2010-11-28T00:00:35 i'm about to test one last parameter change ;-) maybe a miracle happens 2010-11-28T00:04:34 i still like my bot because it looks absolutely disorganized but still manages to win some games. 2010-11-28T00:04:48 reflects my character :D 2010-11-28T00:05:11 mine miracally is ranking 1400ish and I crapped it together in like 12 hours 2010-11-28T00:05:14 all hardcoded random logic 2010-11-28T00:05:30 basically just nukes the closest thing to each planet if its growth is >= 3 2010-11-28T00:07:06 haha, Celica has a nice fireworks mode triggered a few turns before game loss 2010-11-28T00:07:23 http://ai-contest.com/visualizer.php?game_id=8822046 - last few turns are fun to watch.. 2010-11-28T00:08:12 heh, nice 2010-11-28T00:08:18 heheh 2010-11-28T00:08:29 "crapped it together" = newest addition to my phrase library 2010-11-28T00:08:39 "launch the escape pods, initiate self destrcuct sequence" 2010-11-28T00:09:14 hehe 2010-11-28T00:09:15 authorization Picard, Alpha-Alpha-3-0-5 2010-11-28T00:09:57 *** Terr__ has joined #aichallenge 2010-11-28T00:13:30 gah, wish you could upload more than one bot ;). Can not decide which version of mine works best D: 2010-11-28T00:13:52 put them in a cagematch with each other 2010-11-28T00:13:54 the survivor is 2010-11-28T00:14:14 yeah, but it's no good for testing, when they're both pretty much the same. 2010-11-28T00:14:35 *** winexception has quit IRC (Read error: Connection reset by peer) 2010-11-28T00:15:25 bit late for that anyway 2010-11-28T00:15:32 Kingpin, unless you had a chance to test both bots on tcp servers, pick one that dispatches example bots better/quicker 2010-11-28T00:15:51 lets ask for an extension lol 2010-11-28T00:16:02 Phreeza, haha, it's never too late :P. 2010-11-28T00:16:17 i would mind. really need my life back asap! 2010-11-28T00:16:20 j3camero: extension pretty please? :D 2010-11-28T00:16:35 rebelxt_, thanks for the suggestion, I'll try that, wish there were more example bots though 2010-11-28T00:17:13 whats the pairing cutoff? 2010-11-28T00:17:18 example bots beat mine occasionally -_- 2010-11-28T00:18:04 phreeza_: for later in the finals as we limit the tournament to the top N bots 2010-11-28T00:18:28 ah ok 2010-11-28T00:19:57 Top 10 players: asavis(3387), bocsimacko(3379), _Astek_(3362), Slin-.-(3351), _iouri_(3302), dmj111(3237), bix0r4ever(3235), Doraj8(3217), Fill_the_Missing(3203), thedreamer(3200) 2010-11-28T00:20:15 *** flag has quit IRC (Quit: Lost terminal) 2010-11-28T00:20:48 looking forward to seeing how fast elo scores stabilize once a cutoff is in place. was nice getting a rock solid position on tcp servers but it used to take 200+ games normally.. 2010-11-28T00:21:12 wow this guy submitted about 3000 times 2010-11-28T00:21:13 http://ai-contest.com/profile_submissions.php?user_id=10464&page=1 2010-11-28T00:21:25 Raschi 2010-11-28T00:22:28 wow.. was he trying to inflate ELO by re-submitting upon a loss? 2010-11-28T00:23:01 i don't think so 2010-11-28T00:23:23 yeah, I think he was just resubmitting every change 2010-11-28T00:23:24 my guess is he had a hook set up with his git 2010-11-28T00:23:49 phreeza_: ahh interesting theory. a bad idea though ;-) 2010-11-28T00:24:05 looks like he worked on it more or less constantly from Oct 3rd 2010-11-28T00:24:12 Kingpin13: Regarding multiple bots... In the suggestion thread in the forum there's an idea for that 2010-11-28T00:24:19 i guess he deserves 11th then :) 2010-11-28T00:24:42 he's updating like every 10 min o_O 2010-11-28T00:24:44 Basically, allow 2-3 concurrent bots per user, with one of them always marked "primary" and the only one which will make it into the final competition if they walk away 2010-11-28T00:24:55 Allows limited comparison testing 2010-11-28T00:25:19 Terr__, yeah sounds like a good idea, would have been ncie to have that :( 2010-11-28T00:25:47 Terr__: a good idea, but tcp servers would still be a much better tool for comparison testing 2010-11-28T00:27:38 *** delt0r has joined #aichallenge 2010-11-28T00:28:57 *** delt0r___ has quit IRC (Ping timeout: 276 seconds) 2010-11-28T00:30:57 right, i'm uploaded 2010-11-28T00:30:58 *** winexception has joined #aichallenge 2010-11-28T00:31:14 i think the new tree idea was better but i decided against uploading it, should have tried it earlier 2010-11-28T00:32:30 You know what would make the next contest better? 2010-11-28T00:32:32 Two Rounds. 2010-11-28T00:32:36 i'm off to sleep :) good luck everyone 2010-11-28T00:32:40 why two rounds? 2010-11-28T00:32:44 antimatroid1: so was it heuristic bot you uploaded? 2010-11-28T00:32:44 First is optional, but bragging rights anyway 2010-11-28T00:32:51 revelxt_: minimax 2010-11-28T00:32:55 antimatroid1: Procrastinators! 2010-11-28T00:33:18 why not three rounds? for the uber procrastinaros? 2010-11-28T00:33:20 ad infinitum 2010-11-28T00:33:38 antimatroid1: cool. two many pure heuristic bots out there ;) 2010-11-28T00:34:53 i wonder how many generations it would take a GP bot with no knowledge of the game to manage to play correctly to 200 turns 2010-11-28T00:35:03 my strategy is basically as follows, 1. cancel out any of their offensive moves that I can, 2. use minimax tree to pick my other main moves (don't get more than like 2-3 of those turns per move, the new idea gave more), 3. pick redistribution, done 2010-11-28T00:35:10 *** winexception has quit IRC (Ping timeout: 240 seconds) 2010-11-28T00:35:54 i may even try that next time 2010-11-28T00:36:07 its all about blindly spamming ships to their doom for me 2010-11-28T00:36:28 *slams head on desk as he sees where he reversed a heuristic calculation* 2010-11-28T00:36:47 Terr__: you've got 20 minutes! 2010-11-28T00:36:57 fix it 2010-11-28T00:38:30 antimatroid1: how deep can it search? 2010-11-28T00:38:44 not deep at all 2010-11-28T00:38:49 we've hit 14 submissions in the compiler backlog now :/ 2010-11-28T00:38:52 like 4-6 layers max i'd say 2010-11-28T00:39:08 janzert: i uplaoded a few minutes ago and it's already compiled 2010-11-28T00:39:15 Zannick: what do you mean my "no knowledge"? 2010-11-28T00:39:22 yep, it's still storming through them 2010-11-28T00:40:08 sigh: i mean start the generations with bots that only know to output a series of "%d %d %d\n" until a final line of "go\n", each turn 2010-11-28T00:40:27 ok, I see 2010-11-28T00:40:32 GA doesn't sound like it'd work very well to me 2010-11-28T00:40:57 GP is crazy hilarious 2010-11-28T00:41:02 it's a GA that makes programs 2010-11-28T00:41:14 so it might take awhile 2010-11-28T00:41:18 if i did do it, i certainly would be trying to program some heuristics in 2010-11-28T00:41:31 but then again i don't plan to seriously compete in the next one, so 2010-11-28T00:41:54 do we know what the next one is> 2010-11-28T00:41:57 ? 2010-11-28T00:41:58 *** phreeza_ has quit IRC (Quit: Computer has gone to sleep) 2010-11-28T00:42:09 we haven't decided 2010-11-28T00:42:37 I don't know if I'll compete... it takes so much time 2010-11-28T00:42:37 lol that is not promising, my final upload lost it's first nonofficial game to someone ranked in the 200's 2010-11-28T00:42:42 still interesting to watch everyone else 2010-11-28T00:46:35 looks like submission rate is slowing down. I wonder if we'll get another surge at the deadline. 2010-11-28T00:47:04 janzert: hopefully not ;-) 2010-11-28T00:47:33 yeah, I hope too but I'm not sure that aligns with human nature :) 2010-11-28T00:47:39 haha 2010-11-28T00:49:10 bah, cutting it fine, i reuploaded the new oen 2010-11-28T00:49:18 lets see how that goes for a gamble 2010-11-28T00:50:27 so... 2010-11-28T00:50:30 how's everyone? 2010-11-28T00:50:33 @rankings 2010-11-28T00:50:34 amstan: Top 10 players: Slin-.-(3392), bocsimacko(3377), _iouri_(3327), _Astek_(3317), asavis(3303), bix0r4ever(3263), Doraj8(3232), george(3220), dmj111(3219), Fill_the_Missing(3218) 2010-11-28T00:51:49 *** Sylph has joined #aichallenge 2010-11-28T00:51:52 *** roginn has joined #aichallenge 2010-11-28T00:51:59 amstan: great! let's get the party started 2010-11-28T00:52:49 antimatroid1: you are a gambler 2010-11-28T00:53:12 ya, let's do this 2010-11-28T00:53:14 hmmm i have no idea how this will go 2010-11-28T00:53:15 let's see 2010-11-28T00:53:45 http://www.youtube.com/watch?v=wvsboPUjrGc 2010-11-28T00:54:03 ARRRGHGHG 2010-11-28T00:54:06 aha 2010-11-28T00:54:08 10 MINUTES BETWEEN SUBMITS 2010-11-28T00:54:10 o_O 2010-11-28T00:54:14 AARRRGH BROKEN COPY ARRGH 2010-11-28T00:54:21 janzert: oh yeah... good point 2010-11-28T00:54:22 I never hit the submit limit before... 2010-11-28T00:54:26 janzert: can we disable that? 2010-11-28T00:54:30 And now I have to wait until after the contest. 2010-11-28T00:54:33 Huh. 2010-11-28T00:54:41 ooh, yeah that should get turned off 2010-11-28T00:54:49 hmm? 2010-11-28T00:54:51 if you didn't compile that shouldn't have triggered? 2010-11-28T00:54:54 janzert: disable upload limit 2010-11-28T00:55:03 Oh, it's screwed up, but not in compilation 2010-11-28T00:55:04 janzert: maybe leave submissions open 10 minutes extra? 2010-11-28T00:55:07 that'd sort it 2010-11-28T00:55:08 antimatroid1: can compile but still be broken 2010-11-28T00:55:42 nah, we can just take out the limit 2010-11-28T00:55:59 it's not really applicable now 2010-11-28T00:56:19 amstan: I'm afraid of the backlog growing to nasty sizes 2010-11-28T00:56:21 Don't worry on my account: My bot sucks either way 2010-11-28T00:56:25 Just...blarg. 2010-11-28T00:56:28 janzert: make it one then 2010-11-28T00:56:34 1m 2010-11-28T00:56:42 which will cause people to keep resubmitting since they didn't get a response 2010-11-28T00:56:56 2 MINUTES 2010-11-28T00:56:59 !!! 2010-11-28T00:57:13 2! minutes! 2010-11-28T00:57:17 @countdown 2010-11-28T00:57:18 amstan: User error, it's not my fault. 2010-11-28T00:57:20 aww 2010-11-28T00:57:24 :P 2010-11-28T00:57:28 like 2010-11-28T00:57:33 that would have been epic 2010-11-28T00:57:35 lol 2010-11-28T00:57:36 the cs class i TAed in college wouldn't count a hw as a day late until 15 minutes after 2010-11-28T00:57:51 If there's a problem, can't you just default to the last working version? Would that work? 2010-11-28T00:57:53 ok, 2 minutes now 2010-11-28T00:57:53 mega1: you aren't uploading again? 2010-11-28T00:57:57 yeah.. that too 2010-11-28T00:57:59 are we going to see a 2000-way tie for first place after the first rankings update? 2010-11-28T00:58:02 vim submit.php 2010-11-28T00:58:03 @seen mega1 2010-11-28T00:58:03 amstan: mega1 was last seen in #aichallenge 13 hours, 11 minutes, and 39 seconds ago: can't be too soon 2010-11-28T00:58:04 *** sanddbox has joined #aichallenge 2010-11-28T00:58:05 oops 2010-11-28T00:58:10 50 seconds till launch 2010-11-28T00:58:15 i think mega's sleeping 2010-11-28T00:58:25 i guess he's happy with it as is, just checking 2010-11-28T00:58:27 39 2010-11-28T00:58:28 so that's kind of what i recommend, but without telling anyone >.> 2010-11-28T00:58:29 30* 2010-11-28T00:58:33 mega's clearly a robot 2010-11-28T00:58:38 ;-) 2010-11-28T00:58:43 15! 2010-11-28T00:58:47 10! 2010-11-28T00:58:48 20s... :o 2010-11-28T00:58:48 10! 2010-11-28T00:58:53 10 even ;) 2010-11-28T00:58:53 5 2010-11-28T00:58:54 4 2010-11-28T00:58:54 oh god! 2010-11-28T00:58:54 3 2010-11-28T00:58:55 4 2010-11-28T00:58:56 2 2010-11-28T00:58:56 3 2010-11-28T00:58:57 1 2010-11-28T00:58:58 2 2010-11-28T00:58:58 0 2010-11-28T00:58:59 1 2010-11-28T00:59:00 :D 2010-11-28T00:59:01 woo 2010-11-28T00:59:02 @predict 2010-11-28T00:59:02 I predict that jmcarthur will win! 2010-11-28T00:59:04 :P 2010-11-28T00:59:04 happy new year! 2010-11-28T00:59:09 oh wait 2010-11-28T00:59:10 heh 2010-11-28T00:59:13 haha 2010-11-28T00:59:14 Well, we'll see if what got in last fixed the no-op bug :( 2010-11-28T00:59:16 happy 10pm! 2010-11-28T00:59:18 so much for jeff making it back in time :P 2010-11-28T00:59:24 I think there's one thing we can ALL agree on... 2010-11-28T00:59:26 CLOSURE 2010-11-28T00:59:26 (in the time zone i get to fly back to tomorrow) 2010-11-28T00:59:28 :) 2010-11-28T00:59:29 janzert: where did he go? 2010-11-28T00:59:31 backlog size 10 at the deadline 2010-11-28T00:59:34 ABILITY TO SLEEP RESTORED 2010-11-28T00:59:37 amstan: read your emails 2010-11-28T00:59:48 antimatroid1: i have emails? lol 2010-11-28T00:59:57 amstan: shrug left about an hour ago said he'd be back by one 2010-11-28T01:00:20 sites down? 2010-11-28T01:00:26 janzert: weird.. 2010-11-28T01:00:31 antimatroid1: shouldn't be 2010-11-28T01:00:33 everybody refreshed 2010-11-28T01:00:40 Quick idea to plant for participants: Merging code and methods to create the UBERBOT. (Or more likely, the spaghetticodebot.) 2010-11-28T01:00:40 too many connections for mysql 2010-11-28T01:00:42 lol 2010-11-28T01:00:48 need better server <.< 2010-11-28T01:00:54 yeah, lol 2010-11-28T01:00:54 no kidding 2010-11-28T01:00:55 yeah, it's also doing it's on the hour stuff so wait a minute or two 2010-11-28T01:00:57 edit the max connections in config 2010-11-28T01:01:01 Terr__: too much work/wouldn't work 2010-11-28T01:01:05 everyone, relax and stop refreshing ;-)) 2010-11-28T01:01:06 k, it's back 2010-11-28T01:01:24 janzert: you tricky man: If you haven't submitted your final bot do so now! 2010-11-28T01:01:28 antimatroid1: I just feel bad, looking at all this code and knowing it'll be ophaned. 2010-11-28T01:01:57 i'm gonna open-source my terrible thing early dec. 2010-11-28T01:02:09 :) gotta see what's happening with the compilation backlog here 2010-11-28T01:02:22 janzert: of man, everyone's going to the forums now 2010-11-28T01:02:33 Now all I need to do is never ever reveal my bot name to avoid public shame. 2010-11-28T01:02:35 "Submissions will be closing shortly." janzert you are just too nice! ;-) 2010-11-28T01:03:17 i miss my mod powers 2010-11-28T01:03:26 there's a thread marked with unread posts 2010-11-28T01:03:34 but i go there and there aren't any! 2010-11-28T01:03:48 Terr__; all of my programs end up orphaned :P 2010-11-28T01:03:59 Zannick: I get that too though 2010-11-28T01:04:08 Yeah 2010-11-28T01:04:13 the unread stuff isn't quite working right 2010-11-28T01:04:21 Several threads keep bugging me that way. 2010-11-28T01:04:31 my mod powers disappeared for some reason, and then i decided i wanted a new account with this name instead of the other one 2010-11-28T01:04:32 i hope to god my new bot was bug free 2010-11-28T01:04:41 i literally through that different game tree together today 2010-11-28T01:04:43 so it goes 2010-11-28T01:04:47 Zannick: you're from google, right? 2010-11-28T01:04:52 antimatroid1: I know from the devil mine isn't ;) 2010-11-28T01:05:50 Something keeps screwing up my evaluation so that it likes high-pop low-growth distant neutral planets ;) 2010-11-28T01:05:54 amstan: aye 2010-11-28T01:06:02 Zannick: k, you can have mods back 2010-11-28T01:06:06 hahaha 2010-11-28T01:06:14 what a great reason ;) 2010-11-28T01:06:17 lol 2010-11-28T01:06:30 i don't think i left anyone a mod 2010-11-28T01:06:40 Nothing like a last minute untested code change submit. 2010-11-28T01:06:43 Well, stuff like that anyway. At this point a "fresh start rewrite" looks pretty good. 2010-11-28T01:06:47 nvm, i guess jmcarthur and a1k0n_ were 2010-11-28T01:06:58 McLeopold: that's my style 2010-11-28T01:07:05 i mean, it's okay with me if i get mod powers for being a googler, or for being a contributor, or for planning to be an admin/contributor in the future 2010-11-28T01:07:09 And mine, I just did it. 2010-11-28T01:07:10 3 months of work and a last minute huge change, hopefully it pays off aha 2010-11-28T01:07:17 My "ragebot-esque random lunatic" version would've probably performed better because it'd get all the folks who didn't code anti-rage-ish behavior. 2010-11-28T01:07:20 2 minutes past the dealine I think... 2010-11-28T01:07:27 decent effort :) 2010-11-28T01:07:38 feeling as unconfident as me? :P 2010-11-28T01:07:49 Zannick: i think you were mod before i did that behind the scenes stuff, so you should have been left as a mod 2010-11-28T01:07:56 Did anybody end up using "crazy math" as a method? 2010-11-28T01:07:59 It turns out, I suck at AI. 2010-11-28T01:08:00 i think the first thing i coded in my bot was anti-ragebot 2010-11-28T01:08:01 Massive matrix transformations etc? 2010-11-28T01:08:02 you take rage bot, but make it rage at entirely random planets 2010-11-28T01:08:03 thats mine 2010-11-28T01:08:20 Ken69267: Platonic anger, Mr. Tulip? 2010-11-28T01:08:21 amstan: yeah, i was. but i'm on a new account "Zannick" now so as not to confuse people 2010-11-28T01:08:31 what was the old one? 2010-11-28T01:08:36 Actually, I think I just tuned by bot to play well on only certain map types. 2010-11-28T01:08:44 amsta: bswolf 2010-11-28T01:08:45 amstan: bswolf 2010-11-28T01:08:50 :p 2010-11-28T01:08:53 ha, someone else remembers 2010-11-28T01:08:53 Stalemate: http://ai-contest.com/visualizer.php?game_id=8849319 2010-11-28T01:08:55 i'm not really sure how one would do any kind of matrix transformations 2010-11-28T01:08:57 Zannick: right.. yeah.. you were a mod 2010-11-28T01:09:02 medal of honour* 2010-11-28T01:09:08 most of my crazy methods were too slow 2010-11-28T01:09:14 linear programming was the slowest i tried 2010-11-28T01:09:28 red should send 24 at th 23 2010-11-28T01:09:30 McLeopold: bot building on the edge! i decided against last minute change 2010-11-28T01:09:45 CryBaby: i lol'd 2010-11-28T01:09:49 chaotic could have easily won 2010-11-28T01:10:23 chaotic should have just taken over that 23 neutral 2010-11-28T01:10:26 janzert: need any help? 2010-11-28T01:10:36 submissions have finally worked the the final surge backlog 2010-11-28T01:10:41 I'm pleased to say my hybrid knapsacky narrowing-tree minimax future-steps-modeling crazy soft reference caching abomination at least runs. 2010-11-28T01:10:46 amstan: not yet anyway 2010-11-28T01:11:24 You just described my bot. 2010-11-28T01:11:28 If the code is a waste, I might try eeking out some value by posting a "Here's how it was supposed to work" thing on a blog. 2010-11-28T01:11:38 McLeopold: 'Cept yours WORKS, right? ;) 2010-11-28T01:11:39 amstan: cool, thanks 2010-11-28T01:11:53 It plays a game, yes. 2010-11-28T01:12:03 It also times out on the main servers. :( 2010-11-28T01:12:24 doh 2010-11-28T01:12:28 :( 2010-11-28T01:12:30 ouch 2010-11-28T01:12:44 *** jaman4dbz has joined #aichallenge 2010-11-28T01:12:54 My bot never connected to the TCP server nor was uploaded prior to yesterday: I worked for too long in a vacuum. 2010-11-28T01:12:57 too bad mega's sleeping 2010-11-28T01:13:01 Like way too long. 2010-11-28T01:13:05 he'll wake up in an hour or so 2010-11-28T01:13:12 I lost a game I was winning to someone who was rank 22 earlier today. 2010-11-28T01:13:15 Architecture astronautics dooms another of my projects. 2010-11-28T01:13:45 Terr__: i did that for tron 2010-11-28T01:13:51 was alright for that, not for this :P 2010-11-28T01:14:06 didn't wait till the very last day, but worked for most of the time locally 2010-11-28T01:14:28 now that this contest is over, is minecraft really really worth checking it out? lol 2010-11-28T01:14:36 haha, that's my plan 2010-11-28T01:14:37 yes! 2010-11-28T01:14:38 apparently yes 2010-11-28T01:14:38 haha 2010-11-28T01:14:41 Submissions are now really completely closed 2010-11-28T01:14:47 :) 2010-11-28T01:14:48 no! stay away, less you become victim to it's addiction. 2010-11-28T01:14:49 antimatroid1: I imagine it was because it was less "fuzzy" in terms of reactions 2010-11-28T01:14:54 and all submissions have compiled 2010-11-28T01:15:00 antimatroid1: Easier to code in a unit test when the "right" action is much easier to verify 2010-11-28T01:15:01 @topic delete 6 2010-11-28T01:15:01 amstan: (topic []) -- Returns the topic for . is only necessary if the message isn't sent in the channel itself. 2010-11-28T01:15:04 great, downloading minecraft then ;) 2010-11-28T01:15:05 @topic remove 6 2010-11-28T01:15:06 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask. || http://tinyurl.com/aichallengefinale || Ignore PMs from borderland, his links are of a malicious nature!" 2010-11-28T01:15:08 @topic remove 5 2010-11-28T01:15:09 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask. || http://tinyurl.com/aichallengefinale" 2010-11-28T01:15:10 @topic remove 4 2010-11-28T01:15:10 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask." 2010-11-28T01:15:19 Terr__: yeah, in tron you knew and checked all moves they could make, in this you're making a hell of a lot of guesses 2010-11-28T01:15:27 "Ignore PMs from borderland, his links are of a malicious nature!" 2010-11-28T01:15:29 i lol'd 2010-11-28T01:15:30 rebelxt_: Think of it like legos, and remember it's an alpha 2010-11-28T01:15:39 that's what makes this game way better than tron 2010-11-28T01:15:44 @topic add Contest is now closed, Results Wednesday 2010-11-28T01:15:44 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask. || Contest is now closed, Results Wednesday" 2010-11-28T01:15:57 burny: agreed 2010-11-28T01:16:17 Does anyone have any really good resources that explain the practical uses of linear algebra, eigenvectors, etc? 2010-11-28T01:16:23 "Sorry, submissions are now closed. Good luck in the finals." /cry 2010-11-28T01:16:28 When I go to wikipedia all the math symbols put me to sleep. 2010-11-28T01:16:34 :( i love maths 2010-11-28T01:16:38 haha, it has uses in graphics. 2010-11-28T01:16:43 pseudocode > math geek symbols 2010-11-28T01:16:46 @topic add Go play Minecraft! 2010-11-28T01:16:46 *** contestbot changes topic to "Official Google AI Challenge: http://ai-contest.com/ || Channel Logs: http://contestbot.hypertriangle.com/ || No stupid questions. Don't ask to ask. || Contest is now closed, Results Wednesday || Go play Minecraft!" 2010-11-28T01:16:47 oh and eigenvectors are very useful in graph theory 2010-11-28T01:16:51 it has uses in everything! 2010-11-28T01:16:54 one can never know too much linear algebra, true fact 2010-11-28T01:17:02 i agree with this statement 2010-11-28T01:17:04 lol at topic 2010-11-28T01:17:07 jaman4dbz: Yeah, and I've got a shitload of graphs in my bot :( 2010-11-28T01:17:11 calculus on the other hand... ewwwwww 2010-11-28T01:17:20 ehh that sucks I was trying to make a last minute edit and the file i sent didn't compile 2010-11-28T01:17:22 calculus of variations was fun 2010-11-28T01:17:29 awww 2010-11-28T01:17:31 Snowplan: it'll default to the previous one 2010-11-28T01:17:38 Snowplan: if you had a compile error 2010-11-28T01:17:40 google isn't doing very well :P 2010-11-28T01:17:47 jaman4dbz: I took calculus... didn't like it, but I get what it does. All the set/combinatorics/etc. is far more daunting to me 2010-11-28T01:17:51 Terr__: were you creating subgraphs? 2010-11-28T01:17:58 i considered trying something like that, but meh 2010-11-28T01:18:09 Id like it to be 3 because the last one crashed 2010-11-28T01:18:11 haha graphs and subgraphs... You didn't really need any of that. 2010-11-28T01:18:11 antimatroid1: OK, so all planets were a heavily connected graph (no X/Y positions) 2010-11-28T01:18:15 o well 2010-11-28T01:18:16 *** exaide has joined #aichallenge 2010-11-28T01:18:21 Snowplan: you want too much now 2010-11-28T01:18:24 antimatroid1: And I have a "timeline" class which is basically a mesh of ndoes 2010-11-28T01:18:26 yeah, so why even store a graph? 2010-11-28T01:18:26 lol 2010-11-28T01:18:34 *** roginn has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100423140709]) 2010-11-28T01:18:36 it's completely connected, i just stored distances at the start 2010-11-28T01:18:45 antimatroid1: Distances are in a table, yeah 2010-11-28T01:19:02 antimatroid1: Each grid-node was a time/space position, like Planet X turn Y 2010-11-28T01:19:11 *** Snowplan has left #aichallenge 2010-11-28T01:19:18 yeah, mine was futureStates[planet][time] 2010-11-28T01:19:25 I'm curious, did anyone else just transform the backpack problem for choosing planets to attack? 2010-11-28T01:19:26 antimatroid1: Any node combined with a Set of fleet-moves could be used to calculate the next, etc. 2010-11-28T01:19:37 *transform = reduce... ya know what i mean 2010-11-28T01:19:52 antimatroid1: States of a node were stored in a Map,Outcome> 2010-11-28T01:19:58 Top 10 players: Slin-.-(3416), asavis(3369), bocsimacko(3349), xpctxpeer(3322), CorwinAlex(3313), _iouri_(3279), bix0r4ever(3268), _Astek_(3245), dmj111(3225), george(3218) 2010-11-28T01:20:00 when are old games being deleted? 2010-11-28T01:20:09 antimatroid1: And at each stage, the fleet-set was pruned so that "unimportant" fleets were not used 2010-11-28T01:20:23 jaman4dbz: there was a post about it in the forum early in the competion, so I imagine many people tried it 2010-11-28T01:20:35 makes sense 2010-11-28T01:20:45 antimatroid1: There were a few other tricks, but I think I made a very effective way to say: "What would everything look like if I did this" 2010-11-28T01:20:57 antimatroid1: The problem is my code doesn't ask good questions ;) 2010-11-28T01:20:59 *** medrimonia has quit IRC (Ping timeout: 272 seconds) 2010-11-28T01:20:59 *** Neverstu has joined #aichallenge 2010-11-28T01:21:21 my problem was I used signed ints and turned them into unsigned ints 'cause im smart. 2010-11-28T01:21:24 i have a function to update the future states of a planet from a starting time 2010-11-28T01:21:29 *** Mjothvitnir has joined #aichallenge 2010-11-28T01:21:38 then i store a vector of earliest times that the state information has changed 2010-11-28T01:21:46 then when i update i just run through the states that need changing 2010-11-28T01:21:55 antimatroid1: But what about modeling "speculative" moves? 2010-11-28T01:22:14 game tree 2010-11-28T01:22:24 I think my max future modeling ended up being 2x the max travel distance 2010-11-28T01:22:36 i just throw moves down each branch, copy state, add moves, update, then continue 2010-11-28T01:22:46 i just did up to maxdistance in the end 2010-11-28T01:22:57 i originally did up to 2*maxdistance 2010-11-28T01:23:08 *** medrimonia has joined #aichallenge 2010-11-28T01:23:14 Well, I had a minimax-ish thing that had depth 1x, and then at each step it would proceed by 1x assuming no changes in order to calculate how "happy" it was. 2010-11-28T01:23:14 but i figured the games changed so much by then anyway, why even bother 2010-11-28T01:23:52 Ironically when I started, I said: "Steer, don't plan!" and ended up making a planning system. Discipline fail. 2010-11-28T01:24:23 hehe 2010-11-28T01:24:36 you can't control the code! It does what it wants! 2010-11-28T01:25:01 man I'm so disappointed... I had so many plans... and just got stuck on infinite bugs... I fail at debugging. 