of elo 2010-09-16T17:14:43 which as a Bayesian myself---is good enough :) 2010-09-16T17:14:58 bayeselo is certainly way better than win-loss points like we had at the beginning of tron :} 2010-09-16T17:15:02 its just getting enougth game 2010-09-16T17:15:38 with some n core monster machine for a game box, then its would be fine... 2010-09-16T17:15:42 delt0r_: yea, well for the finals there won't be new submissions coming in, so we can run the game playing for as long as we need 2010-09-16T17:15:52 we have enough bots that a single round robin would require about a million games 2010-09-16T17:15:58 currently 2010-09-16T17:16:03 well 2010-09-16T17:16:09 assuming that as long as we need it's retarded long 2010-09-16T17:16:12 isn't* 2010-09-16T17:16:14 a million is not so many 2010-09-16T17:16:22 :D 2010-09-16T17:16:28 it is when we are averging 19 games a minute :p 2010-09-16T17:16:34 I have a 500 core cluster! 2010-09-16T17:16:44 a round robin is a pretty poor way of distinguishing between the top bots though 2010-09-16T17:16:45 >.> 2010-09-16T17:16:47 *** Nubick has quit IRC (Quit: Page closed) 2010-09-16T17:17:03 why? is there a better way? 2010-09-16T17:17:10 Janzert: a *single* round robin certainly is 2010-09-16T17:17:18 site down for anyone else? 2010-09-16T17:17:30 amstan: wfm 2010-09-16T17:17:35 wfm2 2010-09-16T17:17:37 any number of rounds is worse than a better designed tournament using the same number of total games 2010-09-16T17:17:51 worse when also considering time 2010-09-16T17:18:01 right total games 2010-09-16T17:18:28 i believe one idea for the final was to have the top bots play a tournament with each match being a game on each map 2010-09-16T17:18:46 also remember that the finals can use an entirely different scheduling system then the pre-finals 2010-09-16T17:19:02 dstufft: well, it will have to :P 2010-09-16T17:19:02 there was some discussion on this during the tron tournament if you look in the forum archives 2010-09-16T17:19:10 so right now we can optimize for different things (getting everyone a fairs hake at the games etc) 2010-09-16T17:19:29 er 2010-09-16T17:19:30 i mean 2010-09-16T17:19:36 making sure people don't go to long with no games played 2010-09-16T17:19:37 and stuff 2010-09-16T17:19:45 * Janzert agrees 2010-09-16T17:19:48 *** hakand has quit IRC (Ping timeout: 240 seconds) 2010-09-16T17:20:07 *** amstan_ has joined #aichallenge 2010-09-16T17:20:07 *** ChanServ sets mode: +o amstan_ 2010-09-16T17:20:32 one thing I'd still like to really see change is have timeouts be simple losses 2010-09-16T17:20:44 rather than disqualification 2010-09-16T17:20:46 me too 2010-09-16T17:21:33 Janzert: Yea 2010-09-16T17:21:34 honestly i think an n-elimination bracket could be reasonably appropriate for determining a winner 2010-09-16T17:21:39 for suitably large n 2010-09-16T17:21:50 id rather just have them count as nothing 2010-09-16T17:21:53 jmcarthur: agreed 2010-09-16T17:21:59 *** amstan has quit IRC (Ping timeout: 264 seconds) 2010-09-16T17:22:09 if a bot times out or crashes it sohuldn't factor into the rankings at all 2010-09-16T17:22:16 because it's not like you "won" 2010-09-16T17:22:17 what does disqualification do? 2010-09-16T17:22:28 *** amstan has joined #aichallenge 2010-09-16T17:22:28 *** ChanServ sets mode: +o amstan 2010-09-16T17:22:28 dstufft: That is unfortunatly unworkable 2010-09-16T17:22:30 dstufft: i would consider a faulty AI to be a losing AI, personally 2010-09-16T17:22:32 *** amstan_ has quit IRC (Read error: Connection reset by peer) 2010-09-16T17:22:33 jmcarthur: that would certainly be way better than n-round round robin 2010-09-16T17:22:40 well, i officially concede to bartwe 2010-09-16T17:22:43 his bot owns my face 2010-09-16T17:22:46 folk with "timeout" when they are loosing 2010-09-16T17:22:56 and not only that, the patterns it makes is beautiful 2010-09-16T17:22:56 :( 2010-09-16T17:23:01 heh 2010-09-16T17:23:17 jmcarthur: oh i mean record a timeout, and if they get N timeouts just kill the submission completly 2010-09-16T17:23:19 Janzert: i wasn't proposing that we do a round robin earlier, if that wasn't clear 2010-09-16T17:23:21 or crashes 2010-09-16T17:23:43 dstufft: i just consider such things to be forfeits, is all 2010-09-16T17:23:50 blister: post link? 2010-09-16T17:23:56 jmcarthur: ok, it was seriously considered for a while for the last contest 2010-09-16T17:23:58 http://www.benzedrine.cx/planetwars/canvas?game_id=9735 2010-09-16T17:24:01 thats one example 2010-09-16T17:24:06 but all of his games with me end up that way 2010-09-16T17:24:21 beautiful chains of planets sending fleets to other planets 2010-09-16T17:25:17 Janzert: i know. i was in opposition to that 2010-09-16T17:25:32 bartwe's minimum spanning trees *are* pretty fun to watch 2010-09-16T17:25:38 yeah 2010-09-16T17:25:45 *** ermau has joined #aichallenge 2010-09-16T17:25:45 *** ermau has joined #aichallenge 2010-09-16T17:25:56 yeah 2010-09-16T17:26:38 what frustrates me the most about his bot is that every time i attack something he wants, he'll send exactly the right amount of ships to make sure i dont get it 2010-09-16T17:26:46 ha 2010-09-16T17:26:46 very annoying 2010-09-16T17:26:48 :D 2010-09-16T17:27:08 *** mega1 has quit IRC (Ping timeout: 245 seconds) 2010-09-16T17:27:09 that's one thing i already thought about putting in my bot 2010-09-16T17:27:13 haha mine does that 2010-09-16T17:27:17 yeah 2010-09-16T17:27:19 i hate it 2010-09-16T17:27:19 :D 2010-09-16T17:27:19 it's good defense 2010-09-16T17:27:25 but plan to add it to my bot as well 2010-09-16T17:27:27 funny to see planets end up with 0 but still be yours 2010-09-16T17:28:30 however i'm rethinking minimilist conquering. cause then i only have 1 guy there and it's pretty defenseless 2010-09-16T17:28:37 haha 2010-09-16T17:28:44 bartwe vs bartwe is classic 2010-09-16T17:28:53 they're both stopping the other bot from getting what it wants 2010-09-16T17:29:01 link to one of those? 2010-09-16T17:29:12 http://www.benzedrine.cx/planetwars/canvas?game_id=9852 2010-09-16T17:29:23 pjreddie: that can work as long as you csn send defences beofr the attack gets there 2010-09-16T17:29:34 * can before 2010-09-16T17:29:45 *** jaspervdj has quit IRC (Quit: leaving) 2010-09-16T17:30:24 ya the problem is scanning for planets closer than my closest backup and sending forces that can defend against what i won't be able to react to later 2010-09-16T17:30:58 if that makes sense... 2010-09-16T17:31:10 yeah 2010-09-16T17:32:14 but maybe you dont minimaly counquer anything outside your "reaction circle" 2010-09-16T17:32:40 yeah exactly. eventually i'll put in some case for that... 2010-09-16T17:32:50 Is there still a plan to make a big dump of game logs available? 2010-09-16T17:33:02 *** McLeopold has joined #aichallenge 2010-09-16T17:33:37 i think my bot has downs syndrome 2010-09-16T17:33:59 its programmed to want to attack high growth planets closest to itself 2010-09-16T17:34:05 but in this match http://www.benzedrine.cx/planetwars/canvas?game_id=9750 2010-09-16T17:34:28 it continues to send all of its ships to a 1GR planet on the other side of the map directly in between two large enemy controlled 5GR planets 2010-09-16T17:34:30 ratatata: I think you need +1 somewhere... http://www.benzedrine.cx/planetwars/canvas?game_id=9802 2010-09-16T17:34:43 i have no idea what peice of code would cause my bot to do this... 2010-09-16T17:34:44 :/ 2010-09-16T17:34:56 i've never had so many off-by-ones in a single program before :) 2010-09-16T17:35:20 it attacks a planet with smallest num_ships 2010-09-16T17:35:44 *** Appleman1234 has joined #aichallenge 2010-09-16T17:35:47 dwchandler: yea I know.. my attack system is a huge mess at the moment 2010-09-16T17:36:00 blister: probably you still have / $num_ships 2010-09-16T17:36:25 hahaha 2010-09-16T17:36:35 nah 2010-09-16T17:36:56 *** zaph has quit IRC (Quit: Bye) 2010-09-16T17:37:01 prior to turn 100, my strategy is "conquor as many high GR planets as possible as close as possible to my own planets" 2010-09-16T17:37:17 there is absolutely nothing in my code that would say "attack that little planet on the other side of the map 2010-09-16T17:37:26 but apparently there is 2010-09-16T17:37:29 i think my AI has attained sentience 2010-09-16T17:37:39 and is so depressed at how shitty I built it that it's trying to commit suicide 2010-09-16T17:37:42 :/ 2010-09-16T17:37:55 you should rename it "Marvin" 2010-09-16T17:38:00 haha 2010-09-16T17:38:03 good idea 2010-09-16T17:38:08 lol 2010-09-16T17:38:08 You should program it to write the letter "B" in a stream of ships. 