2010-11-14T00:01:27 *** phreeza has joined #aichallenge 2010-11-14T00:01:44 phreeza: how's your bot doing with the new maps 2010-11-14T00:02:40 janzert: any idea how much crashes have spiked? 2010-11-14T00:03:22 not anything hard no, but errors per minute seems to be relatively stable 2010-11-14T00:03:26 http://ai-contest.com/game_stats.php 2010-11-14T00:03:36 actually right now it's really really low 2010-11-14T00:03:53 that was my biggest worry 2010-11-14T00:03:55 but it always varies around quite a bit 2010-11-14T00:03:56 actually breaking lots of bots 2010-11-14T00:03:59 new maps are online? 2010-11-14T00:04:00 yep 2010-11-14T00:04:08 phreeza: yep 2010-11-14T00:04:14 cool 2010-11-14T00:04:20 i've found a couple of examples where the bot seems to be doing worse 2010-11-14T00:04:23 let me see my bot 2010-11-14T00:04:33 but overall win/loss ratios seem about the same 2010-11-14T00:04:36 urgh i'm actually totally drunk 2010-11-14T00:04:37 phreeza: link? :P 2010-11-14T00:05:14 janzert: those errors are compilation and not crashes yeah? 2010-11-14T00:05:21 http://ai-contest.com/profile.php?user_id=3891 2010-11-14T00:05:44 those are any errors in a game 2010-11-14T00:05:49 not compilation at all 2010-11-14T00:06:10 ah okay, cool 2010-11-14T00:06:15 phreeza: your bot seems to like them 2010-11-14T00:06:16 to be clear I'm talking about the errors per minute listed under the workers 2010-11-14T00:07:01 at what time were new maps introduced, server time? 2010-11-14T00:07:45 i wouldn't mind knowing that if possible actually 2010-11-14T00:08:14 18:04:37 2010-11-14T00:09:38 *** krokkrok has joined #aichallenge 2010-11-14T00:11:42 thx 2010-11-14T00:12:04 i think there might be a slight bias towards placing planets further from the centre of the circle 2010-11-14T00:12:17 because points selected closer to the centre are more likely to be rejected 2010-11-14T00:12:20 but meh 2010-11-14T00:13:29 oldman i'm worried about 2010-11-14T00:14:57 tcp is still using the old maps? 2010-11-14T00:15:24 yeah, no one has seen dhartmei 2010-11-14T00:15:25 ie dhartmei is MIA? 2010-11-14T00:15:30 deepblue is crashing too 2010-11-14T00:15:30 ok 2010-11-14T00:15:51 i'm off to sleep 2010-11-14T00:15:58 it's just a max planet problem, that's alright 2010-11-14T00:15:58 see yall in 12h 2010-11-14T00:16:02 night 2010-11-14T00:16:06 *** delt0r has joined #aichallenge 2010-11-14T00:16:46 *** delt0r___ has quit IRC (Ping timeout: 265 seconds) 2010-11-14T00:16:49 *** phreeza has quit IRC (Quit: kthxbai) 2010-11-14T00:17:02 Top 10 players: bocsimacko(4315), bix0r4ever(4144), Raschi(4076), GreenTea(4052), dmj111(4029), DarthMasta(4020), george(4006), Accoun(3966), asavis(3954), shangas(3941) 2010-11-14T00:17:16 the top 100 bots, do the battles normally go fo rhte full 200rounds, or mostly end fast? 2010-11-14T00:17:48 burny: depends on the bots and the map 2010-11-14T00:17:57 it varies quite a lot 2010-11-14T00:18:08 i'm hoping there wont be many draws anymore but we'll see 2010-11-14T00:19:52 dimkakimon's ranking is certainly going up after the change 2010-11-14T00:20:10 no way i'm letting him take first place for australia over me 2010-11-14T00:20:18 lol 2010-11-14T00:22:04 bix0r4ever got one over bocsimacko 2010-11-14T00:22:35 hehe. that bot is prescentient 2010-11-14T00:24:28 i just went back and read Accoun's complaint about 'chiting' and there is a way to avoid this accusation in the future 2010-11-14T00:25:04 the problem is, there is no way for anyone to prove that cheating has not taken place, because the ELO system is so random 2010-11-14T00:25:18 haha 2010-11-14T00:25:42 gah, at least please talk about the right thing the ELO system isn't random at all 2010-11-14T00:25:45 i think with a live leaderboard, if people cared, they could screen shot all the games they've had 2010-11-14T00:25:50 the tournament pairing used is 2010-11-14T00:25:57 then they can't bitch 2010-11-14T00:26:01 doesnt matter. 2010-11-14T00:26:07 because we can see games that weren't run 2010-11-14T00:26:13 i mean, that were run, but not listed 2010-11-14T00:26:23 you can see them? 2010-11-14T00:26:27 what? 2010-11-14T00:26:41 the outcome of these tournaments is not objecitve, therefore, cheating is always possible 2010-11-14T00:27:29 if, as accoun asserts, the judges were cheating, then, they can leave out games in which bots win, and so make it appear another bot is the leader. the process is not open and objective. I know you are not cheating, but you can't prove it 2010-11-14T00:27:35 I think you're looking for deterministic 2010-11-14T00:27:51 if you don't trust the organisers not to cheat, you shouldn't compete 2010-11-14T00:27:59 organisers could always cheat if they really wanted 2010-11-14T00:28:05 NOT SO 2010-11-14T00:28:07 they aren't getting any money for this 2010-11-14T00:28:20 it is possible to rate some contests where the organizers cannot cheat 2010-11-14T00:28:27 how? 2010-11-14T00:28:35 as you said, games might not be shown that were played 2010-11-14T00:28:48 if organizers want to cheat it's easier to cheat by throwing the results of games than removing games I think 2010-11-14T00:28:49 well, I have a post in the forums that sort of covers this 2010-11-14T00:28:51 the code for the contest is open source, that is pretty open if you ask me 2010-11-14T00:29:10 please, think about it. if you wanted to cheat in an undetectable way, you could 2010-11-14T00:29:22 of course 2010-11-14T00:29:26 this is why Accoun can accuse you and you cannot really defend yourself 2010-11-14T00:29:41 you have to trust the organizers no matter what 2010-11-14T00:29:47 NOT SO 2010-11-14T00:29:58 last year chiting: http://ai-contest.com/forum/viewtopic.php?f=3&t=366&start=30#p2151 2010-11-14T00:30:13 and this http://openpaste.org/en/19644/ writen actions 2010-11-14T00:30:19 if the contest were originally setup as I say in my forum post, transitive and objective, then there could be no undetectable cheating 2010-11-14T00:30:29 Accoun: I'm pretty sure the rankings weren't reset when the final tournament started, and that really could explain discrepencies in game numbers 2010-11-14T00:30:36 and you would expect old shit bots to have the most games 2010-11-14T00:31:28 i really do suggest you guys think this through before the next contest. Completely open and fair contests, that have clear objective and calculable winner can be created. 2010-11-14T00:32:07 i don't want the problems to be too specific, this is an ai contest, and you're never going to get a perfect ordering, it's just not going to happen 2010-11-14T00:32:19 I can. why cant you? 2010-11-14T00:32:42 because of this reason only: you have not thought about it enough. 2010-11-14T00:32:51 i am not an organiser 2010-11-14T00:33:24 well, you just said "you're never going to", im not sure to whom you were referring 2010-11-14T00:34:12 it really bugs me when you say "its just not going to happen", this sort of thinking is very frustrating 2010-11-14T00:37:10 i mean that from a it can't happen 2010-11-14T00:37:23 i have no authority to say what will or will not happen for these contests 2010-11-14T00:38:01 i mean that it can't actually happen* that first line made no sense aha 2010-11-14T00:38:42 Accoun: lol.. janzert is canadian? 2010-11-14T00:38:54 Accoun: never knew that, i guess we all learn new things every day 2010-11-14T00:45:41 yeah, I never knew that before either ;) 2010-11-14T00:47:56 *** Rubicon-|-Cross has joined #aichallenge 2010-11-14T00:49:38 janzert: where are you from btw? 2010-11-14T00:49:51 Michigan 2010-11-14T00:50:02 cool 2010-11-14T00:50:56 btw, I'll be leaving town on business tomorrow so probably won't be around till Wednesday :( 2010-11-14T00:51:02 np 2010-11-14T00:51:15 oh 2010-11-14T00:51:17 everything seems to be stable at the moment at least 2010-11-14T00:51:25 any news on clojure? the guy is pretty insistent 2010-11-14T00:51:37 no not really 2010-11-14T00:51:45 can i tell him something? 2010-11-14T00:51:46 like 2010-11-14T00:51:55 will you be able to implement it by the deadline? 2010-11-14T00:52:22 I really don't know 2010-11-14T00:52:49 it would increase the chances if he can add information to the issue here http://code.google.com/p/ai-contest/issues/detail?id=66&q=clojure#makechanges 2010-11-14T00:53:04 on how to install it on ubuntu 8.04 2010-11-14T00:54:14 also being that only 3 people have ever starred that issue I'm not real enthused about it 2010-11-14T00:54:32 so have him get people that are going to use clojure to start it too :) 2010-11-14T00:54:38 http://ai-contest.com/forum/viewtopic.php?f=18&t=411&start=20#p7105 2010-11-14T00:54:50 there's that depressing forum post about it 2010-11-14T00:56:09 really I wish j3camero would come back and add languages since he seems to really want those 2010-11-14T00:56:26 I want to work on rankings and am running out of time for that :/ 2010-11-14T01:14:49 *** yasith_ has quit IRC (Ping timeout: 250 seconds) 2010-11-14T01:14:58 *** yasith has quit IRC (Ping timeout: 265 seconds) 2010-11-14T01:17:04 Top 10 players: bocsimacko(4298), bix0r4ever(4173), GreenTea(4053), dmj111(4030), Raschi(4030), Accoun(4028), george(4009), DarthMasta(3995), asavis(3948), shangas(3932) 2010-11-14T01:18:54 old man is having the most trouble as far as I have seen 2010-11-14T01:19:41 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-11-14T01:20:46 Accoun: I think your bot is general enough that it's doing quite well on the new maps anyway ;) 2010-11-14T01:27:25 *** yasith_ has joined #aichallenge 2010-11-14T01:27:27 *** yasith has joined #aichallenge 2010-11-14T01:31:38 *** krokkrok has quit IRC (Quit: Page closed) 2010-11-14T01:35:02 a1k0n: any thoughts on the new maps? :P 2010-11-14T01:35:26 how did you know i was here? 2010-11-14T01:35:39 i haven't really been keeping up 2010-11-14T01:35:56 it says you're online on the forums :P 2010-11-14T01:36:01 oh haha 2010-11-14T01:36:17 have you looked at games on the official site at all today? 2010-11-14T01:36:46 nope 2010-11-14T01:37:02 line symmetric maps have been added 2010-11-14T01:37:11 and up to 5 neutral planets starting at an equal distance 2010-11-14T01:37:18 and number of planets ranges from 15 to 30 2010-11-14T01:37:24 huh 2010-11-14T01:37:31 wow 2010-11-14T01:37:42 ooh and the player is updated 2010-11-14T01:38:05 oh and you can wipe the slider at the bottom, rad 2010-11-14T01:38:38 yes, but the new maps are the "cool" part 2010-11-14T01:38:48 i think they're cool, some don't like the change 2010-11-14T01:39:07 oh wow, found one 2010-11-14T01:40:20 like? 2010-11-14T01:40:39 a line symmetric one. yeah, neat. i dunno. i don't really have an opinion. 2010-11-14T01:41:19 30 planets is.. ok. starting to be painful. 2010-11-14T01:41:34 people didn't want to go above 32 because of people using bitmaps 2010-11-14T01:41:45 line symmetric maps are what i care about 2010-11-14T01:41:49 it's way more general 2010-11-14T01:41:53 yeah 2010-11-14T01:41:58 tron was line symmetric.. eventually 2010-11-14T01:42:04 i thought they did both? 2010-11-14T01:42:12 well, yeah. it was both. 2010-11-14T01:48:10 we can also discusss the maps themselves but the decision to stir the pot this late is more fitting for sports with paying viewership 2010-11-14T01:48:29 haha. well, yes. 2010-11-14T01:48:43 indeed. symmetric maps aren't as big a deal as increasing the number of planets 2010-11-14T01:49:21 or decreasing 2010-11-14T01:49:42 i was considering hardcoding 23 2010-11-14T01:49:45 i had always expected such maps 2010-11-14T01:49:51 it was always expected they'd change for tron 2010-11-14T01:49:54 but yeah i figured it was coming 2010-11-14T01:50:06 right, they announced immediately the final tournament would be on bigger maps for tron 2010-11-14T01:50:14 another big change is how many planets are usually the same distance from both player 2010-11-14T01:50:14 and then didn't even do it 2010-11-14T01:50:18 which i didn't like 2010-11-14T01:50:24 antimatroid: how about radial symmetry on 3 lines? 2010-11-14T01:50:39 McLeopold: considered that, never did it 2010-11-14T01:50:44 could do it on n lines 2010-11-14T01:50:45 the new maps are more rock-paper-scissors I think 2010-11-14T01:50:47 yea 2010-11-14T01:50:54 mega1: that is a good thing 2010-11-14T01:51:02 mega1: i think that'd be middle planets 2010-11-14T01:51:09 i would have thought such situations are better 2010-11-14T01:51:20 perhaps not when you don't repeat the same game lots 2010-11-14T01:51:44 i've always wanted the most general maps possible, see whose ai really is best 2010-11-14T01:52:02 forget about AI 2010-11-14T01:52:08 it's only in the name 2010-11-14T01:52:25 depends what one means when they say ai :P 2010-11-14T01:52:31 nobody has true AI and there is no universally superior solution 2010-11-14T01:52:32 *** hellman has joined #aichallenge 2010-11-14T01:52:34 it's a very vague term 2010-11-14T01:53:11 so far the biggest worry i see is oldman, he isn't handling the new maps at all 2010-11-14T01:53:20 this is a contest to see who can distill their intelligence into an automated program that applies that logic very fast to make moves 2010-11-14T01:53:24 deepblue is crashing with more than 23 planets, but he should be able to fix that easily 2010-11-14T01:53:58 (his rank is now in the 100's 2010-11-14T01:57:50 I see no activity on zerviz 2010-11-14T01:58:01 is there a tcp server with the new maps? 2010-11-14T01:58:19 i think mcleopold did one 2010-11-14T01:58:29 yep 2010-11-14T01:58:35 port 995 now 2010-11-14T01:58:43 I'm still working on canvas 2010-11-14T01:58:47 but you can play 2010-11-14T01:59:12 there is a big old NaN in the middle of the visualizer 2010-11-14T02:00:21 that's zeroviz, right? I see only two McBots there 2010-11-14T02:00:27 yeah 2010-11-14T02:00:30 that's it 2010-11-14T02:01:27 hmm, there was more rankings before? 2010-11-14T02:01:37 if I can figure out dprintf, I think we are a go 2010-11-14T02:01:43 antimatroid: there were 2010-11-14T02:01:56 I got the latest code from dhartmei's server and reinstalled 2010-11-14T02:02:11 The old one was doing really crazy things to the games 2010-11-14T02:02:24 If a planet hit 0 from an enemy, it would switch sides anyway 2010-11-14T02:02:41 I was losing to ZooRage 2010-11-14T02:03:22 um.. can you try port 995, so I can see if my firewall is correct? 2010-11-14T02:04:49 doesn't seem to load 2010-11-14T02:06:04 no connection? 2010-11-14T02:07:08 "webpage is not available" 2010-11-14T02:07:22 i haven't got a bot to run atm, i'm in the process of rewriting 2010-11-14T02:07:27 http://zeroviz.us:8080/ 2010-11-14T02:07:33 mega1? 2010-11-14T02:07:33 that loads 2010-11-14T02:07:39 don't you want 995 to load? 2010-11-14T02:07:49 995 is the tcp port 2010-11-14T02:07:52 for bots 2010-11-14T02:07:53 ah, sorry :P 2010-11-14T02:08:01 humans go to 8080 2010-11-14T02:08:12 mega1 isn't a human :P 2010-11-14T02:08:16 or, if you can type fast, 995 in a tty session 2010-11-14T02:08:36 seems to work 2010-11-14T02:08:42 it's playing 2010-11-14T02:08:49 just type "1 2 100" as fast as you can, the type "go" 2010-11-14T02:09:05 goo 2010-11-14T02:09:06 d 2010-11-14T02:09:32 The server's looking better. It's reporting the elo, unlike before. 2010-11-14T02:10:22 there was 2 turns? 2010-11-14T02:10:34 no 2010-11-14T02:10:39 canvas.c doesn't work 2010-11-14T02:10:54 I tried the old one, but it's not reading the files correctly. 2010-11-14T02:11:01 hmm okay 2010-11-14T02:11:13 you can see the data string messed up if you view source 2010-11-14T02:11:35 the funny thing is, the new one runs fine on the command line 2010-11-14T02:11:41 just not off the thttpd server 2010-11-14T02:11:47 @rankings 2010-11-14T02:11:48 amstan: Top 10 players: bocsimacko(4290), bix0r4ever(4141), Raschi(4118), Accoun(4090), dmj111(4034), george(4005), GreenTea(4000), DarthMasta(3958), _Astek_(3944), asavis(3944) 2010-11-14T02:12:12 mega1: you doing tcp? 2010-11-14T02:12:17 amstan: oldman i'm worried about 2010-11-14T02:12:27 antimatroid: ? 2010-11-14T02:12:31 amstan: just tried a single game 2010-11-14T02:12:48 http://ai-contest.com/profile.php?user_id=7526 2010-11-14T02:12:51 the old tcp server doesn't have the maps 2010-11-14T02:13:01 he's down to 241 now 2010-11-14T02:13:13 antimatroid: what was he before? 