2010-11-28T01:25:18 i just got a new game? 2010-11-28T01:25:19 "Maybe being ignorant of the field will let me do great because I don't know what 'can not be done'!" --> Nope. 2010-11-28T01:25:21 I murdered a small tree to write down all of my algorithms... that tree died for nothing! 2010-11-28T01:25:31 janzert: shouldn't games have stopped running? 2010-11-28T01:25:34 ah dang... lol 2010-11-28T01:25:35 infinite bugs :( were they at least countable? 2010-11-28T01:25:38 jaman4dbz: I have all these scribbled "timelines" of fleet movements 2010-11-28T01:25:42 Graph paper FTW 2010-11-28T01:26:04 Terr__: haha awesome 2010-11-28T01:26:06 X axis : planets, Y axis: time. Variously-angled lines connecting points along planet-lines. 2010-11-28T01:26:09 fyi i fail at IRC 2010-11-28T01:26:43 Not exactly a "parallel graph", but it looked very similar 2010-11-28T01:26:44 mine were just ideas into ensuring all of my planets were defended... and then ideas to tricky enemys who would do very robotic things like going for the perfect move. 2010-11-28T01:26:50 ha, just in time i made first for australia at closing of submissions aha 2010-11-28T01:26:54 I really wasn't thinking very systematically =P 2010-11-28T01:27:08 nice! Congrats dude. 2010-11-28T01:27:19 Terr__: it sounds like we did a lot of things the same 2010-11-28T01:27:21 Ex: http://filer.case.edu/~dbh10/eecs466/pGraph.png 2010-11-28T01:27:27 McLeopold used excel 2010-11-28T01:27:29 I was like 36th in Canada... not quiet what i was hoping. 2010-11-28T01:27:31 i really should have too, but oh wel 2010-11-28T01:27:44 wow 2010-11-28T01:27:53 It looks like that 2010-11-28T01:28:19 oh maybe not 2010-11-28T01:28:34 i considered trying discrete 3d graphs like that 2010-11-28T01:28:38 but never tried it 2010-11-28T01:28:50 the final open period rankings aren't quite finished running yet 2010-11-28T01:28:58 did you guys test against alternate/legacy versions of your bots? i did that for a while but found the data to be almost useless. winning against myself does not mean anything as it turns out. 2010-11-28T01:29:07 antimatroid1: the Y axis was time (one turn per unit) while the X axis had no real measurement, all "travel times" were expressed by the slope of the connecting lines 2010-11-28T01:29:14 I used that alot neverstu 2010-11-28T01:29:17 that was all i had =p 2010-11-28T01:29:25 how else would I know to even bother submitting. 2010-11-28T01:29:49 good night all 2010-11-28T01:29:51 I tried so hard to convince my friends to join this competition, but they are SOOOO lazy. 2010-11-28T01:29:56 see how things look in the mornings... 2010-11-28T01:30:03 night random guy i never said hello to. 2010-11-28T01:30:09 heh 2010-11-28T01:30:11 i kept it up for major regression checking, but generally then i had to re-create official matches or hit the TCP server to get reasonable data 2010-11-28T01:30:12 My god... I CAN PLAY VIDEO GAMES AGAIN! 2010-11-28T01:30:18 *** davidjliu has quit IRC (Quit: Page closed) 2010-11-28T01:30:19 night 2010-11-28T01:30:22 (Not as good as someone talking about their signfiicant other, but still!) 2010-11-28T01:30:30 Niters. 2010-11-28T01:30:45 haha ya Terr__, now i can finally fix my gaming desktop, my server, reinstall ubuntu, redo my website, and maybe start applying for jobs xD 2010-11-28T01:30:55 *** Rubicon-|-Cross has joined #aichallenge 2010-11-28T01:31:01 i can do many things now 2010-11-28T01:31:05 i don't even know where to start 2010-11-28T01:31:08 Neverstu: i only used old versions for regression testing. tcp servers gave much better data but still noisy 2010-11-28T01:31:15 i've been enslaved for three months 2010-11-28T01:31:17 jaman4dbz: Note down some things from the bot in case you get an interview question about "hard problems and how you solved them' 2010-11-28T01:31:28 *** Corg has joined #aichallenge 2010-11-28T01:31:40 Now I can go finish my accounting test which is due soon... 2010-11-28T01:31:48 @Terr__ thanks for the advice 2010-11-28T01:31:49 jaman4dbz: You have no gotten any error messages recently, so here's a random one just to let you know that we care. 2010-11-28T01:31:51 what are you doing accounting for? :P 2010-11-28T01:31:53 antimatroid1: I've not been unable to do other things, but I *have* been feeling guilt and apprehension that sucked all joy from them. 2010-11-28T01:32:00 I'm going to try everything in IRC untill i figure out how to send a private message, lol. 2010-11-28T01:32:10 *** Appleman1234 has quit IRC (Ping timeout: 250 seconds) 2010-11-28T01:32:17 /msg 2010-11-28T01:32:18 jaman4dbz: /msg user messgae 2010-11-28T01:32:48 jaman4dbz: if you wanna adress a line to someone while still talking inside a channel, you previs your message with "nickname:" 2010-11-28T01:32:54 rebelxt_: yep, lots of noise from all sources unfortunately 2010-11-28T01:33:10 amstan: address* 2010-11-28T01:33:14 @amstan: got it. 2010-11-28T01:33:15 jaman4dbz: No! 2010-11-28T01:33:21 haha 2010-11-28T01:33:21 =( 2010-11-28T01:33:27 as much as i enjoyed playing 42 versions of Accoun's bot... 2010-11-28T01:33:28 jaman4dbz: ideally without the @ in front, 2010-11-28T01:33:31 contestbt is very eager 2010-11-28T01:33:33 @contestbot is weird 2010-11-28T01:33:34 Terr__: Run as fast as you can and don't look back. 2010-11-28T01:33:36 amstan: ah... 2010-11-28T01:33:43 jaman4dbz: just use tab complete 2010-11-28T01:33:45 Neverstu: haha yeah accoun and iouri had a lot of bots on tcp 2010-11-28T01:33:46 makes it easy to highlight 2010-11-28T01:33:46 @it doesn't recognize itself? 2010-11-28T01:33:47 Terr__: User error -- Replace user. 2010-11-28T01:33:49 contestbot doesn't need the @ if you are talking to it, Terr__ 2010-11-28T01:33:50 sigh: Run as fast as you can and don't look back. 2010-11-28T01:34:04 o.o 2010-11-28T01:34:04 ^^ see 2010-11-28T01:34:05 sigh: It shouldn't be reacting to every @-prefixed word though 2010-11-28T01:34:13 @why not 2010-11-28T01:34:14 antimatroid1: Run as fast as you can and don't look back. 2010-11-28T01:34:14 Well, every first-of-sentence @prefix 2010-11-28T01:34:18 I mean, wtf? 2010-11-28T01:34:25 @@@ 2010-11-28T01:34:25 it shouldn't but it does 2010-11-28T01:34:25 antimatroid1: User error -- Replace user. 2010-11-28T01:34:27 yo dog 2010-11-28T01:34:33 damn 2010-11-28T01:34:35 Now THAT'S an AI project 2010-11-28T01:34:35 i don't see why it shouldn't 2010-11-28T01:34:40 Fixing up cnotestbot 2010-11-28T01:34:57 ya contestbot seems rigid... he needs to socialize more. 2010-11-28T01:35:03 amstan: Vat iz zee *purpose* of ze bot respondink to every amperzand? 2010-11-28T01:35:06 contestbot: sigh is awesome, right? 2010-11-28T01:35:06 i know... 2010-11-28T01:35:23 contestbot is turning incomplete 2010-11-28T01:35:24 McLeopold: I have no idea what you mean. 2010-11-28T01:35:24 my feeder algorithm was bolted on in a very bad way...did anyone have feeders/supply lines emerge from offense naturally? 2010-11-28T01:35:26 contestbot: Are you lying? 2010-11-28T01:35:27 Terr__: An error occured while trying to show the previous error. 2010-11-28T01:35:30 :) 2010-11-28T01:35:30 i already have an ircbot that works, though i know supybot has been around for a lot longer 2010-11-28T01:35:37 i was hoping for that, but i never found a good formulation 2010-11-28T01:36:02 Neverstu: i sort of did 2010-11-28T01:36:04 Neverstu: My attempts to make "PlanetRank" kept hitting a wall. 2010-11-28T01:36:07 Neverstu: I have an idea for that, but I never got around to it. 2010-11-28T01:36:23 Terr__, planetrank? 2010-11-28T01:36:23 On account of being an eigenvector ignoramus ;) 2010-11-28T01:36:30 if i added fleets at time 0 and sent them later, i would see if i could send them to a friend closer to the enemy and have the fleet make it's final journey in time 2010-11-28T01:36:34 btw, the final rankings have from the open tournament have finished 2010-11-28T01:36:36 sanddbox: Like Google's "PageRank" but not at all like it. 2010-11-28T01:36:56 not sure how backlinks apply to planets xD 2010-11-28T01:36:59 janzert: ) 2010-11-28T01:37:04 i ranked planets with a simple formula 2010-11-28T01:37:12 i snuck into 41st 2010-11-28T01:37:27 sanddbox: Well, a 5 planet right next to a 4 planet is more powerful than one farther away 2010-11-28T01:37:28 ya I was sorta wondering how others ranked their planets as well 2010-11-28T01:37:30 Distance(home.PlanetID(), expansion.PlanetID())) - (2 * expansion.GrowthRate()) + (expansion.NumShips() / 10)) 2010-11-28T01:37:41 sanddbox: Because the advantage is worn down by the travel time 2010-11-28T01:37:41 Just an example 2010-11-28T01:37:44 but i only found out about this two days ago so it's not like i had time to make it advanced 2010-11-28T01:37:54 ah 2010-11-28T01:37:55 my ranking of planets was as follows, maximise growth, then minimise average distance to enemies 2010-11-28T01:38:05 Terr__, that's why i just incorporated distance into my ofrmula 2010-11-28T01:38:07 mine literally just found the worth of the planet according to how long it took me to get there 2010-11-28T01:38:12 Terr__: I tried eigenvector centrality, but couldn't figure ut a ggood way of using the data 2010-11-28T01:38:15 *** gump_ has quit IRC (Ping timeout: 240 seconds) 2010-11-28T01:38:19 jaman4dbz, that's pretty primitive 2010-11-28T01:38:27 *** margoth has joined #aichallenge 2010-11-28T01:38:28 so you'd call the function giving it the two planets, and it would know how long it would take, and therefor how much growth it would get out of it. 2010-11-28T01:38:29 sigh: I tried egenvector centrality, but then I realized I didn't really know what that meant ;) 2010-11-28T01:38:29 i just factored in growth, distance, and ships defending 2010-11-28T01:38:34 I have no planet rankings, maybe that's my problem. 2010-11-28T01:38:36 Terr__: haha 2010-11-28T01:38:39 so you'd say 20 turns and if it took 5 turns it would only get 15 turns worth. 2010-11-28T01:38:55 sigh: This is one of the few times I look back and say: "Gosh, I wish I took all those hard CS courses" 2010-11-28T01:38:57 Terr__: lmao, awesome! 2010-11-28T01:39:00 the only planet ranking i did was on turn 1 - i calculated a weighting of each planet to the chord between each pair of planets. this gives a rough idea of which planets will be useful "on the way" to other planets. 2010-11-28T01:39:00 Neverstu: same for me.. ship re-distribution is its own separate algorithm using enemy proximity and ship counts into account 2010-11-28T01:39:05 As opposed to a slightly moer wishy-washy degree. 2010-11-28T01:39:19 Neverstu: haha one of my profs from Windsor did that 2010-11-28T01:39:34 for all other rankings, the planets were ranked according to the attack plans for them, not in isolation 2010-11-28T01:39:39 Terr__: apparently its some alien searching algorithm or something 2010-11-28T01:39:39 games not being played anymore? 2010-11-28T01:39:44 Terr__: it's like... assume your randomly go from planet to planet, following the links with the highest weight more often... where are you likely to end up 2010-11-28T01:39:55 that's one way of looking at it 2010-11-28T01:40:07 burny: it's settign up for the final contest 2010-11-28T01:40:32 a big thanks to the organizers, it was great fun to participate - great distraction from writing boring business software 2010-11-28T01:40:40 ^^ word to that 2010-11-28T01:40:43 sigh: Yeah, I'm just sayinig I don't have the terminology down. Maybe it's just in my head, but it seems like 99% of the resources I can find are all geared for math-heads rather than coders. 2010-11-28T01:40:43 especially janzert 2010-11-28T01:40:44 But ya, that pairing algorithm... my old prof spent like 3 hours making a bot and beat the one I spent 2 days on... I was impressed. 2010-11-28T01:40:50 *** Zepp has quit IRC (Ping timeout: 250 seconds) 2010-11-28T01:41:01 sigh: And there's a very different sort of vocabulary, mental model, and presentation that goes with that. 2010-11-28T01:41:04 no offense amstan, i'm very grateful of you too, but janzert has done buttloads this contest 2010-11-28T01:41:05 margoth: +1 ;-) 2010-11-28T01:41:11 ahh.. what's the expected time it'll take to complete the final? 2010-11-28T01:41:11 Terr__, what is the use of being a computer programmer if you're not a computer scientist? 2010-11-28T01:41:15 *** yasith has quit IRC (Ping timeout: 272 seconds) 2010-11-28T01:41:19 jaman4dbz: interesting! i didn't do any rigorous proofs of it, but it seemed to help indeed 2010-11-28T01:41:21 antimatroid1: oh, np, i agree 2010-11-28T01:41:22 Terr__: yeah, it can be that way... the math is not actually that hard, but you need to have some basic linear algebra under your belt 2010-11-28T01:41:28 janzert's awesome 2010-11-28T01:41:48 seriously, all you really need is a good understanding of algebra, logic, and possibly statistics 2010-11-28T01:41:50 Thanks 2010-11-28T01:41:53 sanddbox: I knoes. I got my degree in Informatics, mainly because at the time it seemed like web/db/visualization stuff was going to be the interesting bits 2010-11-28T01:41:55 I think hard CS courses hurt me, lol. 2010-11-28T01:42:01 but quit distracting me by making my window flash :P 2010-11-28T01:42:10 Terr__: all that maths crap is usually just notation 2010-11-28T01:42:14 sanddbox: Later I had to teach myself things like formal grammars (ANTLR) etc. 2010-11-28T01:42:16 I wasted days trying to think of an effecient way for me to reduce the backpack problem to attacking from multiple planets to multiple planets 2010-11-28T01:42:21 if you don't understand something, look up what it means then presto :) 2010-11-28T01:42:22 FYI I don't think it's feasible... 2010-11-28T01:42:33 antimatroid1: And I hateses it. It burneses poor gollum. 2010-11-28T01:42:40 if it was feasible though it would be the gaurenteed best planets to take =P 2010-11-28T01:42:44 gaurenteed as in proven. 2010-11-28T01:42:46 i love notation 2010-11-28T01:42:57 i'm only a freshman in HS (:/) but i plan to go into AI 2010-11-28T01:43:06 i do wish there was a standard relation for "is a sequence in" 2010-11-28T01:43:09 it's really the only CS field that still has a lot of room for work 2010-11-28T01:43:11 lol your crazy sanddbox 2010-11-28T01:43:17 you're* 2010-11-28T01:43:21 <3 2010-11-28T01:43:21 that's pretty early. 2010-11-28T01:43:28 ya ya... pft. 2010-11-28T01:43:38 sanddbox: do maths and cs for undergrad, don't try to specialise that early 2010-11-28T01:43:42 Neverstu: you should do a write-up on your algorithm later. i'd be interested to read how you took attack plans into account when ranking planets etc 2010-11-28T01:43:43 also sanddbox that is not at all true. 2010-11-28T01:43:52 sanddbox: Get a job doing intranet site utilities and crap and THEN tell me algebra/logic/statistics is all you need ;P 2010-11-28T01:44:08 their are tons of fields were CS can evolve 2010-11-28T01:44:08 *** Corg has quit IRC (Quit: Page closed) 2010-11-28T01:44:08 i made the mistake of trying to adapt the "knapsack" algorithms to ship packing. huge waste of mental energy. when i went off and rolled my own solution it worked much better 2010-11-28T01:44:08 Terr__: ew, intranet site utilities 2010-11-28T01:44:10 *** krokokrusa has joined #aichallenge 2010-11-28T01:44:17 for example UX still has tons of room to evolve... we have a lot of Microsofts mistakes to fix. 2010-11-28T01:44:27 the only type of web work i would do would be pen testing 2010-11-28T01:44:33 sanddbox: Muwhahaha, and stakehoders who whine when a button says "OK" instead of "Okay" because it's NOT IN THE SPEC! Hah. 2010-11-28T01:44:37 rebelxt_: you as well! i was thinking of doing a blog post about it, i need to think of some pretty pictures to make it interesting 2010-11-28T01:45:45 ...bleh. 2010-11-28T01:45:48 *** McLeopold has quit IRC (Quit: Page closed) 2010-11-28T01:46:04 any hints as to what the next ai challenge will be? 2010-11-28T01:46:04 would be interesing to see if anyone managed to take knapsack to the next level.. i.e. using integer programming for move scheduling, ship distribution 2010-11-28T01:46:06 :D 2010-11-28T01:46:40 Jenga 2010-11-28T01:46:56 rebelxt_: I wanted to, but I couldn't see a clear way to express it and use a solving library. (It's probably doable, just not by me.) 2010-11-28T01:46:58 i guess many of use are using knapsack for the first turn.. but it got so messy trying to use it with move scheduling required to do attack coordination 2010-11-28T01:47:01 sanddbox: i would guess somewhere near the start of next year 2010-11-28T01:47:04 rebelxt_, how exactly does the knapsack problem relate to this contest? 2010-11-28T01:47:16 antimatroid1, wat 2010-11-28T01:47:21 the next ai challenge is the start of next year? 2010-11-28T01:47:29 the last one was earlier this year 2010-11-28T01:47:32 i mean what the challenge actually is 2010-11-28T01:47:34 not when it begins 2010-11-28T01:47:34 i'm not sure if knapsack as such is applicable 2010-11-28T01:47:35 :P 2010-11-28T01:47:36 rebelxt_: I wonder if flow algorithms... nah, everything is too dependent on +/-1 values 2010-11-28T01:47:47 oh, it hasn't even been considered properly 2010-11-28T01:47:54 you can use it to expend all your ships but that might not be the best move 2010-11-28T01:48:31 sandbox, many are using knapsack to pick best planets to hit on first turn 2010-11-28T01:48:44 margoth: sure, but I don't think *anyne* found the best move :P 2010-11-28T01:48:53 margoth: 2010-11-28T01:49:10 there is no "best" move, only moves that might be better than others 2010-11-28T01:49:12 margoth: The "best" move depends on what the opponent does... and you can't see that or react except one turn later 2010-11-28T01:49:23 er, that was to sigh 2010-11-28T01:49:36 * Terr__ wonders about sleeping 2010-11-28T01:49:39 * Terr__ is out of practice 2010-11-28T01:49:44 is anyone using simulated annealing to calculate best possible way to expand (or simple hill climbing)? 2010-11-28T01:50:33 anyway, I tried to use knapsack to produce different solution candidates but couldn't make it produce any improvement 2010-11-28T01:50:38 No, but I did seek out dubiously-legal torrents of scanned AI/algorithm e-books. And stayed at a Holday Inn Express. 2010-11-28T01:50:39 "best possible way" - i mean to get a good approximation of course since it's NP hard problem like TSP 2010-11-28T01:50:58 Terr__: lol 2010-11-28T01:51:47 Terr__: "Artificial Intelligence, A Modern Approach"? 2010-11-28T01:51:59 Hmmmm... Supreme Commander 2 on Steam for 3.75 2010-11-28T01:52:02 I decided to abandon complex far reaching strategies as I had limited time and I assumed that whatever the opponent does will destroy my plans anyhow 2010-11-28T01:52:03 rebelxt_: XD 2010-11-28T01:52:06 that's the one i torrented about a month ago 2010-11-28T01:52:09 Lemmecheck. Gotta remote into the other comp. 2010-11-28T01:52:14 i *hate* reading on my computer, though 2010-11-28T01:52:15 Home VPN = the win 2010-11-28T01:52:23 that's why my bot doesn't save any calculations to the next turn 2010-11-28T01:53:01 rebelxt_: No, but I may have gone through it from the library.not sure. 2010-11-28T01:53:12 *** yasith has joined #aichallenge 2010-11-28T01:53:27 I sorta missed a lot, but I can say that the knapsack provided a perfect solution, but only in the context you gave it. 2010-11-28T01:53:28 anyone know when we can start seeing the final things: and when it'll be done? 2010-11-28T01:53:49 a great book. i even thought it was gonna help "solve" planet wars ;-)) 2010-11-28T01:53:51 sanddbox: I agree to a point: A book is awesome for absorbing information, but when it comes to troubleshooting etc. the ability to do a fulltext search is *invaluable* 2010-11-28T01:53:52 i feel like a crack addict, always refreshing my profile page 2010-11-28T01:54:00 I'm a little under half way through the list of things need to start the finals running 2010-11-28T01:54:06 @rankings 2010-11-28T01:54:07 Terr__, when it comes to troubleshooting, fuck ebooks 2010-11-28T01:54:09 for example if you had a function that calculated how many ships you could gain from a planet if you tried to take it and plugged that into the knapsack algorithm, then you would have the maximum solution for that number of turns. 2010-11-28T01:54:10 amstan: Top 10 players: asavis(3371), Slin-.-(3353), bocsimacko(3345), _iouri_(3282), bix0r4ever(3271), _Astek_(3257), smloh1(3246), dmj111(3229), george(3215), Doraj8(3189) 2010-11-28T01:54:11 i'm heading to google 2010-11-28T01:54:15 sanddbox: If it's, say, a book on Samba, I'd want it digitally so I can debug why the hell this server won't let people log in with permissions to the yada yada 2010-11-28T01:54:45 sanddbox: Depends on the usefulness of the error message... if any. 2010-11-28T01:55:12 *** sigh has quit IRC (Remote host closed the connection) 2010-11-28T01:55:16 Supreme Commander 2: Is it worth it for $3.75 assuming you're a strategy wuss and would just play offline? 2010-11-28T01:55:21 I didn't notice... did any country have two or more people in the top ten? It seemed pretty well spread out. 2010-11-28T01:55:50 if you would play offline 2010-11-28T01:55:51 Hah, for everyone who needs a "fix" for their sudden withdrawal symptoms: http://store.steampowered.com/app/44200/ 2010-11-28T01:55:51 pirate it 2010-11-28T01:55:55 "is game X worth <$5" -> categorically "yes" 2010-11-28T01:56:00 ukraine 2010-11-28T01:56:27 i'm going out for a little bit 2010-11-28T01:56:28 back later 2010-11-28T01:56:29 the way the ships collide with each other in Galcon-proper now seems so wrong to me 2010-11-28T01:56:32 Terr__: i was just playing Galcon on android/iphone minutes ago ;-) a bit rusty but so much fun 2010-11-28T01:56:34 sanddbox: But I'm employed now... 2010-11-28T01:56:37 haha 2010-11-28T01:56:51 Terr__, so? 2010-11-28T01:56:59 being employed is great I miss it... you get to buy stuff rather than pirate it! 2010-11-28T01:57:19 Neverstu: They're retargetable too, aren't they? 2010-11-28T01:57:21 wat 2010-11-28T01:57:27 i see it this way 2010-11-28T01:57:35 unemployed: have no money, pirate 2010-11-28T01:57:35 Terr__: yes, insanity!! 2010-11-28T01:57:38 employed: have money, pirate 2010-11-28T01:58:07 fleet re-direction rocks. makes games so dynamic 2010-11-28T01:58:19 I suppose their is no reason NOT to wear a pirate hat all of the time. 2010-11-28T01:58:22 if money be problem, at least pay for indie games 2010-11-28T01:58:22 rebelxt_: Makes it less interesting AI-wise though 2010-11-28T01:58:51 rebelxt_: Less chaos-theory, more "put cloud in right place" 2010-11-28T01:58:53 Terr__: yep, unfortunetly 2010-11-28T01:59:16 now, if passing fleets would fire at one another, _that_ would make it ai-worthwhile 2010-11-28T01:59:22 albeit a bit more complex 2010-11-28T01:59:33 complex... 2010-11-28T01:59:34 yes. 2010-11-28T01:59:54 one of the hooks that got me into this contest was how simple it was =P 2010-11-28T02:00:11 margoth: From what I know of Galcon there are... aspects to it which an AI player could do too much more effectively than a human player. 2010-11-28T02:01:19 margoth: Similar to bots having unbeatable aim in FPS games, a galcon bot could micromanage things so that their ships dove into a planet at the exact moment of weakness... or continually hovered between two destinations... or could super-accurately estimate the normally-hidden numbers of ships at planets. 2010-11-28T02:01:46 looks like mega1 is up 2010-11-28T02:01:52 yeah, this game was a great choice: it's very easy to get started but there is more depth than one can manage 2010-11-28T02:02:11 amstan: you mean awake? yes 2010-11-28T02:02:22 mega1: nervous? 2010-11-28T02:02:30 never :-) 2010-11-28T02:02:58 I only checked once during the night. 2010-11-28T02:03:21 mega1: good time to share your general approach? i'm super curious 2010-11-28T02:03:37 I'm afraid you'll have to wait for the blog post :-) 2010-11-28T02:03:42 *** Appleman1234 has joined #aichallenge 2010-11-28T02:03:45 mega1: only once? nerves of steel! 2010-11-28T02:03:53 mega1: Good idea: Easier to express on a blog. 2010-11-28T02:04:12 rebelxt_: simple exhaustion helps too 2010-11-28T02:04:21 mega1: hehe, already subscribed to your blog 2010-11-28T02:04:22 mega1: Gotta watch out for kids these days... they're all saying things like: "OMG plz xplain alg here on twitr plz!" 2010-11-28T02:04:36 I spent the last few days mindlessly tuning stuff 2010-11-28T02:04:49 mega1: "can u put sudocod in 140 chars?" 2010-11-28T02:04:58 heh 2010-11-28T02:05:07 Actually 2010-11-28T02:05:17 That gives me a great idea for a silly prop at some sort of programmer/IT convention. 2010-11-28T02:05:23 the Sudo Cod. 2010-11-28T02:06:06 I think I'll go to sleep now... 2010-11-28T02:06:23 the conversation has obviously become to awesome to keep watching. 2010-11-28T02:06:29 night all. 2010-11-28T02:06:57 mega1, promise a blog post as soon as tourney is over 2010-11-28T02:07:09 yup, I do 2010-11-28T02:07:14 *** jaman4dbz has quit IRC (Quit: Page closed) 2010-11-28T02:07:34 great! 2010-11-28T02:07:59 i am excited to have my mind blown 2010-11-28T02:08:03 mine is not worth a write-up but git commit log is a funny read, as always 2010-11-28T02:08:14 rebelxt_: yay 2010-11-28T02:08:20 rebelxt_: github? 2010-11-28T02:08:32 amstan: yeah, private repo atm 2010-11-28T02:08:57 beautiful winter morning here, missed most of the sleep -> make -k coffee 2010-11-28T02:09:02 i found github's prices to be quite high 2010-11-28T02:09:20 I wish I could install something github-esque on my web host 2010-11-28T02:09:32 i fear change and still use svn 2010-11-28T02:09:37 Free an open-source, obviously ;) 2010-11-28T02:09:45 i agree they are overpriced :( i was just too hooked to it 2010-11-28T02:09:47 Neverstu: Ditto. 2010-11-28T02:09:52 * Ken69267 uses bzr 2010-11-28T02:09:54 * Ken69267 hides 2010-11-28T02:10:07 Neverstu: I have just enough Windows machines in my life that I want to see git get some better tools/integration there 2010-11-28T02:10:12 Ken69267: RCS! 2010-11-28T02:10:24 Terr__: very true 2010-11-28T02:10:33 Actually, I have one project where I'm still just making a bi-weekly zip file and moving it to an external drive... 2010-11-28T02:10:46 though we use Perforce at work and love the crap out of it 2010-11-28T02:11:02 Neverstu: At my last job they transitioned from Perforce to SVN while I was there 2010-11-28T02:11:07 even P4V, after some teething issues 2010-11-28T02:11:11 Neverstu: Worked great for me ;) 2010-11-28T02:11:31 heh 2010-11-28T02:12:01 I think the problem with git is--as many have mentioned--that there's a sort of mental-model mismatch 2010-11-28T02:12:17 Terr__: how so? 2010-11-28T02:12:19 And right now it just isn't quite compelling enough for me to move my one-person stuff to it 2010-11-28T02:12:37 amstan: Compared to the models people ingrain from SVN 2010-11-28T02:12:58 i find it perfect for one person stuff, i don't have to worry about managing servers and stuff, just git init 2010-11-28T02:13:00 amstan: Causes pain to SVN-users. 