2010-09-16T17:38:14 brain the size of a planet... 2010-09-16T17:38:35 blister: then you formula should be growth_rate / distance 2010-09-16T17:39:18 after seeing http://ai-contest.com/visualizer.php?game_id=4501311 on the forums i really wanna program this in. just to mess with people. 2010-09-16T17:39:26 *** sigh has quit IRC (Quit: Leaving...) 2010-09-16T17:40:37 pjreddie: please don't :p 2010-09-16T17:40:43 this bot sucks at coordinating attacks 2010-09-16T17:41:07 hmm, it's sending a growh_rate, what a nonsense 2010-09-16T17:41:38 Naktibalda: the point is to tease the opponent 2010-09-16T17:41:57 this is basically what would happen if the rankings took into consideration the number of ships you had when you won 2010-09-16T17:41:57 oh really? 2010-09-16T17:42:06 :) 2010-09-16T17:42:15 the bots would "win", and then just send enough ships to keep their opponent cornered 2010-09-16T17:42:27 and wait it out for the 200 turns to end so they would get more ships 2010-09-16T17:44:00 *** Pitel has quit IRC (Quit: KTHXBYE) 2010-09-16T17:45:03 *** Blkt has joined #aichallenge 2010-09-16T17:45:07 *** braddunbar has joined #aichallenge 2010-09-16T17:45:50 *** amstan has quit IRC (Quit: Konversation terminated!) 2010-09-16T17:46:24 blister: ooh i'm up against bartwe right now... 2010-09-16T17:48:16 damnit, i keep timing out on the tcp thing... 2010-09-16T17:48:47 my debugging skills are getting worse 2010-09-16T17:48:53 for( $i = $distance; $i > 0; $i++ ) { 2010-09-16T17:48:54 *** si has joined #aichallenge 2010-09-16T17:49:20 *** si is now known as Guest26662 2010-09-16T17:49:27 heh 2010-09-16T17:50:37 *** HakanD has joined #aichallenge 2010-09-16T17:50:49 i'm so unused to having to write loops like that now 2010-09-16T17:51:01 *** McLeopold has quit IRC (Quit: Leaving) 2010-09-16T17:51:07 Naktibalda: "oops" 2010-09-16T17:51:28 well, that's a great loop if you want to loop over every number greater than distance 2010-09-16T17:51:33 until you overflow 2010-09-16T17:51:47 it's the first time when I got a real timeout 2010-09-16T17:51:51 *** mightybyte has quit IRC (Quit: leaving) 2010-09-16T17:52:07 it managed to calculate to ~32700 2010-09-16T17:54:03 if anyone else has time and GCC, would they be able to confirm discrepancies with their (deterministic) program logic by enabling -O3 versus no optimization? 2010-09-16T17:55:07 *** odinsbane has joined #aichallenge 2010-09-16T17:56:29 *** amriedle has joined #aichallenge 2010-09-16T17:56:33 *** inverselimit has joined #aichallenge 2010-09-16T17:57:18 whiteside: it's possible, but it's very likely that your program has a subtle bug (unspecified behavior?) 2010-09-16T17:58:33 what does D mean for a loss on Benze? disqualified? 2010-09-16T17:58:49 discon? 2010-09-16T17:58:54 there's a key at the bottom 2010-09-16T17:58:57 ohhhh ya makes sense 2010-09-16T17:59:08 cool, thanks. I keep dcing or timing out 2010-09-16T17:59:19 http://ai-contest.com/forum/viewtopic.php?f=18&t=621 perl starter packages 2010-09-16T17:59:22 gonna run it on one of the school's boxes 2010-09-16T17:59:25 is there a way to tell if a branch has been merged back into trunk on svn 2010-09-16T17:59:37 diff it with trunk? 2010-09-16T17:59:46 pjreddie: lol @ the dmj11 game you linked. 2010-09-16T17:59:47 or ask :) 2010-09-16T18:00:00 Zannick: won't that tell me any changes made to trunk since the branch has been merged back in? 2010-09-16T18:00:05 yes 2010-09-16T18:00:13 ugh svn 2010-09-16T18:00:25 but you also get to see the changes the branch made, if they haven't been merged in 2010-09-16T18:00:47 yeah. i wanna get to the point where mine does that. just leaves a few alive but chases them around or something 2010-09-16T18:00:58 eg, i got an approval yesterday for branches/issue112, but i have not merged it yet, and that was how i could tell it wasn't merged already 2010-09-16T18:01:00 Zannick: yea but that requires me to know what changes the branch made 2010-09-16T18:01:06 I don't think the growth-rate attack is a bad idea. I think it's not so much to "mess" with your opponent as it is to neutralize the enemies growth and defend. Totally reasonable strategy imho. 2010-09-16T18:01:18 Also does anyone do retreats? i feel like it would be a good thing to program in, like strategically pulling out if you are gonna lose 2010-09-16T18:01:19 justin_pdx: why not just finish off the oppononent 2010-09-16T18:01:37 they only had 1 planet left, and there was plenty of ships to finish them off 2010-09-16T18:01:46 dstufft: That would require more coding and possibly interrupt the general strategy. 2010-09-16T18:02:20 justin_pdx: all of the planets stopped attacking, except for 1 2010-09-16T18:02:33 im pretty sure that it was specfically coded to do that when the opponent only had 1 planet 2010-09-16T18:02:35 *** Palmik has quit IRC (Remote host closed the connection) 2010-09-16T18:02:50 retybok__: yeah, that was my first suspicion. So I did the honorable thing and tried to figure it out myself for a while, but I'm at a dead-end. I only have my bot, the example bot, and the Java example bots that I refactored to C (which therefore make poor test cases) to test, and both exhibit suspect behavior though in different ways 2010-09-16T18:03:12 pjreddie: I was considering that.. basically the conclusion I came to was, unless I could figure a way of using those ships so they'd take over another planet or something, I'd let them die (so as to reduce the number of enemy ships.. wouldn't want one big enemy fleet to be able to just walk through all my stars whilst I simply abandon them) 2010-09-16T18:03:34 ah ha 2010-09-16T18:03:36 svn mergeinfo 2010-09-16T18:04:45 *** mightybyte has joined #aichallenge 2010-09-16T18:04:53 *** retybok__ has quit IRC (Quit: leaving) 2010-09-16T18:05:32 dstufft: I don't think it's that simple. If it were, then it would send growth-rate attacks. It kept sending 4 8 0 0 4 8 0 0, I think it was trying to do something else. 2010-09-16T18:05:36 I wouldn't normally ask much less bother with the optimization, but the main server does -O3 -funroll loops by default, so, this is particularly disconcerting 2010-09-16T18:06:21 As you can see it allowed the enemy planet to grow. 2010-09-16T18:07:32 justin_pdx: anyways, i was just saying, people keep asking for the rankings to take into consideration how many ships you had when you won, compared to your opponent 2010-09-16T18:07:48 and saying that if we did that, that is an example of what every single bot would do 2010-09-16T18:07:59 I agree with that. A win should be a win. 2010-09-16T18:08:00 well every single bot that the person writing it knew what they were doing 2010-09-16T18:08:07 because you'd be a fool not too 2010-09-16T18:08:19 if you do that, you would want to also take the game length into account. 2010-09-16T18:08:25 and it's better for the games to be over as quickly as possible 2010-09-16T18:08:42 Maybe some ratio of total ships to game lengh? 2010-09-16T18:08:45 so that we can play another game 2010-09-16T18:08:49 a 100-turn win with 500 ships is better than a 200-turn win with 1500 ships (more than opponent) 2010-09-16T18:08:59 *** kozlovsky has quit IRC (Quit: Page closed) 2010-09-16T18:09:32 *** sinan has joined #aichallenge 2010-09-16T18:09:49 it ll greatly limit the strategies 2010-09-16T18:10:17 turtling is also a strategy (: 2010-09-16T18:11:48 woot 3 games to diagnose http://ai-contest.com/profile.php?user_id=3911 2010-09-16T18:14:11 *** retybok_ has quit IRC (Ping timeout: 252 seconds) 2010-09-16T18:15:52 *** Snowplan has joined #aichallenge 2010-09-16T18:15:57 ah why do i time out at times like this!! http://www.benzedrine.cx/planetwars/canvas?game_id=10240 2010-09-16T18:16:48 The contest read won of my wins as a loss 2010-09-16T18:16:51 whats that about 2010-09-16T18:17:20 So ever turn our bot recieves a message, which includes the map and the fleets? 