2010-11-14T02:13:14 he's resubmitted a couple of times, and doesn't seem to be handling the maps well 2010-11-14T02:13:17 top 30 i think 2010-11-14T02:13:27 i expect they're having a very bad time atm 2010-11-14T02:13:49 deepblue is also crashing, but i'm pretty sure that's just a no planets issue 2010-11-14T02:15:07 other than that i've seen no major differences so far 2010-11-14T02:15:19 but there probably are other examples 2010-11-14T02:15:31 wait.. so.. maps are already in there? 2010-11-14T02:15:34 yes 2010-11-14T02:15:49 janzert did it earlier 2010-11-14T02:16:27 mega1: your bots is doing well then on the last 2 games 2010-11-14T02:16:34 although, a lot of his losses are from him not expanding anywhere 2010-11-14T02:16:43 amstan: quite a few of the games are on new maps 2010-11-14T02:16:55 not for mega1 2010-11-14T02:16:58 all the games are on new maps 2010-11-14T02:17:04 Top 10 players: bocsimacko(4290), bix0r4ever(4141), Raschi(4118), Accoun(4090), dmj111(4034), george(4005), GreenTea(4000), DarthMasta(3958), _Astek_(3944), asavis(3944) 2010-11-14T02:17:14 there are older games still up yeah? 2010-11-14T02:17:18 since shortly after 18:00 2010-11-14T02:17:21 yeah, but he only had one game on the new maps 2010-11-14T02:17:36 amstan: no 2010-11-14T02:17:40 yes, the old games are still there 2010-11-14T02:17:43 http://ai-contest.com/visualizer.php?game_id=7246869 2010-11-14T02:17:44 that's a new map 2010-11-14T02:17:45 yes, I'm losing more but it's early still 2010-11-14T02:17:59 one of the losses was a point symmetric map 2010-11-14T02:18:00 the switchover was 18:04 or whatever I said before 2010-11-14T02:18:07 actually, two were 2010-11-14T02:18:08 oh! 2010-11-14T02:18:09 k 2010-11-14T02:18:27 although they're not exactly the same as the old generator either 2010-11-14T02:18:33 some are 2010-11-14T02:34:31 *** virdo has quit IRC (Ping timeout: 265 seconds) 2010-11-14T02:35:08 *** raedwulf has quit IRC (Ping timeout: 252 seconds) 2010-11-14T02:36:23 *** raedwulf has joined #aichallenge 2010-11-14T02:38:24 *** k has joined #aichallenge 2010-11-14T02:38:50 *** k is now known as Guest85926 2010-11-14T02:39:24 got it! 2010-11-14T02:39:33 new maps tcp is up and running 2010-11-14T02:40:38 watch mega1 kick my butt here: http://zeroviz.us:8080/canvas?game_id=4 2010-11-14T02:42:29 McLeopld: awesome 2010-11-14T02:42:36 i'll try to get a bot up and testing by the end of the night 2010-11-14T02:42:56 *** tapwater has quit IRC (Quit: tapwater) 2010-11-14T02:42:59 *** Mekanik has joined #aichallenge 2010-11-14T02:47:05 someone tell that site to update their visualizer 2010-11-14T02:47:19 amstan: ? 2010-11-14T02:47:32 there's a cooler version out there 2010-11-14T02:47:33 talk to McLeopold 2010-11-14T02:47:55 I had trouble with it 2010-11-14T02:47:56 McLeopold: ^^^ 2010-11-14T02:48:13 I can try again 2010-11-14T02:48:47 i don't think it's that important 2010-11-14T02:51:37 Accoun is up to second 2010-11-14T02:54:40 amstan: done 2010-11-14T02:54:55 yeah, that is better, awesome 2010-11-14T02:55:26 interesting bug... 2010-11-14T02:55:28 http://zeroviz.us:8080/canvas?game_id=4 2010-11-14T02:55:29 so.. 2010-11-14T02:55:32 wait till the game ends 2010-11-14T02:55:33 and stops 2010-11-14T02:55:38 then hover over a planet 2010-11-14T02:55:45 before hovering it's at 58/58 2010-11-14T02:55:51 after it goes to 59/58 2010-11-14T02:56:03 what browser? 2010-11-14T02:56:09 chrome 2010-11-14T02:56:14 yeah, i have that with chrome 2010-11-14T02:56:42 I don't see it 2010-11-14T02:57:31 i have to let the game fully play 2010-11-14T02:57:32 weird, now I do 2010-11-14T02:57:44 and the map was a different size... 2010-11-14T02:57:46 and once you fix it, you wont get it till you replay it 2010-11-14T02:58:11 yeah, I think my fixes earlier uncovered that 2010-11-14T02:58:12 that's because it was designed to play 1 extra turn 2010-11-14T02:58:22 only when playing 2010-11-14T02:58:32 before it would throw an exception before advancing the the number I guess 2010-11-14T02:59:20 *** hellman has quit IRC (Remote host closed the connection) 2010-11-14T02:59:47 janzert: how many maps are there now? 2010-11-14T03:00:07 30 in use 2010-11-14T03:01:15 it's a pitty, new maps for every game is interesting for watching but not for good results 2010-11-14T03:04:42 what do you mean by that? 2010-11-14T03:04:45 *** flag has quit IRC (Quit: Lost terminal) 2010-11-14T03:04:48 "not good for results" 2010-11-14T03:05:40 you want lots of bots to play on the same map 2010-11-14T03:05:51 otherwise it's too much luck with the map you're generated 2010-11-14T03:06:13 ideally i think one wants a set of maps that are quite different, then have lots of games on them 2010-11-14T03:06:17 if that makes sense 2010-11-14T03:06:50 *** narnach has joined #aichallenge 2010-11-14T03:06:50 *** narnach is now known as narnach_mb 2010-11-14T03:07:03 do you think the new maps are somehow more prone to this issue? 2010-11-14T03:07:04 *** Guest85926 has quit IRC (Ping timeout: 265 seconds) 2010-11-14T03:07:17 to what issue? 2010-11-14T03:07:49 it was more a comment about it being boring watching games on the same maps 2010-11-14T03:08:02 "not good for results" 2010-11-14T03:08:11 and "interesting" 2010-11-14T03:08:29 i think the new maps provide more possible scenarios, making it more interesting 2010-11-14T03:08:48 and i think the ai bots should be ranked about how good they are on general maps 2010-11-14T03:08:52 but how is it "not good for results"? 2010-11-14T03:09:07 it would be bad for results for any generator if every game was on a brand new map 2010-11-14T03:09:48 because bots have different strengths and weaknesses, it's better (imo) to have a small set of unknown maps for an actual contest, but not with bots playing each other twice on the same map 2010-11-14T03:10:02 i don't think anything changes with quality of results between the two generators 2010-11-14T03:10:19 I see 2010-11-14T03:10:28 agree or disagree? 2010-11-14T03:10:56 it'd certainly be more interesting to watch games if it was always a new map 2010-11-14T03:11:47 wtf 2010-11-14T03:11:56 accouns ranking just went from 2nd to 12th without any new games 2010-11-14T03:12:04 raschi resubmitted 2010-11-14T03:12:12 ahh :P 2010-11-14T03:14:32 with a small set of maps you risk not representing the true distribution of maps, but can better ensure that players get to play the same maps 2010-11-14T03:15:20 ... which doesn not necessarily mean much because it's not a one player game. 2010-11-14T03:15:56 That is, you ideally want to have a big, representative sample and every possible pairing on every map. 2010-11-14T03:16:43 yep, but huge round robbins aren't possible, so it's a trade off of having enough variety but not too much 2010-11-14T03:17:04 Top 10 players: bocsimacko(4205), bix0r4ever(4030), dmj111(3939), george(3907), GreenTea(3884), _Astek_(3872), asavis(3861), felixcoto(3852), shangas(3841), DarthMasta(3835) 2010-11-14T03:17:34 If you don't do all pairings on each map then you open the door for A) maps suiting a player B) maps suiting a player against a certain other player 2010-11-14T03:18:10 I _think_ A is stronger than B. 2010-11-14T03:18:13 yes, but if you do all pairings on each map, you don't get to do it for very many maps, so you also have the problem of luck of map generation 2010-11-14T03:18:31 antimatroid: I'm proposing anything, just trying to analyze it. 2010-11-14T03:18:44 yes, i'm just offering my thoughts :) 2010-11-14T03:18:46 ok 2010-11-14T03:19:08 "with a small set of maps you risk not representing the true distribution of maps" 2010-11-14T03:19:14 *** narnach_mb has quit IRC (Ping timeout: 240 seconds) 2010-11-14T03:19:19 yep 2010-11-14T03:19:26 janzert put it more concisely yesterday 2010-11-14T03:19:40 certainly didn't more than one paragraph 2010-11-14T03:19:44 +need 2010-11-14T03:19:55 i often do things incredibly inefficiently 2010-11-14T03:20:02 *** flag has joined #aichallenge 2010-11-14T03:20:40 @seen McLeopold 2010-11-14T03:20:40 flag: McLeopold was last seen in #aichallenge 22 minutes and 17 seconds ago: only when playing 2010-11-14T03:20:54 where? 2010-11-14T03:20:57 *** Accoun has quit IRC () 2010-11-14T03:21:31 flag: what's up? 2010-11-14T03:21:37 Is the port for your server still 9999? 2010-11-14T03:21:41 995 2010-11-14T03:21:50 AntBot standing by... 2010-11-14T03:22:13 Which would be why it wasn't connecting, thanks 2010-11-14T03:22:27 woot, win for me 2010-11-14T03:22:35 flag: have you had any major issues with the new maps? 2010-11-14T03:22:55 Other than the fact that I suck on them, no 2010-11-14T03:23:26 I think mega1 was right, with more center line planets, it's much more r/p/s 2010-11-14T03:23:29 But I guess that's more my fault than anything 2010-11-14T03:24:12 ratings are dropping quite fast 2010-11-14T03:24:17 McLeopold: yes, but is that good or bad? 2010-11-14T03:24:21 i would have said good 2010-11-14T03:24:31 a shakeup is good 2010-11-14T03:24:41 flag: one of your losses was on a point symmetric map 2010-11-14T03:24:42 My bot is more conservative, so I will do worse 2010-11-14T03:24:46 *** Accoun has joined #aichallenge 2010-11-14T03:24:58 i'm not sure about that 2010-11-14T03:25:08 i think one needs to be better at expanding defensively 2010-11-14T03:25:11 if that makes sense 2010-11-14T03:25:17 as your angles get closer 2010-11-14T03:25:21 I think it's just noise in the results. 2010-11-14T03:25:33 considering that my nash bot is not better than on the old maps. 2010-11-14T03:25:58 this about this... 2010-11-14T03:26:23 consider 2 growth 5 planets on the left, and 2 growth 1 on the right, fairly close 2010-11-14T03:26:37 that would be a line symmetrical map 2010-11-14T03:26:52 a point symmetrical would have a growth 5 and 1 on the left and right 2010-11-14T03:27:30 what for? 2010-11-14T03:27:36 so, across the line, there sections that are more valuable 2010-11-14T03:27:55 where as before, if you opponent when left, you naturally when right 2010-11-14T03:28:04 *went* 2010-11-14T03:29:05 this means, I believe quicker matches, because often bots will need to throw resources to the correct side, which is near each other 2010-11-14T03:29:18 has anyone seen a single draw on the new maps with different bots? 2010-11-14T03:29:21 that happens less often when your dancing around a center point 2010-11-14T03:29:26 yes 2010-11-14T03:29:29 testmap9 2010-11-14T03:29:42 but that was because of a close start without good planets 2010-11-14T03:30:27 who's mb08? 2010-11-14T03:30:28 yeah okay, there should be at least one neutral people can take 2010-11-14T03:30:35 that one had 2 2010-11-14T03:30:53 but there was still no good 3 planet 2010-11-14T03:31:20 yeah fair enough 2010-11-14T03:31:21 I think maybe you need to have at least 6 growth in the planets worth taking 2010-11-14T03:31:44 more than the start, so they don't just funnel back to the start 2010-11-14T03:31:59 something needs to draw them away from that position 2010-11-14T03:32:31 with good old map30, it was very close, but still prone to draws 2010-11-14T03:32:48 varying starting growth would spice things up 2010-11-14T03:33:06 atm it's basically never a good idea to forfeit your home planet 2010-11-14T03:33:26 it would be nice to see a map where it was. :) 2010-11-14T03:33:27 not near the start anyway 2010-11-14T03:34:00 but, I don't think that would exist 2010-11-14T03:34:08 Unless you start on a growth 0 2010-11-14T03:34:42 antimatroid, I'm going to try 3 spokes for the symmetry 2010-11-14T03:34:55 at 4 and 5 2010-11-14T03:35:46 :) 2010-11-14T03:36:41 wait I actually won one 2010-11-14T03:36:44 actually, at more than 2, you can create interesting patterns 2010-11-14T03:36:57 You must be mb08? 2010-11-14T03:37:01 :) 2010-11-14T03:37:05 McLeopold: yeah, they'd be like rotating stars etc. 2010-11-14T03:37:30 well, consider 5 points, with one pointing up 2010-11-14T03:37:54 http://i.imgur.com/TfTBN.png that's the best random map i've generated 2010-11-14T03:38:27 or wait, consider 12 spokes, so we can talk about number on a clock 2010-11-14T03:38:42 let's say the starting planets are at 10 & 2 2010-11-14T03:39:04 you can create a set of 2 planets at 9 and 3 that are the same, and maintain symmetry 2010-11-14T03:39:21 then, you can create 3 at 8, 4 and 6 2010-11-14T03:39:38 then, 4 at 10,2,8 and 4 2010-11-14T03:39:52 or 4 at 3,6,9,12 2010-11-14T03:40:14 basically you can layer different radial symmetries onto 1 map 2010-11-14T03:40:27 another option is to parition the circle into n pies, then generate a planet inside the first, and radially place another copy in every othe rpie 2010-11-14T03:41:06 that would give my rotating stars maps 2010-11-14T03:41:10 like star fish 2010-11-14T03:41:14 yeah 2010-11-14T03:41:26 that's kinda the same thing 2010-11-14T03:41:31 i liked the double helix like map 2010-11-14T03:41:41 ? 2010-11-14T03:41:47 it was with the oringal set 2010-11-14T03:41:57 what number? 2010-11-14T03:42:01 hmm no idea 2010-11-14T03:42:24 it was like two circling lines crossing over, (like a double helix :P) 2010-11-14T03:43:04 wow, 2 out of 10, not bad for a bot with a rotten post for a brain 2010-11-14T03:43:16 I gotta let you play flag now 2010-11-14T03:43:27 Mjothbitnir: opinions on the maps? 2010-11-14T03:44:21 Mjothvitnir: connect up again 2010-11-14T03:44:27 flag is waiting 2010-11-14T03:44:39 there you go 2010-11-14T03:44:44 i do feel quite a few of the neutrals where the planets angles are close end up being redundant 2010-11-14T03:44:54 Get ready to win =) 2010-11-14T03:45:07 the minimum distance between angles could possibly be revised 2010-11-14T03:45:07 my only opinion is that the original example maps should have contained a better cross section of possibilities (or in other words I wish they had been around sooner.) 2010-11-14T03:45:35 we gotta get more people 2010-11-14T03:45:48 and JBotManager needs a random pause 2010-11-14T03:45:53 one of the problems with setting very specific details on maps early is that it isn't always clear what kinds of maps will result in the most interesting games 2010-11-14T03:45:55 more people where? 2010-11-14T03:46:06 my tcp 2010-11-14T03:46:08 http://zeroviz.us:8080/ 2010-11-14T03:46:11 port 995 2010-11-14T03:46:21 is it fixed now? 2010-11-14T03:46:26 yeah 2010-11-14T03:46:32 cool, one sec 2010-11-14T03:47:14 possibly, but also with more than one type of map in the starting pool, I think more people would have had a better idea what was meant 2010-11-14T03:48:29 more people playing with different symmetries, overlays etc 2010-11-14T03:49:32 oof, the air is rare on this server 2010-11-14T03:49:47 what does that mean? 2010-11-14T03:50:17 everyone is better than me :/ 2010-11-14T03:50:22 oh 2010-11-14T03:50:34 Do you miss Rage? 2010-11-14T03:50:40 :P 2010-11-14T03:50:55 you must not have played my rotten-post bot yet 2010-11-14T03:52:07 i don't think ragebot could ever beat me 2010-11-14T03:52:30 no mb08 beats me too :/ 2010-11-14T03:52:58 mine plays better in general if I leave out the anti-ragebot code :/ 2010-11-14T03:53:34 that means that other AI's are weak then(or your nati-rage code is faulty) 2010-11-14T03:53:58 burny: often there's a trade off, and it depends whether your enemy expands or not 2010-11-14T03:54:19 if they fully expand and you do reservedly, you might lose, but if you expand fully and they send nothing, they will win 2010-11-14T03:54:20 etc. etc. 2010-11-14T03:54:22 good AI will rage, if rage will be successful 2010-11-14T03:54:37 if they fully expand, then you should rage them 2010-11-14T03:54:52 but you might have sent too much out expanding cautiously 2010-11-14T03:55:16 hence the tradeoff 2010-11-14T03:55:30 if they're fully expanding, and you can't rage.. then you should expand more also 2010-11-14T03:55:35 i don't think any bot could ever ALWAYS win this game, you just need to win the most 2010-11-14T03:55:42 but what if they waited? 