2010-11-28T02:13:51 *** bemile has joined #aichallenge 2010-11-28T02:14:02 *** jukkerknott has joined #aichallenge 2010-11-28T02:14:52 amstan: yeah using git locally and backing up using dropbox works great for one person projects 2010-11-28T02:16:17 *** NannerBflizzle has joined #aichallenge 2010-11-28T02:16:26 *** bemile has quit IRC (Client Quit) 2010-11-28T02:16:47 hi? 2010-11-28T02:16:51 :> 2010-11-28T02:17:10 blaarg 2010-11-28T02:17:13 ello ;-) 2010-11-28T02:19:58 Top 10 players: asavis(3371), Slin-.-(3353), bocsimacko(3345), _iouri_(3282), bix0r4ever(3271), _Astek_(3257), smloh1(3246), dmj111(3229), george(3215), Doraj8(3189) 2010-11-28T02:20:17 I think I go now and sleep, or at least do things while muttering that I oughg to sleep. 2010-11-28T02:20:20 BBL. 2010-11-28T02:20:38 *** CryBaby has left #aichallenge 2010-11-28T02:20:50 *** aizu has quit IRC (Read error: Connection reset by peer) 2010-11-28T02:21:21 *** aizu has joined #aichallenge 2010-11-28T02:21:43 night, all. figured janzert can't work any faster if i stay up ;-) 2010-11-28T02:21:50 *** aizu has quit IRC (Read error: Connection reset by peer) 2010-11-28T02:21:55 later guys 2010-11-28T02:22:16 cya 2010-11-28T02:22:19 *** rebelxt_ has quit IRC (Quit: Page closed) 2010-11-28T02:23:03 trivia, there were over 4.6 million games played during the open contest 2010-11-28T02:23:03 *** aizu has joined #aichallenge 2010-11-28T02:23:39 janzert: that is awesome. did you get more participants than you projected? 2010-11-28T02:24:29 no idea, I wasn't really around till after the contest started 2010-11-28T02:25:16 Probably a dumb question, but how do I use the contestBot that amstan put up? 2010-11-28T02:25:49 NannerBflizzle: depends on what you want to do 2010-11-28T02:26:22 use the plugin 2010-11-28T02:26:30 oh 2010-11-28T02:26:36 @rankings 2010-11-28T02:26:37 or get rankings 2010-11-28T02:26:37 amstan: Top 10 players: asavis(3371), Slin-.-(3353), bocsimacko(3345), _iouri_(3282), bix0r4ever(3271), _Astek_(3257), smloh1(3246), dmj111(3229), george(3215), Doraj8(3189) 2010-11-28T02:26:41 that's mostly what it does 2010-11-28T02:26:46 @help rankings 2010-11-28T02:26:46 amstan: (rankings [how many]) -- Get the top $(how many) players. 2010-11-28T02:27:09 how do i "install" the plugin? 2010-11-28T02:27:23 NannerBflizzle: again.. what are you trying to do? 2010-11-28T02:27:30 you want your own bot or something? 2010-11-28T02:28:23 *** Zepp has joined #aichallenge 2010-11-28T02:28:29 I'm just trying to use it but I think I just figured it out 2010-11-28T02:28:49 I don't need to install this plugin to use it here :\ sorry lol 2010-11-28T02:28:55 *** KP13 has joined #aichallenge 2010-11-28T02:31:31 *** winexception has joined #aichallenge 2010-11-28T02:32:16 *** Kingpin13 has quit IRC (Ping timeout: 255 seconds) 2010-11-28T02:36:25 *** Ice_Harley has quit IRC (Ping timeout: 265 seconds) 2010-11-28T02:41:38 *** Mistmanov has joined #aichallenge 2010-11-28T02:47:25 finals map set is now in svn 2010-11-28T02:49:00 *** sigh has joined #aichallenge 2010-11-28T02:49:41 so games starting any min then? 2010-11-28T02:50:00 not quite but it's getting closer 2010-11-28T02:50:23 you rock! 2010-11-28T02:51:27 *** NannerBflizzle has quit IRC (Quit: Page closed) 2010-11-28T02:56:34 any official word on how long the finals will go on for? 2010-11-28T02:56:53 burny: topic says: Contest is now closed, Results Wednesday 2010-11-28T02:57:30 thanks 2010-11-28T03:00:57 oh boy, so i missed nothing? 2010-11-28T03:00:58 awesome 2010-11-28T03:01:13 missed nothing? 2010-11-28T03:01:18 no games yet 2010-11-28T03:01:21 i went out for a bit 2010-11-28T03:01:31 ah, so did I 2010-11-28T03:01:48 i made a big risk with my submission 2010-11-28T03:01:56 hopefully i at least get 1st for australia 2010-11-28T03:03:22 I'm hoping to land finnish #1, a long shot 2010-11-28T03:04:42 ha, i gave up that hope a while ago 2010-11-28T03:05:14 what was the risk? a big last-minute change? 2010-11-28T03:05:23 antimatroid1: you wanted finnish #1? o_O 2010-11-28T03:05:49 yeah 2010-11-28T03:06:11 i spent some time in first for tron before the final contest 2010-11-28T03:06:12 yeah, finnish #1 is the best! 2010-11-28T03:06:21 it was really disappointing when we got disqualified 2010-11-28T03:06:54 antimatroid1: do you realise that finnish is a nationality and not a verb? 2010-11-28T03:07:16 :P i'm not an english major 2010-11-28T03:07:26 *** sanddbox has quit IRC () 2010-11-28T03:07:26 neither am I :P 2010-11-28T03:07:26 :P 2010-11-28T03:07:32 doh! i forced myself to submit yesterday and not touch it today. the rankings were a complete mess today, impossible to figure out if you've made any improvements 2010-11-28T03:07:52 I actually rolled back my last submission 2010-11-28T03:08:27 my own tests kinda said it wasn't as good as the latest but... with no proper benchmarking I decided to rely on hunch 2010-11-28T03:08:39 probably a wise choice 2010-11-28T03:08:48 *** Terr__ has quit IRC (Ping timeout: 265 seconds) 2010-11-28T03:10:54 i implemented a completely new tree idea today and submitted it without any real testing :P 2010-11-28T03:11:33 it shouldn't be that different to my others anyway 2010-11-28T03:11:49 holy cow 2010-11-28T03:12:29 it seemed better, so i thought fuck it :p 2010-11-28T03:12:29 i had a number of bugs that i left in intentionally, because they turned other bugs into nice features :) "fixing" them would upset the whole house of cards 2010-11-28T03:12:57 i also went for an agressive bot 2010-11-28T03:13:18 antimatroid1: where do you think your bot will place? 2010-11-28T03:13:23 any chance at top 10? 2010-11-28T03:13:26 in that it will send offensive moves to arrive whenever (provided it's not as they claim a neutral 2010-11-28T03:13:32 sigh: doubtful 2010-11-28T03:13:35 top 30 would be nice 2010-11-28T03:13:41 ah, too bad 2010-11-28T03:13:45 i can beat iouri, accoun etc. sometimes 2010-11-28T03:13:48 col 2010-11-28T03:13:50 *cool 2010-11-28T03:13:51 but it loses to some crap bots too 2010-11-28T03:14:30 i have no idea what the competition is like on the main site though, i haven't had a bot run on there in like a month 2010-11-28T03:14:40 the rock-paper-scissors nature of this problem is a bit irritating. it's hard to get any real consistency (though clearly some have done it) 2010-11-28T03:15:31 Neverstu: there was never much stability in the rankings 2010-11-28T03:15:39 except for first place* 2010-11-28T03:15:42 :) 2010-11-28T03:15:48 indeed :) 2010-11-28T03:15:58 well i'm gunning for bocsi to win 2010-11-28T03:16:27 I hope there he has some competition 2010-11-28T03:16:32 *** Azrathud has joined #aichallenge 2010-11-28T03:16:40 he does now i think 2010-11-28T03:16:41 if only to see some awesome games 2010-11-28T03:16:54 i don't think any bots are anywhere near optimal 2010-11-28T03:16:58 it's a really really hard problem 2010-11-28T03:17:06 yeah 2010-11-28T03:17:11 it's remarkable that at first it seems so simple too 2010-11-28T03:17:16 yep 2010-11-28T03:17:24 was a wondeful choice of game 2010-11-28T03:17:28 heh, I never thought it was simple o_O 2010-11-28T03:17:28 only after starting to analyze the tree you realize how unbelieveably hopeless it is 2010-11-28T03:17:58 lol, the actual game tree is so big you can't even touch the first layer 2010-11-28T03:18:13 well "simple" is an overstatement, but i think given the number of contestants it certainly has a casual appeal, as if a few hours of work could generate a decent bot 2010-11-28T03:18:54 it's simple in that it has a fairly low barrier for entry 2010-11-28T03:19:17 *** Whalefreezer has joined #aichallenge 2010-11-28T03:19:18 yeah, it's easy to get going, hard to do well 2010-11-28T03:19:20 which is ideal 2010-11-28T03:19:20 simple for someone to get a bot going which doesn't look completely randm :P 2010-11-28T03:19:24 ah yes, that's a more concise way to put it 2010-11-28T03:19:29 yeah 2010-11-28T03:19:59 Top 10 players: 2010-11-28T03:20:07 heh 2010-11-28T03:20:08 woot, I'm tied for first! 2010-11-28T03:20:14 hot damn 2010-11-28T03:20:58 i hit top 10 a few days ago, but that was all due to ratings instability for new submissions :/ 2010-11-28T03:21:04 for a few minutes there it was wonderful 2010-11-28T03:21:08 :) 2010-11-28T03:21:22 where were you before they shut off the games? 2010-11-28T03:21:26 my highest was 58th or so 2010-11-28T03:21:41 42 i believe 2010-11-28T03:21:49 cool 2010-11-28T03:22:06 you? 2010-11-28T03:22:35 hmmm... somewhere around 100... I haven't worked on my bot for over a month and my last submission was 2 weeks ago 2010-11-28T03:22:44 wish I had had more time 2010-11-28T03:22:51 mega1: Are you planning to release your code? 2010-11-28T03:23:10 Zoo: he's said in that past that he is 2010-11-28T03:23:19 not sure when though 2010-11-28T03:23:22 thats lovely 2010-11-28T03:23:33 I wish spaceinvaders could too =/ 2010-11-28T03:23:37 i had a co-worker to compete with; after he beat me once i became obsessed 2010-11-28T03:24:17 Zoo: the generator code, or the soupy product? 2010-11-28T03:24:48 the implementation of the genetic part on planet wars 2010-11-28T03:25:03 that would be neat to see 2010-11-28T03:25:04 Im mostly interested in how they managed to pull it off and make it effective 2010-11-28T03:25:38 yeah it would 2010-11-28T03:25:52 apparently there were a few others who had genetic bots aswell 2010-11-28T03:26:45 i thought of parameterizing key parts of mine, then evolving those parameters, but it seemed so hopeless to get real data 2010-11-28T03:28:41 Yeah I doubt they got any quick effects either. Im imagining they used clouding and ran it for a month before they managed to get any kind of logic in it 2010-11-28T03:30:20 they were in the top 100 right before the end 2010-11-28T03:30:42 hidden contest? 2010-11-28T03:30:50 Accoun_: it hasn't started yet :) 2010-11-28T03:30:52 *** Accoun_ is now known as Accoun 2010-11-28T03:30:55 it will be live 2010-11-28T03:32:35 *** Whalefreezer has left #aichallenge 2010-11-28T03:42:54 first worker instance is coming up now 2010-11-28T03:43:02 omg omg omg omg 2010-11-28T03:43:11 janzert: <3<3<3 2010-11-28T03:44:05 be another little bit before it starts playing games but should be less than 10 minutes 2010-11-28T03:44:37 *** virdo has quit IRC (Ping timeout: 260 seconds) 2010-11-28T03:45:12 and the games have begun 2010-11-28T03:45:47 \o/ 2010-11-28T03:46:00 goodbye sleep! 2010-11-28T03:46:10 none showing up in ranks yet though 2010-11-28T03:46:22 first game of the finals is http://ai-contest.com/visualizer.php?game_id=8857501 2010-11-28T03:46:35 burny: no it will be a while yet before the first rankings update 2010-11-28T03:46:58 gg :) 2010-11-28T03:47:02 oh ya, forgot about that 2010-11-28T03:48:05 *** virdo has joined #aichallenge 2010-11-28T03:49:07 *** Mjothvitnir has quit IRC (Quit: Page closed) 2010-11-28T03:49:28 the "next game will be played within xxxx next games" thingie is disabled for the finals? or it doesn't kick in untill everyone has had a few games? 2010-11-28T03:49:52 Mistmanov: it's disabled at least for now 2010-11-28T03:49:56 k =) 2010-11-28T03:50:17 probably actually it'll stay disabled 2010-11-28T03:50:57 since it would be hard to make it work right once we start only playing the top x bots 2010-11-28T03:51:34 yeah, i don't see much need for it 2010-11-28T03:52:00 janzert: speaking of, can you provide more details about how the pairings will work? or should i wait patiently for the inevitable forum discussions? :) 2010-11-28T03:52:23 you can look at create_matchups.py in the repo to see exactly how they work :) 2010-11-28T03:52:27 @repo 2010-11-28T03:52:28 janzert: repo = http://code.google.com/p/ai-contest/ 2010-11-28T03:52:30 in the future there needs to be a "latest games" page 2010-11-28T03:52:51 janzert: looking now thanks! 2010-11-28T03:52:53 it's basically the same as has been though 2010-11-28T03:53:24 or perhaps the schedule of "pairing cutoffs"? 2010-11-28T03:53:56 Neverstu: that is still dependent on how many games we manage to fit into the next few days 2010-11-28T03:54:50 I can say it won't start until at least the bottom of the rankings have shook themselves out into the groups we saw before and it will hopefully be gradual 2010-11-28T03:54:53 janzert: is it flexible about extending if not enough games have played? or MUST results be released on the first? 2010-11-28T03:55:22 antimatroid1: that's up to Jeff but I think we'll probably make it 2010-11-28T03:55:31 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-11-28T03:55:39 janzert: thanks! 2010-11-28T03:56:05 i don't see why it isn't run for at least a week or two, but doesn't matter now 2010-11-28T03:56:39 *** Ice_Harley has joined #aichallenge 2010-11-28T03:56:48 I think possibly too many people will lose interest if it drags out too long 2010-11-28T03:57:01 but I wasn't actually around for that decision either 2010-11-28T03:58:22 *** zcheng has joined #aichallenge 2010-11-28T04:01:14 *** Palmik has joined #aichallenge 2010-11-28T04:03:20 ah, my first glorious victory agaisnt a starter package bot 2010-11-28T04:03:55 I really think you could've cut down dramatically on the server load by just removing everyone ranked starter package and below 2010-11-28T04:04:28 or just removing everyone who hasnt committed for 2 months 2010-11-28T04:04:45 that'd kill 2000 or so :P 2010-11-28T04:04:45 *** jukkerknott has quit IRC (Remote host closed the connection) 2010-11-28T04:04:47 *** Utkarsh has joined #aichallenge 2010-11-28T04:04:50 i browsed the bottom of the rankings a few times and was slightly appalled 2010-11-28T04:05:13 who submits a bot that crashes every single time...and leaves it like that? 2010-11-28T04:05:24 someone who doesn't care :p 2010-11-28T04:06:01 Neverstu: quite a lot of people made bots that works only for 23 or less planets 2010-11-28T04:06:11 so they kind of got crippled with the new map generator 2010-11-28T04:06:24 or left maxdistance at 25 2010-11-28T04:06:26 antimatroid1: ha! i had forgotten about that 2010-11-28T04:06:39 like me 2010-11-28T04:06:57 there are still a lot of bots at the very bottom that have always just crashed immediately on the first turn. 2010-11-28T04:06:57 wasn't max distance originally 34? 2010-11-28T04:07:26 People like me dont browse the forums too much hence dont hear about map generator changes until they come on irc and ask about them :) 2010-11-28T04:07:30 jaznert: how often do rankings update? 2010-11-28T04:07:47 Zoo: you can mostly thank/hate me for that change :) 2010-11-28T04:07:47 *** zcheng has left #aichallenge ("Leaving") 2010-11-28T04:08:14 its no problem 2010-11-28T04:08:21 people seem divided on the thanking/hating part :P 2010-11-28T04:08:42 yeah I imagine its because people dont know their code and cant adapt :P 2010-11-28T04:09:02 the line-symmetric maps caused a bit of headache, but ultimately my bot was better for it 2010-11-28T04:09:06 i think it's made it better, and imo it's boring when people are trying to optimise just for specific maps 2010-11-28T04:09:22 it certainly made it tougher 2010-11-28T04:09:52 having said that, my bot fails miserably when there are lots of takable planets 2010-11-28T04:10:22 what kind of starting move do you use? 2010-11-28T04:10:32 i have my first win under my belt 2010-11-28T04:10:38 the new generator occasionally spits out extremely sparse maps - i found these to be very illuminating in terms of middle-planet strategy 2010-11-28T04:11:11 *** yasith_ has joined #aichallenge 2010-11-28T04:11:33 i basically have 3 steps, 1. cancel out as many of their attacks as i can, 2. use a game tree to pick moves (can't do more than a few moves), 3. redistribute ships closer to the front line 2010-11-28T04:11:58 http://ai-contest.com/visualizer.php?game_id=8858447 2010-11-28T04:12:00 yeah :) 2010-11-28T04:12:22 oh, I ment how do you choose the planets to conquer at the start of the match 2010-11-28T04:12:48 ^^ game tree part handles it 2010-11-28T04:13:43 I see 2010-11-28T04:13:47 *** yasith has quit IRC (Ping timeout: 245 seconds) 2010-11-28T04:14:11 the game tree i entered works as follows, pick moves I would like to make to each planet (sending ships from wherever at any time), then create children, do the same for them at each child, then run minimax, best the best two moves, then extend from that terminal node, rinse and repeat 2010-11-28T04:14:50 antimatroid1: what is your eval function> 2010-11-28T04:14:50 ? 2010-11-28T04:14:52 makes sense 2010-11-28T04:16:30 *** yasith_ has quit IRC (Ping timeout: 265 seconds) 2010-11-28T04:16:47 sigh: my ships at time 199 - their ships at time 199 2010-11-28T04:17:03 ah, nice and simple 2010-11-28T04:17:11 i like nice and simple evaluation functions 2010-11-28T04:17:19 :) 2010-11-28T04:17:28 did you try anything else? 2010-11-28T04:17:32 nope 2010-11-28T04:17:41 our evaluation function for tron was coolest 2010-11-28T04:17:54 preferencing went something like.. 2010-11-28T04:18:35 *** Mekanik has joined #aichallenge 2010-11-28T04:19:48 certain loss < separated with less squares reachable than them < together with less squares closest to us < certain draw < separated with same amount of squares available < together with same amount of squares accessible < separated with more squares than them < together with more squares then them < certain win 2010-11-28T04:19:59 Top 10 players: 2010-11-28T04:20:27 o_O 2010-11-28T04:20:45 and it would take the soonish guaranteed win, so would just pounce them when it got the chance 2010-11-28T04:21:35 why is separated with more squares than them < together with more squares then them 2010-11-28T04:21:51 i assumed we could fight better than them 2010-11-28T04:22:07 path filling was easier, and we were quite strong 2010-11-28T04:22:08 then why the opposite way around when losing? 2010-11-28T04:22:20 it might not have been :P 2010-11-28T04:22:25 sorry, i did that off the top of my head 2010-11-28T04:22:29 :P 2010-11-28T04:22:43 seperated < together, on both sides 2010-11-28T04:22:51 eagle-eyed sigh 2010-11-28T04:23:23 *** KP13 has quit IRC (Ping timeout: 272 seconds) 2010-11-28T04:23:31 http://csclub.uwaterloo.ca/contest/visualizer.php?game_id=4127671 2010-11-28T04:23:34 that game was half decent 2010-11-28T04:23:55 whoa that website is still up? 2010-11-28T04:24:10 yep :P 2010-11-28T04:24:11 http://csclub.uwaterloo.ca/contest/visualizer.php?game_id=4127689 2010-11-28T04:24:21 that game shows where he'd take the "first possible win" 2010-11-28T04:24:30 cool :) 2010-11-28T04:24:55 well no, it could have won earlier, but didn't see it :| 2010-11-28T04:25:29 huh? that second game your bots just collide head on? 2010-11-28T04:25:33 what's the result there? 2010-11-28T04:25:40 no, watch again, they had no move 2010-11-28T04:25:40 i win 2010-11-28T04:26:02 oh, i see... you had a square to move into 2010-11-28T04:26:08 yep 2010-11-28T04:26:29 he won only to live 1 more turn and die :) 2010-11-28T04:26:46 the game was to just outsurvive them :P 2010-11-28T04:27:44 *** Neverstu has quit IRC (Quit: Leaving) 2010-11-28T04:28:03 there was some obvious stupidity in our evaluation from that game though 2010-11-28T04:28:08 it could have won much earlier 2010-11-28T04:36:50 initial rankings are up, not that they are at all sensible :) 2010-11-28T04:37:45 how do some people have 10elo, and others have 9? 2010-11-28T04:38:27 100 people tied for first :P 2010-11-28T04:38:35 @rankings 2010-11-28T04:38:36 sigh: Top 10 players: xargs000(10), p26chan(10), k5_flik(10), alihaji484(10), AshBot(10), Loopholes(10), Boysen4254(10), alien21(10), mitbal(10), Jarod42(10) 2010-11-28T04:38:43 because bayeselo is basically trying to throw up at this data :P 2010-11-28T04:40:46 so 10 vs 9 is just random? 2010-11-28T04:41:06 or is based lightly on elo before the finals? 2010-11-28T04:41:30 sigh: aussie is doing well :P 2010-11-28T04:41:36 haha 2010-11-28T04:41:42 http://ai-contest.com/profile.php?user_id=12465 2010-11-28T04:41:44 ahaahaha 2010-11-28T04:42:00 burny: just minor random variation in the maximization-minorization formula I'm pretty sure 2010-11-28T04:42:18 definitely nothing left over from the data before the finals, it's not getting any of that at all 2010-11-28T04:42:50 minorization? 2010-11-28T04:43:09 some bots have two games 2010-11-28T04:43:20 it probably depends on what your opponents results were in their other games 2010-11-28T04:43:29 sigh: see the paper at the bottom of the bayeselo page http://remi.coulom.free.fr/Bayesian-Elo/ 2010-11-28T04:44:15 thanks, I'll have a look 2010-11-28T04:44:49 http://ai-contest.com/profile.php?user_id=9915 2010-11-28T04:44:50 ouch 2010-11-28T04:46:54 i've posted the main outline of my strategy in the forums 2010-11-28T04:49:09 yay, my first game of finals was victorious 2010-11-28T04:49:18 :) 2010-11-28T04:49:20 mine too 2010-11-28T04:49:50 my first two were :) 2010-11-28T04:51:03 I only got 1 so far :( 2010-11-28T04:51:48 you're in 45th place though :P 2010-11-28T04:52:06 lol ya.. that's accurate 2010-11-28T04:52:15 i'm at 1877 with 2 wins aha 2010-11-28T04:52:31 1505 :D 2010-11-28T04:52:36 *** narnach has quit IRC (Ping timeout: 255 seconds) 2010-11-28T04:53:04 I must say, that I'm kinda surprised we didn't go with an equal distribution pairing 2010-11-28T04:53:14 burny: it was done last time, it sucked 2010-11-28T04:53:28 how so? 2010-11-28T04:53:34 top bots played very few games against each other 2010-11-28T04:54:05 most games are between players with a huge imbalance in strength 2010-11-28T04:54:14 so don't really mean much of anything 2010-11-28T04:54:20 I meant: that bots all play same number of games.. not games against all ratings equally 2010-11-28T04:55:23 Howdy 2010-11-28T04:55:26 it has begun :D 2010-11-28T04:55:30 Erro323: :) 2010-11-28T04:55:30 oh, it should work out to that 2010-11-28T04:55:56 the first player is chosen from the least number of games played 2010-11-28T04:55:56 second game victorious too :)) 2010-11-28T04:55:57 https://github.com/Error323/googleai <---- my sourcecode =) 2010-11-28T04:55:57 *** jaspervdj has joined #aichallenge 2010-11-28T04:55:57 so there will be some small variance in number of games played but it should stay pretty close 2010-11-28T04:55:57 should we maybe wait to share? 2010-11-28T04:55:57 why? 2010-11-28T04:56:13 not sure 2010-11-28T04:56:18 ahhh 2010-11-28T04:56:24 i guess it's fine now, i thought maybe people should wait till results are finalised 2010-11-28T04:57:01 error you may as well start a thread then 2010-11-28T04:57:02 well i figured its ok to do now 2010-11-28T04:57:04 i'll chime in 2010-11-28T04:57:28 yeah ok 2010-11-28T04:57:45 *** jmpespxo1eax has joined #aichallenge 2010-11-28T04:57:55 *** mega1_ has joined #aichallenge 2010-11-28T04:58:24 mega1_: excited? :P 2010-11-28T04:59:32 janzert: why is the countdown of games gone :( 2010-11-28T04:59:44 hehe 2010-11-28T05:00:14 *** Zepp_ has joined #aichallenge 2010-11-28T05:00:18 *** narnach has joined #aichallenge 2010-11-28T05:00:30 because the old way wouldn't work anymore and I'm not sure how to make it work well once not all the bots are still playing 2010-11-28T05:00:41 *** AlliedEnvy_ has joined #aichallenge 2010-11-28T05:01:10 I'm sure the difficulty could be overcome but I probably won't I want to lighten the load on the server so more games can be played 2010-11-28T05:01:33 boo, I fell from 45th to 695, without playing a single game 2010-11-28T05:01:34 *** ulope_ has joined #aichallenge 2010-11-28T05:01:40 *** jmpespxoreax has quit IRC (Ping timeout: 240 seconds) 2010-11-28T05:01:40 *** mega1 has quit IRC (Ping timeout: 240 seconds) 2010-11-28T05:01:40 *** dmj111` has quit IRC (Ping timeout: 240 seconds) 2010-11-28T05:01:48 *** aizu has quit IRC (Ping timeout: 272 seconds) 2010-11-28T05:01:49 *** Zepp has quit IRC (Ping timeout: 272 seconds) 2010-11-28T05:01:52 *** AlliedEnvy has quit IRC (Ping timeout: 240 seconds) 2010-11-28T05:02:02 i went up to 211 :P 2010-11-28T05:02:44 I'm down to 2226 :D 2010-11-28T05:03:24 http://ai-contest.com/visualizer.php?game_id=8861416 lol he's just giving everything away xD 2010-11-28T05:03:50 haha 2010-11-28T05:04:06 hah 2010-11-28T05:04:30 *** RubiconCross has joined #aichallenge 2010-11-28T05:04:34 http://tinyurl.com/antimatroidsbot 2010-11-28T05:04:41 that's my code, exactly as submitted 2010-11-28T05:05:07 "Sorry, the page (or document) you have requested is not available." 2010-11-28T05:05:15 *** _Utkarsh <_Utkarsh!~admin@117.201.93.181> has joined #aichallenge 2010-11-28T05:05:32 try now? 2010-11-28T05:05:44 yeah, works now 2010-11-28T05:06:07 kk, i hate the sharing set to people with the link, but of course it doesn't count if it redirects from tinyurl :P 2010-11-28T05:06:12 had* aha 2010-11-28T05:07:01 it's a bit of a mess, but meh 2010-11-28T05:07:03 it works 2010-11-28T05:07:31 *** Utkarsh has quit IRC (*.net *.split) 2010-11-28T05:07:31 *** ulope has quit IRC (*.net *.split) 2010-11-28T05:07:57 *** krokokru1a has joined #aichallenge 2010-11-28T05:08:01 *** Rubicon-|-Cross has quit IRC (Ping timeout: 240 seconds) 2010-11-28T05:08:03 *** AlliedEnvy_ has quit IRC (Ping timeout: 264 seconds) 2010-11-28T05:08:03 *** krokokrusa has quit IRC (Ping timeout: 264 seconds) 2010-11-28T05:08:19 *** AlliedEnvy_ has joined #aichallenge 2010-11-28T05:08:32 yay, 2/2 2010-11-28T05:09:02 ofcourse, both peopel I beat have suck ratings though :\ 2010-11-28T05:09:17 * burny has a flash back to the fireswamp 2010-11-28T05:09:32 got my third win :) 2010-11-28T05:09:50 2nd win, still ranked 2226 2010-11-28T05:09:54 janzert: ahh ok 2010-11-28T05:10:06 I guess it knows my bot is going to suck :P 2010-11-28T05:10:40 i have no idea how my bot will do against the other top 200 bots 2010-11-28T05:10:47 http://ai-contest.com/forum/viewtopic.php?f=17&t=1177 2010-11-28T05:10:51 it better at least get top 50 :\ 2010-11-28T05:11:10 same here anti.. I only got to see 1 game before final hehe 2010-11-28T05:12:18 I'm thinking I'm gona be aroudn 100-120 though 2010-11-28T05:12:31 MAYbe as high as 50-100, but I doubt it 2010-11-28T05:12:51 i think my bot is crash free 2010-11-28T05:12:57 and i'm pretty sure it shouldn't eve time out 2010-11-28T05:12:58 burny: what's ur botname on tcp? 2010-11-28T05:13:07 it's not on tcp 2010-11-28T05:13:16 in the past 2010-11-28T05:13:21 *** krokokru1a is now known as krokokrusa 2010-11-28T05:13:24 never was 2010-11-28T05:15:13 Error323: it's a one week effort 2010-11-28T05:15:18 and a pretty decent one 2010-11-28T05:15:31 Error323: mmm? what is 2010-11-28T05:15:42 heh 2010-11-28T05:15:48 janz: how hard would it be to make a chart of elo vs time per game(time, rather than turns) 2010-11-28T05:16:12 i doubt they recrd that 2010-11-28T05:16:16 burnys bot 2010-11-28T05:16:27 burny: not going to happen now 2010-11-28T05:16:53 burny: yeah, not possible game time length isn't recorded 2010-11-28T05:17:56 well, time to get to bed 2010-11-28T05:18:27 as much fun as it is to watch my bot slowly win games, as it drops in the ranks.. 