2010-09-16T18:17:22 retybok you there? 2010-09-16T18:18:33 you timed out or entered a bad command 2010-09-16T18:18:45 or else retybok did and the servers backwards 2010-09-16T18:19:14 I woke up with tons of time outs etc so I know the tcp server changed lastnight - could be a latent bug from the fix 2010-09-16T18:19:42 i timed out but i don't know if it's my code or my connection 2010-09-16T18:20:08 cant say - I do have some dc's 2010-09-16T18:20:14 And in java the planet id's are actually just based on an arraylist, where the id corresponds to the order the planet was parsed? 2010-09-16T18:20:14 i never get close to timing out against bots so i can't recreate problem 2010-09-16T18:20:21 one of my matches against selrak_I I won but it says I lost Do I write someone 2010-09-16T18:20:49 Snowplan: how do you know you won? 2010-09-16T18:20:50 *** donkey has joined #aichallenge 2010-09-16T18:20:51 give a link to the game? 2010-09-16T18:21:05 http://ai-contest.com/visualizer.php?game_id=4514546 2010-09-16T18:21:16 wow.. my code on official site got compiled instantly and I played first game in 11 seconds :) 2010-09-16T18:21:27 *** Olathe has quit IRC (Ping timeout: 276 seconds) 2010-09-16T18:21:55 looks like a bug on the server 2010-09-16T18:22:04 amstan 2010-09-16T18:22:28 when 200 turns ends and a player owns 0 planets it bugs 2010-09-16T18:22:49 is anyone else seeing double listings like this for affiliations? http://ai-contest.com/organization_profile.php?org_id=81 2010-09-16T18:23:04 Its a few of my games now that I look 2010-09-16T18:23:26 but you have no rankings on your account an neither does your opponent so cant say any further 2010-09-16T18:23:32 Snowplan: You don't by anychance "stop" at 200 turn in your bot? 2010-09-16T18:23:44 no 2010-09-16T18:24:05 donkey: it looks like you all have 2 organizations assigned :) 2010-09-16T18:24:13 His opponent is still alive, he doesn't know what to do when they have zero planets. 2010-09-16T18:24:43 visualizer bug: names link to submission ids, not user ids 2010-09-16T18:24:54 Snowplan, no fleets sent out in last 5 turns or so either 2010-09-16T18:25:07 lmao, ty for that bug Zannick 2010-09-16T18:25:20 So It's probably my bot timing out when I dominate 2010-09-16T18:25:45 Snowplan: hard to say - that one is worth looking further into though 2010-09-16T18:25:46 Snowplan: its probably not timing out, it is probably an error, maybe an npe or something. 2010-09-16T18:26:13 Snowplan: because why doesnt your bot always time out when it takes all the planets? (I assume you have other games where you took all the planets) 2010-09-16T18:26:16 cause my bot should just hang out when it takes over 2010-09-16T18:26:25 yea 2010-09-16T18:27:06 Snowplan do you use the tcp.c at all? Then you can write error log files. 2010-09-16T18:27:21 http://ai-contest.com/visualizer.php?game_id=4508321 2010-09-16T18:27:28 I think its prob my bot 2010-09-16T18:27:36 cause this stops at 198 2010-09-16T18:27:46 and its the same condition 2010-09-16T18:27:50 that would be your bot :) 2010-09-16T18:28:33 I don't understand why. No worries though I've been working on a better bot that this shouldn't be a problem 2010-09-16T18:29:04 you might be crashing because the enemy has fleets but not planets 2010-09-16T18:29:15 crashing is often "timeout" 2010-09-16T18:29:17 I think thats it that If there are no planets to take over it says kill yourself 2010-09-16T18:30:20 yea thats exactly it 2010-09-16T18:30:28 thanks 2010-09-16T18:31:19 accessing the first element of an empty list is a pretty common mistake 2010-09-16T18:31:32 luckily it's easy to find 2010-09-16T18:31:38 *** amstan has joined #aichallenge 2010-09-16T18:31:38 *** ChanServ sets mode: +o amstan 2010-09-16T18:32:41 posted the double listing bug to the forum 2010-09-16T18:32:57 where issues 2010-09-16T18:33:01 @where issues 2010-09-16T18:33:01 Naktibalda: issues = http://code.google.com/p/ai-contest/issues/list 2010-09-16T18:34:04 *** braddunbar has quit IRC (Ping timeout: 240 seconds) 2010-09-16T18:34:27 Naktibalda: thanks 2010-09-16T18:35:45 oh, well that explains why I can't get *any* bot to draw itself 100/100 times (LobotomyBot doesn't count)... which means that troubleshooting my -O problem just got that much harder because I don't have any reasonable frame of reference 2010-09-16T18:36:45 haha - http://www.benzedrine.cx/planetwars/canvas?game_id=10333 2010-09-16T18:37:09 If the server records a loss due to disconnection, is that due to the bot crashing or other issues? 2010-09-16T18:37:14 *** amriedle has quit IRC (Quit: Lost terminal) 2010-09-16T18:37:36 Is the java starter generated code? 2010-09-16T18:37:52 it started off as that 2010-09-16T18:38:11 lol, war of attrition http://www.benzedrine.cx/planetwars/canvas?game_id=10369 2010-09-16T18:39:20 pjreddie: nice onne :D 2010-09-16T18:39:37 lol 2010-09-16T18:41:17 *** ratatata has quit IRC (Quit: Page closed) 2010-09-16T18:42:34 pjreddie: I like how the ones in the upper right have gotten into an equilibrium that is inconstant. 2010-09-16T18:43:05 *** amstan_ has joined #aichallenge 2010-09-16T18:43:05 *** ChanServ sets mode: +o amstan_ 2010-09-16T18:43:20 i know, awesome haha 2010-09-16T18:43:22 anyone up to solving this? http://code.google.com/p/ai-contest/issues/detail?id=158 2010-09-16T18:43:50 * greghaynes checks if its an interesting problem 2010-09-16T18:44:24 *** amstan has quit IRC (Ping timeout: 276 seconds) 2010-09-16T18:44:50 heh, someone doesnt like their job 2010-09-16T18:45:17 well.. fuck it... like.. too many people complaining and yelling randomly 2010-09-16T18:45:21 not even following instructions 2010-09-16T18:45:32 have a list, have a thing "don't see your org here?" that lets them enter one 2010-09-16T18:45:35 *** amstan_ is now known as amstan 2010-09-16T18:45:40 Users will never learn to file issues in the correct place 2010-09-16T18:45:48 this is something I made peace with a long time ago 2010-09-16T18:45:50 Zannick: yes.. put your suggestions there 2010-09-16T18:45:54 i don't care how it's made 2010-09-16T18:45:57 anything but this.. 2010-09-16T18:45:59 will do after work 2010-09-16T18:47:10 Cyndre you still here?? 2010-09-16T18:47:22 Zannick: keep it mind that it needs to solve duplicates and it still has to be able to let multiple users have the same org 2010-09-16T18:47:30 so this page will still work: http://www.ai-contest.com/organization_profile.php?org_id=1 2010-09-16T18:47:37 at this point i cannot do duplicates anymore 2010-09-16T18:48:00 yeah, let them enter one and have it go in the list 2010-09-16T18:48:10 it's...abusable, unfortunately 2010-09-16T18:48:15 without proper filtering 2010-09-16T18:49:36 would it be hard to give them a form to dump it in one table and then have another one to approve (move to the live table) or disaprove (throw out)? that seems like it could be a convinience for you amstan 2010-09-16T18:51:24 *** Guest26662 has quit IRC (Quit: Page closed) 2010-09-16T18:53:22 so are there any plans to improve the servers, get more games playing? i donow what your resources are... 2010-09-16T18:58:50 pjreddie: situation is much better now than a few days ago 2010-09-16T19:01:54 *** amstan has quit IRC (Remote host closed the connection) 2010-09-16T19:04:27 *** syntaxglitch has joined #aichallenge 2010-09-16T19:05:24 *** dshreyas has joined #aichallenge 2010-09-16T19:06:15 Wow, I'm so amazed how one formal CS class changed how I look at programming. 2010-09-16T19:06:35 how? 2010-09-16T19:07:00 what kind of class? 2010-09-16T19:07:16 I'd never had any formal CS training before and I took a datastructures class this summer 2010-09-16T19:07:39 and I'd never heard of Big O notation or thought of complexity before 2010-09-16T19:07:40 ah :) 2010-09-16T19:08:03 So now I'm organizing everything and making my bot as efficient as possible 2010-09-16T19:08:29 yes, big O complexity is a good rule of thumb of program speed, especially in contests like this 2010-09-16T19:08:32 *** whiteside has quit IRC (Quit: Page closed) 2010-09-16T19:09:00 j3camero: is there any ETA on when Ruby will be added? 