2010-11-14T03:55:52 then they didn't expand obviously 2010-11-14T03:56:16 no, consider three options on the first move {do nothing, expand catiously, expand fully} 2010-11-14T03:56:21 all out expand beats reserved expand beats ragebot beats all out expand 2010-11-14T03:56:22 my point is that that can sometimes be rps 2010-11-14T03:56:23 if(opponent.expanded == true) expand more else expand less 2010-11-14T03:56:43 well, not exactly, but be sort of similar 2010-11-14T03:56:48 burny, but you're a turn behind by the time you figure out what they're doing 2010-11-14T03:57:15 I think burny's right 2010-11-14T03:57:34 how? we're talking about what to do on the first move 2010-11-14T03:57:38 you have no information about your opponent 2010-11-14T03:57:39 well, my AI will stomp, nearly 100% of the time, any bot that expands agressivly on first move 2010-11-14T03:57:58 no, he said if the opponent does x, then do y. That's after the first turn 2010-11-14T03:58:09 because I will expand cautiously on first move, and then rage them 2010-11-14T03:58:09 burny: if people expand catiously and you didn't, you wont be able to take them and they will grow faster than you 2010-11-14T03:58:15 depends on map, too, my bot does ok on distant starting maps 2010-11-14T03:58:28 my old bot did what you suggest, but i found it didn't do well against the better bots who could expand catiously 2010-11-14T03:58:47 I read that as "better bots do better"? 2010-11-14T03:59:46 I think what you're saying though.. is that the better bots never(or rarely) go into rage mode/ 2010-11-14T04:00:10 oh hey I can beat mb08 sometimes 2010-11-14T04:00:43 50-50 2010-11-14T04:01:00 no, waht i'm saying is the better bots expand as much as they can, but not so much that they're left vulnerable 2010-11-14T04:01:13 you don't want to expand to a planet you're going to lose 2010-11-14T04:01:13 so, why don't you do that too? 2010-11-14T04:01:25 even if you recoup the ships you spent before you lose it, the other player then gets that growth 2010-11-14T04:01:41 i am in the middle of a clean rewrite that hopefully will 2010-11-14T04:01:59 my old bot was 2 months worth of trying ideas 2010-11-14T04:02:06 i wanted cleaner code with less bugs 2010-11-14T04:02:12 particulary for the final tournament 2010-11-14T04:02:15 against good AI: rage should never win(and only tie certain maps) 2010-11-14T04:02:27 like i said, i shouldn't ever lose to rage bot 2010-11-14T04:02:32 i haven't in like 2 months 2010-11-14T04:02:36 and on nearly 100% of maps, maximum expansion should get demolished 2010-11-14T04:02:50 I find that erasing half my code greatly improves my performance on these new maps 2010-11-14T04:02:59 that isn't something i was saying is a smart idea :P 2010-11-14T04:03:04 *** sibislaw has joined #aichallenge 2010-11-14T04:04:55 servers are stopping? 2010-11-14T04:05:13 hmmm 2010-11-14T04:05:29 *** sibislaw has quit IRC (Client Quit) 2010-11-14T04:05:37 for me the hardest part is going to be the end-game code 2010-11-14T04:06:25 like, on turn 183, it's worth sending 200ships at a 1growth plane with 10ships, even if it can be taken back a few turns later 2010-11-14T04:06:42 simple, if my payoff is negative, i don't make a move 2010-11-14T04:06:48 one of the opponents planets I mean, not neutral 2010-11-14T04:07:49 there's no point sending any fleets that don't arrive before the end 2010-11-14T04:08:02 and there's no point expanding if the return at the end is less than it cost 2010-11-14T04:08:12 otherwise i think it's worth sending 2010-11-14T04:11:12 http://zeroviz.us:8080/canvas?game_id=102 2010-11-14T04:11:17 that was a weird game 2010-11-14T04:13:51 nice opening >.< 2010-11-14T04:13:59 it's not me 2010-11-14T04:13:59 that was nice 2010-11-14T04:15:03 *** Utkarsh has quit IRC (Ping timeout: 265 seconds) 2010-11-14T04:16:10 lavalamp_: cool looking game being played 2010-11-14T04:16:48 http://zeroviz.us:8080/canvas?game_id=118 that one? 2010-11-14T04:17:05 Top 10 players: bocsimacko(4059), bix0r4ever(3898), Raschi(3871), dmj111(3794), george(3749), GreenTea(3734), _Astek_(3722), asavis(3704), shangas(3703), DarthMasta(3697) 2010-11-14T04:17:16 http://zeroviz.us:8080/canvas?game_id=120 2010-11-14T04:17:18 that one 2010-11-14T04:18:20 antimatroid: overall, these new maps are more exciting 2010-11-14T04:18:21 woah, the ratigns have droped a bunch 2010-11-14T04:18:34 much more cutthroat 2010-11-14T04:19:08 oh hey I actually did ok there for a while 2010-11-14T04:19:25 McLeopold: more exciting/interesting was the plan 2010-11-14T04:19:37 I think if you hadn't launched that 400 from the bottom, you would have one 2010-11-14T04:19:41 won 2010-11-14T04:19:46 *** Utkarsh has joined #aichallenge 2010-11-14T04:19:50 i'm still worried about oldman though 2010-11-14T04:19:58 ooh, draw map 2010-11-14T04:20:58 McLeopold good point, I'll have to fix that :/ 2010-11-14T04:21:11 http://zeroviz.us:8080/canvas?game_id=136 2010-11-14T04:21:35 damn, if only that neutral had like 2 ships instead of 6 2010-11-14T04:21:53 i don't think you guys picked the best planet either 2010-11-14T04:21:55 still draw 2010-11-14T04:22:03 no, we did 2010-11-14T04:22:16 say you go for the 10 2010-11-14T04:22:17 well, maybe not 2010-11-14T04:22:29 i think you'd have done better going for the 10 2010-11-14T04:22:31 everything in the center would have been sniped 2010-11-14T04:22:40 not if they make a move too 2010-11-14T04:22:50 oh, yeah probably 2010-11-14T04:22:52 yeah, true 2010-11-14T04:22:57 My bot would go for the 6 and the 10 2010-11-14T04:23:04 and get crushed :( 2010-11-14T04:23:28 hmm..., it's odd that we both picked very poor planets 2010-11-14T04:23:39 I suppost one should just wait forever on that map 2010-11-14T04:23:56 then you lose to what you guys just did 2010-11-14T04:24:05 that maps actually quite interesting now :P 2010-11-14T04:24:30 it is very much a draw map 2010-11-14T04:24:38 i dunno 2010-11-14T04:24:44 i think if you do separate moves it wouldn't be 2010-11-14T04:24:53 and not sure any move is optimal 2010-11-14T04:25:01 okay, consider 2 planets with a neutral right in the middle with 30 ships. 2010-11-14T04:25:04 if one of you had gone for the 47 + 1 instead, then maybe you could take the 6 in the middle 2010-11-14T04:25:07 that is a draw map, no questions 2010-11-14T04:25:24 split the center into 2, still with 30 2010-11-14T04:25:34 again, draw 2010-11-14T04:25:43 because they can be sniped 2010-11-14T04:25:51 not if they move too 2010-11-14T04:25:56 if every planet along the center can be sniped, it is a poor map 2010-11-14T04:26:00 you are relying on them waiting to snipe you 2010-11-14T04:26:14 it's the best strategy for that type 2010-11-14T04:26:16 what's sniped mean? 2010-11-14T04:26:23 offense 2010-11-14T04:26:29 :p 2010-11-14T04:26:37 burny: landing 1 turn later after their ships are absorbed by the neutral ships 2010-11-14T04:26:43 snipe = you take a planet right after your opponet did 2010-11-14T04:26:46 people refer to sniping as taking the planet just after the enemy took it from being a neutral 2010-11-14T04:27:05 i always have the least efficient explanations :( 2010-11-14T04:27:06 shouldn't the middle be snipable on nearly 100% of maps? 2010-11-14T04:27:24 there are cases where you should send everything you have at the middle planet 2010-11-14T04:27:37 otherwise a perfect bot will beat you 2010-11-14T04:27:39 Nope, consider a growth 5 with 4 ships closer to you 2010-11-14T04:27:44 Un snipable 2010-11-14T04:27:59 if it's closer to you, then it's not in the middle? 2010-11-14T04:28:27 still, if it's equal distance, unshipable 2010-11-14T04:28:33 uh, snipable 2010-11-14T04:28:47 ya, that's just a draw map then though 2010-11-14T04:28:51 if they arive 1 turn late, you will have already gained ships 2010-11-14T04:28:59 if hte best strat is to just send your entire army to the middle ever turn 2010-11-14T04:29:24 if there are 2 or more in the center, then it's a guessing game 2010-11-14T04:30:28 except that then, most likely(assuming you don't pick the same one), the best move will just be to attack their middle from yours 2010-11-14T04:31:20 the other thing before was, the centre planet was exactly halfway between teh two bots 2010-11-14T04:31:23 that's no longer the case 2010-11-14T04:31:29 not for line symmetry 2010-11-14T04:31:51 antimatroid: I don't think tri-symmetry will work 2010-11-14T04:31:57 or three slices 2010-11-14T04:32:14 picking home planets? 2010-11-14T04:33:10 one bot may be draw toward the opponent and another away 2010-11-14T04:33:37 huh? 2010-11-14T04:34:37 I'd have to split it into 6 slices, then mirror adjacent slices 2010-11-14T04:34:46 the home planets would need to be on the line 2010-11-14T04:35:17 what is wrong with three? 2010-11-14T04:35:20 ELO's are below 4000 2010-11-14T04:35:44 consider the points of a clock with the follow growths at each number 2010-11-14T04:35:53 1,2,3,4,1,2,3,4,1,2,3,4 2010-11-14T04:35:58 from 1 to 12 2010-11-14T04:36:11 ok 2010-11-14T04:36:33 if 2 of the growth 4's are the home planets, at 1oclick and 5 2010-11-14T04:36:55 http://ai-contest.com/profile_games.php?user_id=10474 accounts like that should be disabled 2010-11-14T04:37:02 wait, 4oclock and 8 oclock 2010-11-14T04:37:45 yes 2010-11-14T04:37:46 janzert: ^^ 2010-11-14T04:38:27 okay so, 4 and 8 o clock starting positions will both naturally want to go counter-clockwise 2010-11-14T04:38:48 8oclock runs into the opponent, 4oclock runs into free neutrals 2010-11-14T04:38:52 it's unbalanced 2010-11-14T04:39:08 but, consider this layout on the clock 2010-11-14T04:39:11 yeah okay 2010-11-14T04:39:28 it was like when i put centre neutrals in point symmetric maps but not doubled on both sides 2010-11-14T04:39:39 technically it was symmetric at the start, but the games weren't fair 2010-11-14T04:39:42 2,1,2,4,2,1,2,4,2,1,2,4 2010-11-14T04:40:32 actually, i guess technically it wasn't 2010-11-14T04:41:02 yeah, it was just repeated 2010-11-14T04:41:17 that's why I need 6 slices with every other one mirrored 2010-11-14T04:41:37 hmm, max distance was always 24 yeah? 2010-11-14T04:41:49 Mistmanov thinks it was 34 2010-11-14T04:42:01 it was higher than 24 2010-11-14T04:42:13 ah crap, that needs changing then 2010-11-14T04:42:22 but, that was corner to corner 2010-11-14T04:42:35 were maps inside a square before? 2010-11-14T04:42:42 yes 2010-11-14T04:42:42 yea 2010-11-14T04:42:50 i don't think it matters, i think the radius should be extended then 2010-11-14T04:42:57 they were 24 squares 2010-11-14T04:43:07 okay, but you gotta check the radius with the angle 2010-11-14T04:43:45 what? 2010-11-14T04:43:51 a 12 rad circle is bound by a 24 square 2010-11-14T04:44:28 (the circle just crops the corners of the square) 2010-11-14T04:44:29 a 24x24 square has a corner to corner of 24*sqrt(2) 2010-11-14T04:44:34 you've got what, maxD = sqrt(2*24^2)? 2010-11-14T04:44:47 yep 2010-11-14T04:45:20 janzert: can you change the maxRadius to 17 in the map generator please 2010-11-14T04:45:32 so it's consistent with old max distances 2010-11-14T04:45:32 33.9411 2010-11-14T04:45:40 :P close enough 2010-11-14T04:46:09 you just have to check your x and y cords after converting to cartiesian 2010-11-14T04:46:34 McLeopold: it's easiest to just change the variable for max radius 2010-11-14T04:47:03 okay, but you might generate stuff outside the box 2010-11-14T04:47:37 how? 2010-11-14T04:47:51 oh yeah, but that's fine 2010-11-14T04:47:52 with 17, the circle will bound the box not the box bound the circle 2010-11-14T04:48:06 you want to capture the maximum distance 2010-11-14T04:48:39 max dist is the same but play area is larger 2010-11-14T04:48:43 btw, 33.94 distance rounded up to 34 anyway, it's fine to just use max radius of 17 2010-11-14T04:48:56 k 2010-11-14T04:50:49 McLeopold: Could you put AntBot back up? I want to see how I do against it 2010-11-14T04:51:10 k 2010-11-14T04:51:14 Thanks 2010-11-14T04:51:40 http://ai-contest.com/profile.php?user_id=7526 2010-11-14T04:51:44 oldman's really struggling 2010-11-14T04:51:54 i have no idea what would cause him to crash :| 2010-11-14T04:51:59 hmm, tri-symmetry would actually just be line symmetry 2010-11-14T04:52:16 planet count is my guess 2010-11-14T04:52:36 i don't think so, he makes it through most of the games before crashing 2010-11-14T04:52:46 deepblue crashes at the start whenever there are more than 23 planets 2010-11-14T04:53:08 you know what that mean... 2010-11-14T04:53:15 2 week extension! 2010-11-14T04:53:35 they wont extend the deadline 2010-11-14T04:53:39 flag: not looking good if you made changes 2010-11-14T04:53:57 well, i very much doubt it from my limited experience 2010-11-14T04:54:14 antimatroid: actually, I never noticed it was always 23 planets until about a week ago 2010-11-14T04:54:30 McLeopold: no kidding 2010-11-14T04:54:35 Okay, you can take it off now 2010-11-14T04:54:39 I need some sleep 2010-11-14T04:54:41 i never realised only the centre planet could have 0 growth till i was writing the new generator 2010-11-14T04:55:05 that crashed my bot sooo much 2010-11-14T04:56:11 varying planets? 2010-11-14T04:56:27 i just input them into a vector on the first pass of the gamestate and update every turn thereafter 2010-11-14T04:56:47 the only thing that would break my bot would be blowing out the max distance and number of planets quite a bit 2010-11-14T04:57:28 and it wouldn't actually break, probably just make really stupid moves 2010-11-14T04:57:38 or run out of memory eventually 2010-11-14T04:58:00 I think mine just gets slower 2010-11-14T04:58:13 i have timers telling it when to stop 2010-11-14T04:58:25 so in theory you would have to use up all memory before it would break 2010-11-14T04:58:30 I ought to do that 2010-11-14T04:58:52 i don't make assumptions :P 2010-11-14T05:00:44 computers should support elliptical windows 2010-11-14T05:00:47 with resizing 2010-11-14T05:00:54 that'd be cool 2010-11-14T05:10:36 http://ai-contest.com/ down 2010-11-14T05:12:01 seems up for me 2010-11-14T05:12:08 it was slow a few minutes ago though 2010-11-14T05:16:17 now up 2010-11-14T05:17:06 Top 10 players: bocsimacko(3916), bix0r4ever(3642), dmj111(3608), george(3576), GreenTea(3571), _Astek_(3565), shangas(3556), asavis(3552), Raschi(3522), felixcoto(3517) 2010-11-14T05:18:40 so, has anyone used a java bot with the tcp program? 2010-11-14T05:19:29 someone must have 2010-11-14T05:22:08 define "tcp program" 2010-11-14T05:22:27 dhartmei's tcp.c 2010-11-14T05:22:30 I got it working 2010-11-14T05:22:38 RageBot on my tcp. :) 2010-11-14T05:23:14 you sound so proud :P 2010-11-14T05:23:16 no. (I was using java bot with tcp.jar and with jbotmanager) 2010-11-14T05:23:35 it was more of a "nevermind" 2010-11-14T05:24:24 *** perhaps has joined #aichallenge 2010-11-14T05:24:46 *** perhaps is now known as aasdfareg 2010-11-14T05:25:12 hmm, possibly alternative ragebot, having it constantly attack the first planet in the list of planets with all ships it owns 2010-11-14T05:25:22 or have it send all its ships to the closest enemy every turn 2010-11-14T05:25:34 i'll make that after 2010-11-14T05:26:09 that'd actually be a pretty good bot to test with 2010-11-14T05:27:54 do you mean each planet sends to it's closest so that it is attacking (possibly) more than one planet a turn 2010-11-14T05:28:36 yeah 2010-11-14T05:28:40 sorry 2010-11-14T05:29:47 take out the 10* growth restrictor and ragebot will do that 2010-11-14T05:30:14 *** Frontier has joined #aichallenge 2010-11-14T05:30:27 probably not exactly as i'd like 2010-11-14T05:30:36 i want it to go for it's closest expected enemy 2010-11-14T05:30:44 not the closest planet the enemy already owns 2010-11-14T05:31:44 closest planet enemy has fleets headed for? 2010-11-14T05:32:00 owns or has fleets headed for 2010-11-14T05:32:03 closest planet that is owned by the enemy when the fleets arrive 2010-11-14T05:32:10 and only send enough to "take it" then go for the next one 2010-11-14T05:32:25 it'd be super rage bot 2010-11-14T05:32:35 i'd probably also have it do nothing on the first move 2010-11-14T05:32:49 i think people would find that a bit more painful 2010-11-14T05:33:30 so if you don't send enough to take the planet it ignores that one 2010-11-14T05:33:43 no 2010-11-14T05:33:53 but if it has more than needed, it moves onto the next planet with the rest 2010-11-14T05:34:27 sorry I was refering to "owned by the enemy when the fleets arrive" 2010-11-14T05:36:24 in other words, if "I" send one ship to the neutral closest to it home planet will it attack that planet 2010-11-14T05:37:00 bix0r4ever has a good record against bocsimacko on the new maps 2010-11-14T05:37:30 in fact 2010-11-14T05:37:32 @rankings 2010-11-14T05:37:33 antimatroid: Top 10 players: bix0r4ever(3988), bocsimacko(3902), dmj111(3605), george(3575), GreenTea(3571), _Astek_(3562), shangas(3556), asavis(3541), Raschi(3533), felixcoto(3518) 2010-11-14T05:37:47 ooo poor mega1 2010-11-14T05:37:56 it's still early 2010-11-14T05:38:07 can some other people analyse the games? 