2010-11-28T05:20:00 Top 10 players: csr_22(1319), noripcord7(1246), fontanini(1229), arekkap(1227), rastafarra(1227), .mamay.(1225), dustypup(1218), EricManley(1216), b1uespeck(1214), pomppis(1178) 2010-11-28T05:20:00 burny: don't worry bocsimacko is down to 2070th ;) 2010-11-28T05:20:46 2799 for me, woot! 2010-11-28T05:21:17 http://ai-contest.com/visualizer.php?game_id=8862368 2010-11-28T05:21:30 how does zheka manage to lose that :| 2010-11-28T05:21:36 finalyl someone in the top20 that belongs in top20? 2010-11-28T05:21:40 dunno but thank fuck :P 2010-11-28T05:21:56 what a comeback 2010-11-28T05:22:12 he pretty much just spits on you at the start 2010-11-28T05:22:25 not so much on.. as beside 2010-11-28T05:22:33 i don't want to be coming back yet :( 2010-11-28T05:22:48 haha, attacking the top right from the bottm left 2010-11-28T05:22:54 that did it 2010-11-28T05:23:21 well, just attacking stuff on the opposite side of the board in various places 2010-11-28T05:23:35 man, that was a close one to ruining my rank early 2010-11-28T05:23:46 I imagine he adds score based on purely on shipcount 2010-11-28T05:24:43 I duno.. looks like you dominated that one really 2010-11-28T05:24:51 actually, i'm not sure i played that badly there, he made stupid moves the whole time 2010-11-28T05:25:09 and i waited to ruin them 2010-11-28T05:25:26 it was sort of dangerous, but wasn't terrible 2010-11-28T05:25:29 could have done a lot better(had hundred+ at base whole game) 2010-11-28T05:25:38 he was well ahead on ship count and production 2010-11-28T05:25:43 turn 4 kinda makes it for him 2010-11-28T05:26:14 why turn 4? 2010-11-28T05:26:15 if the opponent made bad moves at the start, just to throw you off, it coulda totally stomed you 2010-11-28T05:26:26 sigh: look where he tries to expand 2010-11-28T05:26:58 if planet can recoop it's production*2 before the enemy gets to it, it's better to take it, rather than wait for enemy to snipe it 2010-11-28T05:27:09 moves before that allow him to get ahead of antimadroid 2010-11-28T05:27:15 i'll be much more comfortable when i'm playing against better ranked bots 2010-11-28T05:27:24 it's not got rage bot checks or anything anymore 2010-11-28T05:27:24 ll 2010-11-28T05:27:27 *lol 2010-11-28T05:27:30 although i do try to not make snipable moves 2010-11-28T05:27:56 bury: yes, but it's messy to do that when incoming fleets might be arriving before 2010-11-28T05:28:00 i couldn't be bothered dealing with all of that 2010-11-28T05:28:07 there's a 13cost 3grow right beside your base, that you ignored for like 30 turns 2010-11-28T05:28:48 if i took that first turn and they did nothing, they'd snipe my 24 2010-11-28T05:28:48 took enemy 31moves to get to it: it coulda produced 90ships by then 2010-11-28T05:28:53 and the 77 5 growth looked bette 2010-11-28T05:29:08 then like i've said before, i'm severely capped on the number of actual moves my bot can make each move 2010-11-28T05:29:17 what's iouri's username? 2010-11-28T05:29:17 hence i'll be more comfortable once my opponents are better 2010-11-28T05:29:22 _iouri 2010-11-28T05:29:42 look out for asvaisus or whatever 2010-11-28T05:29:53 thanks 2010-11-28T05:29:56 not sure how to actually spell it 2010-11-28T05:30:00 oh, so if someone just expands like made and then pukes at you, you'll lose for sure? 2010-11-28T05:30:28 don't think so? 2010-11-28T05:36:51 oops, just realized I was sorting the player order the wrong direction so it wanted to play the player with the most games first instead of last 2010-11-28T05:36:54 all fixed now 2010-11-28T05:37:16 -_- 2010-11-28T05:37:22 :P 2010-11-28T05:37:37 bocs up to 282 now 2010-11-28T05:37:40 so someone has a lot of games? 2010-11-28T05:37:46 fortunately I caught it with when they had only gotten 5 games for the leader 2010-11-28T05:38:01 ah, that's not too bad 2010-11-28T05:38:25 nah, it'll even out again quite quick 2010-11-28T05:38:35 burny: link for bocs user profile? 2010-11-28T05:38:46 wouldn't that have made it so one person would likely have like 5000 games now? 2010-11-28T05:39:05 sigh: search for bocs in the user search on the sidebar 2010-11-28T05:39:05 http://ai-contest.com/profile.php?user_id=8565 2010-11-28T05:39:24 oh haha, when was that added? 2010-11-28T05:39:37 last week? 2010-11-28T05:39:41 not too long ago 2010-11-28T05:39:55 :) 2010-11-28T05:40:25 damn 2010-11-28T05:40:27 manwe beat me 2010-11-28T05:41:51 i threw the game too 2010-11-28T05:41:53 http://ai-contest.com/visualizer.php?game_id=8863622 2010-11-28T05:41:56 stupidest move ever 2010-11-28T05:42:42 that's an interesting map 2010-11-28T05:42:43 * krokokrusa is running local tournament between his 2 bots, ragebot, error323, and antimatroid :) 2010-11-28T05:42:58 let me know the results :) 2010-11-28T05:43:13 ouch, it didn't notice all those nasty ships sitting next it? 2010-11-28T05:43:25 janzert: I think it decided that planet was a lost cause 2010-11-28T05:43:28 janzert: nope, could never get it to do that well 2010-11-28T05:43:47 sigh: yeah the game tree would have, but still :( 2010-11-28T05:44:25 I remember similar stuff happened when I tried to implement code to abandon planets 2010-11-28T05:44:41 my guess is the game tree is sayin, i can take that, they can then take me, they were going to take me if i did nothing, hence doing my move is better 2010-11-28T05:44:59 a talking tree! 2010-11-28T05:45:02 although it probably wasn't 2010-11-28T05:45:49 it's alright, if i'm going to lose to someone, i'd rather it be someone like manwe who is going to be ranked reasonably high 2010-11-28T05:46:00 :) 2010-11-28T05:46:35 kinda surprised already starting to see some familiar names in the top 100 2010-11-28T05:46:52 I see in the source code that you use the command "md5sum" rather than pythons hashlib 2010-11-28T05:46:55 is it really faster? 2010-11-28T05:47:18 proc = Popen(["md5sum", filename], stdout=PIPE) 2010-11-28T05:47:36 bobng: not sure where you're seeing that but I'm pretty certain it wasn't written by me 2010-11-28T05:47:48 submission_hash.py 2010-11-28T05:47:51 or maybe that was, where is this 2010-11-28T05:47:55 yeah it was 2010-11-28T05:47:56 # suprisingly in testing this is slower than doing the file hash in python 2010-11-28T05:47:56 # this hash value will include the full absolute path to the file 2010-11-28T05:48:08 oh wait, wat, its slower? 2010-11-28T05:48:29 just sec let me open it and refresh my memory 2010-11-28T05:48:43 this has caught my interest. lemme run a test and see 2010-11-28T05:49:32 yes, that was slower calling md5sum instead of using the python hash 2010-11-28T05:49:56 jan, after about 8games, teh top100 list should have most of the correct people 2010-11-28T05:50:02 look at def hash_file() and you see I ended up using the internal sha hash 2010-11-28T05:50:30 *** krokokrusa has quit IRC (Read error: Connection reset by peer) 2010-11-28T05:50:34 oh right, I see now 2010-11-28T05:50:42 i don't know about 8 games 2010-11-28T05:50:45 were you really surprised? 2010-11-28T05:51:01 and after 10games, it should resemble the correct order also 2010-11-28T05:51:39 burny: that depends on a lot of variable though 2010-11-28T05:51:47 bobng: yes, a bit 2010-11-28T05:52:16 *** krokokrusa has joined #aichallenge 2010-11-28T05:52:17 since I didn't think reading the whole file into python then hashing it would be as fast as md5sum should have been able to do it 2010-11-28T05:52:43 blah.. I still don't have a 3rd game 2010-11-28T05:52:52 it only takes 2010-11-28T05:52:55 I have 3 :) 2010-11-28T05:52:56 oh i do 2010-11-28T05:52:59 1.1 seconds to hash a 100mb file on my machine 2010-11-28T05:53:01 also a win :D 2010-11-28T05:53:09 from mister epic fail 2010-11-28T05:53:10 :P 2010-11-28T05:53:17 Pythons hashlib is actually a C module, so it should be pretty fast 2010-11-28T05:53:24 pretty sure the pairing, isn't giving a game to the person with least games 2010-11-28T05:53:53 janzert said he fixed that 2010-11-28T05:53:55 http://ai-contest.com/forum/viewtopic.php?f=17&t=1177&p=7809#p7809 2010-11-28T05:53:58 :) 2010-11-28T05:54:08 I updated my post with rank,language etc 2010-11-28T05:55:26 heh, I eat my words. Spawning md5sum is faster than doing it in python 2010-11-28T05:55:30 by a huge factor. 2010-11-28T05:55:35 servers all broked? 2010-11-28T05:55:40 >>> test_two() 2010-11-28T05:55:40 9b5183f05b62ca114c467945467050be 100mb.test 2010-11-28T05:55:40 0.351421117783 2010-11-28T05:56:06 compared to 1.30111789703 2010-11-28T05:56:13 when using hashlib. 2010-11-28T05:57:29 bobng: hmm, there must be some other factors since I had the reverse 2010-11-28T05:57:58 well I was reading 100 bytes at a time and updating the hash object, that might be why. 2010-11-28T05:58:02 burny: actually I broke the create_matchup script when adding some logging with my last fix to pairing 2010-11-28T05:58:27 bobng: ouch, yeah that's gotta be slowing it down 2010-11-28T05:58:36 but in any case it academic. I was doing it on a 100mb file though 2010-11-28T05:58:54 perhaps on a smaller file the overhead of starting a md5 process will make it worthless 2010-11-28T05:59:04 out of curiosity try my actual hash_file_sha from submission_hash.py 2010-11-28T05:59:29 and yeah, this is almost exclusively used on quite small files 2010-11-28T05:59:51 sure, give me a sec 2010-11-28T06:01:19 *** choas has joined #aichallenge 2010-11-28T06:04:54 *** fgump has joined #aichallenge 2010-11-28T06:07:09 yay i'm #100 2010-11-28T06:07:10 lol 2010-11-28T06:07:39 i'm at 182 2010-11-28T06:07:53 i'm rooting for manwe now :) 2010-11-28T06:07:58 the better he does, the better i do 2010-11-28T06:08:17 error: when I try to compile your code on cygwin with just "make", it first says somethign like unrecognized option -rdynamic, then gives a boatload of other errors, and then just quits :p 2010-11-28T06:08:43 Mistmanov: just remove rdynamic from the makefile 2010-11-28T06:09:37 that takes care of the rdynamic error =) 2010-11-28T06:09:59 Mistmanov: let me know the result please :D 2010-11-28T06:10:08 still a boatload of errors to go :p 2010-11-28T06:10:19 what errors 2010-11-28T06:10:24 execinfo.h: no such file or directory 2010-11-28T06:10:32 ahhh myes 2010-11-28T06:10:52 *** fawek has joined #aichallenge 2010-11-28T06:11:06 that's linux specific 2010-11-28T06:11:09 ah k 2010-11-28T06:11:13 ok here's what you can do 2010-11-28T06:11:22 jan: should save a screenshot of the top100 list at about 8am and 10am(for the 6game and 8game count?).. I'm kinda curious of how accurate my prediction is, on what it'll look like at that point 2010-11-28T06:11:23 Seems like a lot of people are using regular knapsack01 as a starting move :| quite surprising considering they seem to be top 50ish 2010-11-28T06:11:28 janzert: my tests show that md5 is faster than sha1 2010-11-28T06:11:31 in the makefile remove -DDEBUG 2010-11-28T06:11:34 "alot" being the ones who posted their source 2010-11-28T06:11:34 do we have a copy of the board before rseting it also? 2010-11-28T06:11:47 specifically that 2010-11-28T06:11:47 >>> def hash_md5(): 2010-11-28T06:11:47 ... f = open("100mb.test") 2010-11-28T06:11:47 ... hash = md5.new() 2010-11-28T06:11:48 ... while True: 2010-11-28T06:11:48 ... d = f.read(BUFFER_SIZE) 2010-11-28T06:11:48 ... if not d: break 2010-11-28T06:11:48 ... hash.update(d) 2010-11-28T06:11:48 ... return hash.hexdigest() 2010-11-28T06:11:56 is faster than a near exact copy of your sha1 hash function 2010-11-28T06:12:09 at least with a BUFER_SIZE of under 10,000 2010-11-28T06:13:03 need to use similiar buffer sizes to compare fairly of course, but yeah I'm not too surprised 2010-11-28T06:13:47 yeah, they both use the same buffer size 2010-11-28T06:14:08 and also md5sum seems to be consistently just a bit faster than hashlib 2010-11-28T06:14:14 >>> run_func(hash_sha) 2010-11-28T06:14:14 0.461477994919 2010-11-28T06:14:18 >>> run_func(hash_md5sum) 2010-11-28T06:14:18 0.355353116989 2010-11-28T06:14:22 >>> run_func(hash_md5) 2010-11-28T06:14:22 0.393083095551 2010-11-28T06:14:28 so meh. 2010-11-28T06:14:37 interesting 2010-11-28T06:14:44 lols.. http://ai-contest.com/visualizer.php?game_id=8862223 2010-11-28T06:14:47 not that it makes much difference of course. 2010-11-28T06:14:50 *** smloh1 has joined #aichallenge 2010-11-28T06:15:03 so I wonder if there is a crossover point in size of files or just was different for me 2010-11-28T06:15:04 *** fawek has quit IRC (Ping timeout: 245 seconds) 2010-11-28T06:15:17 oh, you are running each with a cold disk cache right? 2010-11-28T06:15:19 just over 4600 bots in the end 2010-11-28T06:15:23 quite a nice improvement on tron 2010-11-28T06:15:36 antimatroid1: how much did tron have? 2010-11-28T06:15:42 like 750 i think 2010-11-28T06:16:07 but even here, it's only really competitive in the top 100 2010-11-28T06:16:32 bobng: to clear the cache under linux for testing you can use "sync; echo 3 > /proc/sys/vm/drop_caches" before each test 2010-11-28T06:16:48 as root 2010-11-28T06:16:56 the top 300ish, seem to all have logic for defending planets 2010-11-28T06:17:17 and most of those, also logic for avoiding snipable planets 2010-11-28T06:17:32 but yeah, it doesn't really make much a difference at this point 2010-11-28T06:18:10 sec, lemme clear the cache 2010-11-28T06:18:23 the last hour had 4400gmes, and I've only had 1 game per hour 2010-11-28T06:18:33 shouldn't it be an average of nearly 2games per hour? 2010-11-28T06:19:12 bah, write error. whatever, it is purely academic 2010-11-28T06:19:25 yeah 2010-11-28T06:19:25 and i'm just procrastinating 2010-11-28T06:19:29 hehe 2010-11-28T06:19:58 i'm dreading my next game being a loss 2010-11-28T06:20:01 Top 10 players: dabino(1433), bpellom(1405), hutabhu(1375), rshoesmi(1375), sarigalin(1374), SloppyBot(1364), certhia(1362), tunasilver(1354), lasantha(1337), lavalamp(1332) 2010-11-28T06:20:14 er, I guess the first couple hours, had a bunch less games per hour 2010-11-28T06:20:40 *** Appleman1234 has quit IRC (Remote host closed the connection) 2010-11-28T06:20:47 woot 4th game 2010-11-28T06:20:52 burny: the first couple hours had 0 games per hour 2010-11-28T06:21:17 *** fawek has joined #aichallenge 2010-11-28T06:21:40 but yes, I've been gradually bringing more workers online since games started as well 2010-11-28T06:22:02 ok.. teh second couple hours then :P 2010-11-28T06:22:16 where is the stats page 2010-11-28T06:22:25 @gamestats 2010-11-28T06:22:26 janzert: gamestats = http://ai-contest.com/game_stats.php 2010-11-28T06:22:27 http://ai-contest.com/game_stats.php 2010-11-28T06:22:30 thanks 2010-11-28T06:22:42 woah, thats a lot of gpm 2010-11-28T06:22:56 I remember when it was like 1 game pm 2010-11-28T06:23:10 so is the average game like, 5seconds or so? 2010-11-28T06:23:16 some nice comments in mogron's commit log :) "30e205a no idea what i changed. am now at 617/700 in run_tests5.sh" 2010-11-28T06:23:19 with only the web server running. 2010-11-28T06:23:37 or are multiple games running on each server, with multithreads etc.? 2010-11-28T06:23:40 where did all this extra computing power come from? 2010-11-28T06:23:46 burny: right around there 2010-11-28T06:24:04 2 simultaneous games per server generally 2010-11-28T06:24:28 *** Appleman1234 has joined #aichallenge 2010-11-28T06:24:30 bobng: same as it has been, amazon ec2 :) 2010-11-28T06:24:41 Jeff approved firing up more for the finals 2010-11-28T06:24:42 is there memory limits? 2010-11-28T06:24:45 oh nice 2010-11-28T06:24:56 burny: there are always memory limits 2010-11-28T06:24:58 you know you could each get sign up and get one of their free vps's 2010-11-28T06:25:16 or, what happens if one bot allocates tons of memory, and thrashes it, while it's opponent trys to maximize it's usage of processing time 2010-11-28T06:25:20 i don't know how much memory running the engine would take, but they come with like 600mb 2010-11-28T06:25:29 *** fawek has quit IRC (Ping timeout: 245 seconds) 2010-11-28T06:25:34 it's the cpu that would be more of a problem 2010-11-28T06:25:53 yeah, i use the cpu pretty much the whole time 2010-11-28T06:25:54 the micro's are allowed to burst but if they stay at high load for a long period they get throttled back 2010-11-28T06:26:01 if the system has 600meg,s and I allocate a gig of int's, and sort it.. 2010-11-28T06:26:01 which is only like 0.5 seconds against good bots 2010-11-28T06:26:26 that'll hurt the performance of the opponent(and the other saps, that happen to be on the same server) 2010-11-28T06:26:50 janzert: cool how many servers? 2010-11-28T06:27:03 5 right at the moment 2010-11-28T06:27:20 I'll probably kick another one off in a bit 2010-11-28T06:27:54 dimkadimon is having a rough go :( 2010-11-28T06:28:18 how much $$ 2010-11-28T06:28:19 anyone remember where he was ranked? 2010-11-28T06:28:20 is this costing? 2010-11-28T06:28:38 I wonder how many of the top50 bots are gona drop a bunch, due to being excessivly tweaked for the test maps 2010-11-28T06:28:42 roughly 1.50 per worker per hour 2010-11-28T06:29:21 a draw with asavis 2010-11-28T06:30:20 janzert: lower end of the top 100 2010-11-28T06:30:22 interesting final moves there by both of you :) 2010-11-28T06:31:11 that was a draw map, I guess 2010-11-28T06:32:29 dimka is having a rough time 2010-11-28T06:32:30 bocs would have lost, if asavis made an attack just hoping for the best 2010-11-28T06:32:38 he is my unofficial rival, but damn that's rough aha 2010-11-28T06:33:38 it is a pretty bad map 2010-11-28T06:33:52 oh actually nm, even if he lost the planet in the final turn: wouldn't have lost any growth right? 2010-11-28T06:34:03 at least it was a draw with another top ranked bot 2010-11-28T06:34:18 mega_1: I think it's a draw map too 2010-11-28T06:34:44 pretty sure my bot would lose on that map 2010-11-28T06:35:03 cause once population was up to about 200, it'd want to attack something 2010-11-28T06:35:40 *** Ice_Harley has quit IRC (Ping timeout: 265 seconds) 2010-11-28T06:35:42 burny: i don't think you're going to get your correct top 100 after 7 games 2010-11-28T06:36:44 anti: I said the top100 would have most the right people after 6.. and be in roughly the right order after 8 2010-11-28T06:36:50 yay, i won my 6th game 2010-11-28T06:36:55 against euthanasia aha 2010-11-28T06:37:07 burny: the right people is what i mean 2010-11-28T06:37:07 it 2010-11-28T06:37:11 s still nowhere near that 2010-11-28T06:37:26 I think we'll have to wait at least 20 games before having most of the right people in top 100 2010-11-28T06:37:42 *** asdfghv has joined #aichallenge 2010-11-28T06:37:47 it's only at about 4 games right now 2010-11-28T06:37:48 I think therell be surprises 2010-11-28T06:37:57 Zoo: I hope so =) 2010-11-28T06:38:03 burny: i've had 6 games 2010-11-28T06:38:16 oh, there will 2010-11-28T06:38:19 I'm sure some people still only have 3 2010-11-28T06:38:23 i hope there are surprised, but i hope bocsi gets his win 2010-11-28T06:38:26 i think he deserves it 2010-11-28T06:38:33 Doraj8 has 5 wins and he's #813, but I think he's at least a top 50 bot 2010-11-28T06:39:37 well, anyone have a list of the top100, to compare it to ina bout 2 hours? 2010-11-28T06:39:50 Error323: you were @ #37 for a while ;) 2010-11-28T06:39:52 nope sorry 2010-11-28T06:40:15 Error323: ouch, you lost :P 2010-11-28T06:40:33 *** BaronTrozo has joined #aichallenge 2010-11-28T06:40:43 BaronTrozo: wtf is rufes band? 2010-11-28T06:40:56 there's so damn many of you :P 2010-11-28T06:41:16 the final open rankings output from bayeselo is at http://ai-contest.com/final-open-ratings.txt 2010-11-28T06:41:32 cool janz 2010-11-28T06:41:37 as you can see though a lot of the top bots at that point had hardly any games 2010-11-28T06:41:54 lol, my being in 41 there was off 2 wins 2010-11-28T06:42:01 one against deepblue crashing 2010-11-28T06:42:17 The Rufes Band is agroup of friends taht have work in a comon Company, but we don't want to metion the name of the company 2010-11-28T06:42:34 Microsoft? :) 2010-11-28T06:42:38 ahh that makes sense, well you guys win for a group i think :P 2010-11-28T06:42:46 hope so 2010-11-28T06:42:48 :P 2010-11-28T06:43:13 Not a big company 2010-11-28T06:43:22 why the anonymity? 2010-11-28T06:43:34 that is one impressive company 2010-11-28T06:43:52 we are not happy with our company so we don't want publicity 2010-11-28T06:43:54 They don't want to let their boss know what they spent their time on during the last 2 months :P 2010-11-28T06:44:07 ahh, i almost felt the same about my uni 2010-11-28T06:44:13 antimatroid1: Zoo yeehhh :/ it happens 2010-11-28T06:44:24 well you guys should start your own company then 2010-11-28T06:44:52 *** mceier has joined #aichallenge 2010-11-28T06:44:53 haha 2010-11-28T06:44:55 http://ai-contest.com/visualizer.php?game_id=8865197 2010-11-28T06:45:00 nicely done 2010-11-28T06:45:06 BaronTrozo: well ur bot was good on tcp, I hate you ;) 2010-11-28T06:45:23 yeah, it seemed to beat me more than i beat it 2010-11-28T06:46:09 not so good 2010-11-28T06:46:13 yay, victories 5/5, #174 2010-11-28T06:46:29 i'm at 307 thanks to losing to manwe 2010-11-28T06:47:12 yeah I lost my first aswell damnit 2010-11-28T06:47:19 *** heikoh has joined #aichallenge 2010-11-28T06:47:29 antimatroid1: i'm sure manwe will be up there so you shouldn't have anything to worry about 2010-11-28T06:48:05 smloh1: yep, i'm now rooting for him :) 2010-11-28T06:48:43 i just need to beat sigh dimkadimon and smiley 2010-11-28T06:48:49 yeah, based on tcp server i'm hoping to end/stay in top 100 2010-11-28T06:48:50 smiley will probably be the biggest challenge 2010-11-28T06:49:10 pesky australian competitors 2010-11-28T06:49:17 dimkadimon is the worst, he goes to anu and used java 2010-11-28T06:49:52 lol 2010-11-28T06:50:12 i think you'll have a harder time with smiley 2010-11-28T06:50:44 yeah, i meant worst as in people i want to anialate the most 2010-11-28T06:50:52 smiley will certainly place higher than him 2010-11-28T06:51:01 *** RubiconCross has quit IRC (Read error: Connection reset by peer) 2010-11-28T06:53:12 manwe beat raschi 2010-11-28T06:53:19 bonus points for me :) 2010-11-28T06:53:57 I wonder how my multiturn knapsack works agaisnt everyone else's regular knapsack01 now that the rankings have reseted :o~ 2010-11-28T06:54:49 well I use knapsack merely on the first turn 2010-11-28T06:55:00 same here 2010-11-28T06:55:01 i don't use knapsack at all 2010-11-28T06:55:02 what is multiturn knapsak? 2010-11-28T06:55:07 well, not first turn but first expansion 2010-11-28T06:55:11 and skip planets that are in the center for capturing so i can snipe them off 2010-11-28T06:55:15 optimise for n turns in the future? 2010-11-28T06:55:20 and im not sure if it can be called knapsack anymore 2010-11-28T06:55:22 antimatroid1: yeah 2010-11-28T06:55:30 i see 2010-11-28T06:55:37 most profit in n turns 2010-11-28T06:55:41 i tried dynamic programming for optimal ships to send for given feasible amounts 2010-11-28T06:55:44 but it was REALLY slow 2010-11-28T06:55:52 no way it would work for move selection in a game tree 2010-11-28T06:56:01 looks like roughly 30% of the original top100 are in the top100 now.. and about 50% of the top100, were in the top200 2010-11-28T06:56:34 my version of knapsack uses dynamic programming 2010-11-28T06:56:42 knapsack is dp :P 2010-11-28T06:57:12 I figure he means variating conditions 2010-11-28T06:58:03 http://ai-contest.com/visualizer.php?game_id=8870086 kthxbye 2010-11-28T06:58:23 quick 2010-11-28T06:58:36 quickest here is 13 turns 2010-11-28T06:58:52 and a 12 turn loss agaisnt accoun =/ 2010-11-28T06:59:07 on tcp that is 2010-11-28T06:59:24 did i fail on the first move or is that a draw map too? 2010-11-28T07:00:03 janzert: how are the draws looking so far? 2010-11-28T07:00:14 dunno, you can take something crappy like the 4-neutral, i suppose? 2010-11-28T07:00:23 i thnk it's 0 growth 2010-11-28T07:00:28 oh 2010-11-28T07:00:34 either way, looks like draw 2010-11-28T07:00:52 that's 2 draw maps we've seen already D: 2010-11-28T07:00:53 highest is 12.9% with this map http://ai-contest.com/visualizer.php?game_id=8870321 2010-11-28T07:00:56 the 19's: are they 5grow? 2010-11-28T07:01:02 fuck that's three 2010-11-28T07:01:12 hmmm 2010-11-28T07:01:23 http://spz.nofate.com/planetwars/visualizer/?game_id=8870086&server=Official 2010-11-28T07:01:38 @svn 2010-11-28T07:01:39 Error323: I do not know about 'svn', but I do know about these similar topics: 'spec' 2010-11-28T07:01:47 @spec 2010-11-28T07:01:48 Error323: spec = http://ai-contest.com/specification.php 2010-11-28T07:01:57 @code 2010-11-28T07:01:58 Error323: I have no idea what you mean. 2010-11-28T07:02:00 @repo 2010-11-28T07:02:00 @repo 2010-11-28T07:02:00 Mistmanov: repo = http://code.google.com/p/ai-contest/ 2010-11-28T07:02:01 krokokrusa: repo = http://code.google.com/p/ai-contest/ 2010-11-28T07:02:03 * antimatroid1 goes to watch a whole bunch of games to look at maps 2010-11-28T07:02:03 ah 2010-11-28T07:02:04 ty 2010-11-28T07:02:46 *** _Utkarsh <_Utkarsh!~admin@117.201.93.181> has quit IRC (Ping timeout: 245 seconds) 2010-11-28T07:02:59 anti, I think taking the 19 if it's 5grow, is possibly the best first move, but most likely, it's a draw map 2010-11-28T07:03:09 antimatroid1: I got the same map: http://www.ai-contest.com/visualizer.php?game_id=8870454 2010-11-28T07:03:43 bocs is in the top100 now 2010-11-28T07:04:03 sigh: interesting, maybe i was just lucky i played a crap bot on that map 2010-11-28T07:05:02 actually, sigh: if they didn't send their move on the second turn 2010-11-28T07:05:06 could they have owned you? 2010-11-28T07:05:15 *** Appleman1234 has quit IRC (Quit: Leaving) 2010-11-28T07:05:20 3/4 turn sorry 2010-11-28T07:05:36 I haven't calculated, but I shouldn't think so 2010-11-28T07:05:47 *** rock69 has joined #aichallenge 2010-11-28T07:05:48 oh.. ya.. for sure taking the 19 is the best first move 2010-11-28T07:05:59 janzert: I'm playing against your bot on arimaa^^ 2010-11-28T07:06:10 medrimonia: that was fast 2010-11-28T07:06:11 how is it? 2010-11-28T07:06:16 i'll get stuck into that soon 2010-11-28T07:06:36 *** aizu has joined #aichallenge 2010-11-28T07:06:39 i've just checked the leaderboard. what happened to bocsimacko and everybody else? everyhing has changed! 2010-11-28T07:06:47 I'll watch the game^^ 2010-11-28T07:06:51 rock69: the final tournament has started 2010-11-28T07:07:02 the leaderboard is still just resolving things 2010-11-28T07:07:02 sigh's opponent there was lousy 2010-11-28T07:07:22 had advantage in the middle, and didn't even defend their base 2010-11-28T07:07:22 oh, ok. had no idea :) 2010-11-28T07:07:25 you played against me antimatroid1:? 