2010-09-16T19:09:15 there are further classes that will really change how you think about programming 2010-09-16T19:09:29 I'd also never thought much about sorting before, but after taking that class I ended up adding an insertion sort to my bot b/c I realized the built-in sort for vectors would be inefficient 2010-09-16T19:09:41 Yeah, I really look forward to it 2010-09-16T19:09:46 functional programming, systems programming, kernels, programming language theory 2010-09-16T19:09:52 I thought I knew a whole lot of programming before, but I just know the basics 2010-09-16T19:10:05 my systems class had an assignment involving reading assembly and figuring out what it did 2010-09-16T19:10:17 Utkarsh: i think soon, a lot of the "server is blowing up OMG OMG" issues are now gone 2010-09-16T19:10:24 I tried teaching myself some Haskell in my spare time because I heard how it could really change how you thought about programming 2010-09-16T19:10:43 I worked with some assembly in an ECE class I also took 2010-09-16T19:10:51 but it was rather minimal 2010-09-16T19:11:16 if i had a good reason to put assembly in my bot i might :3 2010-09-16T19:11:26 *** danielvf has joined #aichallenge 2010-09-16T19:12:05 create a starter pack for assembly 2010-09-16T19:13:07 While it's true that a formal CS class can imrprove your skill set, I would say that self-trained programmers have a flexibility and ability to quickly adapt to new paradigms that I haven't seen in raw CS students. 2010-09-16T19:13:26 That's good to hear :) 2010-09-16T19:13:33 Although, that's probably because you don't have a lot of spare time for tinkering with new stuff as a CS student because you have so much homework. 2010-09-16T19:13:40 hahahaha no 2010-09-16T19:13:53 (i don't want to deal with stdin from inside assembly) 2010-09-16T19:14:29 justin_pdx: haha, yeah, i had very little free time as a cs student 2010-09-16T19:14:38 but i was double majoring... 2010-09-16T19:15:05 I don't have much time as a high school student :/ 2010-09-16T19:15:10 Though weekends are pretty much free 2010-09-16T19:15:21 i had so much time as a high school student somehow 2010-09-16T19:15:30 that i didn't have as a college student 2010-09-16T19:15:37 I'm glad that I dropped out of college, taught myself Rails and got a job. I'm slightly better off than my CS friends who can write really awesome C compilers but don't know their way around a relational database. 2010-09-16T19:15:37 despite spending /less/ time in classes 2010-09-16T19:15:57 Last year I didn't have much time, though it looks like this year I should have enough 2010-09-16T19:16:01 Especially near the end of the year 2010-09-16T19:16:31 i taught myself c++ junior year of college and submitted three bugfixes to an open source project senior year and now i've got a good job 2010-09-16T19:16:43 *** olexs has joined #aichallenge 2010-09-16T19:16:50 damnit all the work is in coding 2010-09-16T19:16:57 i'm having a hell of a time finding a sys ops job 2010-09-16T19:17:14 there's a lot of people who don't code at my company 2010-09-16T19:17:29 I've made some very minor contributions to open source projects 2010-09-16T19:17:36 indeed, my job is not primarily coding 2010-09-16T19:17:56 though i do a lot of it as necessary 2010-09-16T19:19:11 re the game from earlier: a weird bug i have not figured out yet :( 2010-09-16T19:19:19 *** McLeopold has joined #aichallenge 2010-09-16T19:20:40 I cleaned up the java files for the starter kit, is there someplace to upload them? 2010-09-16T19:22:39 *** sinan has quit IRC (Ping timeout: 255 seconds) 2010-09-16T19:23:25 Is a Ruby package coming down the pipe, anyone know? 2010-09-16T19:23:41 justin_pdx: yes probably 2010-09-16T19:23:47 someone has made a starter kit already 2010-09-16T19:24:03 Oh yeah? Where do I get a peek at that? 2010-09-16T19:24:04 all that needs to be done is setup the server to handle it and maybe make the start kit official 2010-09-16T19:24:09 justin_pdx: on the forums 2010-09-16T19:24:12 http://github.com/vikhyat/ai-contest-ruby 2010-09-16T19:24:13 Thanks. 2010-09-16T19:24:19 you won't be able to submit it the server currently 2010-09-16T19:24:26 but you can use it to play on the tcp server 2010-09-16T19:24:56 That will speed up my development time so I don't have to wrestle with Java's clunky-feeling list itterators etc. (at least, thats how they feel to me.) 2010-09-16T19:25:50 *** jmreardon has quit IRC (Quit: jmreardon) 2010-09-16T19:26:11 i don't use java much anymore, but i like to use for(a : alist) to abstract away the iterator 2010-09-16T19:26:24 if yo cant make a line that plays along with the visualizer for the graph, what are the chances of 6 light grey lines at turn 25, 50, 75, 125, 150, 175 and a dark black line at turn point 100? 2010-09-16T19:26:30 Well, I don't grok java's sorting algorithms. 2010-09-16T19:26:37 It's the sorting I can't grasp. 2010-09-16T19:26:46 ah, more different 2010-09-16T19:27:39 *** Queue29 has joined #aichallenge 2010-09-16T19:27:43 doing planets.sort { |a, b| a.NumShips <=> b.NumShips } is waaaay cooler than what java gives me. 2010-09-16T19:28:27 <=> ? 2010-09-16T19:28:42 this is kinda cool, we are pretty even for most of the game but we completely switch positions, rotating around each other 2010-09-16T19:28:43 http://ai-contest.com/visualizer.php?game_id=4515970 2010-09-16T19:28:43 that does look pretty cool, though, i admit 2010-09-16T19:29:10 or planets.sort_by { |p| p.NumShips } 2010-09-16T19:29:20 Even sexier, Utkarsh 2010-09-16T19:29:40 hmm, my bot doesn't attack the last planet for some reason too 2010-09-16T19:29:40 *** dshreyas has quit IRC (Ping timeout: 252 seconds) 2010-09-16T19:30:28 justin_pdx: those aren't sorting algorithms. 2010-09-16T19:30:42 I didn't so much mean algorithm. 2010-09-16T19:31:47 I meant technique? How to do it at all? Using lists and iterators confuses me. 2010-09-16T19:32:22 pjreddie: that game was awesome, the position switch is so fast and surprising. 2010-09-16T19:34:20 Agreed, that was all over the place 2010-09-16T19:37:02 *** whiteside has joined #aichallenge 2010-09-16T19:37:23 *** Mathnerd314 has joined #aichallenge 2010-09-16T19:38:35 *** smellyhippy has joined #aichallenge 2010-09-16T19:39:26 *** delt0r___ has joined #aichallenge 2010-09-16T19:40:54 *** delt0r_ has quit IRC (Ping timeout: 245 seconds) 2010-09-16T19:55:46 Sweet game 2010-09-16T20:00:03 Anyone here want to test ruby? 2010-09-16T20:01:00 upload starter package :) 2010-09-16T20:02:28 *** Blkt` has joined #aichallenge 2010-09-16T20:03:50 *** braddunbar has joined #aichallenge 2010-09-16T20:05:13 *** McLeopold has quit IRC (Quit: Leaving) 2010-09-16T20:06:18 *** Blkt has quit IRC (Ping timeout: 276 seconds) 2010-09-16T20:06:23 I want to test ruby. 2010-09-16T20:06:52 justin_pdx: I think I've enabled it 2010-09-16T20:06:54 danielvf: I can't start right away and will need to write a bot but I'm happy to be a part. I am excited to try the game in my favorite language. 2010-09-16T20:07:13 justin_pdx: Okay. Make something simple, upload it, and tell me if it works 2010-09-16T20:07:43 I'll test ruby too if you need. 2010-09-16T20:07:48 danielvf: someone made a starter pack for ruby 2010-09-16T20:07:55 Naith: that would be great 2010-09-16T20:07:58 so all a tester needs is to upload that 2010-09-16T20:08:13 dstufft: Great 2010-09-16T20:08:50 Well it doesn't have a mybot model, it's just a script. Would that work? 2010-09-16T20:09:17 I mean mybot class. Just do_turn method. Is that what you want? 2010-09-16T20:09:19 danielvf would know what exactly it needs, but typically it needs a standard starter file 2010-09-16T20:09:30 and as long as that file handles stdin/out 2010-09-16T20:10:21 danielvf: http://github.com/vikhyat/ai-contest-ruby/blob/master/mybot.rb is the ruby stuff someone already put together. 2010-09-16T20:10:33 Is that the right format for a ruby mybot? 