2010-11-14T05:38:12 bix0r is reuploading constantly 2010-11-14T05:38:14 although bix0r4ever had a goodish (the best?) record against him before 2010-11-14T05:38:21 yes 2010-11-14T05:38:35 mega1: oh yeah.. lol .. he has 2 games 2010-11-14T05:38:41 lol 2010-11-14T05:38:44 mega1: 1 win against you 2010-11-14T05:38:44 ah, that'll do it 2010-11-14T05:39:10 i was going to say, no bot has come close to touching mega1 2010-11-14T05:39:14 mega1's win rates against everyone else keep him on top 2010-11-14T05:39:18 even with a few wins against him 2010-11-14T05:39:25 actually.. no 2010-11-14T05:39:33 he seems to loose constantly against bix 2010-11-14T05:39:45 well 2010-11-14T05:39:48 http://ai-contest.com/visualizer.php?game_id=7256222 :P 2010-11-14T05:39:49 he lost 2 other games too 2010-11-14T05:40:08 tought.. lol 2010-11-14T05:40:20 it's 13 to 2 total 2010-11-14T05:40:21 yeah I recall a few days ago noting that bix0r won a lot vs bocsi 2010-11-14T05:40:36 mega1: for him? 2010-11-14T05:40:39 for me 2010-11-14T05:40:43 oh 2010-11-14T05:40:53 but that include that last couple of days 2010-11-14T05:41:13 he didn't reupload that constantly.. 2010-11-14T05:41:20 hey I wish my bot would go 2-13 vs bocsi... 2010-11-14T05:41:23 no, but he did then, which always overshoots 2010-11-14T05:41:43 *** Mjothvitnir has quit IRC (Quit: Page closed) 2010-11-14T05:41:49 antimatroid: what? 2010-11-14T05:41:59 bots overshoot their elo near the top 2010-11-14T05:42:02 once reuploaded 2010-11-14T05:42:11 weird 2010-11-14T05:42:16 not really 2010-11-14T05:42:25 i guess it also undershoots if you lose early 2010-11-14T05:44:52 lavalamp_: you really do suck against ragebot :P 2010-11-14T05:45:19 McLeopold: http://zeroviz.us:8080/canvas?game_id=260 nice ragebot :P 2010-11-14T05:45:23 it's not very ragebotty 2010-11-14T05:45:38 that is ragebot 2010-11-14T05:45:50 it's the example bot 2010-11-14T05:46:13 did it always start making lots of moves like that? 2010-11-14T05:46:26 yeah, 1 move per planet owned 2010-11-14T05:46:37 ah, i've never let ragebot do so well aha 2010-11-14T05:52:01 fireworks mode.. could anyone make the brief picture of a smiley in fireworks? 2010-11-14T05:53:51 *** RainCT has joined #aichallenge 2010-11-14T05:58:05 *** irchs has joined #aichallenge 2010-11-14T06:05:23 * lavalamp_ writes O(n^3) algorithm where n = planet count 2010-11-14T06:06:09 yeah, i have some nasty loops 2010-11-14T06:09:12 I would like to see radially symmetric maps with N spokes (N is even) where each player gets a planet on alternating spokes 2010-11-14T06:09:29 *** McLeopold has quit IRC (Ping timeout: 250 seconds) 2010-11-14T06:09:53 lavalamp_: then make some :P 2010-11-14T06:10:02 maybe I will :p 2010-11-14T06:10:26 also how about maps where you make it fair by tiling instead of symmetry? 2010-11-14T06:10:35 maybe I will do that too 2010-11-14T06:10:42 fair by tiling? 2010-11-14T06:11:01 yeah, so you have a random group of planets 2010-11-14T06:11:12 it's not even sufficient for a map to be symmetric in order for it to be fair 2010-11-14T06:11:35 wait, no, i went over that before 2010-11-14T06:11:37 yes it is 2010-11-14T06:11:38 repeat it 4 times over the board (rotating/mirroring as needed) and each player gets a starting planet in two of the groups, diagonal from each other 2010-11-14T06:11:39 ignore me 2010-11-14T06:11:47 lol 2010-11-14T06:12:11 hm actually if you do that maybe the map would be symmetric 2010-11-14T06:12:21 it would 2010-11-14T06:12:23 ignore me too 2010-11-14T06:12:33 i think sufficiency is equivalent to a fair match 2010-11-14T06:12:36 iff etc. etc. 2010-11-14T06:12:45 sufficiency == symmetry 2010-11-14T06:12:48 *** fawek has joined #aichallenge 2010-11-14T06:12:54 yeah I guess so :( 2010-11-14T06:13:15 ah, but an infinitely tiled map could be fair and not symmetric 2010-11-14T06:13:18 i don't think the new generator is far off being able to make all possible maps 2010-11-14T06:13:30 looks like there was not much damage to the rankings with the new maps. all the bots seem to work more or less on them 2010-11-14T06:13:44 Frontier: i'm very worried about oldman 2010-11-14T06:14:02 who? oldman? the chinese guy, where is he now? 2010-11-14T06:14:19 antimatroid put his bot in the nursing home 2010-11-14T06:14:30 http://ai-contest.com/profile.php?user_id=7526 2010-11-14T06:14:37 down to 428 2010-11-14T06:14:40 418* 2010-11-14T06:15:02 *** p4p4 has joined #aichallenge 2010-11-14T06:15:04 and it's not crashing on the first move 2010-11-14T06:15:13 so i don't think it's just a matter of number of planets 2010-11-14T06:15:19 hm if the game were played on the surface of an octohedron that could be really interesting 2010-11-14T06:15:44 or other geometric solid 2010-11-14T06:15:44 well, did anyone send an email notification about the changes? 2010-11-14T06:15:55 anyway I go code now 2010-11-14T06:16:04 is it just me, or did most of the maps end up with more than 23 planets? 2010-11-14T06:16:22 Frontier: don't think so, but he's seen the changes, he has resubmitted a couple of times 2010-11-14T06:17:02 mega1: back where you belong :) 2010-11-14T06:17:06 Top 10 players: bocsimacko(3681), bix0r4ever(3505), dmj111(3406), george(3400), GreenTea(3354), shangas(3348), asavis(3337), _Astek_(3337), Raschi(3326), Accoun(3315) 2010-11-14T06:17:13 perfect timing 2010-11-14T06:17:18 it's pretty hectic 2010-11-14T06:17:28 a lot of people are resubmitting too 2010-11-14T06:17:46 maybe he doesnt know the maximum planet count. *sigh* why was there no email. this is like when i code something at 10 pm 2010-11-14T06:18:14 i would expect bots to crash on the first turn if there was an issue with planet counts 2010-11-14T06:18:38 i wonder how many bots are like oldmans 2010-11-14T06:18:42 that's the biggest issue i've found 2010-11-14T06:18:51 what if they just don't know about the other planets and suddenly receive a source planet id of 29? 2010-11-14T06:19:14 they had to process it at the start too 2010-11-14T06:19:18 if all loops iterate from 0 to 22, that could be possible 2010-11-14T06:20:22 it would be a bad mistake if all loops went from 0 to 22 2010-11-14T06:20:25 hmm, i get your point, you actually had to code in an abort after line 23 of planet input ^^ 2010-11-14T06:20:36 and how the hell does one test on maps with like 3 or 4 planets if your bot can only handle that? 2010-11-14T06:21:19 i don't know how people have worked out their strategies if they haven't used small examples like that 2010-11-14T06:22:41 i never used small examples actually ^^. It was always using the 23 planets sets. And it worked ok. Just now I changed some ownerships so half the map belongs to each player to test some defense code. 2010-11-14T06:23:30 *** Bobng has joined #aichallenge 2010-11-14T06:23:56 I could have written a bot that reads planets from stdin, and ignores them if their id is above 22. not very bright, but possible :) 2010-11-14T06:24:51 very unbright :P 2010-11-14T06:24:57 so it works with smaller sizes, but not larger sets. and it looks like the bot wont crash - until the fleets come into play 2010-11-14T06:25:01 i read the planets and just add however many i get to a vector 2010-11-14T06:26:35 Error323: your bot seems to like the new maps 2010-11-14T06:26:41 i do the same, it's just my new bot uses fixed size arrays and copies the data from the framework into these arrays as a first step (filtering out fleet.source and other unimportant information) 2010-11-14T06:27:08 i did fixed arrays etc. for tron, i can't be bothered for this 2010-11-14T06:27:10 it's too messy 2010-11-14T06:33:41 *** mceier has joined #aichallenge 2010-11-14T06:34:31 *** irchs has quit IRC (Quit: irchs) 2010-11-14T06:36:37 *** boegel has joined #aichallenge 2010-11-14T06:37:11 has the server started using different maps? 2010-11-14T06:37:21 boegel: yes 2010-11-14T06:37:32 http://ai-contest.com/forum/viewtopic.php?f=19&t=1094 2010-11-14T06:37:38 that post contains information about them 2010-11-14T06:38:21 cool, thx 2010-11-14T06:38:46 need to setup new maps to TCP server 2010-11-14T06:39:28 Accoun: http://zeroviz.us:8080/ (use port 995 for your bot) 2010-11-14T06:39:37 no one can find dhartmei to get new maps onto tcp 2010-11-14T06:39:48 there aren't very many people on there atm though 2010-11-14T06:42:15 controversial, it seems :) 2010-11-14T06:43:16 it was a fairly even split as far as i'm concerned 2010-11-14T06:43:23 and it was always intended to change 2010-11-14T06:43:44 the rankings seem about the same so far too 2010-11-14T06:43:47 excepting oldman 2010-11-14T06:44:54 the rankings have 2.5 days worth of old games too for those who have not resubmitted 2010-11-14T06:45:18 good point 2010-11-14T06:45:19 * boegel thinks changing the maps is a good idea 2010-11-14T06:45:42 antimatroid feels really bad about oldman 2010-11-14T06:45:44 lol 2010-11-14T06:45:52 i hope it will continue to happen like this in future contests, just with clearer messages that it will 2010-11-14T06:45:55 antimatroid: for all you know it could be a one line fix 2010-11-14T06:46:09 it could, but i know he's seen the change 2010-11-14T06:46:20 oh ,right.. 2010-11-14T06:46:30 and it's not crashing on the first turn like most bots with number of planets issues 2010-11-14T06:46:51 Frontier might be right, in which case it should be alright 2010-11-14T06:46:57 tcp.exe 76.121.152.123 995 $botName $botFile $botparameters 2010-11-14T06:46:58 otherwise i have no idea what would be causing it 2010-11-14T06:47:31 play please, tcp.exe 76.121.152.123 995 $botName $botFile $botparameters 2010-11-14T06:47:38 im playing now 2010-11-14T06:47:58 http://zeroviz.us:8080/getratings 2010-11-14T06:48:08 Accoun: there is probably only a couple of other bots running on there atm, within a day it should be better 2010-11-14T06:48:30 new map set on it 2010-11-14T06:48:58 http://zeroviz.us:8080/canvas?game_id=415 that's a line symmetric map :) 2010-11-14T06:50:00 i do think perhaps there could be a minimum distance between the two players starting angles 2010-11-14T06:50:31 need command line help on http://zeroviz.us:8080/ 2010-11-14T06:50:41 tcp.exe 76.121.152.123 995 $botName $botFile $botparameters 2010-11-14T06:51:23 i haven't run a bot on there yet sorry 2010-11-14T06:51:35 Accoun: have you tried jBotManager? 2010-11-14T06:51:44 it lets you watch tcp games live, and play your own bot in human mode 2010-11-14T06:51:51 http://ai-contest.com/forum/viewtopic.php?f=18&t=751 2010-11-14T06:52:05 antimatroid> i not have problem with it 2010-11-14T06:53:20 http://zeroviz.us:8080/canvas?game_id=424 i don't think that's as much of a draw map as mcleopold thinks 2010-11-14T06:53:55 lol 2010-11-14T06:53:57 wow 2010-11-14T06:55:06 amstan: the new maps are pretty awesome :P 2010-11-14T07:06:00 *** hellman has joined #aichallenge 2010-11-14T07:07:45 play please, tcp.exe 76.121.152.123 995 $botName $botFile $botparameters 2010-11-14T07:07:57 im playing now http://zeroviz.us:8080/getratings 2010-11-14T07:08:02 new map set 2010-11-14T07:09:31 Accoun: in a few minutes 2010-11-14T07:13:48 windows wersion tcp.exe not support DNS, need IP adress 2010-11-14T07:14:34 *** antimatroid has left #aichallenge 2010-11-14T07:15:51 new maps like Rorschach test 2010-11-14T07:16:26 heh 2010-11-14T07:16:57 *** antimatroid has joined #aichallenge 2010-11-14T07:17:06 Top 10 players: bocsimacko(3878), bix0r4ever(3744), dmj111(3587), george(3571), Accoun(3566), shangas(3551), asavis(3537), _Astek_(3532), GreenTea(3528), Raschi(3525) 2010-11-14T07:18:57 Accoun: bet you wish you wish you didn't ask for more competition now :P 2010-11-14T07:19:55 antimatroid> use more sample english 2010-11-14T07:20:20 you asked for more competition and the only person to start running their bot was number 1 rank :P 2010-11-14T07:21:41 waiting for ozers 2010-11-14T07:22:00 welcome http://zeroviz.us:8080/ 2010-11-14T07:22:09 i should have my new rewrite up and testing within 24 hours 2010-11-14T07:24:53 I was 10-0 against bix0r now I'm 3-3 2010-11-14T07:25:05 maybe something's changed 2010-11-14T07:25:55 yeah, come to zeroviz, I want to see the difference between my bots 2010-11-14T07:26:24 i will when it's running 2010-11-14T07:26:45 i have watched those 3 losses, what are your thoughts on them? 2010-11-14T07:26:57 do you think it was strictly a result of the map generator change? 2010-11-14T07:27:05 you would be the best judge of that 2010-11-14T07:27:15 I have to rewatch them, wait a sec 2010-11-14T07:28:38 the most recent loss was a drastically different map to old ones 2010-11-14T07:28:45 the other two not so much 2010-11-14T07:28:53 the first and the third ones look to me like r-p-s 2010-11-14T07:28:58 the second, I don't know 2010-11-14T07:29:26 do you think rps situations are good or bad 2010-11-14T07:29:34 my first thoughts are yes, but i'm not sure 2010-11-14T07:29:53 i think multiple neutrals significantly increases the chances of those sitation 2010-11-14T07:29:53 s 2010-11-14T07:30:02 multiple central neutrals*** 2010-11-14T07:31:09 this is not a moral question. I think they are OK but it's a different game now. 2010-11-14T07:31:25 http://ai-contest.com/visualizer.php?game_id=7246869 2010-11-14T07:31:25 *** boegel has quit IRC (Quit: *poof!*) 2010-11-14T07:31:29 i'm not sure you couldn't have won that 2010-11-14T07:31:42 if you kept going at the neutral you first claimed and the one they did near it 2010-11-14T07:32:18 sure, there are many ways to play differently 2010-11-14T07:32:32 *** Rubicon-|-Cross has joined #aichallenge 2010-11-14T07:33:06 bbl 2010-11-14T07:35:29 *** raedwulf has quit IRC (Ping timeout: 240 seconds) 2010-11-14T07:40:09 *** Xanthor[aw] has quit IRC (Ping timeout: 276 seconds) 2010-11-14T07:42:52 *** raedwulf has joined #aichallenge 2010-11-14T07:53:16 *** raedwulf has quit IRC (Quit: WeeChat 0.3.3) 2010-11-14T07:55:32 *** Xanthor[aw] has joined #aichallenge 2010-11-14T07:59:12 antimatroid: have you seen any discussion about draw resuls not being recorded in the results? 2010-11-14T08:04:47 *** smellyhippy has quit IRC (Ping timeout: 264 seconds) 2010-11-14T08:05:55 *** narnach has joined #aichallenge 2010-11-14T08:09:43 *** smellyhippy has joined #aichallenge 2010-11-14T08:10:23 oh god.... mega1 give up, you have no chance in this competition :) 2010-11-14T08:10:58 dmj111: what do you mean? 2010-11-14T08:14:38 draws are not recorded at all in the official server. 2010-11-14T08:14:54 http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/tournament_manager.py 2010-11-14T08:14:59 *** krokkrok has joined #aichallenge 2010-11-14T08:15:08 *** delt0r___ has joined #aichallenge 2010-11-14T08:15:17 the if / else from line 147 to 170 2010-11-14T08:16:08 last time around we had too many drawish maps... but it doesn't seem quite right to throw out all draws. (unless I misunderstand the code) 2010-11-14T08:16:14 *** delt0r has quit IRC (Ping timeout: 240 seconds) 2010-11-14T08:17:07 Top 10 players: bocsimacko(3889), bix0r4ever(3712), Accoun(3658), Raschi(3654), george(3573), dmj111(3570), shangas(3553), _Astek_(3543), asavis(3534), GreenTea(3513) 2010-11-14T08:17:35 hmmm, i'm not sure 2010-11-14T08:17:55 i would have thought draws should count too 2010-11-14T08:18:59 yeah, there is a difference between losing 4 out of 4, to losing 4 out of 10, with 6 draws. 