2010-11-28T07:07:46 where? 2010-11-28T07:07:54 *** rock69 has quit IRC (Client Quit) 2010-11-28T07:08:06 you told me: "that was fast" 2010-11-28T07:08:10 What was fast? 2010-11-28T07:08:16 medrimonia: cool, beat OpFor around a bit 2010-11-28T07:08:33 *** Utkarsh has joined #aichallenge 2010-11-28T07:08:37 someone wana save a copy of the top100 in about an hour, and again another 2hours after that? 2010-11-28T07:08:49 http://fhuizing.no-ip.org/final_maps.zip these are the official final maps 2010-11-28T07:08:59 didn't you just ask me like yesterday what arimaa is? 2010-11-28T07:09:01 *** smloh1 has quit IRC (Ping timeout: 265 seconds) 2010-11-28T07:09:26 Error323: how are you in the top 100 atm and i'm not :P 2010-11-28T07:09:29 ah yes antimatroid1^^ 2010-11-28T07:09:40 I'm beginning to be interested^^ 2010-11-28T07:09:51 I've won 6/6 and still 518 :P 2010-11-28T07:10:08 medrimonia: cool, i expect i will too 2010-11-28T07:10:21 is arimaa the next game? 2010-11-28T07:10:34 antimatroid1: dunno, just wait :) 2010-11-28T07:10:45 sigh: no, it's another ai thing 2010-11-28T07:10:59 http://en.wikipedia.org/wiki/Arimaa 2010-11-28T07:11:03 *** Muffi has joined #aichallenge 2010-11-28T07:11:08 I know the game :) 2010-11-28T07:11:11 anti: they lost against a higher ranked person than you did: and also beat higher ranked people than you 2010-11-28T07:11:19 hi 2010-11-28T07:11:22 snif I've got to eat, I'll must beat OpFor2008 later 2010-11-28T07:11:37 will there be any chance before the contest ends to fix a bug in the code? 2010-11-28T07:11:46 Muffi: no 2010-11-28T07:11:48 no 2010-11-28T07:11:51 argl 2010-11-28T07:11:53 Muffi: sorry all submissions are now closed 2010-11-28T07:11:58 those days are gone =) 2010-11-28T07:12:05 *** Rubicon-|-Cross has joined #aichallenge 2010-11-28T07:12:08 *** smloh1 has joined #aichallenge 2010-11-28T07:12:09 ...two characters :-\ 2010-11-28T07:12:13 :( 2010-11-28T07:12:17 ok, thanks 2010-11-28T07:12:18 i know how it feels 2010-11-28T07:13:26 *** Appleman1234 has joined #aichallenge 2010-11-28T07:13:52 *** CACA has joined #aichallenge 2010-11-28T07:13:59 hello 2010-11-28T07:14:02 2nd question. If that contest starts again some day, will it be the same game? Is it worth to keep the code? 2010-11-28T07:14:19 it'll be another game 2010-11-28T07:14:23 ok 2010-11-28T07:14:25 the last game was tron 2010-11-28T07:14:35 i wouldn't just delete the code though 2010-11-28T07:15:16 please tell me is it possible somewhow to upload corrected code - i made stupid division by zero, and now there is 3 crashes (but ship count is on my side :-( 2010-11-28T07:15:22 please 2010-11-28T07:15:23 plase 2010-11-28T07:15:26 hihi 2010-11-28T07:15:30 i asked the same 2010-11-28T07:15:45 no 2010-11-28T07:15:45 http://ai-contest.com/visualizer.php?game_id=8870418 2010-11-28T07:15:47 ouch 2010-11-28T07:16:21 that would have hurt bocsi temporarily too 2010-11-28T07:16:26 admins, pleaseeeeeeeeee 2010-11-28T07:16:48 CACA: sorry, it doesn't work to have people changing now 2010-11-28T07:16:58 but hey, a off by one when to capture a planet and anyhow rank 482 2010-11-28T07:17:10 could be worse ;-) 2010-11-28T07:17:18 CACA: Sorry the deadline is well and truely past 2010-11-28T07:17:31 Muffi: i got disqualified from tron for timing out on a single game, i expect to finish top 10 :( 2010-11-28T07:17:32 lol anti.. wtf was that? 2010-11-28T07:17:43 some poor person that was the second to last submission had there submission fail to compile :( 2010-11-28T07:17:58 burny: ? 2010-11-28T07:18:00 but admins can change rank counting rulesss. 2010-11-28T07:18:03 for example 2010-11-28T07:18:12 if there is division by zero 2010-11-28T07:18:12 rank counting rules? 2010-11-28T07:18:14 asavis just randomly decided to give the game away? 2010-11-28T07:18:17 janzert: to compile _in time_, I believe 2010-11-28T07:18:28 just ignore it- count ships BEFORE crash ?? 2010-11-28T07:18:43 but you can cry seeing all the time sending ships that ends with 0 units for the enemy *cry* 2010-11-28T07:18:46 krokokrusa: ? no it had a compilation error 2010-11-28T07:18:55 so teh best strat is to not attack anything.. and once opponent takes something, to crash? 2010-11-28T07:18:57 undefined variable if I remember right 2010-11-28T07:19:20 I figure it might be a good idea to get some sponsors for the next competition 2010-11-28T07:19:20 lol my boss is at #1 2010-11-28T07:19:25 janzert: oh, ok. I got he wrong idea from the forum 2010-11-28T07:19:31 this thing got huge publicity this time :P 2010-11-28T07:19:33 Error323: :P 2010-11-28T07:19:50 CACA: sorry, it's not going to change. It would be unfair to everyone that doesn't crash. 2010-11-28T07:20:02 Top 10 players: SloppyBot(1837), CorwinAlex(1793), Rexxar(1791), lavalamp(1779), ssieb0(1769), nextgeneration(1756), luismi(1756), Neverstu(1750), cobracom(1721), goldman(1715) 2010-11-28T07:20:11 cu 2010-11-28T07:20:24 *** Muffi has quit IRC (Quit: Page closed) 2010-11-28T07:21:08 why un-fair ? because their strategy is poor ??? ???? ??? 2010-11-28T07:21:34 CACA: you've had 3 months to eliminate bugs / get your bot ready, at some point there needs to be a deadline 2010-11-28T07:21:55 no caca: because there was plenty of time to test the stuff: and because if the rules were that way then the best strategy would be.. 2010-11-28T07:22:16 *** Appleman1234 has quit IRC (Remote host closed the connection) 2010-11-28T07:22:30 doTurn() {if (opponents.planetsOwned() > 1) return x/0;} 2010-11-28T07:23:15 fairly aggressive? http://ai-contest.com/visualizer.php?game_id=8870384 2010-11-28T07:23:19 some people didn't has such big time interval as 3 months - we somethimes had to work on our jobs ... 2010-11-28T07:23:58 i'm now 8/9 games 2010-11-28T07:24:01 that's not a bad start 2010-11-28T07:24:07 is the pairing right now purely random? or rank-based? 2010-11-28T07:24:13 i'm less nervous now 2010-11-28T07:24:19 rank based i think 2010-11-28T07:24:44 which is why i'm less nervous now 2010-11-28T07:25:03 i think i've done a sufficient amount to get to play the better bots soon enough 2010-11-28T07:25:10 smloh1: still rank based, but that is basically random 2010-11-28T07:25:20 the first games were of course purely random 2010-11-28T07:25:29 it's risky business having more games against bad bots for this 2010-11-28T07:25:37 *** aizu has quit IRC (Read error: Connection reset by peer) 2010-11-28T07:26:05 *** aizu has joined #aichallenge 2010-11-28T07:26:28 Error323: do not fear, your boss lost 2010-11-28T07:26:29 i guess there's a bit of luck there, since a loss to a good bot will hurt you less than a loss to a bad bot 2010-11-28T07:26:37 *** aizu has joined #aichallenge 2010-11-28T07:26:39 antimatroid1: yep 2010-11-28T07:26:46 well I know my bot is better 2010-11-28T07:26:47 yeah, but if you lose early, then you probably have to play more bad bots 2010-11-28T07:26:51 so thats good 2010-11-28T07:26:54 so you have a higher chance of losing more games to them 2010-11-28T07:26:56 @admins: could you change rank rules - to ignore crashes for All players and count ships until crash ? 2010-11-28T07:26:57 CACA: User error -- Replace user. 2010-11-28T07:26:59 which would really hurt in the long run i think 2010-11-28T07:27:34 CACA: of course not, that would mean i'd crash my bot as soon as my shipcount is higher 2010-11-28T07:27:37 becausxe in the real world - war is not lost, because of 'envifonment crash' :-P 2010-11-28T07:27:41 CACA: sorry as I already said, no we aren't going to do that 2010-11-28T07:27:44 well if you approached any kind of limit it probably wouldn't be a problem, but that isn't happening here 2010-11-28T07:28:08 :-( 2010-11-28T07:28:13 it is lost when your gun jams or engine on your plane quits though 2010-11-28T07:28:56 I propose drop everyone's scores when somebody divides by zero 2010-11-28T07:29:08 that's what happens in real world 2010-11-28T07:29:53 "so we were working on this super awesome new weapon, but our scientists made a miscalculation that caused a black hole to suck up our entire country. Still, since we had more tanks before the incident, I think you should surrender to us." 2010-11-28T07:30:16 i'd allow that argument 2010-11-28T07:30:22 xD 2010-11-28T07:30:22 just for its absurdity :P 2010-11-28T07:30:49 i propose - lets restart all competition then :-P 2010-11-28T07:30:58 CACA: sorry, never going to happen 2010-11-28T07:30:59 poor asavis.. loosing to a sub100 bot 2010-11-28T07:31:11 learn a lesson, don't make the mistake next time 2010-11-28T07:31:14 that's gota be an upseting loss 2010-11-28T07:31:18 i had to do the same last time, i know it sucks 2010-11-28T07:31:32 but it's not fair on other people to add you in now or restart 2010-11-28T07:31:42 lol, I'm 85, right behind bocsimacko 2010-11-28T07:31:46 ok, ok I was a stupid noob :-( 2010-11-28T07:31:49 burny: yep, he is a top 10 bot :| 2010-11-28T07:32:06 :( 2010-11-28T07:32:14 that bug looks like, a unintitlized variable, or buffer overflow, or some such 2010-11-28T07:32:23 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-11-28T07:32:42 i'm still not sure if my bot will crash at all, up until a few days ago it would crash randomly (like once every 10 games) 2010-11-28T07:32:48 but im so SAD and angry :-( 2010-11-28T07:32:49 but then the last few days it was gone, kapoof 2010-11-28T07:32:51 just see this 2010-11-28T07:32:53 http://ai-contest.com/visualizer.php?game_id=8865161 2010-11-28T07:33:06 looks like it's just about at the point where all bots have 6 games now 2010-11-28T07:33:11 or this 2010-11-28T07:33:12 http://ai-contest.com/visualizer.php?game_id=8866685 2010-11-28T07:33:31 thsi 2010-11-28T07:33:32 http://ai-contest.com/visualizer.php?game_id=8871031 2010-11-28T07:33:33 heh, your bot is fun to watch 2010-11-28T07:33:38 caca.. how about this? http://ai-contest.com/visualizer.php?game_id=8870418 2010-11-28T07:33:48 CACA: last contest i was disqualified for timing out, at least that hasn't happened 2010-11-28T07:34:08 janzert: will drawmaps be removed at some point? 2010-11-28T07:34:39 before i didn't think it'd be needed, but some of those maps before were pretty awful 2010-11-28T07:34:50 Error323: probably not, if some map is getting larger than 50% draws I could probably talked into it though 2010-11-28T07:35:14 yeah ok 2010-11-28T07:35:20 that'll never happen though :P 2010-11-28T07:35:23 although, i think it's possible alright to test some bots on draw maps, and a lot of people will fail on them anyway 2010-11-28T07:35:30 i don't trust my bot on a draw map at all, but we'll see 2010-11-28T07:35:34 or well... it might but is highly unlikely 2010-11-28T07:35:40 top draw percentage is 14% right now so :) 2010-11-28T07:36:06 http://ai-contest.com/visualizer.php?game_id=8873696 2010-11-28T07:36:10 that map 2010-11-28T07:36:34 CACA's bot is beautiful :) 2010-11-28T07:36:44 ahhh, yeah i would probably draw with people on that 2010-11-28T07:36:45 like blooming roses 2010-11-28T07:36:49 that centre planet is very appealing 2010-11-28T07:36:58 Zoo: i agree 2010-11-28T07:37:56 wow nextgeneration won from manwe 2010-11-28T07:37:57 nice 2010-11-28T07:38:10 *** sigh has quit IRC (Remote host closed the connection) 2010-11-28T07:38:22 bad! 2010-11-28T07:38:35 i want manwe to do well 2010-11-28T07:38:43 ah i'm starting to fight better bots, i'm my latest opponent's first loss 2010-11-28T07:38:54 not sure I mentioned it here, current map set is now posted on the forums under the submissions closed announcement thread 2010-11-28T07:38:57 wh000t 6 servers up 2010-11-28T07:39:08 how many maps? 2010-11-28T07:39:14 75 2010-11-28T07:39:20 cool 2010-11-28T07:40:01 http://ai-contest.com/visualizer.php?game_id=8874380 2010-11-28T07:40:03 holy fuck 2010-11-28T07:40:19 asavis is really doing badly 2010-11-28T07:40:34 told you therell be surprises 2010-11-28T07:40:36 pwned 2010-11-28T07:40:47 blame the tcp server and people optimizing there :) 2010-11-28T07:41:03 i've only played there in the last month 2010-11-28T07:41:10 hence my worry about the start of the contest 2010-11-28T07:41:14 or the dangers of the new submissions on the final day possibly 2010-11-28T07:41:27 yeah, mine is a brand new tree i came up with today 2010-11-28T07:41:30 s/of the/of three 2010-11-28T07:41:33 but i knew the risk when i submitted 2010-11-28T07:41:45 *** aizu has quit IRC (Read error: Connection reset by peer) 2010-11-28T07:42:07 antimatroid1: heh, yeah that's a scary way to do it 2010-11-28T07:42:08 *** aizu has joined #aichallenge 2010-11-28T07:42:15 you wouldn't believe 2010-11-28T07:43:05 I think Im going to play antimatroid1 next 2010-11-28T07:43:08 well however it turns out, I have to say it was awesome and I wanna thank the devs a lot for that! Very nice work! 2010-11-28T07:43:22 with an hour left i uploaded this new one, then decided no, uploaded my old one with 30 minutes left, then that lost its first game to someone in the 200's and i was like ah crap now what, then with 11 minutes left i resubmitted my new bot 2010-11-28T07:43:43 antimatroid1: that's very dangerous based on one loss 2010-11-28T07:43:48 yeah, janzert thanks heaps, i'm not sure what it would be like atm without you 2010-11-28T07:43:53 antimatroid1: hehe, what a tossup 2010-11-28T07:43:53 *** medrimonia has quit IRC (Ping timeout: 265 seconds) 2010-11-28T07:43:57 Error323: very dangerous, i was torn and that was my decider 2010-11-28T07:44:00 thanks 2010-11-28T07:44:53 i think this new one could be better if i had thought of it earlier, but oh well 2010-11-28T07:45:47 k, right now is when I think most of the top100 peeps, should be in the top100 2010-11-28T07:45:56 in any case- AI comp is VERY fun, i wish same ai comp spirit for next year !!!! 2010-11-28T07:46:18 burny: i'm still not in it >_< 2010-11-28T07:46:19 hehe 2010-11-28T07:46:20 good job devs 2010-11-28T07:46:31 weren't you in it earlier? 2010-11-28T07:46:41 burny: yea got kicked out again 2010-11-28T07:47:17 I'm still at position 600, while I was 70 in the latest ranking (and more like 50 when taking into account the ranking-boost for new submissions) 2010-11-28T07:47:18 mega1_: is comming =) 2010-11-28T07:47:18 burny: i certainly hope to be in the top 100 and i'm not atm 2010-11-28T07:47:28 http://ai-contest.com/profile.php?user_id=3945 2010-11-28T07:47:32 i'm at 239 with that record 2010-11-28T07:47:33 have won all my games, but like all of them were starter bot or barely better 2010-11-28T07:47:48 yeah that makes it go slow Mistmanov, but u'll get there :P 2010-11-28T07:47:51 I enjoyed this challenge as well. Unfortunately didn't hear about it untill last Wednesday, so didn't have much time to come up with something good. Will be on the lookout for the next challenge and step in a lot earlier! 2010-11-28T07:48:04 heikoh: :) 2010-11-28T07:49:01 man, dimka has his 4th loss 2010-11-28T07:49:11 antimatroid1: where was he ranked then 2010-11-28T07:49:15 bye 2010-11-28T07:49:29 i had my second loss :( 2010-11-28T07:49:38 *** CACA has quit IRC (Quit: Page closed) 2010-11-28T07:50:03 same 2010-11-28T07:50:24 ahh, got my first loss vs JimRogerz 2010-11-28T07:51:04 why are so many yop10 bots losing against sub 100s? 2010-11-28T07:51:21 too much fine tuning against strictly top100's, ad on specific maps? 2010-11-28T07:51:31 Error323: just above 100 i think 2010-11-28T07:51:32 burny: possibly 2010-11-28T07:51:40 burny: exactly that 2010-11-28T07:51:43 looks like the crash bots are just about sorted out to the bottom anyway 2010-11-28T07:52:02 janzert: rip them out :) 2010-11-28T07:52:04 janzert: people are having a lot of trouble with lower ranked bots 2010-11-28T07:52:11 i think a lot of people haven't played them in a while 2010-11-28T07:52:19 yeah, I've noticed 2010-11-28T07:52:30 i don't have a problem with it, just a comment :) 2010-11-28T07:52:35 trapped in local optima 2010-11-28T07:52:38 ranking is one very tough business 2010-11-28T07:52:41 I think if people lose against weaker bots.. that should be a reasonable component of their score.. just like losing on a tie map 2010-11-28T07:52:53 burny: agreed 2010-11-28T07:52:56 ofc it is 2010-11-28T07:53:09 if you assume your opponent doesn't suck.. it allows you to make more powerful moves 2010-11-28T07:53:36 unfortunately i think it confirms my point about the order in which you win games is not independent of your final ranking, but i can't think of a better ranking method 2010-11-28T07:54:08 part of the problem is during the open contest the resubmissions have to play near ranked bots to validate they deserve the rank but it means everyone nearby never plays bots further away 2010-11-28T07:54:25 everyone playing against everyone on every map.. but that's a lot of games -_- 2010-11-28T07:54:32 yeah, i think in the future it would be good if elos were completely reset for each resubmission? 2010-11-28T07:54:35 janzert: yeah thats exactly the problem ppl were having 2010-11-28T07:54:35 I think next contest the resubmission needs to be paired somewhere between where the old submission was and the middle of the rankings 2010-11-28T07:54:49 i think they should always start in the middle 2010-11-28T07:55:02 and a lot of people can game the current system beofre the actual contest 2010-11-28T07:55:03 centeroid of the ranking 2010-11-28T07:55:12 like it was at the start 2010-11-28T07:55:13 well that's the way it used to be and we heard endless griping about it :P 2010-11-28T07:55:21 not by me ^_^ 2010-11-28T07:55:23 which is fine by me, they only let themselves down when they finish lower than they got their hopes up for 2010-11-28T07:55:25 but ye i understand 2010-11-28T07:55:57 *** BaronTrozo has quit IRC (Quit: Page closed) 2010-11-28T07:56:00 yeah, not complaining, i just like to rip things to pieces analytically if you will 2010-11-28T07:56:22 yep, yep. 2010-11-28T07:56:51 I do think there is probably a fairly happy middle ground between always start in the middle and start where you were 2010-11-28T07:57:08 what if all bots started in the middle? 2010-11-28T07:57:09 if there's a sufficient number of games: the top bots having a few losses to the crap-bots, won't make much difference 2010-11-28T07:57:12 rahter than the bottom 2010-11-28T07:57:26 burny: the problem is how many games can we run in 5 days 2010-11-28T07:57:36 I did debate about suggesting to some of the top people in the last week to submit crash bots for a bit 2010-11-28T07:57:37 yeah 5 days isn't a lot 2010-11-28T07:57:40 and if you have those 2-3 losses and your equivalent other bot doesn't, you lose 2010-11-28T07:58:15 why is it 5 days btw 2010-11-28T07:58:17 so they could see how they were going to do climbing back up but decided that could be seen as unfair help 2010-11-28T07:58:21 that's scary little time 2010-11-28T07:58:33 i'd say a month would be much better :> 2010-11-28T07:58:47 i feel bad for you, i don't have to worry about being impartial much 2010-11-28T07:58:53 Error323: there'd be no one left to hear the result by then ;) 2010-11-28T07:58:58 yes, a month should be enough time to run everyone against everyone on every map 2010-11-28T07:59:10 i'd stick around to watch for a month 2010-11-28T07:59:15 after spending 3 months programming 2010-11-28T07:59:41 i still don't understand the nonlive leaderboard concept even being considered, it's like only broadcasting the results of sports matches at the end 2010-11-28T07:59:48 at current rate, 5days is about 300games per bot 2010-11-28T08:00:05 *** medrimonia has joined #aichallenge 2010-11-28T08:00:08 *** GreenTea has joined #aichallenge 2010-11-28T08:00:47 GreenTea: feeling confident? 2010-11-28T08:01:08 if we punt everyone who is at or below the rank of the default bot, after 48hours, that should get the 'real' bots, like 500games each? 2010-11-28T08:01:10 asavis is really struggling with the lower ranked bots 2010-11-28T08:01:11 smloh1: no, it would take 75 * ((4600 * (4600 - 1)) / 2) / (100gpm * 60 * 60) days 2010-11-28T08:01:45 anyone else feel really sad and kind of empty now? 2010-11-28T08:01:46 Error323: lol 2010-11-28T08:01:56 Hard to say =) 2010-11-28T08:01:58 asdfghv: lol :) 2010-11-28T08:02:01 I always feel sad and empty 2010-11-28T08:02:11 ehhh 2010-11-28T08:02:13 :p 2010-11-28T08:02:20 5509 days 2010-11-28T08:02:36 for each bot to face each bot on each map given 100 games per minute 2010-11-28T08:02:36 asdfghv: i have no idea what to do with myself after the 1st 2010-11-28T08:02:37 ^_^ 2010-11-28T08:02:46 but i think watching the leaderboard for 5 days will keep me entertained 2010-11-28T08:02:58 ah but crash bot games run in way faster than 100 games per minute 2010-11-28T08:03:10 lol, and slow bots take more than a minute per game 2010-11-28T08:03:19 smloh1: yes, but 200 turn games way slower 2010-11-28T08:03:29 that's true, 200 seconds max 2010-11-28T08:03:30 on the subject of bayeselo, you guys might have been interested in the kaggle challenge to make a Better Elo 2010-11-28T08:03:34 100 gpm is already quite an optimistic average 2010-11-28T08:04:06 what about completely random pairings for 2 days, then implementing the better pairing algorithm 2010-11-28T08:04:14 so people could sort out their positions roughly, then let it rip 2010-11-28T08:04:22 not for this contest, but for the future 2010-11-28T08:05:21 or weight losing against a much weaker opponent with the inverse probability of facing the weak opponent. 2010-11-28T08:06:32 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-11-28T08:07:12 it's a bit random whether you lose from weak opponents though. It's like my latest bot destroys my earliest attempts 95 out of 100 games.. but in 5 games the map just favors the coding of the earlier 2010-11-28T08:07:16 roughly 30hours, lets all bots face all toher bots once, on a random map 2010-11-28T08:07:34 if you then have the bad luck to play on that crappy map, and the loss is weighted very heavily 2010-11-28T08:07:35 .. 2010-11-28T08:07:45 you want to run more than a couple of maps 2010-11-28T08:07:57 anti: ideally.. obviously.. 2010-11-28T08:08:20 but 1map against EVERY person, is a fairly good average 2010-11-28T08:08:26 yeah its too biased probably 2010-11-28T08:08:26 hehe 2010-11-28T08:08:37 not 1 specific map.. random one for each pairing 2010-11-28T08:08:53 i think it will work reasonably well now 2010-11-28T08:09:08 burny: that's still 10577700 games 2010-11-28T08:09:14 i'd love to see more graphs and stuff of elo vs time 2010-11-28T08:09:20 ya, I think the way it is now(the original plan) is acceptable 2010-11-28T08:09:43 I personally think that it should have been more of a balance though, between win %, and elo 2010-11-28T08:09:56 (N (N - 1)) / 2, with N being nr o fplayers 2010-11-28T08:10:04 *** Mekanik has quit IRC (Ping timeout: 245 seconds) 2010-11-28T08:10:17 error, ya.. 30hours :P 2010-11-28T08:10:47 73 hours 2010-11-28T08:10:56 ehhh no 2010-11-28T08:10:58 73 days 2010-11-28T08:10:59 =) 2010-11-28T08:11:03 *** medrimonia has quit IRC (Ping timeout: 255 seconds) 2010-11-28T08:11:18 122days at 1game per minute? 2010-11-28T08:11:27 *** medrimonia has joined #aichallenge 2010-11-28T08:11:41 the whole open contest we only played around 4.7 million games 2010-11-28T08:12:03 oh, meh.. ya.. I was off by a factor of 60 lol 2010-11-28T08:12:08 games vs similarly rated opponents are more informative -- it teaches us nothing new to pair bocsimacko vs bot#1000 2010-11-28T08:12:19 ((4600 * (4600-1))/2) / (100 * 60 * 24) = 73 days of continuous play 2010-11-28T08:12:32 asdf.. it does when bosc loses to bot #1000 :P 2010-11-28T08:13:21 sure. but if bot#1000 can beat bosci then surely it can also beat #900-#999 quite easily 2010-11-28T08:13:30 ideally we would like to pair the biggest potential upsets :) 2010-11-28T08:13:38 hehe yeah 2010-11-28T08:13:52 just make bocsimacko face everyone on each map :P 2010-11-28T08:13:58 asdf.. : http://ai-contest.com/visualizer.php?game_id=8870418 2010-11-28T08:14:45 an upset for sure but he's 135th 2010-11-28T08:14:49 was asavis that good unofficial? 2010-11-28T08:14:55 http://ai-contest.com/profile.php?user_id=8348 2010-11-28T08:14:56 was 1st 2010-11-28T08:15:07 or is everyone just drewling on its loss because it drawed against bocsimacko 2010-11-28T08:15:08 http://ai-contest.com/final-open-ratings.txt 2010-11-28T08:15:18 OMG asavis.. 2010-11-28T08:15:21 it was in the top 5, and finished sitting first 2010-11-28T08:15:33 yes asavis is doing quite badly 2010-11-28T08:16:02 well 2010-11-28T08:16:14 my bot is holding strongish for the moment 2010-11-28T08:16:20 last commit of asavis was Nov 27th 19:29:54 2010-11-28T08:16:23 but hasn't really had a challenging person except manwe who it lost to 2010-11-28T08:16:36 its a number game, you can beat a much weaker opponent 95% of the time but if this game you face a map of the 5% u'll drop like a brick right now 2010-11-28T08:16:42 this is 3:29 by Kyiv timezone 2010-11-28T08:16:44 *** aizu has quit IRC (Read error: Connection reset by peer) 2010-11-28T08:17:06 *** aizu has joined #aichallenge 2010-11-28T08:17:06 ouch i just had 2 losses in a row 2010-11-28T08:17:20 damn i'm 105th 2010-11-28T08:17:24 smloh1: i wouldn't worry too much, it looks like a lot of good bots are 2010-11-28T08:17:26 grmbl 2010-11-28T08:17:33 Erro323: rankings are still completely out of whack 2010-11-28T08:17:41 true 2010-11-28T08:17:59 i want my bot to go into the top 100 already 2010-11-28T08:18:07 i'm sicking of risky games against lower ranked bots 2010-11-28T08:18:27 It's really funny how this is completely random now 2010-11-28T08:18:30 i almost lost one 2010-11-28T08:18:38 http://ai-contest.com/visualizer.php?game_id=8862368 2010-11-28T08:18:41 medrimonia: indeed 2010-11-28T08:18:56 how will the pruning process work? 2010-11-28T08:19:04 smloh1: jimrogerz is quite good but uwetudd was #277 2010-11-28T08:19:15 will we just freeze the lowest bots? 2010-11-28T08:19:36 medrimonia: ah, curses on my unreliable reinforcement algorithm! 2010-11-28T08:19:37 smloh1: yep 2010-11-28T08:20:02 Top 10 players: ssieb0(2194), nextgeneration(2172), Jimsbot(2064), shangas(2064), goldman(2057), luismi(2031), _Astek_(2026), Vogons(2022), CorwinAlex(2012), pocreot(2002) 2010-11-28T08:20:07 when that happens how about the matches with them? 2010-11-28T08:20:25 btw guys, publish your code please: http://ai-contest.com/forum/viewtopic.php?f=17&t=1177 2010-11-28T08:20:29 The more interessant is that we will know how will the good bots racts against those who aren't ranked in top 100 2010-11-28T08:20:32 was around 50% of bots, the default? 