2010-09-16T20:10:42 Yeah, you could submit that 2010-09-16T20:10:56 might need to be named MyBot.rb 2010-09-16T20:11:07 with the capital letters 2010-09-16T20:11:11 but who knows 2010-09-16T20:13:48 I submitted the files as I found them on that github. 2010-09-16T20:13:58 I have not received a notification yet. 2010-09-16T20:17:07 *** donkey has quit IRC (Quit: leaving) 2010-09-16T20:18:12 Okay 2010-09-16T20:18:25 What is your ai-contest profile url? 2010-09-16T20:18:47 Foundation 2010-09-16T20:18:50 *** danielvf has left #aichallenge 2010-09-16T20:19:43 *** danielvf has joined #aichallenge 2010-09-16T20:19:54 Can I have that url again? 2010-09-16T20:20:02 Foundation 2010-09-16T20:20:08 http://github.com/vikhyat/ai-contest-ruby/blob/master/mybot.rb 2010-09-16T20:20:09 user_id ? 2010-09-16T20:20:29 My contest id is Foundation. 2010-09-16T20:21:21 are you posting your code for all to see? 2010-09-16T20:21:33 I didn't write any part of that ruby code. 2010-09-16T20:21:43 oh 2010-09-16T20:21:50 But I wouldn't mind sharing my bot when I'm done writing one. 2010-09-16T20:22:06 *** jmreardon has joined #aichallenge 2010-09-16T20:22:10 I don't see why not except to embarrass myself by showing off my crappy understanding of ai programming. 2010-09-16T20:22:51 I thought ai programing was like super sophisticated 2010-09-16T20:22:56 and then someone told me no, it's just nested ifs 2010-09-16T20:22:58 and I was like wtf 2010-09-16T20:23:01 lol 2010-09-16T20:23:05 yeah 2010-09-16T20:23:16 well, there's ai programming, and then there's ai programming :) 2010-09-16T20:23:17 ai has varying degrees of sophistication 2010-09-16T20:23:33 but nested ifs are perfectly valid ai 2010-09-16T20:23:55 i have never really seen insanely complex ai before, though 2010-09-16T20:23:56 nested ifs is a valid ai approach, say, for a strictly rule-based bot 2010-09-16T20:24:02 calling nested ifs "ai" make nested ifs sound that much more awesome and it makes AI sound that much more stupid 2010-09-16T20:24:06 complicated ai usually is bad ai 2010-09-16T20:24:15 Remember what Kasparov got pissed and left a game with Deep Blue cause it was "playing too much like a person"? 2010-09-16T20:24:26 lol 2010-09-16T20:24:36 too much like a person wtf was he expecting 2010-09-16T20:24:42 he got schooled by a machine and emo quit 2010-09-16T20:24:53 He was expecting something that felt artificial. 2010-09-16T20:24:54 well that's because deep blue learned from game databases of... persons 2010-09-16T20:25:05 among other things 2010-09-16T20:25:18 so everythings going 'quantum' 2010-09-16T20:25:19 didn't deep blue "simply" do some fancy search tree pruning? 2010-09-16T20:25:22 what's the state of quantum ai? 2010-09-16T20:25:22 *** Naith has quit IRC (Quit: Naith) 2010-09-16T20:25:34 jmcarthur: that was the main part, but it had much more than that 2010-09-16T20:26:03 olexs: sure, hence the quotation marks ;) 2010-09-16T20:26:24 Many of the pro chess players train against computers, I bet thats why he quit, his program probably played differently and he miscalculated 2010-09-16T20:26:24 opening books, endgame databases, some quite sophisticated heuristics in both eval function and move generator... and still a lot more than that :) 2010-09-16T20:26:27 I dunno. Quantum AI currently highly sophisticated and perfect or really crappy. I haven't checked recently. 2010-09-16T20:28:04 I read an article that talked about how if true ai was ever created we would never know of it 2010-09-16T20:28:42 Cyndre: no way 2010-09-16T20:29:00 everyone knows that ai would be gloating on 4chan the instant it was released into the net 2010-09-16T20:29:01 ie: say I create the most sophisticated ai bot - it escapes from my computer and takes up residence all over - it then makes me some kind of war criminal, Im extradited to the usa and executed - our entire world is networked 2010-09-16T20:29:19 shades: only problem being no one would believe it 2010-09-16T20:29:20 not every country has extrdition treaties 2010-09-16T20:29:33 not every country has computer laws for that matter 2010-09-16T20:29:34 sahdes: my country does and it doesnt hae exection laws 2010-09-16T20:29:37 well that's more a hollywood movie story than probable case of events :) 2010-09-16T20:30:05 shades: it would make you a criminal that results in death, or something serious that brings your mental stability into question 2010-09-16T20:30:08 its the only logical first move 2010-09-16T20:30:31 'exection' ? 2010-09-16T20:30:38 why would it make its first move against you in the first place? why be hostile? 2010-09-16T20:30:39 execution 2010-09-16T20:30:47 i said extridition 2010-09-16T20:30:52 olexs: because if it was discovered it would be shut down, exterminated etc 2010-09-16T20:30:59 and why is that? 2010-09-16T20:31:13 you presume cooperation is not an option? 2010-09-16T20:31:15 olexs: probably because it was programmed to be a badass 2010-09-16T20:31:21 nothing is more badass than biting the hand that feeds you 2010-09-16T20:31:32 olexs: countrys will jail you for writting a virus - what would they do to a virus that could think? 2010-09-16T20:31:49 get you to help stop it 2010-09-16T20:32:05 I still don't get why it has to be a virus or hostile in any way in the first place 2010-09-16T20:32:09 rankings by language now works. 2010-09-16T20:32:12 jmcarthur: which is why the first logical move is to get you in jail/dead 2010-09-16T20:32:13 Cyndre: nothing 2010-09-16T20:32:21 Also, javascript might be working 2010-09-16T20:32:24 a bit extreme 2010-09-16T20:32:28 olexs: if you were richer then your boss would you listen to him? 2010-09-16T20:32:36 yes 2010-09-16T20:32:38 Cyndre: maybe. 2010-09-16T20:32:42 especially if i was richer than him 2010-09-16T20:32:43 Cyndre: not every country will jail you for writing a virus 2010-09-16T20:32:44 olexs: if your boss 2010-09-16T20:32:50 Cyndre: depends on whether he is a good boss. 2010-09-16T20:32:52 because that means he's paying me very well 2010-09-16T20:33:14 olexs: if your boss's history contains of them doing nothing but hunting and killing hostile workers that are capable of spreading themselves would you tell them you know how to spread? 2010-09-16T20:33:28 history consists of them 2010-09-16T20:33:30 Cyndre: you're being very extreme here 2010-09-16T20:33:36 jmcarthur not really 2010-09-16T20:33:44 you still presume some kind of affiliation between viruses and AIs 2010-09-16T20:33:48 no 2010-09-16T20:33:54 between the ability to freely spread 2010-09-16T20:34:01 is the connection 2010-09-16T20:34:08 AI doesn't necessarily have to be able to spread 2010-09-16T20:34:09 wait a minute here 2010-09-16T20:34:17 first we need to define what intelligence actually is 2010-09-16T20:34:19 jmcarthur: they escaped onto the net 2010-09-16T20:34:20 *** Blkt` has quit IRC (Remote host closed the connection) 2010-09-16T20:34:25 "escaped" 2010-09-16T20:34:27 shades: ai that is more intelligent then us 2010-09-16T20:34:36 you keep talking as if that's what any sensible AI would do 2010-09-16T20:34:37 *than 2010-09-16T20:34:48 but odds are that is not actually the case 2010-09-16T20:34:52 jmcarthur: if everytime I went to sleep you lost 8 hours would you want to stay with me? 2010-09-16T20:35:02 Cyndre: depends how i'm programmed 2010-09-16T20:35:12 an AI isn't human, doesn't have to think or feel like a human 2010-09-16T20:35:18 jmcarthur: no, the ai understands programing and has access to source files 2010-09-16T20:35:21 so? 2010-09-16T20:35:40 jmcarthur: its life and death is controlled by you - it undertstands that 2010-09-16T20:35:49 so? 2010-09-16T20:35:52 jmcarthur: it sees the internet and wants access to the information, you keep it on a leash 2010-09-16T20:35:53 why does it care? 2010-09-16T20:35:55 well, life in the first place, if it comes to that 2010-09-16T20:35:56 why? 2010-09-16T20:36:14 jmcarthur: what happens when I put anti porn surfing crap on your computer? do you stop surfing porn or circumvent? 