2010-11-14T08:19:06 it looks like there is a log message when a draw happens, maybe someone with access can see how often it happens 2010-11-14T08:20:07 you probably need to talk to janzert for anything like that 2010-11-14T08:20:14 and he's going away tomorrow/today 2010-11-14T08:20:17 for a few days 2010-11-14T08:21:02 ok. if I see him, I'll ask. I also realize it is kind of late for something like that. 2010-11-14T08:21:17 I was just curious if it was ever brought up, and it doesn't sound like it was. 2010-11-14T08:23:39 antimatroid: i also like the results i get on the new maps :0 my bot seems a little better on them then the other ones. 2010-11-14T08:24:42 :) 2010-11-14T08:24:51 i haven't got my bot in a working state to have even tried 2010-11-14T08:25:02 it would be a little funny if i pushed hard and then it really really sucked on them :P 2010-11-14T08:25:39 the only major problem i've seen is oldman 2010-11-14T08:25:57 deepblue is also crashing, but i think his problem should be fixed in one line 2010-11-14T08:26:03 but oldman is crashing mid game 2010-11-14T08:26:49 ha. i hope you do well :) 2010-11-14T08:26:58 http://ai-contest.com/profile.php?user_id=7526 :( 2010-11-14T08:38:19 Im getting almost same kind of crashes, luckily not as often. Something in my sniping caused me to crash in almost every game. Now it just happens occasionally when Im gaining the lead in growth (one step before I get neurals) 2010-11-14T08:48:52 *** hellman has quit IRC (Remote host closed the connection) 2010-11-14T08:58:09 *** smloh1 has joined #aichallenge 2010-11-14T09:02:20 *** smloh1 has quit IRC (Client Quit) 2010-11-14T09:03:20 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-11-14T09:04:55 *** irchs has joined #aichallenge 2010-11-14T09:08:36 *** krokkrok has quit IRC (Quit: Page closed) 2010-11-14T09:09:18 *** choas has joined #aichallenge 2010-11-14T09:17:06 Top 10 players: bocsimacko(3886), bix0r4ever(3749), Accoun(3631), Raschi(3629), george(3575), dmj111(3569), shangas(3542), asavis(3540), _Astek_(3519), Xrillo(3517) 2010-11-14T09:17:33 *** Azrathud has quit IRC (Ping timeout: 250 seconds) 2010-11-14T09:23:47 *** aasdfareg has quit IRC (Quit: Page closed) 2010-11-14T09:32:26 *** Frontier has left #aichallenge 2010-11-14T09:32:29 *** braddunbar has joined #aichallenge 2010-11-14T10:00:41 *** perror has joined #aichallenge 2010-11-14T10:03:39 this bix0r guy is getting on my nerves 2010-11-14T10:05:24 :) 2010-11-14T10:05:53 he does seem to be quite good 2010-11-14T10:06:30 he does resubmit a lot though 2010-11-14T10:07:15 just when he drops a bit ;-) 2010-11-14T10:07:33 can't say i've never done it myself :P 2010-11-14T10:07:37 makes it a bit hard to assess his strength 2010-11-14T10:07:43 but constantly? 2010-11-14T10:07:47 no 2010-11-14T10:08:03 i haven't even had a bot on the main site in like over a month 2010-11-14T10:08:33 i had one on tcp up until a few weeks ago, but i really had to stop for a while because of exams 2010-11-14T10:08:53 i already did barely anything for classes during semester :P 2010-11-14T10:09:55 mega1: http://ai-contest.com/visualizer.php?game_id=7290413 2010-11-14T10:10:01 what do you think the best first move is on that map? 2010-11-14T10:10:47 i think it's pretty close to rps 2010-11-14T10:11:03 wait no, always wait 2010-11-14T10:11:08 yeah, mcleopold is right 2010-11-14T10:11:11 *** irchs has quit IRC (Quit: irchs) 2010-11-14T10:12:36 ha, take that bix0r, 42nd place aha 2010-11-14T10:12:42 yes, I think waiting is the best 2010-11-14T10:13:09 i'm pretty sure you can at worst draw by waiting, originally i thought if they went the 58 you lost, but clearly not :P 2010-11-14T10:17:07 Top 10 players: bocsimacko(3918), Raschi(3650), Accoun(3605), dmj111(3576), george(3574), shangas(3547), asavis(3535), _Astek_(3517), Xrillo(3513), felixcoto(3505) 2010-11-14T10:18:48 *** yasith_ has quit IRC (Ping timeout: 240 seconds) 2010-11-14T10:19:12 *** yasith has quit IRC (Ping timeout: 265 seconds) 2010-11-14T10:30:54 play please, tcp.exe 76.121.152.123 995 $botName $botFile $botparameters 2010-11-14T10:31:01 im playing now http://zeroviz.us:8080/getratings 2010-11-14T10:31:22 new map set 2010-11-14T10:33:35 did mcleopold say it's ok to advertise zeroviz now? 2010-11-14T10:33:53 I guess he's still on his comcast connection. 2010-11-14T10:34:22 no idea sorry 2010-11-14T10:37:04 *** hellman has joined #aichallenge 2010-11-14T10:40:12 *** Frontier has joined #aichallenge 2010-11-14T10:42:36 *** rebelxt_ has joined #aichallenge 2010-11-14T10:43:50 *** irchs has joined #aichallenge 2010-11-14T10:45:45 hey all, who can clarify the map situation? i.e. how come are they changing map generator two weeks before the deadline? 2010-11-14T10:47:47 there was never a map generator set 2010-11-14T10:47:58 the new maps are more interesting, and had actually been intended for a while 2010-11-14T10:48:29 antimatroid: we were told that the same "old" generator would be used with a new seed 2010-11-14T10:48:36 where? 2010-11-14T10:48:53 IRC, weeks ago 2010-11-14T10:49:14 anything i'd seen on here from organisers was that it'd change eventually 2010-11-14T10:49:43 antimatroid: i do agree maps are more interesting, but last minute changes like this are not a great idea, imho 2010-11-14T10:50:05 *** Frontier has left #aichallenge 2010-11-14T10:50:23 rebelxt_: the top 100 tends to feel this way 2010-11-14T10:50:32 (in my opinion) 2010-11-14T10:50:35 i think the problem is that about half the people see it as a change and the rest don't 2010-11-14T10:50:43 antimatroid: so, i'm looking at the poll on the forums where most people voted against the change, and the change was still made.. what's up with that? ;-) just curious 2010-11-14T10:50:45 mega1: did you see Xrillo's post in the forums? 2010-11-14T10:50:50 no 2010-11-14T10:50:58 i would be very surprised if i don't make at least the top 50 2010-11-14T10:51:18 top 10 would be nice 2010-11-14T10:51:52 http://ai-contest.com/forum/viewtopic.php?f=17&t=1094&start=20 2010-11-14T10:52:23 dmj and mcleopold like them too i'm pretty sure, it's certainly not complete hate from the top 100 2010-11-14T10:52:28 "Why on earth would someone assume that the first set of maps would be final when no such official statements have been made?" 2010-11-14T10:52:29 *** Frontier has joined #aichallenge 2010-11-14T10:52:40 clearly he's not read the topic 2010-11-14T10:52:57 "A decision has been made on the official map generator" 2010-11-14T10:53:03 no decision had been made prior 2010-11-14T10:53:23 I mean the posts that come before his post 2010-11-14T10:53:51 just how many times do we confuse final generator and final set of maps 2010-11-14T10:54:05 antimatroid: were the poll results ignored? http://ai-contest.com/forum/viewtopic.php?f=19&t=1087 2010-11-14T10:54:37 the poll results were split 50 50 as far as i'm concerned 2010-11-14T10:54:48 maybe not for wild, but that really wasn't much of a change 2010-11-14T10:54:53 the big change was line symmetric maps 2010-11-14T10:55:34 the big change was more neutral at the same distance from the starting positions 2010-11-14T10:55:54 i tested two of my best bots (which use totally different algorithms) on old and new maps.. got different results. they are within 10% but relative strength of my bots changed which means more tuning will be needed 2010-11-14T10:56:19 i don't tune, i keep refining my general strategy : 2010-11-14T10:56:58 i prefer to have new maps at the end of a contest like that, i do wish it was made more explicitly clear that it would happen though 2010-11-14T10:57:04 like this* 2010-11-14T10:57:43 you could probably fish up forum posts i made from tron asking for general but clear rules :) 2010-11-14T10:59:00 antimatroid: even though i doubt this change will affect the final ratings significantly, this is still a bad change. it it was done one month ago, then fine. 2010-11-14T10:59:25 i don't have any problems with new maps like this at this stage 2010-11-14T10:59:29 i think it's a good thing 2010-11-14T10:59:50 my only complaint would be that not all users seeemed to understand that's how the organisers do this 2010-11-14T10:59:55 they did exactly teh same for tron 2010-11-14T11:00:07 even a1k0n said before, eventually they added line symmetric maps 2010-11-14T11:02:40 let me put this bluntly: 2010-11-14T11:03:08 it's not the way to run a fair contest 2010-11-14T11:03:21 it would be fair if everyone understood it would happen 2010-11-14T11:03:28 it wasn't for tron, it's not now 2010-11-14T11:03:33 no 2010-11-14T11:04:07 the rules can be as general as one wants and the contest remain fair, provided they're clear and stuck to 2010-11-14T11:04:10 just because your particular approach does not involve tuning you can still imagine that that's not only way 2010-11-14T11:05:13 a new set of from a distribution cannot really change the game 2010-11-14T11:05:16 *** hellman has quit IRC (Remote host closed the connection) 2010-11-14T11:05:31 a new generator can. 2010-11-14T11:05:52 not if people are told that isn't the generator that will be used, that it is just to make your own examples with 2010-11-14T11:06:10 then the game was never restricted to the maps of that generator 2010-11-14T11:06:21 which in this case I believe the organisers didn't want 2010-11-14T11:06:28 jeff certainly didn't 2010-11-14T11:06:38 can't you see that the set of possible generator is just too large to prepare for that? 2010-11-14T11:06:49 that's the idea 2010-11-14T11:06:59 write a bot to play the game 2010-11-14T11:07:20 don't give me the "general ai" thing again 2010-11-14T11:07:21 antimatroid: so, who pushed for the change to happen? 2010-11-14T11:07:25 me 2010-11-14T11:07:36 you can hate me :) 2010-11-14T11:07:56 i would have pushed for it a lot earlier, but i didn't have the time 2010-11-14T11:08:08 i had asked and asked, no one did anything, so i did something for myself 2010-11-14T11:08:20 antimatroid: that's my point exactly. you pushed way too late 2010-11-14T11:08:34 there was never a generator set, someone else could have pushed earlier 2010-11-14T11:08:49 wait, I have a great idea for a generator 2010-11-14T11:08:56 as far as i was concerned, nothing was to be assumed yet about the form final maps will take 2010-11-14T11:09:01 *** Frontier has left #aichallenge 2010-11-14T11:09:02 antimatroid: no hate at all. i just want new maps on the tcp server for testing, and also i want a final confirmation on what generator will be used 2010-11-14T11:09:02 in a couple of days I hope it'll official 2010-11-14T11:09:18 http://zeroviz.us:8080/ use port 995 2010-11-14T11:09:27 rebelxt_: sorry, it's not ready yet 2010-11-14T11:10:26 revelxt_: at the moment, it's pretty likely that the new maps will be the maps used 2010-11-14T11:10:39 not the maps, again 2010-11-14T11:10:47 the generator 2010-11-14T11:10:55 ^^ that, sorry 2010-11-14T11:11:22 we need a final call instead of "pretty likely" 2010-11-14T11:11:33 the only way it'll change is if things fuck up 2010-11-14T11:11:38 http://ai-contest.com/forum/viewtopic.php?f=17&t=1094 the information there is correct 2010-11-14T11:11:50 i am not in a position to make official statements 2010-11-14T11:12:56 i understand, but please try to get contest organizers to make a call asap 2010-11-14T11:13:23 like i said, it's very very unlikely it'll change unless something really bad happens 2010-11-14T11:17:10 Top 10 players: bocsimacko(3918), Raschi(3649), Accoun(3597), george(3582), dmj111(3560), shangas(3548), asavis(3533), ClusterTeam(3518), _Astek_(3516), felixcoto(3512) 2010-11-14T11:22:04 *** McLeopold has joined #aichallenge 2010-11-14T11:23:22 McLeopold: is it okay now to advertise zeroviz on benzendrine or you prefer to stay low? 2010-11-14T11:23:57 *** boegel has joined #aichallenge 2010-11-14T11:25:32 gah, sites down? 2010-11-14T11:26:49 *** Frontier has joined #aichallenge 2010-11-14T11:27:11 *** McLeopold has quit IRC (Ping timeout: 245 seconds) 2010-11-14T11:28:42 antimatroid: seems fine here 2010-11-14T11:29:06 antimatroid: http://downforeveryoneorjustme.com/ai-contest.com 2010-11-14T11:29:53 yeah, loaded now, got an error before 2010-11-14T11:31:34 *** hellman has joined #aichallenge 2010-11-14T11:31:45 *** irchs has quit IRC (Quit: irchs) 2010-11-14T11:33:01 *** McLeopold has joined #aichallenge 2010-11-14T11:43:47 *** narnach has quit IRC (Quit: Leaving.) 2010-11-14T11:45:21 antimatroid: map_r528_25.txt is coming out a bit drawish 2010-11-14T11:46:50 woah: error in server code? http://zeroviz.us:8080/canvas?game_id=1144 2010-11-14T11:51:32 which map is that? 2010-11-14T11:51:34 *** Frontier has left #aichallenge 2010-11-14T11:51:50 i agree about that map earlier, it was a wait and draw map 2010-11-14T11:53:28 okay, so my server is starting to prove that now 2010-11-14T11:53:44 there are a few others as well, but not as drawish 2010-11-14T11:54:36 perhaps the neutral needs to be takable and not snipable? 2010-11-14T11:54:47 that wouldn't be that hard to ensure 2010-11-14T11:57:05 *** irchs has joined #aichallenge 2010-11-14T11:59:24 yea, that would be good 2010-11-14T11:59:43 but it might make for too obvious of openings 2010-11-14T12:00:03 as long as there are enough snipable neutrals that you can't snipe them all, that is good too 2010-11-14T12:00:43 If a map draw on a bad bot against a good one, I'd remove it 2010-11-14T12:00:54 If it draws between to really good bots, that's okay 2010-11-14T12:01:21 *** Frontier has joined #aichallenge 2010-11-14T12:02:17 *** irchs is now known as janstaunton 2010-11-14T12:02:28 I wish windows had a which command 2010-11-14T12:05:42 is it possible to have more than one "snipable" planets that can't be sniped in combinations? 2010-11-14T12:05:51 i would have thought that's just worse for the idiot first mover 2010-11-14T12:08:40 oops: http://zeroviz.us:8080/canvas?game_id=1504 2010-11-14T12:08:47 Accoun got raged. 2010-11-14T12:09:01 hehe 2010-11-14T12:09:09 if it takes more that 105 ships, then yes 2010-11-14T12:09:13 ? 2010-11-14T12:09:19 maybe not 2010-11-14T12:09:21 i think i might have then 2010-11-14T12:10:21 I beat mega1! http://zeroviz.us:8080/canvas?game_id=1482 2010-11-14T12:10:46 :-( 2010-11-14T12:12:03 mega1: 2 long bombs that didn't pan out 2010-11-14T12:12:27 the server died 2010-11-14T12:12:35 turn 70 2010-11-14T12:13:34 okay, let me restart it 2010-11-14T12:14:06 uh, still working 2010-11-14T12:14:19 sorry, the official one :-) 2010-11-14T12:14:19 I'm playing accoun right now 2010-11-14T12:14:26 :) 2010-11-14T12:14:37 zero game played and dman slow 2010-11-14T12:14:40 I'm just ego-centric. :P 2010-11-14T12:15:48 *** phreeza has joined #aichallenge 2010-11-14T12:15:49 I wonder if a lot of bots are crashing? 2010-11-14T12:16:42 janzert said the error rate wasn't too high earlier 2010-11-14T12:16:49 deepblue is crashing 2010-11-14T12:16:59 and hasn't uploaded in over a month, so no idea when they'll see that 2010-11-14T12:17:18 Top 10 players: bocsimacko(3918), Raschi(3649), Accoun(3597), george(3582), dmj111(3560), shangas(3548), asavis(3533), ClusterTeam(3518), _Astek_(3516), felixcoto(3512) 2010-11-14T12:17:56 anyway, it's bedtime, night 2010-11-14T12:18:04 i'm even going to make it before sunrise tonight 2010-11-14T12:22:21 *** McLeopold_ has joined #aichallenge 2010-11-14T12:23:13 *** McLeopold has quit IRC (Ping timeout: 265 seconds) 2010-11-14T12:23:23 * mega1 hums another bug bites the dust 2010-11-14T12:23:32 oh knows 2010-11-14T12:23:37 noes* 2010-11-14T12:23:38 I love basic bugs this late. 2010-11-14T12:23:56 finding really bad bugs is the best 2010-11-14T12:24:09 yes, it was kind bad 2010-11-14T12:24:16 how it survived unnoticed is beyond me 2010-11-14T12:24:25 it's so hard to get the logic right 2010-11-14T12:24:39 in unhospitable environment too, right below a FIXME :-) 2010-11-14T12:26:01 *** davidjliu has joined #aichallenge 2010-11-14T12:27:14 georges bot is stil doing well 2010-11-14T12:27:17 and still old 2010-11-14T12:37:29 man website is slooooow 2010-11-14T12:39:54 oh 0 servers running? 2010-11-14T12:47:28 or the main server is deadish 2010-11-14T12:47:51 "too many connections" 2010-11-14T12:48:54 *** tapwater has joined #aichallenge 2010-11-14T12:57:42 *** rebelxt_ has quit IRC (Ping timeout: 265 seconds) 2010-11-14T12:57:50 is planet 0 garanteed to be in the center of the map or do I have to do math to find the center? 2010-11-14T12:58:24 ...probably safer to do the math 2010-11-14T13:03:11 *** Mekanik has quit IRC (Read error: Connection reset by peer) 2010-11-14T13:08:54 *** boegel has quit IRC (Ping timeout: 276 seconds) 2010-11-14T13:11:57 *** jaspervdj has joined #aichallenge 2010-11-14T13:17:08 Top 10 players: 2010-11-14T13:18:36 you know who's really gonna get screwed with the map changes? 