2010-11-28T08:20:47 smloh1: i expect the matches are kept 2010-11-28T08:20:55 i'll be very disappointed if they aren't 2010-11-28T08:21:18 Hmmm Error323: I think my code is really unreadable 2010-11-28T08:21:31 medrimonia: doesn't matter 2010-11-28T08:21:40 it seems to me that roughly 80% of bots can be purged after 24hours 2010-11-28T08:21:56 reading other people's code is the _best_ way to learn programming 2010-11-28T08:21:58 my code isn't very readable 2010-11-28T08:22:26 only 75 maps 2010-11-28T08:22:31 pff 2010-11-28T08:22:36 Accoun: that's plenty 2010-11-28T08:22:36 canadian chiters 2010-11-28T08:23:03 Accoun: I've never seen someone as annoying as you I think 2010-11-28T08:23:28 *** Mekanik has joined #aichallenge 2010-11-28T08:23:31 This message has been submitted successfully, but it will need to be approved by a moderator before it is publicly viewable. You will be notified when your post has been approved. 2010-11-28T08:23:37 im baned on forum 2010-11-28T08:23:44 that isn't banned 2010-11-28T08:23:48 I like Accoun 2010-11-28T08:23:50 it's restricted 2010-11-28T08:23:55 lmao 2010-11-28T08:24:25 all 999 map sets identical and give the same winer 2010-11-28T08:24:25 all 75 map sets sepecial and give special winer 2010-11-28T08:24:29 explain in detail: 2010-11-28T08:24:38 75 map set fast give wrong winer (map set related) 2010-11-28T08:24:38 999 map set give true winer if enougth number of games played, more games more accurate result 2010-11-28T08:24:47 why 999 though, and not 9999? 2010-11-28T08:24:56 it's over 9000! 2010-11-28T08:24:59 haha 2010-11-28T08:25:01 http://ai-contest.com/visualizer.php?game_id=8879702 2010-11-28T08:25:04 Because Accoun know everything, He's God!!! 2010-11-28T08:25:21 ^^^^ 2010-11-28T08:25:29 well he writes a good bot 2010-11-28T08:25:47 Who know how many maps are used? 75? 2010-11-28T08:25:53 protocolon was 31st before 2010-11-28T08:25:57 only 75 2010-11-28T08:25:58 yes 75 maps are in the set 2010-11-28T08:26:02 oO 2010-11-28T08:26:03 they are available in the forums 2010-11-28T08:26:13 they were all generated fresh for the tournament 2010-11-28T08:26:20 it canadin chiters 2010-11-28T08:26:22 This is sad.. 2010-11-28T08:26:27 what is sad? 2010-11-28T08:26:36 they're 75 random maps 2010-11-28T08:26:52 hmm the pairing algorithm, was it based on elo score or on nearness in ranking? 2010-11-28T08:27:07 I agree with Accoun 2010-11-28T08:27:12 *** delt0r___ has joined #aichallenge 2010-11-28T08:27:23 so do I on map amount yes 2010-11-28T08:27:35 but hey, its not the end of the world though 2010-11-28T08:27:42 asdfghv: something to do with pareto distribution to nearest elo or something 2010-11-28T08:27:57 yeah, i agree more maps would be nice 2010-11-28T08:28:02 but i don't think it really matters tbh 2010-11-28T08:28:03 Accoun need to review the meaning of "cheating" 2010-11-28T08:28:37 it's not so small that one bad map is going to ruin you 2010-11-28T08:28:45 ugh http://ai-contest.com/visualizer.php?game_id=8879932 2010-11-28T08:28:46 The fact that there's only 75 maps might be sad, I agree, but it doesn't mean at all that the organizers are cheater 2010-11-28T08:28:50 *** delt0r has quit IRC (Ping timeout: 265 seconds) 2010-11-28T08:29:04 forgot to include the "when you have the choice between taking a neutral and taking an enemy planet, TAKE THE ENEMY" rule -_- 2010-11-28T08:29:10 Mistmanov: ouch 2010-11-28T08:29:17 Mistmanov: ouch 2010-11-28T08:29:17 *** krokokrusa has quit IRC (Ping timeout: 265 seconds) 2010-11-28T08:29:23 there's really not any point in having much more maps than that.. 2010-11-28T08:29:24 people really suck against badly ranked bots ehe 2010-11-28T08:29:30 each bot will only see each map a few times 2010-11-28T08:29:35 last year he do chiting 2010-11-28T08:29:44 Accoun: janzert was a contestant last time 2010-11-28T08:29:47 not an organiser 2010-11-28T08:30:18 Accoun: what's chiting 2010-11-28T08:30:39 woot, I made it into the top100 2010-11-28T08:30:47 Error323: If Accoun doesn't win, that's called "chiting" 2010-11-28T08:30:51 http://ai-contest.com/forum/viewtopic.php?f=3&t=366&start=30#p2151 2010-11-28T08:30:51 http://openpaste.org/en/19653/ 2010-11-28T08:30:56 Error323> ^^^ 2010-11-28T08:31:22 burny: your bot is quite impressive for the time spent on it 2010-11-28T08:31:27 you better not beat me though :P 2010-11-28T08:31:29 ow and 75 maps or 1000 doesn't matter that much Accoun chilll out a bit on the devs they have been working extremely hard. Its not the end of the world ;) 2010-11-28T08:31:31 thanks 2010-11-28T08:32:29 no any reson use only 75 maps except for citing 2010-11-28T08:32:32 no any reson use only 75 maps except for chiting 2010-11-28T08:32:43 Accoun: what is chiting 2010-11-28T08:32:45 ... 2010-11-28T08:32:46 no any reason for use more maps 2010-11-28T08:32:52 Accoun: argue/discuss better methods for ranking if you want, hell we do it here all the time, but at the end of the day, the contest is still fair (which is most important) and the organisers put a lot of effort in, which they do at their own expense 2010-11-28T08:33:02 all 75 map sets sepecial and give special winer 2010-11-28T08:33:08 all 999 map sets identical and give the same winer 2010-11-28T08:33:14 explain in detail: 2010-11-28T08:33:19 75 map set fast give wrong winer (map set related) 2010-11-28T08:33:25 999 map set give true winer if enougth number of games played, more games more accurate result 2010-11-28T08:33:26 Accoun: nonsense, its a random distribution over maptypes from the generator. 2010-11-28T08:33:30 okey? 2010-11-28T08:33:30 is accoun a bot? 2010-11-28T08:33:54 accoun: NO one knows ahead of time which 'special' map types will happen though 2010-11-28T08:34:04 and, bots are SUPOSED to be good at ALL map types 2010-11-28T08:34:08 no any reson use only 75 maps except for chiting 2010-11-28T08:34:16 1000 maps or 75 maps just gives lower amount of types of maps. But the same percentages roughly 2010-11-28T08:34:37 ozer persentage - map set related 2010-11-28T08:34:53 Accoun: it's not going to change, so get over it 2010-11-28T08:34:55 easy ajusted to mekre selected winer 2010-11-28T08:35:05 easy ajusted to make selected winer 2010-11-28T08:35:09 nobody is selecting any winners 2010-11-28T08:35:33 it'd be hilareous if the default ai wins :P 2010-11-28T08:35:40 Accoun you thing winner will be not bocsimacko? )))) 2010-11-28T08:35:47 * think 2010-11-28T08:35:50 haha 2010-11-28T08:36:01 GreenTea: he's been claiming the organisers are cheaters since tron 2010-11-28T08:36:13 well if that's the case, the ranking algo should be revised :P 2010-11-28T08:36:19 no any reson use only 75 maps except for chiting 2010-11-28T08:36:24 actual galcon ai http://ai-contest.com/profile.php?user_id=3907 2010-11-28T08:36:37 Accoun: stop repeating urself -.- 2010-11-28T08:36:45 Accoun said the other day that if he had one "mouth" more, he would win against bocsimacko, because he's a better player than him 2010-11-28T08:36:48 janzert: i don't know about that, the game specifics are different, did he get those right? 2010-11-28T08:36:54 janzert: cool 2010-11-28T08:37:03 good to know we're kicking its ass :P 2010-11-28T08:37:04 I do have to agree though.. that there's no advantage to using 75maps rather than a few hundred maps 2010-11-28T08:37:07 antimatroid1: I don't really know any specifics on it 2010-11-28T08:37:21 all 75 map sets sepecial and give special winer 2010-11-28T08:37:28 but, I also think 75maps is sufficient 2010-11-28T08:37:30 all 999 map sets identical and give the same winer 2010-11-28T08:37:36 antimatroid1: just that the galcon author submitted it as the actual galcon bot 2010-11-28T08:37:38 explain in detail: 2010-11-28T08:37:42 he's had just as long as bocsimacko, if he was a better player he'd already be beating him 2010-11-28T08:37:45 75 map set fast give wrong winer (map set related) 2010-11-28T08:37:54 Accoun you already posted that within the last 15 minutes 2010-11-28T08:37:57 Accoun: unless you can explain why with good arguments in proper english stop repeating, because I can't understand you... 2010-11-28T08:37:59 999 map set give true winer if enougth number of games played, more games more accurate result 2010-11-28T08:38:24 Accoun: the problem is the number of games that can feasibly be played 2010-11-28T08:38:32 but we've been over this, i can't be bothered doing it again 2010-11-28T08:38:46 no matter what, there is always going to be a map set bias 2010-11-28T08:38:46 its like talking to a brick wall 2010-11-28T08:38:46 Accoun let me ask you one question 2010-11-28T08:39:05 no any reson use only 75 maps except for chiting 2010-11-28T08:39:07 Accoun : you are suggesting that by sampling 75 maps with rejection from the map distribution, the outcome can be skewed. However the same argument applies regardless of the number of maps you choose. 2010-11-28T08:39:09 medrimonia: http://ai-contest.com/visualizer.php?game_id=8869599 i'm glad you agreed that it's a draw map 2010-11-28T08:39:10 This guy in you profile http://www.youtube.com/watch?v=p-5nXqUF29Q, this is you? 2010-11-28T08:39:12 error, if 999maps has 37 'draw maps'... 75maps will most likely have either 8% more or 28% less 2010-11-28T08:39:44 999 maps will have many more draw maps 2010-11-28T08:40:00 75 maps already have quite some draw maps 2010-11-28T08:40:09 I just raed ahat.. and picked some numbers 2010-11-28T08:40:10 u a realy so stuped or it sumulation? 2010-11-28T08:40:11 however you define draw anyway 2010-11-28T08:40:19 antimatroid1: Yep, it's sad, but if someone makes an error, he'll loose 2010-11-28T08:40:29 error, for any definable 'type' of map.. 2010-11-28T08:40:51 a smaller set of maps, will have a less 'average' representation OF that type, vs a larger set 2010-11-28T08:41:08 actualy any top bot make luse any ozer top bot on same map 2010-11-28T08:41:09 But antimatroid1: I think that it's normal that there's some draw maps, just to see if the people know how to deal with this cases. 2010-11-28T08:41:22 medrimonia: i agree 2010-11-28T08:41:39 and lower ranked bots wont all handle it, is it reasonable to let them assume such maps don't exist 2010-11-28T08:41:40 burny: it will be less accurate yes 2010-11-28T08:41:45 woah, two losses for bocsi 2010-11-28T08:41:49 shit 2010-11-28T08:41:50 really?? 2010-11-28T08:42:04 holy shit 2010-11-28T08:42:05 actualy any top bot make luse to any ozer top bot on same map- by selection map set can make any winer 2010-11-28T08:42:05 it's all over the show :\ 2010-11-28T08:42:09 no any reson use only 75 maps except for chiting 2010-11-28T08:42:11 whoot, strange really! 2010-11-28T08:42:25 that's strange, those are top 100 bots aren't they? 2010-11-28T08:42:30 error. ya.. it really doesn't matter though, since we were given the generator.. and the ai needs to be good on all types 2010-11-28T08:42:39 in fact i thought protocolon was top 10 for a good while 2010-11-28T08:42:46 in fact actualy any top bot make luse to any ozer top bot on same map- by selection map set can make any winer 2010-11-28T08:42:49 no any reson use only 75 maps except for chiting. yea yea. we know))) 2010-11-28T08:42:51 smloh1: did you just ask who bocsimacko is? 2010-11-28T08:43:00 it'd be a bit different if you weren't given the entire map info(and had to explore).. or weren't given the generator 2010-11-28T08:43:00 no i mean the ones he lost to 2010-11-28T08:43:36 i don't like having a definitive map generator right from the start 2010-11-28T08:43:37 i thought he might have lost if he'd overfitted for top-performing bots, but these are by no means lower than 100-ranked 2010-11-28T08:43:40 canadin chiting 2010-11-28T08:43:49 it isn't always clear what maps will work best till people have toyed with the problem 2010-11-28T08:43:50 I don't recall kevlar being that good? 2010-11-28T08:44:03 maybe i'm just remembering from earlier in the competition 2010-11-28T08:44:06 yeah, i've never seen kevlar 2010-11-28T08:44:15 photocolon is ranked highly 2010-11-28T08:44:25 colocon* TIL 2010-11-28T08:44:25 was 68th with 12games before final 2010-11-28T08:44:27 kavlar was #68 2010-11-28T08:44:33 kevlar* 2010-11-28T08:44:34 ah ok 2010-11-28T08:44:38 thats not bad 2010-11-28T08:44:38 rankings at the end were retarded 2010-11-28T08:44:43 true 2010-11-28T08:44:44 i was 41st 2010-11-28T08:44:48 i was 200th 2010-11-28T08:44:50 :P 2010-11-28T08:44:51 and i hadn't been on the site in like a month 2010-11-28T08:44:53 also.. a single loss, is to be expected 2010-11-28T08:44:55 which is retarded ofc :P 2010-11-28T08:45:17 if the same(or all) crappy bots, beat the same top bots often.. then that's different 2010-11-28T08:45:31 actualy any top bot make luse to any ozer top bot on same map- by selection map set can make any winer 2010-11-28T08:45:37 but it's expected that a 1500elo will beat a 2500elo like 1 in 1000 games 2010-11-28T08:45:46 if bocsi doesn't win i hope shangas does 2010-11-28T08:45:51 cause the bot is almost a month old 2010-11-28T08:46:29 I hope a bot with evolutionary programming or any other type of non-heuristic stuff wins 2010-11-28T08:46:51 does my game tree count? :P 2010-11-28T08:47:03 spaceinvaders was at 300 last I checked 2010-11-28T08:47:09 yeah he was doing badly 2010-11-28T08:47:13 ouch.. I just beat a rank47 2010-11-28T08:47:17 http://ai-contest.com/profile.php?user_id=4038 2010-11-28T08:47:22 back up to 204 2010-11-28T08:47:26 burny: nice 2010-11-28T08:47:42 they crashed 2010-11-28T08:47:48 *** Ice_Harley has joined #aichallenge 2010-11-28T08:47:49 ya 2010-11-28T08:47:52 127 now, just won last though 2010-11-28T08:48:00 I have to attemp a genetic approach on this in the upcoming weeks 2010-11-28T08:48:05 whooot 7 servers up 2010-11-28T08:48:09 awesome 2010-11-28T08:48:11 just got to hope enough good ranking bots post their code 2010-11-28T08:48:28 Zoo: there are already some decent bots up there 2010-11-28T08:48:38 Hello Ice_Harley 2010-11-28T08:48:40 yeah there are 2010-11-28T08:49:11 but I want lots of variation in the style of play 2010-11-28T08:49:49 the only person that beat me so far, is still undefeated 2010-11-28T08:49:53 compare mega1 to shangas or george for example 2010-11-28T08:50:20 *** krokokrusa has joined #aichallenge 2010-11-28T08:50:29 george is still undefeated 2010-11-28T08:50:47 I have first lose.. 2010-11-28T08:51:01 me too GreenTea^^ just a few minutes ago 2010-11-28T08:51:03 woo.. I'm temporarily 43rd 2010-11-28T08:51:05 i had mine before, and now i'm stuck outside top 100 2010-11-28T08:51:07 i'm 116 now 2010-11-28T08:51:17 *** protocolocon has joined #aichallenge 2010-11-28T08:51:24 hi everyone 2010-11-28T08:51:28 protocolocon: nice win against bocsi 2010-11-28T08:51:34 hey protocolocon 2010-11-28T08:51:54 hi 2010-11-28T08:52:17 man, why are all you people turning up here at the end of the contest? 2010-11-28T08:52:26 yeah 2010-11-28T08:52:28 wth 2010-11-28T08:52:29 yep, just possitional luck 2010-11-28T08:52:46 I wonder if this contest will one day be eligible to be added in ones CV :P 2010-11-28T08:53:00 Zoo: mega1 put in his 11th from tron 2010-11-28T08:53:09 i would 2010-11-28T08:53:12 it is already, you can see bocsimacko CV in the web 2010-11-28T08:53:14 nice 2010-11-28T08:53:15 well 2010-11-28T08:53:19 *** insectores has joined #aichallenge 2010-11-28T08:53:23 this one has huge bunch of participants 2010-11-28T08:53:33 protocolocon: you've done your homework reading too then? :P 2010-11-28T08:53:37 so even getting in to final 10% is a feat worth mentioning 2010-11-28T08:53:47 getting in top 2% is something to be very proud of 2010-11-28T08:53:56 and winning...thats something entirely different 2010-11-28T08:54:01 it makes me sad to talk about finishing well 2010-11-28T08:54:15 antimatroid1: yep, one needs to know who is winning you 2010-11-28T08:54:15 i so badly wanted top 10 in tron and i didn't even get to finish due to a stupid timeout 2010-11-28T08:54:39 getting into top500 isn't too hard.. and top 1000 is nearly trivial 2010-11-28T08:54:39 of course 2010-11-28T08:54:49 top 1000 is trivial 2010-11-28T08:55:06 *** phreeza has joined #aichallenge 2010-11-28T08:55:38 I don't know for sure, but I think that top ~200 doesn't really require a lot of programming genius, just a lot of free time 2010-11-28T08:55:42 144 gpm - not bad, not bad.. 2010-11-28T08:55:59 (if you have programming genius and little free time, you can ofcourse also get top 200 =)) 2010-11-28T08:55:59 agree 2010-11-28T08:56:11 Mistmanov, you hit the spot. 2010-11-28T08:56:17 yup 2010-11-28T08:56:26 I didn't had enough spare time, so I'll not even reach top 1000. 2010-11-28T08:56:27 there were thousands of people with free time, yet only 200 make it to top 200 2010-11-28T08:56:31 :) 2010-11-28T08:56:37 a few days before final, it was around rank 300. where it started to require more then 10mnutes of work 2010-11-28T08:56:39 thousands of people who didn't care enough :p 2010-11-28T08:56:48 well wouldn't say that 2010-11-28T08:56:50 or just werent skilled enough 2010-11-28T08:56:52 i agree with that^^ 2010-11-28T08:57:07 top 300ish is exponentially harder to gain rank 2010-11-28T08:57:14 don't know the exponent htough :P 2010-11-28T08:57:26 and not to mention not every programmer is cut out to make a good ai dev 2010-11-28T08:57:37 * antimatroid1 is a maths student :) 2010-11-28T08:57:51 and some ai devs tried some really outlandish stuff which may have failed 2010-11-28T08:57:52 * burny is a problem solver :\ 2010-11-28T08:57:56 and game theory nut 2010-11-28T08:57:57 * Mistmanov is a biophycist ^^ 2010-11-28T08:57:58 like the evolutionary programmers 2010-11-28T08:58:16 yeah there are probbly a few of those 2010-11-28T08:58:25 the "tried something cool and failed" 2010-11-28T08:58:29 although, mega1 did his masters in ai 2010-11-28T08:58:39 but if the cared about the ranking, they would then have tried something more conventional 2010-11-28T08:58:45 Mistmanov: It's not only about genius, but I think being in top 200 is already meaning that you can be a little bit perseverant 2010-11-28T08:59:18 and then there's age difference, programming experience etc 2010-11-28T08:59:24 persistent.. or just really good at finding excuses for not having to work at your actual job/classes/etc :p 2010-11-28T08:59:27 I wish I had some more time, to come up with a proper way of sending my troops to the front lines :\ 2010-11-28T08:59:40 <<--- should've been doing something else while working on the contest 2010-11-28T08:59:51 I had to just hack somethign in, in the last 30mins before deadline 2010-11-28T08:59:51 *** protocolocon_ has joined #aichallenge 2010-11-28T09:00:20 i spent 3 months on my bot, then just today implement a new game tree idea and upload that as my final 2010-11-28T09:00:23 I was already happy to be in top 200 one month ago 2010-11-28T09:00:27 *** protocolocon has quit IRC (Quit: Page closed) 2010-11-28T09:00:27 total risk, so far it seems alright though 2010-11-28T09:01:04 i had never programmed before uni and i would say i never actually learnt to program till last yearish 2010-11-28T09:01:12 nice anti.. won't know for sure though, until you break into the top100 more? 2010-11-28T09:01:15 i only know some c++ and that's about it 2010-11-28T09:01:42 *** smloh1 has quit IRC (Quit: Page closed) 2010-11-28T09:01:54 i hope it last canadian google ai chanlenge 2010-11-28T09:01:56 I have a master in AI, unfortunately I only had about 15 hours to come up with a bot. Would be really happy to finish in the top 400. 2010-11-28T09:02:00 *** protocolocon_ is now known as protocolocon 2010-11-28T09:02:05 Accoun: it wont be, but you don't have to compete 2010-11-28T09:02:39 almost time, for when I had my estimate, of the top100 looking roughly correct 2010-11-28T09:02:50 burny: it's all over the show atm 2010-11-28T09:02:58 although astek is now in first who was ranked highly before 2010-11-28T09:02:59 The level has gone crazy during last week too 2010-11-28T09:03:03 gota wait for the next recalc 2010-11-28T09:03:17 medrimonia: you beat george, nice 2010-11-28T09:03:26 trying the `really outlandish stuff' would have been a lot of fun, but would take months of programming / testing to get something nice out of it I guess 2010-11-28T09:03:33 =) I didn't see that 2010-11-28T09:04:00 I'm training myself in arimaa^^ 2010-11-28T09:04:10 i'm waiting to do that 2010-11-28T09:04:16 i'm expecting another loss soon 2010-11-28T09:04:19 i'm going too well 2010-11-28T09:04:26 _iouri_ is a monster 2010-11-28T09:04:31 oh yeah 2010-11-28T09:04:35 iouri is very good 2010-11-28T09:04:35 the current top6, were all in the top25 before 2010-11-28T09:04:37 i beat iouri a few times in the last few days :) 2010-11-28T09:04:45 well so did I, but only a few 2010-11-28T09:04:48 yesterday comletely lose him on tcp server.... 2010-11-28T09:04:53 i also threw some games against him that i should have won :( 2010-11-28T09:04:57 GreenTea: ur quite good too though =) 2010-11-28T09:05:07 *** Appleman1234 has joined #aichallenge 2010-11-28T09:05:32 2 losses so far 2010-11-28T09:05:35 and 148th :/ 2010-11-28T09:05:54 i'm 10/11 and 110th 2010-11-28T09:06:13 _Astek_ is doing well 2010-11-28T09:06:21 well 2010-11-28T09:06:25 hardly good bots 2010-11-28T09:06:28 yeah 2010-11-28T09:06:38 hasn't really had any major challenger yet, and he does have a good bot 2010-11-28T09:06:45 indd 2010-11-28T09:07:01 why do people put _ 2010-11-28T09:07:04 in their names 2010-11-28T09:07:09 how horrible :\ 2010-11-28T09:07:43 I think bocsimacko will be #1, iouri #2 and bix0r4ever #3 2010-11-28T09:07:44 slin had a bad loss too 2010-11-28T09:07:47 or perhaps manwe 2010-11-28T09:07:51 manwe is very good aswell 2010-11-28T09:08:17 http://ai-contest.com/visualizer.php?game_id=8883932 2010-11-28T09:08:19 cool start 2010-11-28T09:08:55 lol 2010-11-28T09:09:08 sort of reminds me of CACA's bot 2010-11-28T09:09:16 hmm manwe is not top 5 I think (or at least I don't hope)^^ 2010-11-28T09:09:17 its a shame that bot crashes so much :P 2010-11-28T09:09:39 medrimonia: no? I think he might be 2010-11-28T09:10:04 he was 18 at unofficial 2010-11-28T09:10:04 and george will probably be in there somewhere 2010-11-28T09:10:13 oh 2010-11-28T09:10:17 george is highly ranked 2010-11-28T09:10:24 And I have been fighting against him the last two weeks 2010-11-28T09:10:25 i don't think manwe is that good errir 2010-11-28T09:11:25 i broke into top 100 2010-11-28T09:11:40 me too i think, when rating is updated 2010-11-28T09:11:47 http://ai-contest.com/visualizer.php?game_id=8885698 2010-11-28T09:11:49 ok, rankings just updated 2010-11-28T09:11:49 win from thedreamer 2010-11-28T09:11:52 that game required a bit more work 2010-11-28T09:11:59 I think that for the first french guy, the fight will be quite hard between Doraj8, Manwe and Me 2010-11-28T09:12:13 i have to fight smiley for australia 2010-11-28T09:12:16 *** BaronTrozo has joined #aichallenge 2010-11-28T09:12:26 medrimonia: it'll either be you or manwe imo, but i think manwe 2010-11-28T09:12:46 Doraj has uploaded a new version just before 2010-11-28T09:12:50 still at 260 -_- 2010-11-28T09:12:57 haven't played a single person in top250 -_- 2010-11-28T09:13:12 he was tenth with 1 games 2010-11-28T09:13:16 14 games* 2010-11-28T09:13:31 sigh is doing well early 2010-11-28T09:13:53 protocolocon: thanks for beating smiley for me 2010-11-28T09:13:53 My bot has always done very bad on tcp because I have a lot of connection issues 2010-11-28T09:14:49 a pleasure :) 2010-11-28T09:15:11 anti: more than 50% of the top10, were in top20: more than 50% of the top20, were in the top50 2010-11-28T09:15:35 i've had 12 games 2010-11-28T09:15:39 but yeah, they're getting there 2010-11-28T09:15:46 but actual decent rankings are a long way off yet 2010-11-28T09:15:51 damnit i'm climbing so slooooooooooow 2010-11-28T09:15:53 there's an average of 9games per person 2010-11-28T09:16:05 er.. almost 10 2010-11-28T09:16:17 some people have only just had their 9th game 2010-11-28T09:16:53 i've had 12 so far 2010-11-28T09:17:00 of which 2 losses -.- 2010-11-28T09:18:18 Hey Accoun nice video on youtube, could you recomend me your hairdresser ???? 2010-11-28T09:18:52 ))) 2010-11-28T09:19:51 :o 2010-11-28T09:20:03 Top 10 players: _Astek_(2521), ssieb0(2490), ChrisH(2415), CorwinAlex(2400), _iouri_(2387), bix0r4ever(2381), Polsky(2376), GreenTea(2356), nextgeneration(2356), jimrogerz(2353) 2010-11-28T09:20:33 *** insectores has quit IRC (Quit: Page closed) 2010-11-28T09:21:07 my friends call me jesus atm 2010-11-28T09:21:21 i was already doing 5 units and attending classes for measure theory before this contest started 2010-11-28T09:21:27 it's been a long 3 months 2010-11-28T09:21:28 woah, i won another one 2010-11-28T09:21:32 *** nibalize1 has quit IRC (Ping timeout: 265 seconds) 2010-11-28T09:21:32 *** Error323 has quit IRC (Ping timeout: 265 seconds) 2010-11-28T09:21:34 *** nibalizer has joined #aichallenge 2010-11-28T09:21:48 *** Error323 has joined #aichallenge 2010-11-28T09:21:54 i beat rs5147 2010-11-28T09:21:59 up to 53rd 2010-11-28T09:22:07 rank28 now, weee 2010-11-28T09:22:08 http://ai-contest.com/visualizer.php?game_id=8886663 2010-11-28T09:22:10 that was too easy 2010-11-28T09:22:35 burny: you beat oldman, well done 2010-11-28T09:22:45 very impressive couple of days effort bot 2010-11-28T09:22:51 thanks 2010-11-28T09:22:55 how strong is oldman? 2010-11-28T09:23:00 quite strong 2010-11-28T09:23:04 top 10 for quite a while 2010-11-28T09:23:16 I see he was top 20, but only had 1 game listed since latest submission 2010-11-28T09:23:21 nice 2010-11-28T09:23:53 seeming like I'm fairly possibly going to be top50 2010-11-28T09:23:57 that'd be crazy 2010-11-28T09:24:00 *** aizu has quit IRC (Read error: Connection reset by peer) 2010-11-28T09:25:06 well i've hit first for australia, for the moment 2010-11-28T09:25:28 pretty certain I don't have a chance at first for canada lol 2010-11-28T09:25:47 omg i win _iouri_ 2010-11-28T09:25:48 no, i think iouri has that in the bag 2010-11-28T09:25:51 oh nice! 2010-11-28T09:26:13 this is so much better with a live leaderboard 2010-11-28T09:26:24 for tron they closed the leaderboard off while the final tournament ran :| 2010-11-28T09:26:28 also very very distracting from doing anything else 2010-11-28T09:27:15 dang it.. I was suposed to go to bed like 6hours ago :( 2010-11-28T09:27:17 *** p4p4 has joined #aichallenge 2010-11-28T09:27:39 think I've spent more time waching the leader boards, and chatting.. then I did making the stupid bot 2010-11-28T09:27:39 http://ai-contest.com/visualizer.php?game_id=8878517 2010-11-28T09:27:40 ffs 2010-11-28T09:27:44 there are so many shit maps 2010-11-28T09:27:46 anyone knows if it will be a final with best players, or it will be tree days tournament?? 