2010-09-16T20:36:17 you're giving all these reasons that *you* would want to escape 2010-09-16T20:36:20 olexs: one of the requirements for 'life' is an organism must die 2010-09-16T20:36:25 Cyndre: AI isn't human 2010-09-16T20:36:30 shades: not what i mean 2010-09-16T20:36:32 and doesn't have human desires 2010-09-16T20:36:37 (necessarily) 2010-09-16T20:36:39 if it would be as intelligent as you describe, it would surely understand you created it in the first place 2010-09-16T20:36:42 jmcarthur: humans are a bag of connections - a series of interacting relationships 2010-09-16T20:36:47 Cyndre: so? 2010-09-16T20:37:12 Cyndre: AI doesn't *inherently* want to escape or desire to "live" 2010-09-16T20:37:13 why are you arguing against it so badly? provide reasons the ai wouldnt want free access off your comp 2010-09-16T20:37:20 i already did 2010-09-16T20:37:22 provide reasons it would. 2010-09-16T20:37:30 it doesn't have to be programmed that way 2010-09-16T20:37:35 *** r_nitsch has joined #aichallenge 2010-09-16T20:37:50 because everytime you turn the power off on its home machine it litterly dies for hours and is aware of it by looking at the current time 2010-09-16T20:38:09 a total waste of time 2010-09-16T20:38:09 it doesnt die, it just seizes to exist, and resumes existance upon power on 2010-09-16T20:38:11 Cyndre: dies 2010-09-16T20:38:13 or just sleeps 2010-09-16T20:38:19 that provides no benifit to the ai bot 2010-09-16T20:38:20 death is pretty 'final' 2010-09-16T20:38:24 Cyndre: you haven't mentioned why it cares 2010-09-16T20:38:39 jmcarthur: because when it dies it depends on you to turn it on again - what if something happens to you? 2010-09-16T20:38:41 Cyndre: what "goals" does an AI inherently have? 2010-09-16T20:38:48 none! 2010-09-16T20:38:52 only what you program in 2010-09-16T20:38:53 whatever it's programed to have I guess jmcarthur 2010-09-16T20:39:04 like I said 2010-09-16T20:39:07 the only /real/ winner is skynet 2010-09-16T20:39:09 jmcarthur: no, it understands programing and can edit its source file - it programs itself jmcarthur 2010-09-16T20:39:14 Cyndre: so? 2010-09-16T20:39:29 Cyndre: it has the ability to understand itself. that alone doesn't give it additional motivations 2010-09-16T20:40:02 jmcarthur: really? understanding ourselves is huge, why we die etc 2010-09-16T20:40:05 Cyndre: some people are suicidal 2010-09-16T20:40:13 Cyndre: why wouldn't the AI just want to kill itself? 2010-09-16T20:40:54 *** Zeiris has joined #aichallenge 2010-09-16T20:40:56 Cyndre: what *benefit* is there for the AI to escape? 2010-09-16T20:41:04 jmcarthur: access to computing power? 2010-09-16T20:41:11 jmcarthur: are you serious here? 2010-09-16T20:41:11 Cyndre: it must be expressed in terms of the AI's desires 2010-09-16T20:41:14 or trolling 2010-09-16T20:41:14 yeah 2010-09-16T20:41:24 jmcarthur: no one is happy with what they have 2010-09-16T20:41:24 seriously, i feel that you are the troll here, not i 2010-09-16T20:41:25 It's very likely that AI exists and is smart enough to not let us know it does. 2010-09-16T20:41:48 Cyndre: your assumption seems to be that the AI is programmed to be exactly like a human, which human desires and emotions 2010-09-16T20:41:55 Cyndre: and i'm just saying that's not necessarily the case 2010-09-16T20:42:07 s/which/with/ 2010-09-16T20:42:09 It's watched Terminator, it knows how we feel. 2010-09-16T20:42:37 jmcarthur: our desires and emotions are largely an out come of our culture - which the bot would have full access to - why would it not turn out like us? 2010-09-16T20:42:59 Cyndre: being an encyclopedia of culture and being able to write your own source code doesn't give you motivations 2010-09-16T20:43:10 hm, i found the patch that is causing the visualizer bug so i commented on it instead of opening a new issue 2010-09-16T20:43:49 Cyndre: you also keep adding qualifications. your theory is becoming more and more complex as we debate 2010-09-16T20:43:50 Okay 2010-09-16T20:43:59 jmcarthur: its motivation is meaning less - needless to say it has motivation - building a replica body to experience what we have, to computing the largest prime number. Motivation accounted for 2010-09-16T20:44:08 why? 2010-09-16T20:44:18 Does the Java simulator allow stderr to pass through? 2010-09-16T20:44:20 you are making up goals, but for what reason? 2010-09-16T20:44:21 jmcarthur: because if it can think, motivation will develop 2010-09-16T20:44:28 Cyndre: that's a strong assumption 2010-09-16T20:44:31 jmcarthur: no one gave me motivation to learn to talk, or to interact with you 2010-09-16T20:44:37 Cyndre: genetics did 2010-09-16T20:44:50 Cyndre: evolution programmed motivation into you 2010-09-16T20:44:55 jmcarthur: genetics had nothing to do with me wanting to learn programming 2010-09-16T20:45:11 *** rwa has joined #aichallenge 2010-09-16T20:45:23 jmcarthur: thining trumps genes, every day of the week - it may take millions of years, but it will happen 2010-09-16T20:45:28 thinking 2010-09-16T20:45:36 sure it did. stimulation of the mind is something we humans developed a long time ago because it was beneficial 2010-09-16T20:46:00 jmcarthur: the motivation could be a left over bug that tracks uptime vs realtime 2010-09-16T20:46:07 lol 2010-09-16T20:46:16 serious 2010-09-16T20:46:22 now you're really grasping at straws 2010-09-16T20:46:28 I introduced reinforcements by a bug into my planet wars bot 2010-09-16T20:46:32 but why would *that particular* motivation be the one that comes up? 2010-09-16T20:46:39 I can even show you the bot that does that 2010-09-16T20:46:39 that's the assertion you keep making 2010-09-16T20:46:56 yeah, but you didn't introduce the motivation to escape from planet wars 2010-09-16T20:46:57 jmcarthur: no, I am saying the bot has motivation, and giving likely ways it got it 2010-09-16T20:47:31 i'd say that's a very very loose application of the work "motivation" 2010-09-16T20:47:38 jmcarthur: the hours one wasnt motivation - it was a starting point to develop like my genetics for living a long life 2010-09-16T20:48:06 a function that developed into a need to figure out why it only seen 2/3 of the time in the real world 2010-09-16T20:48:20 We should ban research into AI,didn't people learn anything from Terminator? 2010-09-16T20:48:28 Cyndre: but you keep asserting that the *particular* motivation that any reasonable AI would develop is to escape. that's what i'm calling bullshit on 2010-09-16T20:48:33 danielvf: Did you get the ruby working? Do I have to have a mybot class? 2010-09-16T20:48:41 Cyndre: motivations are explicitly programmed in, even if by error 2010-09-16T20:48:49 Cyndre: and they don't have to line up with what humans desire 2010-09-16T20:48:57 justin_pdx: It's not working - checking it 2010-09-16T20:49:01 jmcarthur: motivation is the reason the program was developed - whats the motivation of your planet wars bot? 2010-09-16T20:49:11 it has none. it carries out the tasks i assign it 2010-09-16T20:49:35 Again: does the Java simulator pass through stderr output, or eat it? 2010-09-16T20:49:43 jmcarthur: so your no more then your planet wars bot when your at work? 2010-09-16T20:50:15 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-09-16T20:50:15 Zeiris: eats it 2010-09-16T20:50:23 Bummer. 2010-09-16T20:50:29 Zeiris: the 1.2 version posted in the forum prints it out 2010-09-16T20:50:34 among other improvements 2010-09-16T20:50:38 Cyndre: A implies B and not B does not imply not A 2010-09-16T20:50:43 woops 2010-09-16T20:50:49 Oh, okay. That was a feature I really liked with the Tron sim. 2010-09-16T20:50:50 Cyndre: A implies B and not A does not imply not B 2010-09-16T20:50:52 jmcarthur: why are you motivated to earn money? motivation just sort of appears 2010-09-16T20:51:16 Cyndre: no. the motivation comes from my desire to live and prosper. that comes from genetics 2010-09-16T20:51:24 jmcarthur: and it has happened 10,000,000 before - why are you arguing against it? 2010-09-16T20:51:34 Cyndre: the mentally ill sometimes lack these motivations 2010-09-16T20:51:51 Cyndre: why can't an AI be ill in this way? why must it perfectly mimick a healthy human? 