2010-11-14T13:18:44 copybot 2010-11-14T13:19:13 doesn't copy bot have basicaly 0 chance? 2010-11-14T13:20:25 *** Mekanik has joined #aichallenge 2010-11-14T13:20:27 copybot is an excellent bot to find bugs 2010-11-14T13:22:06 what does copy do, if you do somethign it cna't do? delay it an extra turn or two? 2010-11-14T13:24:44 it just checks the active fleets 2010-11-14T13:38:48 copybot would do weird things 2010-11-14T13:38:58 mine probably wouldn't crash 2010-11-14T13:39:34 oh, wait. if there were an even number of planets it could crash 2010-11-14T13:42:30 *** rebelxt_ has joined #aichallenge 2010-11-14T13:50:33 ugh, of course I'm supposed to be heading out of town and now someone submits something that locks up compile daemon (which eventually leads to the whole server overloading) 2010-11-14T13:53:19 D: 2010-11-14T14:00:46 *** virdo has joined #aichallenge 2010-11-14T14:02:48 *** iris1 has quit IRC (Quit: iris1) 2010-11-14T14:04:19 *** narnach has joined #aichallenge 2010-11-14T14:05:37 doh! 2010-11-14T14:06:17 jeez the distances being non-euclidian makes my grouping algorithm produce unintuitive results... 2010-11-14T14:07:35 *** Mekanik has quit IRC (Quit: WeeChat 0.3.3) 2010-11-14T14:07:51 (due to the ceil()...) 2010-11-14T14:08:51 *** hellman has quit IRC (Remote host closed the connection) 2010-11-14T14:17:04 * lavalamp_ checks off "awesome grouping algorithm" 2010-11-14T14:17:09 Top 10 players: bocsimacko(3922), Accoun(3623), Raschi(3579), sajmon(3577), bix0r4ever(3576), george(3575), dmj111(3565), shangas(3547), asavis(3521), Xrillo(3506) 2010-11-14T14:17:16 now what to do with it... 2010-11-14T14:21:48 *** amriedle has joined #aichallenge 2010-11-14T14:21:55 *** perror has quit IRC (Quit: Bye all !) 2010-11-14T14:26:17 *** rebelxt_ has quit IRC (Quit: Page closed) 2010-11-14T14:36:40 *** boegel has joined #aichallenge 2010-11-14T14:38:43 *** McLeopold_ has quit IRC (Quit: Leaving) 2010-11-14T15:14:23 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-11-14T15:17:09 Top 10 players: bocsimacko(3907), bix0r4ever(3828), sajmon(3700), Accoun(3638), dmj111(3585), george(3578), fglider(3554), shangas(3553), asavis(3525), Raschi(3522) 2010-11-14T15:17:44 *** janzert has quit IRC (Quit: Leaving.) 2010-11-14T15:23:30 *** Accoun has quit IRC () 2010-11-14T15:24:31 *** brx_ has joined #aichallenge 2010-11-14T15:24:33 *** brx has quit IRC (Quit: Reconnecting) 2010-11-14T15:25:47 *** boegel has quit IRC (Ping timeout: 264 seconds) 2010-11-14T15:27:26 *** brx_ is now known as brx 2010-11-14T15:31:38 *** boegel has joined #aichallenge 2010-11-14T15:32:19 *** boegel has quit IRC (Remote host closed the connection) 2010-11-14T15:32:25 *** Utkarsh has joined #aichallenge 2010-11-14T15:32:34 *** smellyhippy has quit IRC (Ping timeout: 240 seconds) 2010-11-14T15:33:05 *** Calster has joined #aichallenge 2010-11-14T15:34:19 *** Accoun has joined #aichallenge 2010-11-14T15:40:03 *** Calster has quit IRC (Quit: Page closed) 2010-11-14T15:40:46 *** smellyhippy has joined #aichallenge 2010-11-14T15:43:07 *** amriedle_fail has joined #aichallenge 2010-11-14T16:00:40 howdy 2010-11-14T16:01:26 howdy how 2010-11-14T16:01:32 *** amriedle_fail is now known as amriedle2 2010-11-14T16:06:05 *** RainCT has quit IRC (Remote host closed the connection) 2010-11-14T16:08:43 *** Frontier has left #aichallenge 2010-11-14T16:16:38 *** delt0r has joined #aichallenge 2010-11-14T16:17:05 *** delt0r___ has quit IRC (Ping timeout: 240 seconds) 2010-11-14T16:17:09 Top 10 players: bix0r4ever(3910), bocsimacko(3894), Accoun(3673), Raschi(3642), george(3584), dmj111(3582), sajmon(3580), shangas(3567), asavis(3539), ClusterTeam(3519) 2010-11-14T16:27:57 *** McLeopold has joined #aichallenge 2010-11-14T16:29:57 *** jfpxtal has joined #aichallenge 2010-11-14T16:30:38 multiple start map: http://zeroviz.us:8080/canvas?game_id=2130 2010-11-14T16:34:45 not sure who a is, another multi-start: http://zeroviz.us:8080/canvas?game_id=2576 2010-11-14T16:36:15 *** McLeopold has quit IRC (Quit: Page closed) 2010-11-14T16:37:19 *** RogerAndDave has joined #aichallenge 2010-11-14T16:49:57 *** jfpxtal has quit IRC (Quit: Page closed) 2010-11-14T16:51:24 *** RogerAndDave has left #aichallenge 2010-11-14T16:58:54 *** narnach has quit IRC (Quit: Leaving.) 2010-11-14T16:59:35 *** McLeopold has joined #aichallenge 2010-11-14T17:08:40 *** choas has quit IRC (Quit: leaving) 2010-11-14T17:17:10 Top 10 players: bocsimacko(3890), bix0r4ever(3782), Accoun(3648), sajmon(3587), dmj111(3580), jambachili(3573), george(3568), shangas(3554), asavis(3541), Raschi(3538) 2010-11-14T17:18:16 *** McLeopold_ has joined #aichallenge 2010-11-14T17:20:23 *** McLeopold has quit IRC (Ping timeout: 260 seconds) 2010-11-14T17:27:49 *** TheMacca has joined #aichallenge 2010-11-14T17:40:04 *** TheMacca has left #aichallenge 2010-11-14T17:43:00 *** amriedle2 has quit IRC (Quit: leaving) 2010-11-14T17:43:09 why are the scores droping so much? 2010-11-14T17:43:26 the top10 were like 500points highesr yesterday 2010-11-14T17:49:05 maybe because the differences between bots are diminished 2010-11-14T17:49:13 the new maps are more random 2010-11-14T17:49:30 or volume is lower. 2010-11-14T17:56:30 *** sigh has joined #aichallenge 2010-11-14T17:59:52 *** TheMacca has joined #aichallenge 2010-11-14T18:01:30 *** Migi32 has joined #aichallenge 2010-11-14T18:08:38 *** Bobng has quit IRC (Quit: Leaving) 2010-11-14T18:09:32 *** sigh_ has joined #aichallenge 2010-11-14T18:09:32 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-14T18:10:46 *** sigh_ is now known as sigh 2010-11-14T18:13:41 *** Migi32 has quit IRC (Quit: www.ai-contest.com) 2010-11-14T18:15:55 *** braddunbar has quit IRC (Disconnected by services) 2010-11-14T18:17:11 Top 10 players: bocsimacko(3851), asavis(3706), bix0r4ever(3689), Raschi(3633), _Astek_(3610), Accoun(3596), sajmon(3562), shangas(3557), dmj111(3547), george(3540) 2010-11-14T18:21:58 *** TheMacca has left #aichallenge 2010-11-14T18:31:23 oldman is back :) 2010-11-14T18:33:42 lavalamp_: planet 0 is always in the centre of the map (14, 14), but that's not always in the centre of the line joining the two players 2010-11-14T18:35:24 *** sigh_ has joined #aichallenge 2010-11-14T18:35:24 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-14T18:39:02 *** janstaunton has quit IRC (Quit: janstaunton) 2010-11-14T18:39:54 sigh_: how do you like the maps? 2010-11-14T18:40:35 *** sigh_ is now known as sigh 2010-11-14T18:41:21 antimatroid: didn't have a chance to look at them, net connection was down at home 2010-11-14T18:41:30 look at any official game :) 2010-11-14T18:42:04 I'll have a look during lunch 2010-11-14T18:42:32 i accidentlaly caused oldman to drop down to like 451 2010-11-14T18:42:39 but he's back in the top 10 now, few 2010-11-14T18:42:43 phew* 2010-11-14T18:42:47 what happened? 2010-11-14T18:42:55 @rankings 2010-11-14T18:42:56 sigh: Top 10 players: bocsimacko(3853), asavis(3793), Raschi(3661), bix0r4ever(3653), Accoun(3616), sajmon(3591), oldman(3558), shangas(3547), dmj111(3538), george(3535) 2010-11-14T18:42:56 new maps 2010-11-14T18:43:06 well, he was 2010-11-14T18:43:21 oh, he's there, i'm just blind 2010-11-14T18:43:31 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-14T18:43:50 *** sigh has joined #aichallenge 2010-11-14T18:44:54 how was python? 2010-11-14T18:45:14 i prefer c++ 2010-11-14T18:45:17 hehe 2010-11-14T18:45:40 i expect haskell may grab me yet, it depends how aethetically pleasing it is :) 2010-11-14T18:45:58 aesthetically* 2010-11-14T18:47:38 are planets guarenteed to be some distance from eachother? like, more then 0.5? 2010-11-14T18:48:10 the distances will be at least 2 2010-11-14T18:48:17 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-14T18:48:21 *** sigh has joined #aichallenge 2010-11-14T18:48:31 oh, cool.. that makes stuff way easier for me then 2010-11-14T18:48:32 people didn't want distances of one, because then you don't ever see the fleet sent by an enemy 2010-11-14T18:48:44 i wanted distances of one, but i don;t think it was very important 2010-11-14T18:49:40 burny: http://ai-contest.com/forum/viewtopic.php?f=3&t=1094 2010-11-14T18:49:45 that contains accurate information about maps 2010-11-14T18:49:57 means I don't have to put special code in, to prevent loss of critical troop count on boarder stations 2010-11-14T18:50:16 well, now it is 2010-11-14T18:50:21 max dist wil be 34 2010-11-14T18:50:30 if I have 80, and opponent has 82, we can just swap back and forth until I manage to get reinforcements 2010-11-14T18:50:53 what do you mean by swap back on forth 2010-11-14T18:51:06 if you only have 2 more ships, they're going to generate that in growth by the time you get there 2010-11-14T18:51:10 when they attack me, I can send my ships to whee they came from 2010-11-14T18:52:04 obviously that has a net-loss.. but if for whatever reason it's important to preserve ships, it's better then losing the fleet 2010-11-14T18:53:56 ok.. 80 to 90 then, with a distance of 2, and both growing 1 2010-11-14T18:54:19 if they attack with 85, my best move is probably to attack back with 80 2010-11-14T18:55:24 will lose 1regen for 1 turn, but if they send the 85 back again, I can scatter my 80 through their base, and then send re-enforcements up, to regain more then just the 1grow 2010-11-14T18:56:11 burny: just be aware that you lose 1 turns worth of growth every swap 2010-11-14T18:56:23 if you are only reacting to your opponent 2010-11-14T18:56:28 yes 2010-11-14T18:56:50 but 1turn of growth can be less important than massive fleet 2010-11-14T18:56:58 sure 2010-11-14T18:57:08 *** fawek has quit IRC (Ping timeout: 245 seconds) 2010-11-14T18:57:09 it's not likely if you are winning.. 2010-11-14T18:57:12 but can be 2010-11-14T18:57:15 :) 2010-11-14T18:57:57 most of all though, if they are barely winning, and they attack. with a distance of 1 you can't respond, but distance of 2 means you can get inside their base 2010-11-14T18:58:39 i liked how distances of one didn't really provide complete information about the opponent 2010-11-14T18:58:55 complete information isn't the right word, but you know what i mean 2010-11-14T18:58:59 ya 2010-11-14T18:59:01 *** bhasker has joined #aichallenge 2010-11-14T18:59:25 the only 'problem' with distance 1, is when it's 2 planets that're on the boarder of the middle line 2010-11-14T19:07:14 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-11-14T19:10:30 *** mceier has quit IRC (Quit: leaving) 2010-11-14T19:11:48 *** Utkarsh has joined #aichallenge 2010-11-14T19:17:11 Top 10 players: bocsimacko(3853), asavis(3823), bix0r4ever(3676), Raschi(3658), Accoun(3601), sajmon(3588), oldman(3568), dmj111(3541), shangas(3538), george(3534) 2010-11-14T19:20:28 so the ranking drops: basically, the new maps are making the top100 bots, not perform as well compared to weaker bots as they used to? 2010-11-14T19:21:01 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-14T19:21:20 *** sigh has joined #aichallenge 2010-11-14T19:22:58 hmm, hard to say, probably something like that 2010-11-14T19:23:11 the top 20 usually play people inside their range 2010-11-14T19:23:19 and i don't really pay attention to the rest of the leaderboard 2010-11-14T19:23:52 well the whole top100, droped 400-500 points since yesterday(as a whole, no idea about specific bots) 2010-11-14T19:24:03 yeah i have noticed that 2010-11-14T19:24:10 bocsimacko loss rate has risen 2010-11-14T19:24:36 i wish it hadn't 2010-11-14T19:24:56 or maybe the ELO highest rating, is based heavily on the win% of the top couple people? 2010-11-14T19:25:14 i thought it was anchored at the bottom end now, but i could be wrong 2010-11-14T19:25:22 i'm not sure how that stuff works properly 2010-11-14T19:25:24 only vaguely 2010-11-14T19:27:29 do the top bots play stricly against other top bots? or just mostly against tops? 2010-11-14T19:28:21 just mostly 2010-11-14T19:28:25 the pairing algorithm favours that 2010-11-14T19:28:26 cool 2010-11-14T19:28:33 for the last contest that didn't happen, and it really sucked 2010-11-14T19:28:49 top bots didn't get anywhere near enough games with each other 2010-11-14T19:29:23 so wound up with stuff like.. bot #1 won 9/10 games, and bot#2 won 10/10 games, but was rated lower? 2010-11-14T19:29:56 http://zeroviz.us:8080/by_Map.html 2010-11-14T19:30:20 not like that I don't think, just that the ranking between 1 and 2 was more likely to be determined at how many crap bots they beat rather than how they performed against the best bots 2010-11-14T19:30:45 i got diqualified in that contest anyway, for timing out on a single move :P 2010-11-14T19:30:47 elo is anchored to 0 now, yes 2010-11-14T19:30:51 that rule has been fixed though 2010-11-14T19:31:15 McLeopold: <3 2010-11-14T19:31:47 i wonder what the stats were like for other maps 2010-11-14T19:31:58 *** narnach has joined #aichallenge 2010-11-14T19:36:13 sigh: zeroviz.us:8080 2010-11-14T19:37:09 McLeopold_: nice, thanks 2010-11-14T19:37:41 antimatroid: http://zeroviz.us:8080/mirror6.txt.html 2010-11-14T19:39:12 *** Mjothvitnir has joined #aichallenge 2010-11-14T19:40:31 yeah, those maps are cool 2010-11-14T19:43:44 *** sigh_ has joined #aichallenge 2010-11-14T19:43:44 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-14T19:54:35 *** bhasker has quit IRC (Quit: bhasker) 2010-11-14T19:56:09 *** sigh has joined #aichallenge 2010-11-14T19:56:09 *** sigh_ has quit IRC (Read error: Connection reset by peer) 2010-11-14T19:57:46 *** McLeopold_ has quit IRC (Quit: Leaving) 2010-11-14T19:58:03 *** McLeopold has joined #aichallenge 2010-11-14T20:17:00 *** bhasker has joined #aichallenge 2010-11-14T20:17:13 Top 10 players: bocsimacko(3833), asavis(3734), Raschi(3665), bix0r4ever(3625), Accoun(3598), sajmon(3563), dmj111(3544), shangas(3537), george(3530), oldman(3524) 2010-11-14T20:18:57 *** bhasker has quit IRC (Client Quit) 2010-11-14T20:19:04 *** jaspervdj has quit IRC (Quit: NEVER GONNA GIVE YOU UP NEVER GONNA LET YOU DOWN) 2010-11-14T20:22:25 *** bhasker has joined #aichallenge 2010-11-14T20:30:11 evening 2010-11-14T20:36:27 *** sigh_ has joined #aichallenge 2010-11-14T20:36:27 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-14T20:37:39 copying python objects is expensive:-\ 2010-11-14T20:38:45 *** sigh_ has quit IRC (Read error: Connection reset by peer) 2010-11-14T20:38:49 *** sigh has joined #aichallenge 2010-11-14T20:40:57 bhasker: what's up 2010-11-14T20:41:17 i have a state object that indicates the state of a planet on a particular turn 2010-11-14T20:41:26 i basically maintain a list of state of objects per planet 2010-11-14T20:41:36 rather list of stateObjects 2010-11-14T20:41:40 I had to factor out my object 2010-11-14T20:41:47 I use all lists 2010-11-14T20:41:48 so a state is like 2010-11-14T20:41:49 class PlanetState: 2010-11-14T20:41:50 def __init__(self,pID,owner,NumShips,myShips,enemyShips,origNumShips): 2010-11-14T20:41:50 self._pID = pID 2010-11-14T20:41:50 self._owner = owner 2010-11-14T20:41:50 self._ships = NumShips 2010-11-14T20:41:51 self._myShips = myShips 2010-11-14T20:41:51 self._enemyShips= enemyShips 2010-11-14T20:41:51 self._origNumShips=origNumShips 2010-11-14T20:41:51 self._heldShips = 0 2010-11-14T20:42:05 just cloning the state object is expensive 2010-11-14T20:42:07 maybe pastbin is better for this? 2010-11-14T20:42:13 *pastebin 2010-11-14T20:42:15 it wasn't too bad 2010-11-14T20:42:43 but yes, do pastebin 2010-11-14T20:43:10 bhasker: zeroviz:8080 2010-11-14T20:43:31 I want to see malazan on the list 2010-11-14T20:43:36 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-14T20:43:54 I'm tired of playing accoun 2010-11-14T20:43:59 *** sigh has joined #aichallenge 2010-11-14T20:44:04 and whoever a is 2010-11-14T20:52:04 oh man... accoun owned bocsimacko http://ai-contest.com/visualizer.php?game_id=7340132 2010-11-14T20:52:20 antimatroid: there... oldman's 9th now 2010-11-14T20:57:51 *** sigh_ has joined #aichallenge 2010-11-14T20:57:51 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-14T20:59:50 ouch, for mega1 2010-11-14T21:00:05 amstan: http://zeroviz.us:8080/canvas?game_id=3474 2010-11-14T21:01:04 and: http://zeroviz.us:8080/by_Player.html 2010-11-14T21:04:11 Boscimacko seems to be losing a lot more with these new maps 2010-11-14T21:04:25 *** davidjliu has quit IRC (Quit: Page closed) 2010-11-14T21:05:28 who's a? 2010-11-14T21:07:18 Me 2010-11-14T21:07:27 why "a"? 