2010-11-28T09:28:06 BaronTrozo: it is intended that the lower ranked sections get frozen over time 2010-11-28T09:28:11 http://ai-contest.com/game_stats.php 2010-11-28T09:28:15 it will have information there 2010-11-28T09:28:20 shit, there's like 7 servers 2010-11-28T09:29:24 over time??? is there any web where this is explained 2010-11-28T09:29:24 yeah its awesome 2010-11-28T09:30:18 Error323: http://pastebin.com/dqNsZrJM :) 2010-11-28T09:30:43 hmm 2010-11-28T09:30:46 damnit 2010-11-28T09:30:48 =) 2010-11-28T09:31:04 that's on the 75 final maps i presume? 2010-11-28T09:31:21 wraaaaaaah 101th 2010-11-28T09:31:23 >_< 2010-11-28T09:31:35 i think so, but then it also depends which map we get paired on if we play 2010-11-28T09:32:07 ya 2010-11-28T09:32:14 *** Utkarsh has joined #aichallenge 2010-11-28T09:32:18 nah, it was 20 randomly made maps 2010-11-28T09:32:21 have no fear :) 2010-11-28T09:32:25 oh 2010-11-28T09:32:27 i'd take from that we're about the same 2010-11-28T09:32:48 well 2010-11-28T09:32:51 wouldn't say that 2010-11-28T09:32:56 run it on the 75 maps 2010-11-28T09:32:59 :) 2010-11-28T09:33:21 i didn't run it :P 2010-11-28T09:33:25 oh 2010-11-28T09:33:31 which bot was that then? 2010-11-28T09:33:40 krokokrusa did it 2010-11-28T09:33:43 like he said earlier 2010-11-28T09:33:46 oh 2010-11-28T09:33:52 bix0r into first 2010-11-28T09:33:55 I can run on 75 official maps, but it would take ages :) 2010-11-28T09:33:59 i can't go for him 2010-11-28T09:34:04 he used java, that is not acceptable 2010-11-28T09:34:08 i'll download ur code antimatroid1 2010-11-28T09:34:13 antimatroid1's bot is very cpu heavy 2010-11-28T09:34:20 super heavy 2010-11-28T09:34:25 i'll use my 1 second damnit! 2010-11-28T09:34:42 that wasn't in response to the other day though, it's the nature of a game tree 2010-11-28T09:34:53 antimatroid1: I tried C++ too if you look into my submission's list :) 2010-11-28T09:35:13 java? 2010-11-28T09:35:16 why? 2010-11-28T09:35:34 i pretty much completely rewrote from the start package 2010-11-28T09:35:46 i even changed to handling the input differently 2010-11-28T09:35:47 I just develop a lot faster in Java 2010-11-28T09:36:04 me too 2010-11-28T09:36:13 starter bot in Java spent way too much time parsing 2010-11-28T09:36:27 like 30 ms or so :))) I don't recall 2010-11-28T09:37:19 i never tested if my way was faster, but i would be very surprised if it isn't 2010-11-28T09:37:48 i capped myself processing information to 0.3 seconds, just in case it was horribly inefficient 2010-11-28T09:38:02 i just deal with wrong information if that ever happens, which would be absurd 2010-11-28T09:45:09 damn lost against smloh 2010-11-28T09:45:29 *** p4p4 has quit IRC (Ping timeout: 245 seconds) 2010-11-28T09:46:56 antimatroid1: holy crap ur bot is slow man :P 2010-11-28T09:47:18 Erro323: it uses the full second :P 2010-11-28T09:47:22 shorts game that i have ever see 2010-11-28T09:47:24 http://ai-contest.com/visualizer.php?game_id=8890170 2010-11-28T09:47:24 no shit 2010-11-28T09:47:33 yeah 2010-11-28T09:47:55 i'm 13/14 2010-11-28T09:48:13 i'm stuck in 100-200 atm >_< 2010-11-28T09:48:20 it'll change 2010-11-28T09:48:22 i was before :) 2010-11-28T09:48:25 i'm 43rd atm 2010-11-28T09:49:10 GreenTea: you ukrainians are doing well! 2010-11-28T09:49:21 UKRAINE IS JOKE TO YOU? 2010-11-28T09:49:27 :> 2010-11-28T09:49:27 *** bhasker has joined #aichallenge 2010-11-28T09:49:31 burny: 17th! 2010-11-28T09:49:45 morning 2010-11-28T09:49:53 tomorrow round this time ratings will be much more reliable 2010-11-28T09:50:02 the leaderboard is really volatile right now 2010-11-28T09:50:12 mega1 is at #46 2010-11-28T09:50:16 well 2010-11-28T09:50:20 bhasker: super volatile 2010-11-28T09:50:22 he kinda had two heavy losses 2010-11-28T09:50:29 yea 2010-11-28T09:50:34 clean bill bhasker! nice 2010-11-28T09:50:37 *** dmj111 has joined #aichallenge 2010-11-28T09:50:39 GreenTea: #1 =) 2010-11-28T09:50:40 my bot seems to be having a lucky streak 2010-11-28T09:50:40 yea 2010-11-28T09:50:41 http://ai-contest.com/profile.php?user_id=5822 2010-11-28T09:50:43 :D 2010-11-28T09:50:47 dmj111: excited! 2010-11-28T09:50:52 wow 2010-11-28T09:50:55 though it hasn't played any good bots yet 2010-11-28T09:50:59 bhasker: nice 2010-11-28T09:51:05 congrats greentea 2010-11-28T09:51:11 bhasker: http://ai-contest.com/forum/viewtopic.php?f=17&t=1177 2010-11-28T09:51:13 antimatroid1: the waiting begins. 2010-11-28T09:51:18 i saw that post 2010-11-28T09:51:23 will put up the code 2010-11-28T09:51:29 very nice greentea 2010-11-28T09:51:30 oh iouri published his code!! :D 2010-11-28T09:51:39 Thank you) this is temporary ^_^ 2010-11-28T09:52:08 yes, enjoy! it could last though :) 2010-11-28T09:53:13 >> oh iouri published his code!! :D. interesting to read.. 2010-11-28T09:54:10 ha 2010-11-28T09:54:41 i'm not quite sure how chrisH isn't first 2010-11-28T09:54:51 my stategy is very similar.. 2010-11-28T09:55:18 did anyone else do something along the lines of minimax? 2010-11-28T09:55:48 dmj you did yeah? 2010-11-28T09:56:06 Im not will be surprised if algorithms of all top bots are very similar... 2010-11-28T09:56:40 yeah, i expect they'll be reasonably different 2010-11-28T09:56:58 i don't think i'm going to get top 10 for this though 2010-11-28T09:57:04 might get top 30 2010-11-28T09:57:18 antimatroid1: when I get a chance, I will describe more, but I tried to do a 2-ply best first search, with lots of pruning ahead of tmie. 2010-11-28T09:57:40 burny: you beat jimrogerz too! 2010-11-28T09:58:00 dmj111: i implemented a new game tree idea today, and submitted it :P 2010-11-28T09:58:26 it generates each two layers, picks the best strategy combination, then keeps generating from that 2010-11-28T09:58:39 cool. 2010-11-28T09:59:13 IOURI i heve read your post, where you explained your bot, i did something very similar I expect that my bot have almost the same great results as yours. 2010-11-28T09:59:33 is iouri in here? 2010-11-28T09:59:54 shit i can't compile his coe 2010-11-28T09:59:55 code 2010-11-28T10:00:09 it's windows-specific 2010-11-28T10:00:18 http://ai-contest.com/forum/viewtopic.php?f=17&t=1168 2010-11-28T10:00:25 i explained my bot there, along with other attempted ideas 2010-11-28T10:00:31 uppss sorry he is not here i think i have see him 2010-11-28T10:00:33 it's a tad long looking back 2010-11-28T10:00:50 *** Naktibalda has joined #aichallenge 2010-11-28T10:01:04 preselected map set 2010-11-28T10:01:06 @rankings 2010-11-28T10:01:07 amstan: Top 10 players: GreenTea(2729), _iouri_(2713), ChrisH(2711), bix0r4ever(2631), george(2608), _Astek_(2606), CorwinAlex(2583), Neverstu(2567), Polsky(2556), medrimonia(2552) 2010-11-28T10:01:27 Accoun: you would say that no matter what 2010-11-28T10:01:29 hmm 2010-11-28T10:01:35 =) 2010-11-28T10:01:58 oh man 2010-11-28T10:01:58 *** davidjliu has joined #aichallenge 2010-11-28T10:02:24 Accoun: nice timing, right before amstan wakes up :) 2010-11-28T10:02:30 amstan: rankings are all over the show 2010-11-28T10:02:44 some good bots clearly have only tested against other good bots recently 2010-11-28T10:03:01 poo i lost a game 2010-11-28T10:03:38 poo, my bot just timed out :( 2010-11-28T10:03:43 ouch 2010-11-28T10:04:09 heh, at this rate I'm pretty sure I'll never get into the top 100 2010-11-28T10:04:29 davidjliu: a lot of people are losing etc. 2010-11-28T10:04:33 i wouldn't be too worried yet 2010-11-28T10:04:35 this is quite interesting 2010-11-28T10:04:48 the rankings are nothing like before the closing 2010-11-28T10:04:48 yea 2010-11-28T10:04:54 and yea 2010-11-28T10:04:57 how am i meant to go to bed in the next 5 days> 2010-11-28T10:05:02 lol 2010-11-28T10:05:04 +1 2010-11-28T10:05:46 nice to see uwaterloo in 86th 2010-11-28T10:05:52 here comes mcleopold 2010-11-28T10:05:55 he's right near me 2010-11-28T10:06:07 antimatroid1: rank? 2010-11-28T10:06:21 nooooooo I'm 181 ;_; 2010-11-28T10:06:35 lavalamp is 99th 2010-11-28T10:06:37 48 2010-11-28T10:06:44 Error323: I'm in the 300s :) 2010-11-28T10:06:52 amstan: burny is 13th! 2010-11-28T10:06:58 that is astounding 2010-11-28T10:07:10 and a little depressing for me 2010-11-28T10:07:16 i think its still to randomish atm 2010-11-28T10:07:24 he beat some good bots to get there 2010-11-28T10:07:25 http://ai-contest.com/forum/viewtopic.php?f=17&t=1177&p=7836#p7836 2010-11-28T10:07:36 oldman, jimrogerz 2010-11-28T10:07:52 aww i lost my first game 2010-11-28T10:08:02 imagine if everyone overshot the optimal target 2010-11-28T10:08:10 but they didn't know, cause they all did it gradually 2010-11-28T10:08:17 wow burny is doing quite well 2010-11-28T10:08:30 for 2 weeks of work 2010-11-28T10:08:31 thats awesome 2010-11-28T10:08:51 wow indeed 2010-11-28T10:08:55 raschi is having a hard time too 2010-11-28T10:09:12 antimatroid1: he can't resubmit anymore, lol 2010-11-28T10:09:31 lmao 2010-11-28T10:09:31 haha 2010-11-28T10:09:36 haha 2010-11-28T10:09:40 xD 2010-11-28T10:09:41 what? 2010-11-28T10:09:43 well my code is up now 2010-11-28T10:09:50 if anyone is interested 2010-11-28T10:09:51 antimatroid1: he was doing it like crazy before 2010-11-28T10:09:55 its a pile of mess 2010-11-28T10:09:56 oh yeah 2010-11-28T10:09:56 i'm still 1v1ing against anti =) 2010-11-28T10:10:02 its taking forever 2010-11-28T10:10:09 whilest my pc is quite powerfull 2010-11-28T10:10:25 and my turnspeed is ~0.0004s 2010-11-28T10:10:26 @rankings 2010-11-28T10:10:27 amstan: Top 10 players: ChrisH(2824), _iouri_(2737), GreenTea(2735), bix0r4ever(2717), _Astek_(2663), george(2650), medrimonia(2617), CorwinAlex(2605), mogron(2591), Polsky(2585) 2010-11-28T10:10:27 they changed 2010-11-28T10:10:43 is there anyone who hasn't lost yet? 2010-11-28T10:10:48 yep 2010-11-28T10:10:48 chrisH 2010-11-28T10:10:49 top 2010-11-28T10:10:50 he is owning 2010-11-28T10:10:50 can you query that easily 2010-11-28T10:10:51 ah 2010-11-28T10:10:56 so amstan how are the games being decided now 2010-11-28T10:11:00 as in the pairing 2010-11-28T10:11:05 same as before 2010-11-28T10:11:06 bhasker: i'm not sure 2010-11-28T10:11:11 wow 2010-11-28T10:11:14 yeah, as far as i know.. 2010-11-28T10:11:25 i think the randomness is caused by the few games 2010-11-28T10:11:33 ah k 2010-11-28T10:11:43 which was caused by the reset of rankings at closing 2010-11-28T10:11:57 they should go back 2010-11-28T10:11:59 well i am not complaining lol my bot is doing quite well right now 2010-11-28T10:12:11 mine is not >:( 2010-11-28T10:12:19 bhasker: :) 2010-11-28T10:12:21 yea i saw that 2010-11-28T10:12:26 #181 or so 2010-11-28T10:12:36 you seem to be losing to some good bots though 2010-11-28T10:12:36 though last version beats urs 110/80 bhasker against 0.23 that is 2010-11-28T10:12:43 oh this is not 0.23 2010-11-28T10:12:46 this is the max version 2010-11-28T10:12:49 yeah 2010-11-28T10:12:53 i should get a special rankings mode.. 2010-11-28T10:12:57 for people in here 2010-11-28T10:13:08 max one kind of upsets good bots too 2010-11-28T10:13:10 amstan: oh yeah that would be awesome 2010-11-28T10:13:12 though it loses to bad ones too ll 2010-11-28T10:13:14 lol* 2010-11-28T10:13:17 We'll see what happens with the rankings. This may be reason to not seed new submissions back at the old ranking. 2010-11-28T10:13:18 amstan: what you really need is a page of "latest games" 2010-11-28T10:13:29 dmj111: already discussed :) 2010-11-28T10:13:34 my bad. 2010-11-28T10:13:37 yep 2010-11-28T10:13:44 just pointing out we agree :) 2010-11-28T10:13:50 always restart at ~mediaan 2010-11-28T10:13:53 i think dmj it will settle down to the same, from what I can see right now bots are all over the place 2010-11-28T10:14:03 TCP elo ratings are good indicator of bot strengt 2010-11-28T10:14:08 and i don't see the order match up yet 2010-11-28T10:14:26 the problem with tcp rankings is you are playing against other reasonably good bots 2010-11-28T10:14:27 true 2010-11-28T10:14:35 it doesn't klet you know how you perform against not so good bots 2010-11-28T10:14:38 tcp is only good after atleast a night of play though 2010-11-28T10:14:42 yes but it gives you some kind of idea who you should be rated below 2010-11-28T10:14:46 i beat clusterteam :) 2010-11-28T10:14:51 like i know my bot is not stronger than dmjs/barontrozo etc 2010-11-28T10:14:58 nor accoun 2010-11-28T10:15:06 so i should be rated below them on official 2010-11-28T10:15:19 13 Wins / 1 Draw / 2 Loss at the moment 2010-11-28T10:15:20 right now i am above dmj even 2010-11-28T10:15:40 i'm 14 wins, 2 loses 2010-11-28T10:15:44 just by looking at bots i couldn't beat on tcp i can see that rankings are not stable 2010-11-28T10:15:47 16 games, 12 wins, 4 losses 2010-11-28T10:15:53 5 losses -.- 2010-11-28T10:15:58 bhasker: yep, they'll take a while to settle 2010-11-28T10:15:59 *** Utkarsh has quit IRC (Read error: Connection reset by peer) 2010-11-28T10:16:12 oh i should learn howto count lol 2010-11-28T10:16:15 stuped 75 preselected maps 2010-11-28T10:16:33 oh wait did we change the maps for the tourney? 2010-11-28T10:16:34 Accoun: they're far too shit to have been preselected 2010-11-28T10:16:37 are these completely new maps? 2010-11-28T10:16:40 some of them are awful :( 2010-11-28T10:16:41 Accoun: be quiet, or give a mathmatical proof on why its so bad 2010-11-28T10:16:44 bhasker: yes 2010-11-28T10:16:47 oooh 2010-11-28T10:16:55 then it may change the ordering lol 2010-11-28T10:17:02 *** Ice_Harley has quit IRC (Ping timeout: 265 seconds) 2010-11-28T10:17:03 same genreator 2010-11-28T10:17:06 cause everytime we changed the maps the bot ratings changed a bit 2010-11-28T10:17:11 yea even with same generator 2010-11-28T10:17:23 at least thats what i saw 2010-11-28T10:17:39 its a bit of luck sometimes you just get maps that don't trigger certain buggy behaviour 2010-11-28T10:17:46 like selection of planets on turn 1 2010-11-28T10:17:53 that can make or break a game 2010-11-28T10:18:00 yep 2010-11-28T10:18:07 but that's the whole point 2010-11-28T10:18:12 wtf did someone delete the post? 2010-11-28T10:18:15 and what i like 2010-11-28T10:18:17 with the source code 2010-11-28T10:18:21 i don't see it now 2010-11-28T10:18:23 no, it got moved 2010-11-28T10:18:25 oh 2010-11-28T10:18:25 O_O 2010-11-28T10:18:27 go to announcements i think 2010-11-28T10:18:28 oh 2010-11-28T10:18:35 ah misc 2010-11-28T10:18:46 *** rhinobot has joined #aichallenge 2010-11-28T10:18:49 oh, that thread 2010-11-28T10:18:50 :P 2010-11-28T10:19:08 http://ai-contest.com/visualizer.php?game_id=8867758 2010-11-28T10:19:10 oh man 2010-11-28T10:19:12 look at that! 2010-11-28T10:19:13 misc is not the place amstan:-\ 2010-11-28T10:19:21 bhasker: yeah, i know 2010-11-28T10:19:26 *** fgump has quit IRC (Read error: Connection reset by peer) 2010-11-28T10:19:38 mega1s redistribution code does that 2010-11-28T10:19:41 i have seen his bot move ships 2010-11-28T10:19:42 around 2010-11-28T10:19:46 back and forth 2010-11-28T10:19:50 amstan: there are worse ones :( 2010-11-28T10:20:03 Top 10 players: ChrisH(2824), _iouri_(2737), GreenTea(2735), bix0r4ever(2717), _Astek_(2663), george(2650), medrimonia(2617), CorwinAlex(2605), mogron(2591), Polsky(2585) 2010-11-28T10:20:09 and yea the map is bad 2010-11-28T10:20:14 that was a draw against a 130th! 2010-11-28T10:20:21 lol 2010-11-28T10:20:30 asavis finished first before entries closes 2010-11-28T10:20:37 he's just doing terribly atm 2010-11-28T10:20:46 asavis i think is a top 20 bot at least 2010-11-28T10:20:48 if i remember 2010-11-28T10:20:53 from before 2010-11-28T10:20:54 http://ai-contest.com/final-open-ratings.txt 2010-11-28T10:20:56 i don't remember 2010-11-28T10:20:58 that was rankings at the end 2010-11-28T10:21:02 asavis is number 1 2010-11-28T10:21:03 is his name as-* on tcp? 2010-11-28T10:21:11 i assume so 2010-11-28T10:21:12 probably i think 2010-11-28T10:21:15 if so, then yes his bot was pretty good 2010-11-28T10:21:17 those bots are quite good 2010-11-28T10:21:20 indeed 2010-11-28T10:21:21 he had 10 games at closing, i wouldn't day that's accurate 2010-11-28T10:21:22 8348,asavis,196006 1644 223 223 10 2010-11-28T10:21:34 he had been up near the top for a few days 2010-11-28T10:21:35 no thats not even close to accurate 2010-11-28T10:21:36 yea but amstan if his bots were named as-e/d01/2/3 2010-11-28T10:21:38 at least i'm pretty sure 2010-11-28T10:21:39 they are quite good 2010-11-28T10:21:40 dmj though.. seems to do much better 2010-11-28T10:21:47 108 games, he hwas 8th 2010-11-28T10:21:55 seemed* 2010-11-28T10:22:04 amstan why are we not seeing the game counter 2010-11-28T10:22:04 http://ai-contest.com/visualizer.php?game_id=8894549 2010-11-28T10:22:05 ffs 2010-11-28T10:22:09 you play a next game in #... 2010-11-28T10:22:09 maps are so shit 2010-11-28T10:22:12 ah fuck 2010-11-28T10:22:23 lol.. so many draws 2010-11-28T10:22:31 ouch this map is bad 2010-11-28T10:22:31 *** Migi32 has joined #aichallenge 2010-11-28T10:22:41 bhasker: there's lots of really really bad maps 2010-11-28T10:22:59 probably weed them out and restart tourney 2010-11-28T10:23:04 *** Utkarsh has joined #aichallenge 2010-11-28T10:23:05 can't 2010-11-28T10:23:06 oh man that will definitly screw up rankings 2010-11-28T10:23:12 that would cause the biggest uproar ever 2010-11-28T10:23:13 we can.. 2010-11-28T10:23:19 not restart.. 2010-11-28T10:23:23 but replace the maps.. 2010-11-28T10:23:41 imho that wouldn't be fair 2010-11-28T10:23:44 hmmm, i don't like that 2010-11-28T10:23:49 without a reset can't do that 2010-11-28T10:23:56 i think the maps sort of need to be stuck to 2010-11-28T10:23:57 cause people who lost on the bad maps 2010-11-28T10:23:57 nah imo don't do anything now anymore 2010-11-28T10:24:00 it will affect their rating 2010-11-28T10:24:01 or you will piss ppl off 2010-11-28T10:24:06 handselecting maps -->> I can imagine how a certain person is going to respond ;) 2010-11-28T10:24:17 i would agree with him then 2010-11-28T10:24:26 ok.. just saying.. 2010-11-28T10:24:32 it's an option.. to get more maps 2010-11-28T10:24:36 more maps yes 2010-11-28T10:24:39 replace maps no 2010-11-28T10:24:46 agree 2010-11-28T10:24:46 say like... 999 at least :p 2010-11-28T10:24:47 more maps isn't fair too 2010-11-28T10:24:49 i think adding maps everyone will agree too 2010-11-28T10:24:51 a decent bot should just perform well on any map, given that both players share the same advantage/disadvantages of the map 2010-11-28T10:24:53 Mistmanov: 1999! 2010-11-28T10:24:53 why krokok? 2010-11-28T10:24:55 more yes, replace no ++ 2010-11-28T10:24:59 adding maps should not change anything 2010-11-28T10:25:05 unless someone has already played 2 games on the same map, adding maps won't change anythin 2010-11-28T10:25:09 i think more maps should be considered 2010-11-28T10:25:12 bhasker: because it changes probabilities of getting an old map 2010-11-28T10:25:13 the maps atm are terrible 2010-11-28T10:25:17 Mistmanov: +1 2010-11-28T10:25:25 aah 2010-11-28T10:25:25 true 2010-11-28T10:25:30 that statement is true 2010-11-28T10:25:31 janzert: are you about? 2010-11-28T10:25:31 hmm 2010-11-28T10:25:36 although, with 4000 players, there's a good chance that someone has already played a duplicate map 2010-11-28T10:25:38 there is a high probability that someone has already played on the same map twice 2010-11-28T10:25:48 if, when the new maps are added, not all people played on all maps equally, it's not fair 2010-11-28T10:25:56 krokokrusa is right 2010-11-28T10:25:58 we're stuck with them 2010-11-28T10:25:59 so i guess no way to do it without a full reset 2010-11-28T10:26:07 and we can't do a full reset 2010-11-28T10:26:10 we're stuck with them 2010-11-28T10:26:20 I wouldn't mind a full reset -_- 2010-11-28T10:26:23 lol 2010-11-28T10:26:34 mistmanov: you would understand why it can't be done 2010-11-28T10:26:35 how is your bot doing mist 2010-11-28T10:26:46 where has the unleash ur code thread gone off to? i lost it 2010-11-28T10:26:46 meh, crappy enemy selection :p 2010-11-28T10:26:48 resetting would be like handselecting scores then 2010-11-28T10:26:55 haven't had a single high ranking opponent yet 2010-11-28T10:27:00 and lost 3 out of 14 against bad players 2010-11-28T10:27:06 Mistmanov: good :P 2010-11-28T10:27:08 yeah, unfortunately i think maps must stay :( 2010-11-28T10:27:10 they are so shit 2010-11-28T10:27:11 so I'm stuck in 250-ville 2010-11-28T10:27:14 however, so have I >_< 2010-11-28T10:27:17 ouch 2010-11-28T10:27:20 i'm allowed to say that, cause in a way i made them 2010-11-28T10:27:27 so i'm only insulting myself :) 2010-11-28T10:27:28 Mistmanov: pretty sure that's not the ranking's fault, if your bot was good you wouldn't have lost against those bad players 2010-11-28T10:27:39 well, true :p 2010-11-28T10:27:45 losses were my own fault 2010-11-28T10:27:54 yeah this is optimization against top bots 2010-11-28T10:27:58 but that's more like.. my bot is just REALLY crappy on some map types 2010-11-28T10:28:06 *** BaronTrozo has quit IRC (Quit: Page closed) 2010-11-28T10:28:07 same here 2010-11-28T10:28:13 why would map types affect yoru bots 2010-11-28T10:28:27 the current pairing method has the issue that your final ranking is not independent of the order in which you win your games 2010-11-28T10:28:33 cause ultimately everyone is using some heuristics 2010-11-28T10:28:33 but it's better than the old method 2010-11-28T10:28:36 they just break down sometimes 2010-11-28T10:28:40 yep 2010-11-28T10:28:44 so on some maps you always do horribly 2010-11-28T10:28:52 mostly its a turn 1 problem 2010-11-28T10:28:53 foucist: the point is just that a lot of the generated maps are very uninteresting 2010-11-28T10:28:59 my bot is different 2010-11-28T10:29:00 foucist: for example, sparse maps with small number of planets give more time to search a tree :) 2010-11-28T10:29:06 it doesn't do horribly on some maps :P 2010-11-28T10:29:12 I tried to update my bot to be at least have a good start on each kind of maps, it seems to be helping for the beginning 2010-11-28T10:29:24 yea medri turn 1 makes a lot of difference 2010-11-28T10:29:29 just going for the center or not 2010-11-28T10:29:33 can make or break a game 2010-11-28T10:29:44 oh yeh 2010-11-28T10:29:44 frankly i liked the symmetrical maps from before 2010-11-28T10:30:01 actually i think there should be no planet in center 2010-11-28T10:30:04 oh I like the linesymmertry aswell 2010-11-28T10:30:09 :) 2010-11-28T10:30:16 bhasker: wut 2010-11-28T10:30:22 no man, that's cool stuff =) 2010-11-28T10:30:28 antimatroid1: uninteresting to the poor bot that has to think about it? ;) 2010-11-28T10:30:36 it requires precise calculations 2010-11-28T10:30:40 antimatroid1: all the maps seem equally interesting to me 2010-11-28T10:30:51 I love those mirror maps 2010-11-28T10:30:55 *** jmpespxo1eax is now known as jmpespxoreax 2010-11-28T10:31:02 *** jmpespxoreax has joined #aichallenge 2010-11-28T10:31:05 Mistmanov: doggettx is also gaining damn 2010-11-28T10:31:14 sloppybot is dropping though 2010-11-28T10:31:21 foucist: well, some maps just have 1 relatively simple ideal strategy. First turn expand to 2-3 close planets, and then do nothing for the rest of the game 2010-11-28T10:31:33 and wait for the othet guy to make a mistake 2010-11-28T10:31:36 jmpespxoreax: hey 2010-11-28T10:31:42 hi 2010-11-28T10:31:44 Mistmanov: lol 2010-11-28T10:31:44 ? 2010-11-28T10:31:59 thats true foucist 2010-11-28T10:32:01 Mistmanov: yeah that's no good 2010-11-28T10:32:08 got to go 2010-11-28T10:32:10 foucist: the opposite, some are so trivial half decent bots should draw on them 2010-11-28T10:32:32 brb 2010-11-28T10:32:33 Woot win against jimrogerz =) #5 At the moment 2010-11-28T10:32:37 mega1 has started his march to the top 2010-11-28T10:32:38 #22 now 2010-11-28T10:32:44 damn you guys :P 2010-11-28T10:32:47 medrimonia: nice 2010-11-28T10:32:57 Error323 and I have completely different bots as far as I'm aware, but we had a draw on 8 out of 75 official maps 2010-11-28T10:32:58 i'm at 36, slowly rising 2010-11-28T10:33:04 i hope it keeps rising 2010-11-28T10:33:06 mine hasn't gotten a game in quite sometime:-\ 2010-11-28T10:33:10 stuck at 70 2010-11-28T10:33:27 is tcp updated with these maps? 2010-11-28T10:33:30 I'll be falling down I think later, but It's still quite fun 2010-11-28T10:33:33 Mistmanov: :| that's terrible 2010-11-28T10:33:37 maybe we should start up tcp with just these maps 2010-11-28T10:33:39 but nothing can be done 2010-11-28T10:33:40 and run our bots 2010-11-28T10:33:43 and see how they perform 2010-11-28T10:33:54 so have any of you used EC or NN etc in this? 2010-11-28T10:33:57 yeah we have 8 draws and our bots _are_ completely different 2010-11-28T10:33:59 we would need mcleopold 2010-11-28T10:34:02 instead of heuristics 2010-11-28T10:34:13 I tried alphabeta 2010-11-28T10:34:18 see my alphabeta branch 2010-11-28T10:34:24 foucist: i use a minimax tree kind of 2010-11-28T10:34:28 but it failed at comming up with a proper eval function 2010-11-28T10:34:34 mega1_: what is ur eval function? 2010-11-28T10:34:38 antimatroid1: that sounds like heuristics though? 2010-11-28T10:34:39 tell us pleaaaaaaaaaze 2010-11-28T10:34:52 Oh man I am back! 2010-11-28T10:35:05 j3camero: mm? 2010-11-28T10:35:36 Error323: you don't think of alphabeta as a heuristic thing? 2010-11-28T10:35:39 foucist: in a way yeah, but it's not just, pick move make move, pick move make move either 2010-11-28T10:35:41 added 0.23 version of the bot as well 2010-11-28T10:35:41 https://code.google.com/p/malazan-bot/source/browse/#svn/trunk/0.23 2010-11-28T10:35:44 i figure all the trees and stuff is basically heurisitcs 2010-11-28T10:35:51 foucist: well, yes but on a higher level 2010-11-28T10:36:05 j3camero: mega's going up apparently 2010-11-28T10:36:16 well he should 2010-11-28T10:36:17 @rankings 25 2010-11-28T10:36:18 amstan: Top 25 players: ChrisH(2885), GreenTea(2824), _iouri_(2813), _Astek_(2751), medrimonia(2745), bix0r4ever(2720), george(2715), Neverstu(2698), protocolocon(2696), CorwinAlex(2693), jimrogerz(2680), nikiburny(2679), mogron(2673), ssieb0(2653), felixcoto(2645), savchenko(2645), goldman(2645), shangas(2640), jason_k(2620), MrFlibble(2597), Hazard(2597), bocsimacko(2592), grinblatd(2590), (1 more message) 2010-11-28T10:36:38 i'm astounded at burnys bot 2010-11-28T10:36:53 ooh one more victory 2010-11-28T10:37:21 Mistmanov: i can't imagine that just sitting on 3 planets would be the optimal strategy, the enemy bot would have to be right next door to not even bother colonizing some more planets ? 