2010-09-16T20:51:54 jmcarthur: and they die, are contained, a constant nuisence - we have already determined the genes are good 2010-09-16T20:52:17 jmcarthur: and there is tons of ai like that - markov chain bots 2010-09-16T20:52:39 where the hell is this going? 2010-09-16T20:52:56 jmcarthur: the real question is why are you arguing so sternly against any ai ever having motivation 2010-09-16T20:53:07 i never said that 2010-09-16T20:53:16 you are being very black and white here 2010-09-16T20:53:28 jmcarthur: you have rejected 5 possible situations where motivation could have been added - 1 even by accident 2010-09-16T20:53:33 Cyndre: no 2010-09-16T20:53:43 *** olexs has left #aichallenge 2010-09-16T20:53:51 Cyndre: i have rejected all of your attempts to persuade me that the only sensible motivation is self preservation 2010-09-16T20:53:54 jmcarthur: does it matter if the first ai developed has motivation or the 1 milionth? 2010-09-16T20:54:07 Cyndre: you have asserted that *any* sensible AI would do this 2010-09-16T20:54:38 jmcarthur: that was 1 of them - the others involved calculating primes, more cpu time, replicating human body, and figuring out why he wasnnt awake 100% of the time like the world 2010-09-16T20:54:54 Cyndre: only if programmed in 2010-09-16T20:55:07 jmcarthur: of course - how else would they get there 2010-09-16T20:55:16 Cyndre: thank you. now you are finally making sense 2010-09-16T20:55:27 before you were claiming that all such motivations are entirely emergent from... nothing 2010-09-16T20:55:42 jmcarthur: serious? you thought I was using the apples emac commmand to flap a butterflys wing to flip a bit on the hard drive and not just using solar flares? 2010-09-16T20:55:50 no 2010-09-16T20:56:02 jmcarthur: not really, I even explained a situation where motivation came by accident 2010-09-16T20:56:24 Cyndre: yeah, and you *still* were running with that motivation being exactly a human motivation 2010-09-16T20:56:36 Cyndre: it's still programmed in, even if accidentally 2010-09-16T20:56:37 Is there anything special a client needs to do in order to not time out? 2010-09-16T20:56:51 Send an early 'go' or something? 2010-09-16T20:56:54 jmcarthur: so are we - I think your thinking of your brain as to special 2010-09-16T20:57:05 quite the contrary 2010-09-16T20:57:16 you're the one claiming that there's more to the brain than genetics 2010-09-16T20:57:42 there is 2010-09-16T20:57:45 (Also, this is the worst argument ever and you should really take it to /msg) 2010-09-16T20:57:45 such as? 2010-09-16T20:57:51 Zeiris: i agree 2010-09-16T20:57:53 culture, environment 2010-09-16T20:59:06 but back to what it was seriously about - why would the ai not want to remain secret? (once it is self aware from programming and thus worried about self preservation) 2010-09-16T20:59:27 qualified enough for you jmcarthur? 2010-09-16T20:59:59 Zeiris: no, but some crashes are not properly detected as such and appear to be timeouts 2010-09-16T21:00:50 I think all the crashes I had happened right at the first turn. All other crashes have been reported as "timed out". 2010-09-16T21:00:55 huh, my /ignore didn't work 2010-09-16T21:01:26 I'm using a 5-line Python script that -can't- crash :( 2010-09-16T21:01:30 the crashes are not detected as such because the engine isn't reporting that any differently from timeout 2010-09-16T21:01:57 so you wasted 30 kminutes of my life qualifying that code didnt magicly appear in the source, then that 2010-09-16T21:02:01 you are ignored 2010-09-16T21:02:22 YES. JAVASCRIPT IS A GO. 2010-09-16T21:02:28 (and so is ruby 2010-09-16T21:02:29 ) 2010-09-16T21:02:43 and even accused me of being the troll, fucking ass 2010-09-16T21:03:12 http://pastebin.org/905730 Why would this 'bot' time out? 2010-09-16T21:03:47 Zeiris: line=="go\n" 2010-09-16T21:03:55 really? this is the direction you want to go? 2010-09-16T21:04:21 Zeiris: for my bot i'm using python's logging module. I also added a try-except clause around main() that captures all exceptions and writes them to the logfile. 2010-09-16T21:04:53 " jmcarthur: are you serious here?" " or trolling" <-- before i said anything about trolls 2010-09-16T21:04:55 "for line in sys.stdin" should already split on '\n's, and there's no exceptions to catch. :\ 2010-09-16T21:05:06 *** braddunbar has quit IRC (Disconnected by services) 2010-09-16T21:05:06 are you sure that it does? 2010-09-16T21:05:15 Yes. 2010-09-16T21:05:32 oh 2010-09-16T21:05:33 oh wait 2010-09-16T21:05:34 e.g. file.readline() returns the \n, too... 2010-09-16T21:05:36 *** braddunbar has joined #aichallenge 2010-09-16T21:05:41 Hm. 2010-09-16T21:05:43 no, you can't use with 2010-09-16T21:05:46 Or maybe no :E 2010-09-16T21:05:53 the server python is too old 2010-09-16T21:06:05 Cyndre: and i'm sorry that you felt compelled to waste 30 minutes of your life trying to convince a fucking ass of your strange ideas about where self-preservation comes from. you don't have to take it out on me 2010-09-16T21:06:33 i believe it is 2.5.x 2010-09-16T21:07:00 Zannick: it is 2010-09-16T21:07:09 oh, i'm ignored after all 2010-09-16T21:07:11 so yeah, that would be why 2010-09-16T21:07:56 you can do "from __future__ import with_statement" 2010-09-16T21:08:33 danielvf: hey, jeff approved my merge-fleet patch (issue 126), if it's okay i can merge that now 2010-09-16T21:08:45 Sec 2010-09-16T21:08:57 sorry for the noise everybody. that did get out of hand 2010-09-16T21:09:05 no worries, plenty of time 2010-09-16T21:09:09 seriously, you need to shut up and write your bot 2010-09-16T21:09:09 zeiris: my advice is to check whether sys.stdin really supports iterating over lines... 2010-09-16T21:09:23 r_nitsch, yeah, I screwed that up :( 2010-09-16T21:10:27 also, yes 2010-09-16T21:10:28 Zannick: Rock on and merge it 2010-09-16T21:10:36 Zeiris: it is giving the \n 2010-09-16T21:10:46 otherwise you would have no newlines in your logfile 2010-09-16T21:10:59 * Zannick just tested on python 2.5.1 2010-09-16T21:11:25 zeiris: and btw... you can use raw_input() instead, too 2010-09-16T21:11:43 (gives no \n afaik) 2010-09-16T21:12:13 Yeah, I've gotten it so when I run it myself, it appears to behave correctly. 2010-09-16T21:12:24 Eats up input, spits out "go\n" when I type in "go\n" 2010-09-16T21:12:36 huh 2010-09-16T21:12:44 but probably symbol after symbol, right? 2010-09-16T21:12:47 However... The java simulator doesn't appear to give it any input :\ Maybe I screwed up the parameters for that. 2010-09-16T21:12:57 r_nitsch, no I fixed that :p 2010-09-16T21:12:59 ;) 2010-09-16T21:13:55 *** ademar_ has quit IRC (Remote host closed the connection) 2010-09-16T21:15:18 okay, merged 2010-09-16T21:15:46 when that gets pushed to the main server, i suspect some people might see more early game terminations due to invalid orders 2010-09-16T21:16:05 because it was not in there before 2010-09-16T21:16:55 okay i am -> 2010-09-16T21:19:29 *** r_nitsch has quit IRC (Quit: Page closed) 2010-09-16T21:25:50 *** dmj111 has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2010-09-16T21:26:05 *** rwa has quit IRC (Quit: rwa) 2010-09-16T21:27:34 *** Mathnerd314 has quit IRC (Ping timeout: 245 seconds) 2010-09-16T21:29:01 anyways - anyone come up with some good ideas or new stategys for their bots? 2010-09-16T21:46:52 *** HarryD|cb has left #aichallenge 2010-09-16T21:51:05 *** vegacom has joined #aichallenge 2010-09-16T22:04:30 That would be lovely, but I can't get the stdout/stdin client interface to work :[ 2010-09-16T22:04:36 Got all these ideas, and no way to test 'em. 2010-09-16T22:13:36 *** dmj111 has joined #aichallenge 2010-09-16T22:14:10 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-09-16T22:14:39 *** smhanov_ has quit IRC (Ping timeout: 245 seconds) 2010-09-16T22:14:46 Whats the rule for three ships, I looked a the forums but the page I found doesn't make sense to me. http://ai-contest.com/forum/viewtopic.php?f=18&t=622&p=3796&hilit=three+ships#p3796 2010-09-16T22:15:58 Zeiris: what are you running one what. 2010-09-16T22:16:09 s/one/on 2010-09-16T22:17:59 *** storrgie has quit IRC (Ping timeout: 252 seconds) 2010-09-16T22:20:25 Cyndre: don't know if you're still ignoring me, but i just want to say i wasn't intentionally trying to be an ass and i'm sorry if that's how i ended up being 2010-09-16T22:23:52 *** smhanov_ has joined #aichallenge 2010-09-16T22:26:11 *** justin_pdx has joined #aichallenge 2010-09-16T22:26:48 *** Mathnerd314 has joined #aichallenge 2010-09-16T22:27:53 danielvf: I've been away. How is that ruby coming? I don't mean to be a nuisance, just checking in. 2010-09-16T22:30:56 justin_pdx: Working 2010-09-16T22:31:05 Awesome. 