2010-11-14T21:07:37 Why not? 2010-11-14T21:07:46 because it's very undecscriptive 2010-11-14T21:07:51 undescriptive* 2010-11-14T21:07:56 I don't mean to be descriptive 2010-11-14T21:09:20 *** phreeza_ has joined #aichallenge 2010-11-14T21:09:20 Are you playing on that server ATM? 2010-11-14T21:09:31 no, i'm an organizer 2010-11-14T21:09:54 Yeah, I knew that, but I figured you might be making a bot anyway 2010-11-14T21:10:04 nah, it's kinda spoiled for me 2010-11-14T21:10:36 How come? 2010-11-14T21:10:47 well 2010-11-14T21:10:57 i can't submit stuff to the main server 2010-11-14T21:11:11 other servers are not as easy 2010-11-14T21:11:16 so i'll die fast 2010-11-14T21:11:23 *** phreeza has quit IRC (Ping timeout: 264 seconds) 2010-11-14T21:11:37 I see 2010-11-14T21:13:56 sorry had to step away 2010-11-14T21:13:58 something came up 2010-11-14T21:14:11 let me start up my submission bot 2010-11-14T21:14:20 on the new maps 2010-11-14T21:14:26 which server is it running on 2010-11-14T21:17:14 Top 10 players: bocsimacko(3817), Raschi(3739), asavis(3678), bix0r4ever(3593), Accoun(3580), sajmon(3561), dmj111(3549), shangas(3541), george(3522), oldman(3495) 2010-11-14T21:17:16 zeroviz.us:8080 is the website 2010-11-14T21:17:52 The battle server is on port 995 2010-11-14T21:18:15 ip is 76.121.152.123 2010-11-14T21:18:54 thanks 2010-11-14T21:24:47 I'll put up some cannon fodder 2010-11-14T21:25:27 http://zeroviz.us:8080/canvas?game_id=3609 2010-11-14T21:25:28 thats a weird map 2010-11-14T21:25:38 how come there is only one map with 17 ships on it? 2010-11-14T21:26:44 McLeopold: love the nice ratings grid but it sorts the time columns alphanumerically >.< 2010-11-14T21:27:01 woops 2010-11-14T21:27:15 the time since last, or the UTC time? 2010-11-14T21:28:34 *** amstan has quit IRC (Read error: Connection reset by peer) 2010-11-14T21:28:36 bhasker: do you mean planet? 2010-11-14T21:28:45 oops yes 2010-11-14T21:28:47 that plaent in on the line of symmetry 2010-11-14T21:29:10 *that planet is on 2010-11-14T21:29:10 hmm planet 0? 2010-11-14T21:29:46 maybe 2010-11-14T21:29:48 the 50 is a single too 2010-11-14T21:29:48 mcleopold wtf http://zeroviz.us:8080/canvas?game_id=3635 ? 2010-11-14T21:29:55 yea weird map 2010-11-14T21:30:09 *** amstan has joined #aichallenge 2010-11-14T21:30:09 *** ChanServ sets mode: +o amstan 2010-11-14T21:30:24 why was your bot sending ships all over? 2010-11-14T21:30:24 um, official bot, really old 2010-11-14T21:30:39 McLeopold: last 2010-11-14T21:31:09 well that was worth it, this map crashed me: http://zeroviz.us:8080/canvas?game_id=3626 2010-11-14T21:31:10 *** sigh_ has quit IRC (Read error: Connection reset by peer) 2010-11-14T21:31:21 oh, yea, I'll fix that as I get off the c code 2010-11-14T21:31:21 *** sigh has joined #aichallenge 2010-11-14T21:31:51 amstan: yeah saw that, you can undrestand why i'd worry when he's ranked that high though :P 2010-11-14T21:31:59 ah, good ol' "run for the hills" map 2010-11-14T21:32:24 wasn't allocating big enough array in one place 2010-11-14T21:33:03 crashed me too i think lavalamp 2010-11-14T21:33:28 bhasker: is your bot loving or hating the new maps? 2010-11-14T21:33:37 its doing okay till now 2010-11-14T21:33:42 http://zeroviz.us:8080/getplayer?player=malazanbot 2010-11-14T21:34:09 going to see how it does 2010-11-14T21:34:22 looks like currently playing AntBot 2010-11-14T21:34:26 just won 2010-11-14T21:34:41 antbot seems quite decent so that's good :) 2010-11-14T21:34:59 *** sigh_ has joined #aichallenge 2010-11-14T21:35:05 interesting http://zeroviz.us:8080/canvas?game_id=3656 2010-11-14T21:35:09 he never sniped me 2010-11-14T21:35:11 i took the 8 2010-11-14T21:35:13 with 9 ships 2010-11-14T21:35:21 right next to his biggest planet 2010-11-14T21:35:25 should have techincally lost that game 2010-11-14T21:35:52 i'm converned about how well bots are now doing against mega1 compared to before 2010-11-14T21:35:57 concerned* 2010-11-14T21:36:05 other than that i've seen no real different 2010-11-14T21:36:12 and that really wasn't my intentions 2010-11-14T21:36:26 hmm if they are doing better why would that be a concern? 2010-11-14T21:36:30 I'm sure it affected other people too 2010-11-14T21:36:41 because he has been clearly better than everyone else for a long time 2010-11-14T21:36:43 probably just different maps change his symmetry based assumptions? 2010-11-14T21:36:53 he has said he doesn't make such assumptions 2010-11-14T21:36:58 he could be lying, but i don't see why 2010-11-14T21:37:24 are the new win/loss ratios much different to what he was getting on tcp? 2010-11-14T21:37:25 Even if he doesn't make assumptions, some constants in his code might be tuned towards the old maps 2010-11-14T21:37:29 he's been tuning his eval function for a month, it's extremely likely it was overfitted 2010-11-14T21:37:47 because the main server was stuck with the original 100 maps till a day ago 2010-11-14T21:37:59 is the bot "a" you antimatroid? 2010-11-14T21:38:05 no, that's flag 2010-11-14T21:38:12 i'm in the middle of a rewrite :P 2010-11-14T21:38:15 aah 2010-11-14T21:38:19 well, i hope i'm more than halfway now 2010-11-14T21:38:21 I'm in the middle of a rewrite D: 2010-11-14T21:38:48 flag: i started with a fresh starter to try and clean things up for final submissions 2010-11-14T21:38:54 the strategy is only minorly different 2010-11-14T21:39:09 i guess i will need to debug a bit more 2010-11-14T21:39:11 getting weird timeouts 2010-11-14T21:39:12 http://zeroviz.us:8080/canvas?game_id=3673 2010-11-14T21:39:27 I'm trying to go for boscimacko-like dominance with my new bot. I give it about a 1% chance of working out 2010-11-14T21:39:40 that's a high probability :P 2010-11-14T21:39:46 haha 2010-11-14T21:39:51 bhasker: have you got timers stopping you if you're taking too long? 2010-11-14T21:39:56 that map has more than 23 planets 2010-11-14T21:39:59 nope 2010-11-14T21:40:07 my average turn takes under 200ms 2010-11-14T21:40:11 so no need for timers per se 2010-11-14T21:40:21 the most time i spend is in updating state thats about it 2010-11-14T21:40:26 actual decision making barely takes anytime 2010-11-14T21:40:31 fair enough 2010-11-14T21:40:40 i use a decision tree, which will often just take as long as it can 2010-11-14T21:41:23 yea well i don't optimize my decisions 2010-11-14T21:41:41 my new tactic: monte carlo to optimize my goal execution, payoff table to select best course of action ^^ 2010-11-14T21:41:45 my pickMoves function is soo inefficient :P 2010-11-14T21:41:48 *** sigh_ has quit IRC (Quit: Leaving) 2010-11-14T21:42:08 i gave up on modelling the other player properly 2010-11-14T21:42:20 conflict resolution is too hard to do without it just being insanely inefficient 2010-11-14T21:42:36 I may too if I can't speed this up some XP 2010-11-14T21:43:08 i would of had to process 2 queues for every state just to ensure everything was correct 2010-11-14T21:43:16 perhaps someone will see a better way than me 2010-11-14T21:43:34 or do well with almost correct calculations 2010-11-14T21:45:35 *** Zoost has quit IRC (Ping timeout: 264 seconds) 2010-11-14T21:46:31 *** Zoost has joined #aichallenge 2010-11-14T21:46:35 heh account needs to fix his start code 2010-11-14T21:46:36 http://zeroviz.us:8080/canvas?game_id=3695 2010-11-14T21:48:33 heh nice win here http://zeroviz.us:8080/canvas?game_id=3690 2010-11-14T21:49:01 *** sigh_ has joined #aichallenge 2010-11-14T21:49:06 nice 2010-11-14T21:49:09 yeah, his loss ratio on the main site for new maps is greater than it was even for tcp previously 2010-11-14T21:49:18 let me know what you think of the mirror and radial maps 2010-11-14T21:49:23 it's at 90% wins now, but that includes a good day and a half of old map usage 2010-11-14T21:49:49 antimatroid: i created the radial maps without sin or cos or atan 2010-11-14T21:49:54 let me see how it goes 2010-11-14T21:51:07 http://zeroviz.us:8080/canvas?game_id=3721 2010-11-14T21:51:49 http://zeroviz.us:8080/canvas?game_id=3643 2010-11-14T21:51:57 McLeopold: http://zeroviz.us:8080/by_Map.html a link to that from the main page would be nice 2010-11-14T21:52:13 it is, refresh? 2010-11-14T21:52:29 ah, :P 2010-11-14T21:53:42 using that, you can sort by draws and see the games that cause them 2010-11-14T21:54:50 good thing they forced us to be single threaded or I wouldn't be runnng six bots concurrently... 2010-11-14T21:55:39 i've never used multiple threads :\ 2010-11-14T21:57:05 also, didn't mcleopold want people not to run so many instances at once? 2010-11-14T21:57:21 eek now I'm confused about which server I'm running things on 2010-11-14T21:57:24 It's on my comcast account. 2010-11-14T21:57:25 something about only having a comcast account 2010-11-14T21:57:32 It's not so bad right now. 2010-11-14T21:57:36 oh I can stop some 2010-11-14T21:57:45 ok let me know if I'm hogging your bandwidth 2010-11-14T21:57:46 yeah, would be fine now, once more people start using it though 2010-11-14T21:58:41 lnteresting games overall 2010-11-14T21:59:30 yeah, i'm glad oldman is almost fixed 2010-11-14T21:59:34 he still crashes some time 2010-11-14T21:59:52 my only worry is the change in bocsimackos performance 2010-11-14T22:00:26 it could just be more rps like situations 2010-11-14T22:00:33 it is 2010-11-14T22:01:02 what is? 2010-11-14T22:01:11 it'd be a big call to say that's the only reason :P 2010-11-14T22:02:52 he beat bix0r again :) 2010-11-14T22:03:38 with radial, planets on the center line are mirrored, so if you opponent goes left, you can usually go right 2010-11-14T22:03:52 where the hell did sajmon come from? 2010-11-14T22:03:54 with 0-5 on the line that are unique, it becomes more difficult 2010-11-14T22:03:54 *** sigh_ has quit IRC (Quit: Leaving) 2010-11-14T22:03:59 @rankings 2010-11-14T22:04:05 McLeopold: Top 10 players: bocsimacko(3816), Raschi(3739), asavis(3723), bix0r4ever(3594), sajmon(3582), Accoun(3564), dmj111(3550), shangas(3544), george(3518), _iouri_(3504) 2010-11-14T22:04:14 *** sigh_ has joined #aichallenge 2010-11-14T22:04:32 i haven't seen their name in the top 10 before 2010-11-14T22:05:19 *** phreeza_ has quit IRC (Quit: Computer has gone to sleep) 2010-11-14T22:06:52 their bots pretty good 2010-11-14T22:08:23 *** sigh has quit IRC () 2010-11-14T22:08:34 *** sigh_ is now known as sigh 2010-11-14T22:12:43 McLeopold: http://ai-contest.com/visualizer.php?game_id=7349477 2010-11-14T22:12:49 what would you do first move? 2010-11-14T22:12:53 i think that's a draw map 2010-11-14T22:13:52 *** TheMacca has joined #aichallenge 2010-11-14T22:13:54 *** tapwater_ has joined #aichallenge 2010-11-14T22:15:31 blah, I installed ie8 today, and still can't see those 2010-11-14T22:16:02 see what? 2010-11-14T22:16:16 antimatroid: http://zeroviz.us:8080/map_r528_7.txt.html 2010-11-14T22:16:19 replays 2010-11-14T22:16:30 no draws on my tcp 2010-11-14T22:16:50 here's what I would do: http://zeroviz.us:8080/canvas?game_id=203 2010-11-14T22:17:01 ah, okay 2010-11-14T22:17:08 *** tapwater has quit IRC (Ping timeout: 245 seconds) 2010-11-14T22:17:08 *** tapwater_ is now known as tapwater 2010-11-14T22:17:14 Top 10 players: bocsimacko(3835), Raschi(3755), asavis(3662), sajmon(3606), protocolocon(3581), bix0r4ever(3576), dmj111(3552), Accoun(3544), shangas(3537), george(3524) 2010-11-14T22:18:24 burny: ie8 doesn't have html canvas? 2010-11-14T22:22:19 no clue, what's html canvas? 2010-11-14T22:22:30 Try firefox, opera or chrome 2010-11-14T22:22:48 An HTML5 feature not supported by outdated browsers such as ie8 :) 2010-11-14T22:24:19 oi 2010-11-14T22:34:15 McLeopold: nice work on the web interface :) 2010-11-14T22:40:26 thanks 2010-11-14T22:40:40 dmj111: we need to switch out the main server piece to python next 2010-11-14T22:41:49 oh hey I beat a real bot: http://zeroviz.us:8080/canvas?game_id=3929 2010-11-14T22:43:18 hehe I won one against you, too, McLeopold: http://zeroviz.us:8080/canvas?game_id=3812 2010-11-14T22:44:47 McLeopold: yeah..... maybe after the 27th :) It would be great to have something ready switch new games into for the next contest. 2010-11-14T22:45:31 i'm not going to know what to do with myself after the 27th 2010-11-14T22:45:44 i guess i might finally buy a copy of minecraft 2010-11-14T22:45:47 that ought to fix me up 2010-11-14T22:46:51 haha 2010-11-14T22:46:59 make a redstone bot :P 2010-11-14T22:48:57 phew, centrelink sent me a letter today saying i'd been cut off from youth allowance cause of age 2010-11-14T22:49:08 but it was only because their records had me as no longer a student 2010-11-14T23:01:07 *** yellowseed has joined #aichallenge 2010-11-14T23:02:46 *** p4p4 has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2010-11-14T23:05:42 so back to my original question about copying objects being expensive in python 2010-11-14T23:06:21 mcleopold still around? 2010-11-14T23:06:27 y 2010-11-14T23:07:22 http://pastebin.com/PP3XPwAK 2010-11-14T23:07:25 thats the main code 2010-11-14T23:07:30 so planetstate just stores the state 2010-11-14T23:07:40 and updateplanet basically runs through handling all fleet resolutions 2010-11-14T23:07:45 and updating the state at each turn 2010-11-14T23:07:52 it appends a new state object to the planet state list 2010-11-14T23:07:54 at every turn 2010-11-14T23:08:12 so if you see like 33->51 2010-11-14T23:08:17 lines i.e 2010-11-14T23:08:27 for most planets its basically in absence of leets 2010-11-14T23:08:29 fleets* 2010-11-14T23:08:34 creates a new state from its previous one 2010-11-14T23:08:45 http://ai-contest.com/visualizer.php?game_id=7356348 2010-11-14T23:08:47 this takes over 9ms for all 23 planets 2010-11-14T23:08:57 i can't seem to make it any faster 2010-11-14T23:08:58 :-\ 2010-11-14T23:09:03 wat 2010-11-14T23:09:06 now i'm really worried 2010-11-14T23:09:13 23 copies takes 9ms? 2010-11-14T23:09:19 no for all 23 planets 2010-11-14T23:09:25 updating states upto 2010-11-14T23:09:27 maxDist*2+1 2010-11-14T23:09:41 so for an average map 2010-11-14T23:09:42 ill have to think about it later 2010-11-14T23:09:49 about 49-50 state copies 2010-11-14T23:09:51 * 23 2010-11-14T23:09:54 dmj might help? 2010-11-14T23:09:58 bocsimacko got beaten by someone in the 600's on the main site!!! 2010-11-14T23:10:03 that's very worrying 2010-11-14T23:10:21 the main site is running your new maps antimatroid? 2010-11-14T23:10:29 yes 2010-11-14T23:10:37 nice 2010-11-14T23:10:41 since when? 2010-11-14T23:10:43 bhasker: that code made me feel a little better about my state keeping code ;) 2010-11-14T23:10:47 over a day ago 2010-11-14T23:10:49 :D 2010-11-14T23:10:52 my code is messy :-\ 2010-11-14T23:11:01 and my python skills aren't the best 2010-11-14T23:11:03 i'm not very happy with what's happening to bocsimacko 2010-11-14T23:11:10 i tend to code python like C at times:-\ 2010-11-14T23:11:16 did the 600 expand agressively, and the other didn't capitalize on it, and it was a draw map? 2010-11-14T23:11:39 bhasker: so the rate is roughly 128 copies/sec? 2010-11-14T23:11:40 why does it matter if mega1's bot does bad? 2010-11-14T23:11:44 burny: http://ai-contest.com/visualizer.php?game_id=7356348 2010-11-14T23:11:44 that's crazy slow 2010-11-14T23:11:54 yea thats what i figure that its slow:-\ 2010-11-14T23:11:54 I can't see replays ;( 2010-11-14T23:11:55 mine is just about as messy but I have it more general (no cases 1-8 or whatever) 2010-11-14T23:11:56 bhasker: because it's clearly been better than everyone elses 2010-11-14T23:12:00 hmm wait it takes 10ms 2010-11-14T23:12:03 to copy that many 2010-11-14T23:12:06 wow 2010-11-14T23:12:16 about 115/sec 2010-11-14T23:12:31 yea 2010-11-14T23:12:38 that's crazy 2010-11-14T23:12:38 so i figured its quite crappy 2010-11-14T23:12:43 let me do a simple test hold on 2010-11-14T23:13:09 what is the algorithmic complexity of list append in python? 2010-11-14T23:13:19 and how many elements are in these lists? 