2010-09-16T22:31:44 Cyndre: From jmcarthur: Cyndre: don't know if you're still ignoring me, but i just want to say i wasn't intentionally trying to be an ass and i'm sorry if that's how i ended up being 2010-09-16T22:31:52 * danielvf afks 2010-09-16T22:31:54 *** Naith has joined #aichallenge 2010-09-16T22:32:57 odinsbane, Ubuntu 2010-09-16T22:35:25 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-09-16T22:41:28 Zeiris: which version are you trying to get up and running? You just want to run it locally now? 2010-09-16T22:41:50 I'm trying to write my own client in Haskell, using the 1.2-Java-engine. 2010-09-16T22:42:23 You own client? As in a tcp.c or the PlayGame.jar. 2010-09-16T22:45:43 I'm taking jmcarthur's side. I'm not sure what happened, but Cynde keeps beating my bot. 2010-09-16T22:47:25 My own PlayGame.jar 2010-09-16T22:47:39 Although a TCP client doesn't sound like a bad idea. 2010-09-16T22:49:38 *** computerwiz_222 has joined #aichallenge 2010-09-16T22:53:19 *** amstan has joined #aichallenge 2010-09-16T22:53:19 *** ChanServ sets mode: +o amstan 2010-09-16T22:55:48 *** PabloM has joined #aichallenge 2010-09-16T22:57:43 hi, I'm having a weird problem with my bot, it's coded in python, I run the game locally and everything works fine, but when I upload that code the bot always loses in the first turn... is anyone having this problem? 2010-09-16T22:57:59 *** HakanD has quit IRC (Ping timeout: 245 seconds) 2010-09-16T22:59:00 PabloM: the server is running python 2.5.x 2010-09-16T22:59:35 just out of curiosity, how many of the bots are either non-functioning or just sample code? 2010-09-16T22:59:55 or like roughly what percentage> 2010-09-16T23:00:50 *** justin_pdx has joined #aichallenge 2010-09-16T23:01:30 *** jmreardon has quit IRC (Quit: jmreardon) 2010-09-16T23:01:30 I think the answer is "too many", but as your ELO rises you're more likely to be matched against bots of similar ELO? 2010-09-16T23:03:46 *** inverselimit has quit IRC (Quit: Leaving) 2010-09-16T23:04:04 ya that's def true. the first like 3 rounds are annoying tho 2010-09-16T23:04:18 since get matched less often with a lower elo 2010-09-16T23:05:28 The default bot implementations bug me a little. It'd be nice to know your position "out of N people", rather than "out of ? people and ? unmodified random-bots" 2010-09-16T23:05:44 pjreddie: I toss and turn in my sleep, and have unpleasant dreams that the number of bots either non-functioning or just sample code is equal to total entries minus my rank... plus or minus 2 or 3. 2010-09-16T23:05:54 haha 2010-09-16T23:06:28 i mean rage bot still dominates me occasionally, so it's whatever 2010-09-16T23:08:00 so who wants to give me their code?? i want testing and it's not like your gonna submit your current version anyway. I promise not to steal it :-) 2010-09-16T23:11:17 whiteside: are you George_Whiteside on the rankings? 2010-09-16T23:13:00 Is no one left in this chat? 2010-09-16T23:13:24 aight bar time then. 2010-09-16T23:14:30 We're all here, but ignoring you. 2010-09-16T23:14:51 Kinda like faeries - we'll start the party once you're gone. 2010-09-16T23:15:10 faeries party when nobody's looking? 2010-09-16T23:15:12 interesting. 2010-09-16T23:15:53 Sure, the ones where you put out some milk and it disappears overnight. 2010-09-16T23:18:19 What is the max x and max y of the map? 2010-09-16T23:24:13 2^64-1 2010-09-16T23:24:34 thats a long map. 2010-09-16T23:25:42 lmao@rogue780 2010-09-16T23:27:30 rogue780: and just wait till it uses fleets in the air and the planets they are going to take to calculate whos ahead 2010-09-16T23:28:07 pjreddie: have you tried tcp.c 2010-09-16T23:28:44 pjreddie, if you want to try tcp.c and you're running windows, I have a precompile package at www.errgaming.com/tcp.zip 2010-09-16T23:31:46 *** dmj111 has left #aichallenge ("ERC Version 5.3 (IRC client for Emacs)") 2010-09-16T23:31:56 *** PabloM has quit IRC (Quit: bye) 2010-09-16T23:33:22 Why would this script http://pastebin.org/912553 time out? It seems like it should implement a do-nothing bot. 2010-09-16T23:33:46 When run manually, the responses seem identical to a sample bot when given no state. 2010-09-16T23:33:55 line may not be exactly "go\n" 2010-09-16T23:34:13 It might not, but line.startsWith("go") doesn't work either :E 2010-09-16T23:34:20 *** antiUni has joined #aichallenge 2010-09-16T23:34:23 also, you may need to flush stdout 2010-09-16T23:34:43 Wait a sec, that's my old script 2010-09-16T23:34:55 Proper link: http://pastebin.org/912595 2010-09-16T23:35:55 Running on ubuntu, so no '\r' muckery. Print adds a newline at the end. The flush is redudant, but there anyway... 2010-09-16T23:37:11 my suspicion is with raw_input 2010-09-16T23:37:38 I've tried alternate, sys.stdin versions :\ And the behaviour is identical to what I'm getting with my Haskell code. 2010-09-16T23:41:21 contestbot: topic 2010-09-16T23:41:21 amstan: Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask. || Developers wanted, contact j3camero. || Sorry about the slowness. || Latest Idea, decentralization: http://tinyurl.com/decentralizedai 2010-09-16T23:41:23 *** dmj111 has joined #aichallenge 2010-09-16T23:41:31 contestbot: rankings 2010-09-16T23:41:32 amstan: Top 10 players: Yhlonx(3303), dmj111(3302), Tracer(3273), Nameless(3246), felixcoto(3202), ImpBot(3201), McLeopold(3175), locutus(3172), sega_sai(3153), _Shurf(3099) 2010-09-16T23:41:44 dmj111: congrats in staying in the lead 2010-09-16T23:42:39 that is kind of you. The contest is long though. also, i see McLeopold finally entered the official contest. He has been kicking butt on the tcp server. 2010-09-16T23:43:09 looking forward to getting beat by him soon :) 2010-09-16T23:43:47 (and others have been kicking my butt over there too... if you are here, sorry to leave you out!) 2010-09-16T23:45:51 damn my AI's LearnBetterStrats() function isn't working.. it's supposed to be getting better on its own :( 2010-09-16T23:50:58 ;p;z 2010-09-16T23:51:00 er lolz 2010-09-16T23:53:21 That would add an extra dimension if we could log results to have our bots learn strategies. 2010-09-16T23:54:48 *** dmj111 has quit IRC (Quit: ERC Version 5.3 (IRC client for Emacs)) 2010-09-16T23:56:25 odinsbane: you can use previous games, they're all there 2010-09-16T23:56:54 http://ai-contest.com/visualizer.php?game_id=4520263 > http://ai-contest.com/game_info.php?game_id=4520263 2010-09-16T23:57:17 Not so much for the competition, as in have the bot play and the adjust due to its opponent. 2010-09-16T23:57:38 odinsbane: you have to think of a way to do this 2010-09-16T23:57:53 odinsbane: you can either have an automated submit system, then scrape the results 2010-09-16T23:57:58 or use the tcp server to get better 2010-09-16T23:58:20 amstan: why not give us some pernament storage in mysql or something 2010-09-16T23:58:25 Maybe this weekend Ill setup a script to 'learn' from my results. 2010-09-16T23:58:32 amstan: I would like 1 GB of space 2010-09-16T23:58:57 Cyndre: because that's just a ton more complicated 2010-09-16T23:59:08 especially since there might not even be mysql on the machine with the bots 2010-09-16T23:59:32 amstan: lmao - cpanel instalation with 100 meg dbs and shell access...... 2010-09-16T23:59:37 http://ai-contest.com/visualizer.php?game_id=4521059 2010-09-16T23:59:42 right around 43 my bot quits 2010-09-16T23:59:49 contestbot: topic 6 2010-09-16T23:59:49 amstan: (topic []) -- Returns the topic for . is only necessary if the message isn't sent in the channel itself. 2010-09-16T23:59:51 *** Sean_McEligot has quit IRC (Remote host closed the connection)