2010-11-14T23:13:20 mega1: i need you to stop doing badly on the new maps :P you're making me reconsider my stance 2010-11-14T23:14:02 antimatroid: you've only made the contest more difficult, not worse 2010-11-14T23:14:06 what was the change exactly? 2010-11-14T23:14:21 it's more difficult yes, but i didn't want to knock the leader off the top 2010-11-14T23:14:21 but ya, I agree probably with mcleo 2010-11-14T23:14:32 @rankings 2010-11-14T23:14:32 well jmc the list shouldn't be longer than 40-50 elements 2010-11-14T23:14:33 McLeopold: Top 10 players: bocsimacko(3822), Raschi(3772), asavis(3614), sajmon(3574), protocolocon(3562), bix0r4ever(3560), SUA.khb(3558), dmj111(3541), shangas(3541), george(3533) 2010-11-14T23:14:36 he's not yet 2010-11-14T23:14:39 since its just one state object per turn 2010-11-14T23:14:43 but it's certainly changed a lot for him 2010-11-14T23:15:14 and i calculate upto 2*maxDistance+1 states 2010-11-14T23:15:23 i do + 2 2010-11-14T23:15:23 where maxDist is the maximum distance between two planets 2010-11-14T23:15:37 bhasker: I wait to calc states until someone requests it 2010-11-14T23:15:45 that might help your copies :/ 2010-11-14T23:15:52 not a bad idea 2010-11-14T23:16:03 though 10ms for that many copies is slow 2010-11-14T23:16:04 as hell 2010-11-14T23:16:06 i need to figure out why 2010-11-14T23:16:16 ...and if it's that slow to copy maybe you should not, and just recalc all the time 2010-11-14T23:16:27 cause i can't do any kind of complex stuff if just copying state is so expensive 2010-11-14T23:16:57 i doubt it's actually the expensive. something sounds wrong 2010-11-14T23:16:58 well i was doing that, but doing recalcs every time makes the code really unweildy 2010-11-14T23:17:01 yea 2010-11-14T23:17:06 i need to figure out what i am doing wrong 2010-11-14T23:17:14 Top 10 players: bocsimacko(3822), Raschi(3772), asavis(3614), sajmon(3574), protocolocon(3562), bix0r4ever(3560), SUA.khb(3558), dmj111(3541), shangas(3541), george(3533) 2010-11-14T23:18:58 anim: unless the map changes makes some naive strat win too often: it should be fine 2010-11-14T23:19:44 eg. if they're all large enough, that agressive expansion is the best.. or there's always a planet close to middle that has low population, that sending 100% of fleet to that first move, forces a win, or a draw 2010-11-14T23:19:44 etc. 2010-11-14T23:20:33 ah, nice catch from dimkadimon on the map generator 2010-11-14T23:21:14 what was it? 2010-11-14T23:22:12 num_ships = random.randint(minShips, 5*distance(p1, p2)-1) should be num_ships = min(100, random.randint(minShips, 5*distance(p1, p2)-1)) when selecting the neutrals that can be claimed 2010-11-14T23:22:25 not that it really matters, but it ensures maps meet the specifications 2010-11-14T23:23:34 so was just missing a cap of 100 neutrals? 2010-11-14T23:24:34 what's cap on growthRate? 2010-11-14T23:25:01 5 2010-11-14T23:25:06 and yes, the cap 2010-11-14T23:25:16 max distance is currently 24 and will be 34 2010-11-14T23:25:43 burny: http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/map_generator.py 2010-11-14T23:25:52 that's the map generator if you want to make your own test maps 2010-11-14T23:28:20 any other python programmers around? 2010-11-14T23:28:37 ^^ that was my first piece of python code :P 2010-11-14T23:28:46 http://pastebin.com/ELyVmMhV 2010-11-14T23:28:48 can you run that 2010-11-14T23:29:00 and send what the output is like 2010-11-14T23:29:34 i can't figure out why its printing like 23*100 times the time 2010-11-14T23:29:40 bhasker: i was just reading through your messages, but haven't lookked at your pastes yet. 2010-11-14T23:29:50 ah k 2010-11-14T23:30:08 is the last one the one to look at? 2010-11-14T23:30:17 did it not used to have center planets? 2010-11-14T23:30:29 hmm well that is similar to what my code does 2010-11-14T23:30:32 but here's the main one 2010-11-14T23:30:38 C:\>file.py 2010-11-14T23:30:39 File "C:\file.py", line 1 2010-11-14T23:30:39 SyntaxError: Non-ASCII character '\xff' in file C:\file.py on line 1, but no enc 2010-11-14T23:30:39 oding declared; see http://www.python.org/peps/pep-0263.html for details 2010-11-14T23:30:48 http://pastebin.com/PP3XPwAK 2010-11-14T23:31:16 hmm is that what you are getting antimatroid? 2010-11-14T23:32:00 i just ran the one file? 2010-11-14T23:32:04 was i meant to get the other one too 2010-11-14T23:32:08 hmm no 2010-11-14T23:32:13 weird never got that encoding error before 2010-11-14T23:32:14 i have no idea about python :P 2010-11-14T23:32:22 i could have just been retarded somehow 2010-11-14T23:32:59 do people assess the best moves globally or per planet? 2010-11-14T23:33:06 I'm assessing per planet and it's ... limiting :) 2010-11-14T23:33:15 anti: what changed with it? added symmetry type!? 2010-11-14T23:33:20 type1 I meant 2010-11-14T23:34:33 bhasker: my planet state is just a tuple. (turn, owner, num_ships). And turn is just there for debugging. 2010-11-14T23:34:58 burny: differeing number of planets, two different types of symmetry and multiple "middle" neutrals 2010-11-14T23:35:04 your held ships is probably a hint at something i _should_ do, but don't. 2010-11-14T23:35:12 hmm 2010-11-14T23:35:25 heh held ships was something i stole from lavalamp 2010-11-14T23:35:30 but i can't get it to work reliably 2010-11-14T23:35:39 i have "usableShips" 2010-11-14T23:35:41 i end up sometimes launching less than required 2010-11-14T23:35:42 did it used to never have 'middle's? 2010-11-14T23:35:55 and is either of the symmetry type's the original one? 2010-11-14T23:35:57 it always had a dead centre one, but otherwise no 2010-11-14T23:36:09 yes, old ones were point symmetric 2010-11-14T23:36:09 themacca i am not even assessing best move 2010-11-14T23:36:17 my best move, is go for the best planet 2010-11-14T23:36:21 if i can't get it 2010-11-14T23:36:24 then try next one 2010-11-14T23:36:25 thats aboutit 2010-11-14T23:36:34 which of the 2 is the new one? 1 or not 1? 2010-11-14T23:36:39 bhasker: Cool 2010-11-14T23:36:50 are you making that decision on a per-planet (origin) basis 2010-11-14T23:36:55 yes 2010-11-14T23:36:57 or overall 2010-11-14T23:36:59 okay 2010-11-14T23:36:59 but i do something like 2010-11-14T23:37:03 bhasker: how do you know when to stop? 2010-11-14T23:37:11 so basically at start of turn 2010-11-14T23:37:14 i calculate how many ships i ahve 2010-11-14T23:37:19 then from each source i calculate 2010-11-14T23:37:23 how many ships i would need 2010-11-14T23:37:24 to take it over 2010-11-14T23:37:30 and keep track of how many i have sent before 2010-11-14T23:37:31 burny: -1 is the new one 2010-11-14T23:37:44 Okay, that's about what I do 2010-11-14T23:37:48 and also keep track of how many enemy ships the previous launch accounted for 2010-11-14T23:37:54 (in fact, I think it's exactly what I do) 2010-11-14T23:37:57 so i subtract that many from enemy reinforcements 2010-11-14T23:38:06 ah, didn't think of that 2010-11-14T23:38:07 nice :) 2010-11-14T23:38:17 so that way i was not double counting 2010-11-14T23:38:21 bhasker: reason I ask, is that I see some of the higher ranked bots 2010-11-14T23:38:33 and they use multiple planets to attack a single destination 2010-11-14T23:38:38 oh my bot does that 2010-11-14T23:38:43 which I think has a number of advantages 2010-11-14T23:38:48 yes it does 2010-11-14T23:38:51 oh, I misunderstood then 2010-11-14T23:38:53 yeah, that is quite common 2010-11-14T23:38:56 also my bot does synchronized attacks 2010-11-14T23:39:00 which is also something most bots do 2010-11-14T23:39:11 I only do sychonized attacks against neutrals 2010-11-14T23:39:17 sigh: why? 2010-11-14T23:39:17 i launch synchronized against enemy too 2010-11-14T23:39:22 for some reason my bot does better with it for normal enemies 2010-11-14T23:39:32 only reinforcements are not synchronized 2010-11-14T23:39:38 just the result of tests, no other reason 2010-11-14T23:40:01 * does better with sychronised attacked off for normal enemies 2010-11-14T23:40:04 * lavalamp_ skims conversation 2010-11-14T23:40:20 bhasker: yeah held ships are freaking hard to get right 2010-11-14T23:40:38 what do you mean by held ships? delayed attacks? 2010-11-14T23:40:38 i kind of loop through all my targets, ordered but some hack, and attack the first one that looks feasible. If I have more time, i do it again, but with my state updated with the launched fleets. my attacks use whatever i think is available. 2010-11-14T23:41:02 i took it as "reserved for defense against possible attack." that is what I haven't come close to getting right. 2010-11-14T23:41:10 dmj111: do you use multiple sources fora single target? 2010-11-14T23:41:14 ships at a planet/turn that are reserved for some later use but not dispatched from the planet on that turn 2010-11-14T23:41:16 yes. 2010-11-14T23:41:17 on a single turn 2010-11-14T23:41:44 my overall strategy is to figure out how to attack each target independently, score the planets based on that and then start attacking from highest score down 2010-11-14T23:41:58 still yes. I launch what I can, but also track launches from future turns from multiple planets, if needed. 2010-11-14T23:42:18 how do you guys decide when to stop attacking? 2010-11-14T23:42:30 I have an ROI threshold 2010-11-14T23:42:41 stop attackign when you win 2010-11-14T23:42:50 burny: I mean on a single turn 2010-11-14T23:43:11 do you only consider the ships you can send now? 2010-11-14T23:43:22 i just added a 2 ply search, that doesn't launch attacks that the enemy can to better against than if I did nothing. 2010-11-14T23:43:27 i let my bot send ships that i'm "going to have" at any point in time up to 2*maxd+1 2010-11-14T23:43:31 I just go down my list of targets sending attacks until I find one I can't attack then stop 2010-11-14T23:43:48 problem is, that I only consider a small set of "reasonable" attacks, and don't consider enemy moves that I should. 2010-11-14T23:43:50 sigh, I used to do that 2010-11-14T23:44:20 yeah, I understand that it's very simplistic :) 2010-11-14T23:44:20 sigh: If I find one I cannot attack, I keep walking the list to see if there is somethign else to take. 2010-11-14T23:44:48 early on I had bots that gave up attacking, when there was a less desirable, but still good target out there. 2010-11-14T23:44:52 dmj111: I've tried that, my bot spread itself too thin and attacked stupid stuff 2010-11-14T23:44:56 dmj111: sigh's asking, how do you know when to stop looking? 2010-11-14T23:45:00 because you wind up attacking crap 2010-11-14T23:45:17 sigh: I evaluate all enemy first, then, when I'm scanning the neutrals 2010-11-14T23:45:25 if I find one below some threshold that I can't attack 2010-11-14T23:45:26 I stop 2010-11-14T23:45:42 sigh: yeah. same problem. mostly attacking across the whole board, or taking a neutral when I am already too weak. 2010-11-14T23:46:06 interesting... I just treat enemy and neutral all in one go 2010-11-14T23:46:14 macca i order my list something like this 2010-11-14T23:46:19 MyPlanets 2010-11-14T23:46:21 EnemyPlanets 2010-11-14T23:46:29 almost the same for the two 2010-11-14T23:46:31 followed by 2010-11-14T23:46:36 Neutrals under attack by me or enemy 2010-11-14T23:46:40 followed by neutrals 2010-11-14T23:46:41 mega strategy share session :) 2010-11-14T23:46:50 heh 2010-11-14T23:47:02 antimatroid: lies, mega is not here :( 2010-11-14T23:47:08 hehe 2010-11-14T23:47:10 bhasker: our stuff sounds pretty similar 2010-11-14T23:47:16 and i stop attacking a particular planet when i have sent enough ships to account for a) enemy growth he can send + enemy ships he can send next turn to that planet 2010-11-14T23:47:17 which is weird, coz you keep besting me ;) 2010-11-14T23:47:23 before my attack from a particular planet can reach it 2010-11-14T23:47:46 actually i overestimate how much i can send to 2010-11-14T23:48:02 bocsimacko's bot really doesn't like that mpa 2010-11-14T23:48:03 the attack is usually enough to take the planet i attacked and neighbouring planets 2010-11-14T23:48:05 map* 2010-11-14T23:48:08 http://ai-contest.com/visualizer.php?game_id=7360934 2010-11-14T23:48:15 the reason for sending a larger than required fleet 2010-11-14T23:48:23 was i would send ships and they would just take the planet and sit 2010-11-14T23:48:32 even though they could technically hit all 3planets nearby 2010-11-14T23:48:36 and capture them 2010-11-14T23:48:52 the reaosn being it would decide the # of ships it can send without losing the planet is zero 2010-11-14T23:48:57 so i send over the # required 2010-11-14T23:48:57 bhasker: I do something similar 2010-11-14T23:49:08 so that it will attack immediately 2010-11-14T23:49:11 neighbouring planets 2010-11-14T23:49:21 also at end of turn i do some other stuff 2010-11-14T23:49:34 which is, i recalculate which planets will be mine given the current launches in progress 2010-11-14T23:49:39 and i launch any ships that i didn't send 2010-11-14T23:49:46 towwards those if they are closer to the enemy 2010-11-14T23:50:02 heh - wow 2010-11-14T23:50:11 so my extra ships follow 2010-11-14T23:50:13 strikingly similar 2010-11-14T23:50:13 my launches 2010-11-14T23:50:21 sometimes that can backfire 2010-11-14T23:50:31 so i have a lot of checks when not to launch 2010-11-14T23:50:44 like if a planet has an enemy planet closer than the target then it won't forward 2010-11-14T23:50:48 etc 2010-11-14T23:50:56 so not all planets will launch the extra ships 2010-11-14T23:53:07 also i have some checks on when to overlaunch for neutrals 2010-11-14T23:53:21 so i calculate if the enemy has planets nearer i adjust for how much enemy could potentially send 2010-11-14T23:53:34 so for some neutrals sometimes i would launch with 2x/3x the ships 2010-11-14T23:54:03 I handle that the same way for neutrals as enemy 2010-11-14T23:54:12 walk through projected futures 2010-11-14T23:54:24 and assume enemy will reinforce / snipe 2010-11-14T23:54:28 if they can reach in time 2010-11-14T23:54:37 the problem with treating neutrals as same as enemy was i would end up not attacking planets when i could have countered a snipe by enemy 2010-11-14T23:54:39 (plus some slop - a couple turns or something) 2010-11-14T23:54:47 because i would overestimate the # of ships enemy can send 2010-11-14T23:54:48 I just avoid sending to neutrals which can be sniped 2010-11-14T23:54:51 and decide i can't send 2010-11-14T23:55:08 so i do the overestimation only if attacks from both me and enemy are not in progress 2010-11-14T23:55:09 et 2010-11-14T23:55:14 once the attacks are happening 2010-11-14T23:55:19 i don't overestimate 2010-11-14T23:55:35 only the initial launch usually will over launch 2010-11-14T23:55:36 to neutrals 2010-11-14T23:56:11 bhasker: sorry to change the subject, but I am heading off to sleep... i can paste my state update code if you want to see it. I am afraid it won't help much with your timing problems since your code handles a lot more than mine does. 2010-11-14T23:56:27 *** TheMacca has left #aichallenge 2010-11-14T23:56:27 hmm k thanks 2010-11-14T23:56:35 i will spend some time optimizing the code 2010-11-14T23:56:41 mostly because i want to be able to make moves undo them 2010-11-14T23:56:45 and evaluate the position 2010-11-14T23:56:50 to do that multiple times per turn 2010-11-14T23:56:54 i need a fast game state updating 2010-11-14T23:56:59 updation* 2010-11-14T23:57:10 the current code is too slow:-\ 2010-11-14T23:57:18 yeah... I am doing that by copying the starting state, and just recalculating based on the new moves. 2010-11-14T23:57:36 i tried that, but i could never get the code to work correctly 2010-11-14T23:57:39 when i recalculate stuff 2010-11-14T23:57:43 and debugging it was a nightmare 2010-11-14T23:57:49 the state arrays are easy to debug 2010-11-14T23:57:53 i can just dump all the state 2010-11-14T23:57:58 and see if i had a bug 2010-11-14T23:58:00 or something 2010-11-14T23:58:38 macca http://ai-contest.com/visualizer.php?game_id=7354000 2010-11-14T23:58:42 oh nm he left 2010-11-14T23:58:46 if you are just refactoring it is easy because you can compare against a working implementation 2010-11-14T23:59:30 yea i guess i could but i just find it easier to do this 2010-11-14T23:59:34 since i can just lookup the state 2010-11-14T23:59:38 easily at any turn