2010-11-28T10:37:27 wow 158 games per minute right now 2010-11-28T10:37:37 foucist: if you could snipe all of them it would be 2010-11-28T10:37:38 it is optimal 2010-11-28T10:37:50 most of the other planets are too expensive 2010-11-28T10:37:58 if you try to go for them enemy will just take them 2010-11-28T10:38:02 *** vpetro has joined #aichallenge 2010-11-28T10:38:13 the planet has to generate enough ships before enemy can take it from you 2010-11-28T10:38:18 for it to be profitable 2010-11-28T10:38:28 ok 2010-11-28T10:38:47 well what about tricking the enemy into attacking a planet 2010-11-28T10:38:57 by first weakening it 2010-11-28T10:39:03 then you have burnt ships 2010-11-28T10:39:06 and gotten nothing 2010-11-28T10:39:11 my code wouldn't fall for that 2010-11-28T10:39:12 in which case enemy will attack your home planet 2010-11-28T10:39:25 better trick enemy by weaken you own planet! :) 2010-11-28T10:39:28 and you can't protect against his initial attack 2010-11-28T10:39:29 *** vpetro has quit IRC (Remote host closed the connection) 2010-11-28T10:39:36 krokokrusa: good point 2010-11-28T10:39:43 i guess if you move your ships away from one of your planets 2010-11-28T10:39:45 i refresh, wanting a new result, but only if it's a win, it's terribly daunting 2010-11-28T10:39:50 yet most of teh time i get nothing D: 2010-11-28T10:39:54 so frustrating to immediatly spot all these horrible flaws in your program, after checking hundreds of tcp games for flaws and finding nothing -_- 2010-11-28T10:39:54 then the enemy would potentially launch ships 2010-11-28T10:40:06 yup 2010-11-28T10:40:11 but hard to code such stuff in 2010-11-28T10:40:18 when you are trying to make a generic bot 2010-11-28T10:40:19 *** Auctus has left #aichallenge ("Leaving") 2010-11-28T10:40:22 to detect standoffs 2010-11-28T10:40:35 and them make moves to attract enemy is hard 2010-11-28T10:40:40 #44 now! 2010-11-28T10:40:58 you know, i wonder what strategy a complete minimax system would come up with 2010-11-28T10:40:59 http://ai-contest.com/profile.php?user_id=5822 2010-11-28T10:41:02 1 loss! 2010-11-28T10:41:05 yeah tbh i'd rather just write code that generates all the strategies itself somehow 2010-11-28T10:41:08 it would be awesome to be able to see that 2010-11-28T10:41:10 Error323: not possible 2010-11-28T10:41:11 ever 2010-11-28T10:41:12 evolutionary computation, neural networks, whatever it takes 2010-11-28T10:41:21 rather than coding the strategies directly.. 2010-11-28T10:41:22 17 games 16 wins/1loss 2010-11-28T10:41:24 antimatroid1: i know 2010-11-28T10:41:39 i have to admit, i'm rather sick of minimax 2010-11-28T10:41:46 focuist the problme is the 1 second rule 2010-11-28T10:41:47 it's too simplistic 2010-11-28T10:41:53 amstan: where has the unleash ur code thread gone too? 2010-11-28T10:42:02 it's up the top 2010-11-28T10:42:03 Error323: announcements, on top 2010-11-28T10:42:05 you could probably do a lot more with a bot that takes say 10 seconds or something 2010-11-28T10:42:06 oh 2010-11-28T10:42:14 foucist: have you read this? http://ai-contest.com/forum/viewtopic.php?f=17&t=1136 2010-11-28T10:42:17 Error323: there were 3 threads doing the same thing, imerged them 2010-11-28T10:42:29 wait, maybe not 2010-11-28T10:42:35 amstan: that's the wrong version 2010-11-28T10:42:42 where is my thread 2010-11-28T10:42:43 ? 2010-11-28T10:42:54 Error323: i merged all of them.. 2010-11-28T10:42:56 1 below mcleopold hmm 2010-11-28T10:43:08 ohhh 2010-11-28T10:43:10 srry 2010-11-28T10:43:17 Error323: second page 2010-11-28T10:43:22 yep, got it 2010-11-28T10:43:28 where is rebelxt 2010-11-28T10:43:58 google docs antimatroid??? put it up on code.google.com or something 2010-11-28T10:44:13 gist.github.com 2010-11-28T10:45:26 bhasker: meh :P 2010-11-28T10:45:44 come on one can't even browse your code:p without having to download it:p 2010-11-28T10:45:45 github is most awesome 2010-11-28T10:45:48 put it up somewhere better 2010-11-28T10:45:59 http://ai-contest.com/visualizer.php?game_id=8890691 at turn 37, he attacks the central neutral. A ridiculous move since I can snipe it (and I do, from the planet on the right). But since no one makes such a move, I never spotted the bug where my supply thing captures the planet first from the north.. 2010-11-28T10:46:23 which makes me lose the game ^^ 2010-11-28T10:46:28 where can i do it at code.google? 2010-11-28T10:46:31 iouri's code is nice btw 2010-11-28T10:46:50 mine is awful 2010-11-28T10:46:56 i don't advise anyone actually reads it 2010-11-28T10:47:06 oh second loss:-\ 2010-11-28T10:47:16 just goto code.google.com and create a new project 2010-11-28T10:47:17 damn cfaftw beat me 2010-11-28T10:47:25 click on project hosting 2010-11-28T10:47:41 https://code.google.com/hosting/ 2010-11-28T10:47:47 click on Create a new porject 2010-11-28T10:47:49 project* 2010-11-28T10:48:00 Mistmanov: wow yeah, that's pretty nasty 2010-11-28T10:48:26 bah, what versioning and code license? or doesn't matter? 2010-11-28T10:48:41 oh, i crashed, worst 2010-11-28T10:49:10 versioning just choose svn 2010-11-28T10:49:14 license BSD 2010-11-28T10:49:22 bhasker: ew.. just get github 2010-11-28T10:49:33 i don't use git :- 2010-11-28T10:49:39 *** Mekanik has quit IRC (Quit: WeeChat 0.3.3) 2010-11-28T10:49:47 never bothered with it since i use svn at work 2010-11-28T10:49:55 so just used svn for the project 2010-11-28T10:50:01 git is the most awesome scm ever 2010-11-28T10:50:15 i am never a big fan of trying out new scm systems 2010-11-28T10:50:18 antimatroid1: https://github.com/ 2010-11-28T10:50:21 svn is good enough for me:) 2010-11-28T10:50:33 bhasker: so were zip files before scm's 2010-11-28T10:50:38 mega1 #15 2010-11-28T10:50:41 you gotta try it to see its benefit 2010-11-28T10:50:42 woot 2010-11-28T10:50:49 @rankings 17 2010-11-28T10:50:50 amstan: Top 17 players: _Astek_(2902), GreenTea(2889), george(2872), _iouri_(2845), medrimonia(2837), protocolocon(2833), Neverstu(2820), CorwinAlex(2799), ChrisH(2785), felixcoto(2769), mogron(2763), jimrogerz(2753), savchenko(2740), nikiburny(2739), bocsimacko(2720), jason_k(2704), bix0r4ever(2702) 2010-11-28T10:50:56 at work i can't really use git 2010-11-28T10:51:07 i can't believe my bot crashed, stupid bot 2010-11-28T10:51:14 and for personal code ( which i rarely write) svn is more than enough 2010-11-28T10:51:15 for now 2010-11-28T10:51:21 bhasker: http://ftp.newartisans.com/pub/git.from.bottom.up.pdf :) 2010-11-28T10:51:27 read it, and be amazed 2010-11-28T10:51:28 bhasker: i can't even work out how to upload files :\ 2010-11-28T10:51:29 hehe 2010-11-28T10:51:35 antimatroid 2010-11-28T10:51:38 checout the code 2010-11-28T10:51:43 it will create an empty directory 2010-11-28T10:51:50 copy in your files to that directory 2010-11-28T10:51:56 svn add 2010-11-28T10:52:02 svn commit -m"some random log comment" 2010-11-28T10:52:12 to create subdirectories 2010-11-28T10:52:17 svn mkdir 2010-11-28T10:52:36 to check out the code click on the checkout tab 2010-11-28T10:52:41 it has instructions on how to do it 2010-11-28T10:52:59 bah, way too much effort 2010-11-28T10:53:03 one time 2010-11-28T10:53:12 antimatroid1: ugh.. how many files is it? 2010-11-28T10:53:22 antimatroid1: do you have a zip? 2010-11-28T10:53:29 10 files 2010-11-28T10:53:34 zip? 2010-11-28T10:53:45 http://tinyurl.com/antimatroidsbot 2010-11-28T10:54:05 bah, i hope i don't get too many crashes 2010-11-28T10:54:10 *** krokokrusa has quit IRC (Ping timeout: 260 seconds) 2010-11-28T10:54:45 brb 2010-11-28T10:54:50 error323 will read it someother time looks quite complicated 2010-11-28T10:54:54 i'm going crazy here 2010-11-28T10:54:56 have to wrap my head around git concepts 2010-11-28T10:55:40 bhasker: yes, well its not hard if i can read it you can too i'm sure. But it gives a good base on how it works and why its so much better 2010-11-28T10:55:40 looks interesting 2010-11-28T10:55:46 i had well under 2000 lines of code in the end 2010-11-28T10:55:46 https://github.com/amstan/antimatroid 2010-11-28T10:56:03 amstan: cheers 2010-11-28T10:56:23 amstan: I <3 the server amount 2010-11-28T10:56:34 will it become moaar epic? 2010-11-28T10:56:37 or is this max? :) 2010-11-28T10:57:07 in other news i got a new ipad! 2010-11-28T10:57:11 time to go tinker with it 2010-11-28T10:57:15 now that bot is done with 2010-11-28T10:57:20 git is awesome 2010-11-28T10:57:52 antimatroid1: now with your readme: https://github.com/amstan/antimatroid 2010-11-28T10:57:57 yep 2010-11-28T10:58:06 well as long as you don't put binary data into it 2010-11-28T10:58:13 they should fix that 2010-11-28T10:58:19 ? 2010-11-28T10:58:24 if it could handle binary good aswell then its epic 2010-11-28T10:58:41 git can handle binary just fine, it's just not pretty 2010-11-28T10:58:49 *** medrimonia has quit IRC (Ping timeout: 245 seconds) 2010-11-28T10:58:52 using it to backup my minecraft stuff actually 2010-11-28T10:59:04 amstan: i mean really lots and lots of binary data > 100 gig 2010-11-28T10:59:05 *** perror has joined #aichallenge 2010-11-28T10:59:07 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-11-28T10:59:27 perforce seems to be handling that a lot better 2010-11-28T10:59:28 lol, yeah, i don't see why you would ever use that much data in a scm 2010-11-28T10:59:28 *** Naktibalda has joined #aichallenge 2010-11-28T10:59:46 amstan: game companies 2010-11-28T11:00:02 they must be doing something wrong then.. 2010-11-28T11:00:04 bhasker: forum posts are now updated for you :) 2010-11-28T11:00:14 game assets, like 3D files, photoshop files, content 2010-11-28T11:00:17 haha thanks antimatroid 2010-11-28T11:00:49 thank amstan :P 2010-11-28T11:00:51 error323 don't most game companies use custom asset management tools 2010-11-28T11:00:58 i doub they maintain it in an SCM 2010-11-28T11:01:08 well I'm intern at one 2010-11-28T11:01:12 ah which one? 2010-11-28T11:01:13 and they use perforce 2010-11-28T11:01:18 guerrilla games 2010-11-28T11:01:32 i have been thinking of joining one 2010-11-28T11:01:34 for a lon time 2010-11-28T11:01:35 Error323: many assets i leave outside of the SCM.. depending on how static they are etc 2010-11-28T11:01:36 *** krokokrusa has joined #aichallenge 2010-11-28T11:01:47 except i have no experience that i can write about when it comes to game dev 2010-11-28T11:02:15 Error323: unless an asset is directly changeable i wouldn't put it in a SCM, i'd just use a standard backup process, timestamp it or whatever 2010-11-28T11:02:42 the moment you add it to an scm, you effectively double it's size 2010-11-28T11:02:49 lol, my bot has like a HUGE bug i think 2010-11-28T11:02:56 since you have to store it both in the scm and the working dir 2010-11-28T11:03:00 that's kind of funny that it's doing well 2010-11-28T11:03:06 well that's easy to say for the geeks 2010-11-28T11:03:17 focuist doesn't work in a big company 2010-11-28T11:03:21 with thousands of assets 2010-11-28T11:03:25 but if you don't know squad about this stuff (most artists don't) you just want to be able to "update" 2010-11-28T11:03:25 you need some revisioning 2010-11-28T11:03:28 and be up to date 2010-11-28T11:03:30 yeah 2010-11-28T11:03:34 and not zipfiles 2010-11-28T11:03:41 and you want reproducable builds 2010-11-28T11:03:45 oh wow, my bot is not doing what i think at all aha 2010-11-28T11:03:53 that's incredible 2010-11-28T11:03:59 you don't want wrong versions of binaries/models being pulled in the next time you do a build 2010-11-28T11:04:09 working at a gaming company as well, all assets are put in svn, and indeed most artists don't know anything about such stuff 2010-11-28T11:04:19 which one heikoh 2010-11-28T11:04:24 khaeon games 2010-11-28T11:04:28 i should send my resume to all of you at game companies:) 2010-11-28T11:04:42 * Error323 humble little intern :P 2010-11-28T11:04:49 hehe look at 10th place here: http://kaggle.com/view-postlist/forum-29-rta-freeway-travel-time-prediction/topic-208-rmse-vs-percent-error/task_id-2467 2010-11-28T11:05:02 dhartmei 2010-11-28T11:05:07 i don't think i intend to move to europe:) 2010-11-28T11:05:24 though that will give me an opportunity to see more of europe 2010-11-28T11:05:39 https://github.com/amstan/antimatroid/blob/master/Tree.cc 2010-11-28T11:05:39 lines 81-83 should be outside that for loop, and i am not clearing the nodes that aren't to be added, but i think it's just running my original minimax idea just really weirdly 2010-11-28T11:06:24 lol 2010-11-28T11:06:30 dammn I got raped hard by sajmon 2010-11-28T11:06:32 that's pretty funny :D 2010-11-28T11:06:35 antimatroid1: you can make an account and add comments to your code 2010-11-28T11:06:56 amstan: i'm more worried about it being my submitted bot :P 2010-11-28T11:07:02 but it's currently in 36th place 2010-11-28T11:07:06 3 losses, one a crash 2010-11-28T11:07:36 i'm going to run the current version against a corrected version and see what happens 2010-11-28T11:07:44 with any luck the submitted one is better anyway 2010-11-28T11:07:58 did you guys know you could edit the files directly on github? i only discovered this recently 2010-11-28T11:08:36 @rankings 15 2010-11-28T11:08:38 amstan: Top 15 players: george(2959), _Astek_(2903), GreenTea(2889), medrimonia(2883), _iouri_(2883), protocolocon(2871), Neverstu(2853), CorwinAlex(2831), jimrogerz(2826), ChrisH(2823), felixcoto(2813), mogron(2806), savchenko(2788), nikiburny(2779), bocsimacko(2773) 2010-11-28T11:08:38 *** _W is now known as _W_ 2010-11-28T11:08:47 he's going down 2010-11-28T11:10:47 whois going down 2010-11-28T11:10:59 i gotta say i'm doing pretty horrific on this mapset 2010-11-28T11:10:59 :/ 2010-11-28T11:11:02 so far anyway 2010-11-28T11:11:12 i thikn its getting more stable now 2010-11-28T11:11:17 i see more of the top bots come back into top 100 2010-11-28T11:11:24 anyone want me to speed up contestbot? 2010-11-28T11:11:26 for rankings? 2010-11-28T11:11:31 change params maybe? 2010-11-28T11:11:33 like top 20? 2010-11-28T11:11:46 wow medri #4 2010-11-28T11:11:47 mm no i think top 10 is good 2010-11-28T11:11:56 amstan: i think it's fine as is 2010-11-28T11:11:57 every 15 min though? 2010-11-28T11:12:02 amstan: yes :> 2010-11-28T11:12:14 or well 2010-11-28T11:12:32 is that lower than the frequency the rankings are updated aswell? 2010-11-28T11:12:38 otherwise no 2010-11-28T11:12:40 anyone a fan of steven Erikson here? 2010-11-28T11:12:52 @repeat rankings 900 "echo [rankings]" 2010-11-28T11:12:53 Top 10 players: george(3029), _Astek_(2927), GreenTea(2923), medrimonia(2922), protocolocon(2905), _iouri_(2895), jimrogerz(2884), mogron(2873), savchenko(2851), ChrisH(2847) 2010-11-28T11:12:58 Error323: idk 2010-11-28T11:15:28 it seems if i make the quick fix it makes no moves at all, so meh 2010-11-28T11:17:11 lot of new names in top 100 2010-11-28T11:17:17 and seem to be beating some of the good bots too 2010-11-28T11:18:37 i wish i hadn't crashed before 2010-11-28T11:18:38 mmm 2010-11-28T11:18:41 how is that my bot beat a #27 but my elo/ranking nothing changed:-\ 2010-11-28T11:18:44 that ruined by goodish run 2010-11-28T11:18:47 I got 22 wins vs iouri on the official mapset 2010-11-28T11:18:55 52 losses :/ 2010-11-28T11:19:12 well iouris bot is quite good 2010-11-28T11:19:17 so 22 wins isn't so bad 2010-11-28T11:19:47 http://ai-contest.com/profile.php?user_id=10154 except for this :/ 2010-11-28T11:20:20 ouch 2010-11-28T11:20:59 :( 2010-11-28T11:21:45 *** aizu has joined #aichallenge 2010-11-28T11:21:57 looking at iouri's code makes me feel like a noobie coder:( 2010-11-28T11:22:08 lol 2010-11-28T11:22:14 ye he is a good coder 2010-11-28T11:22:47 though his code in gametimeline isnot really the most understandable 2010-11-28T11:22:56 I wrote a simulator 2010-11-28T11:23:04 ye ai saw your code 2010-11-28T11:23:10 oh wow, I'm +500 ranks from when the deadline hit 2010-11-28T11:23:11 I don't see why ppl didn't do that 2010-11-28T11:23:20 https://github.com/Error323/googleai/blob/heuristics/Simulator.cc 2010-11-28T11:24:01 well my bot does something similar 2010-11-28T11:24:03 gamestate.py 2010-11-28T11:24:08 takes the state and fleets 2010-11-28T11:24:10 and resolves 2010-11-28T11:24:15 and has an evaluate function 2010-11-28T11:24:19 to get a score for the state 2010-11-28T11:24:24 http://ai-contest.com/profile.php?user_id=12203 2010-11-28T11:24:28 could that be avanish dixit? 2010-11-28T11:24:39 no he is my colleague 2010-11-28T11:24:48 works in my team 2010-11-28T11:24:50 ah, that would be avanash 2010-11-28T11:24:55 he didnt spend much time on it though 2010-11-28T11:25:04 though looks like last 4 days he has been active 2010-11-28T11:25:30 mega1 #9 now 2010-11-28T11:25:31 i'm at 31 :( 2010-11-28T11:25:35 medri #2 2010-11-28T11:25:36 that pesky crash!! 2010-11-28T11:25:39 grats 2010-11-28T11:25:53 #51 now 2010-11-28T11:25:54 for me 2010-11-28T11:26:09 bhasker: mmm i wonder wether your bot is better than mine 2010-11-28T11:26:11 it might be 2010-11-28T11:26:19 though iceharley just owned me http://ai-contest.com/visualizer.php?game_id=8904376 2010-11-28T11:26:25 i doubt it error 2010-11-28T11:26:28 probably close to even 2010-11-28T11:26:29 try it out 2010-11-28T11:26:35 or worse 2010-11-28T11:26:37 yeah gotta find ur source 2010-11-28T11:26:48 oh its on the post 2010-11-28T11:26:58 http://www.ai-contest.com/forum/viewtopic.php?f=19&t=1168&start=20#p7836 2010-11-28T11:27:14 bot max ( my last submission) and 0.23 2010-11-28T11:27:25 ok 2010-11-28T11:27:31 well i know about 0.23 2010-11-28T11:27:35 though not on new mapset 2010-11-28T11:27:48 yea i was debating if i should just reupload 0.23 2010-11-28T11:27:53 Top 10 players: george(2999), medrimonia(2973), _iouri_(2941), _Astek_(2929), GreenTea(2920), jimrogerz(2919), Neverstu(2905), protocolocon(2904), bocsimacko(2898), mogron(2867) 2010-11-28T11:27:54 then i decided against it 2010-11-28T11:27:58 cause i like the new version better loo 2010-11-28T11:28:00 lol* 2010-11-28T11:28:04 it does more interesting stuff 2010-11-28T11:28:10 0.23 is a very boring conservative bot 2010-11-28T11:28:36 bhasker: do I need MyBot.py or malazanbot.py? 2010-11-28T11:28:42 both are the same 2010-11-28T11:28:49 i just run it using malazanbot.py 2010-11-28T11:28:55 oh right 2010-11-28T11:28:55 but for official i kept mybot.py 2010-11-28T11:29:00 yeah, i normally have a set arrive time for my fleets, but it is nice and aggressive if i send any fleets now that arrive when they owne the planet (and didn't just claim neutral) 2010-11-28T11:29:20 but sometimes it gets a little too agressive, i added it in again right at the end and it seems to be paying off for the moment 2010-11-28T11:29:33 heh mistakes that help are not bugs but features 2010-11-28T11:29:55 bugs provide character 2010-11-28T11:30:05 though error323 you may want to change logutils.py to not spout debug logs 2010-11-28T11:30:19 oh its running already 2010-11-28T11:30:20 oh well 2010-11-28T11:30:21 otherwise it outputs like over 100mb of logs per game:) 2010-11-28T11:30:33 doesn't matter much :P 2010-11-28T11:30:42 it slows down the bot a lot 2010-11-28T11:30:45 i'll brb gonnna make some food 2010-11-28T11:30:47 smiley isn't doing very well :D 2010-11-28T11:31:11 i am bouncing all over 2010-11-28T11:31:15 went from 70-51 to 62 2010-11-28T11:31:19 *** aizu has quit IRC (Read error: Connection reset by peer) 2010-11-28T11:31:29 though overall elo is still climbing 2010-11-28T11:31:43 mega1 on his relentless march to the top 2010-11-28T11:31:44 #8 2010-11-28T11:31:53 *** aizu has joined #aichallenge 2010-11-28T11:32:15 mega1_: you're awfully quiet there 2010-11-28T11:32:48 *** heikoh has quit IRC (Quit: Page closed) 2010-11-28T11:32:48 bhasker: running with warning logs 2010-11-28T11:32:54 yea better 2010-11-28T11:33:09 antimatroid1: i think he's hitting refresh 100 times a second ;) 2010-11-28T11:33:11 should not output any logs now 2010-11-28T11:33:31 i don't think he has anything to be worried 2010-11-28T11:34:02 unless someone comes out of the blue his bot will pretty much win this competition for sure 2010-11-28T11:34:03 george he needs to worry about 2010-11-28T11:34:12 don't think george's bot is as good as mega1s 2010-11-28T11:34:16 neither 2010-11-28T11:34:18 but it's good 2010-11-28T11:34:29 Mistmanov: damn doggetx keeps on climbing, we must kill him 2010-11-28T11:34:49 whats with you dutch folks fighting amongst yourselves 2010-11-28T11:34:59 :D 2010-11-28T11:35:15 I wanna cut their eyes out and put their head on a stick 2010-11-28T11:35:17 it would be fun to take all open-sourced bots and make 1 bot from them 2010-11-28T11:35:23 but ya know... i love holland ^_^ 2010-11-28T11:35:24 there is sloppybot too 2010-11-28T11:35:25 :P 2010-11-28T11:35:28 for example forwarding input in round-robin fashion 2010-11-28T11:35:29 Rankings are fluctuating a bit. We'll see how long it takes to settle down. 2010-11-28T11:35:31 turn by turn 2010-11-28T11:35:32 yeah I know I beat him for a fact 2010-11-28T11:35:41 well unless he did sth awesome last submission 2010-11-28T11:35:52 *** Ken69267 has quit IRC (Ping timeout: 255 seconds) 2010-11-28T11:36:33 will probably mess up things krokor 2010-11-28T11:36:53 if you line up a bad bot after a good one 2010-11-28T11:37:10 if, say, only 10 bots are used 2010-11-28T11:37:20 then deviation from normal bot's prediction won't be more than 10 turns 2010-11-28T11:37:36 10 turns is enough to make or break a bots game 2010-11-28T11:37:39 heck one bad luanch 2010-11-28T11:37:41 it should be able to patch things up, maybe :) 2010-11-28T11:37:42 and its game over 2010-11-28T11:37:42 :) 2010-11-28T11:37:45 *** Ken69267 has joined #aichallenge 2010-11-28T11:37:51 yeah, probably 2010-11-28T11:38:15 did anyone try making a bunch of different bots 2010-11-28T11:38:28 then playing them against each other and picking the best moves like that? 2010-11-28T11:39:04 *** mceier has quit IRC (Quit: leaving) 2010-11-28T11:39:07 space.invaders did :) 2010-11-28T11:39:08 what would be interesting would be to 2010-11-28T11:39:16 to run all 10 bots decision making 2010-11-28T11:39:18 on each turn 2010-11-28T11:39:20 then pick the best one 2010-11-28T11:39:32 and see if that makes a better bot ^ 2010-11-28T11:39:43 give same state to all 10 bots 2010-11-28T11:39:48 and collect result /evaluate 2010-11-28T11:39:48 and every time bocsimacko will be chosen :) 2010-11-28T11:39:50 bah, i lost another game 2010-11-28T11:40:02 bhasker: lol 2010-11-28T11:40:05 hehe not neccessarily unless we know his eval function 2010-11-28T11:40:19 his eval function surely is not the simple one i am using 2010-11-28T11:40:20 :-\ 2010-11-28T11:40:27 34 wins, 35 losses, rest is draw. you are better by 1 game on this mapset 2010-11-28T11:40:38 haha 2010-11-28T11:40:39 :D 2010-11-28T11:40:55 i wonder how rebel's bot is doing on this mapset 2010-11-28T11:41:03 his and my bot were also very close 2010-11-28T11:41:30 good 2010-11-28T11:41:43 #1,2 of NL are being kicked out 2010-11-28T11:41:51 mist and me are comming in :D 2010-11-28T11:42:14 heh 2010-11-28T11:42:28 lot of good bots from the US:-\ 2010-11-28T11:42:42 well us >> nl 2010-11-28T11:42:54 Top 10 players: george(3055), _iouri_(3035), medrimonia(3007), protocolocon(2994), _Astek_(2988), GreenTea(2985), jimrogerz(2972), bocsimacko(2956), mogron(2953), CorwinAlex(2940) 2010-11-28T11:42:54 hehe i wonder if i should have set my country to india 2010-11-28T11:42:59 given that i am an indian 2010-11-28T11:43:03 then i could be #1 probably 2010-11-28T11:43:08 haha 2010-11-28T11:43:17 http://www.ai-contest.com/country_profile.php?country_id=5 2010-11-28T11:43:38 lot of python lovers in india 2010-11-28T11:43:41 *** jaspervdj has quit IRC (Quit: NEVER GONNA GIVE YOU UP NEVER GONNA LET YOU DOWN) 2010-11-28T11:43:50 almost all of them are from the same college 2010-11-28T11:43:59 IIT is the premier engineering college in india 2010-11-28T11:44:05 there are like 7 of them 2010-11-28T11:44:21 bhasker: do they know each other 2010-11-28T11:44:30 who knows they are probably class mates 2010-11-28T11:44:58 each IIT has over a few hundred students or more 2010-11-28T11:45:23 *** rebelxt_ has joined #aichallenge 2010-11-28T11:45:23 though its weird that none of the other IIT's have any submissions 2010-11-28T11:45:44 oh wow, someone from my uni placed 11 overall 2010-11-28T11:45:49 probably not many people knew about it 2010-11-28T11:46:06 goldman? 2010-11-28T11:46:12 yeah 2010-11-28T11:46:37 whats your official name 2010-11-28T11:46:40 s/placed/is currently placed at/ 2010-11-28T11:46:45 don't see any aizu 2010-11-28T11:46:46 theprotoss 2010-11-28T11:47:01 ah k 2010-11-28T11:47:04 starcraft fan 2010-11-28T11:47:12 for aiur! 2010-11-28T11:47:21 warp fields stabilized 2010-11-28T11:47:26 :) 2010-11-28T11:47:30 my life for aiur 2010-11-28T11:47:49 battlecruiser online 2010-11-28T11:47:56 got a match 2010-11-28T11:48:07 medic reporting! 2010-11-28T11:48:34 i forget been so long 2010-11-28T11:49:27 I've been away from sc2 for the whole holiday break -- I can't wait to get back and play 2010-11-28T11:49:48 * krokokrusa knows only "we require more vespen gas" 2010-11-28T11:50:01 oh, and "need more canadian overlords" 2010-11-28T11:50:27 lol 2010-11-28T11:53:08 *** perror has quit IRC (Quit: Bye all !) 2010-11-28T11:53:23 sc2 never interested me 2010-11-28T11:53:30 oh yea need more overlords 2010-11-28T11:53:40 nuclear launch detected 2010-11-28T11:53:47 oh wait thats supreme commander 2010-11-28T11:53:48 doh 2010-11-28T11:54:06 rebelxt how is your bot doing 2010-11-28T11:55:01 it's not doing well so far :( 104th 2010-11-28T11:55:17 another local tournament results of 5 open-sourced bots, which I was able to compile: http://pastebin.com/jL9KMmEw 2010-11-28T11:55:20 I love StartCraft2 ! 2010-11-28T11:55:38 on 10 generated maps 2010-11-28T11:55:38 ah well you went up against iouri and dmj 2010-11-28T11:55:53 mine isnt there on the list krokor!@ 2010-11-28T11:56:09 I love StartCraft2 and see demos 2010-11-28T11:56:13 sc2 blows 2010-11-28T11:56:21 its like blizzard made it for the koreans 2010-11-28T11:56:23 bhasker: I know :) I started it just before you shared your code 2010-11-28T11:56:24 didnt change much at all 2010-11-28T11:56:49 i play only 4 or 5 StartCraft2 games on betatest 2010-11-28T11:56:51 aww 2010-11-28T11:57:12 I'll try to add your results there, should be easy, w/o changing the maps 2010-11-28T11:57:13 blisard donr do a key remap 2010-11-28T11:57:21 blisard dont do a key remap in SC2 2010-11-28T11:57:54 Top 10 players: george(3109), _iouri_(3058), medrimonia(3031), _Astek_(3030), protocolocon(3022), bocsimacko(3001), GreenTea(2994), jimrogerz(2994), CorwinAlex(2962), felixcoto(2960) 2010-11-28T11:58:05 bhasker: your bot is doing well 2010-11-28T11:58:10 slin took me out, unsurprisingly 2010-11-28T11:58:10 http://sc2ranks.com/ru/5835/GreenTea ;) 2010-11-28T11:58:13 till now yes 2010-11-28T11:58:18 but its bouncing around quite a bit 2010-11-28T11:58:21 bhasker: i feel the same 2010-11-28T11:58:27 http://www.ai-contest.com/profile.php?user_id=3945 2010-11-28T11:58:30 recent stats 2010-11-28T11:59:35 Rank 12 of Division Liberty Xi 2010-11-28T12:00:25 in beta bw platinum gold silver ets divisions 2010-11-28T12:00:37 Division Liberty ??? wtisiit? 2010-11-28T12:00:59 accoun i can never make any sense of what you talk about 2010-11-28T12:01:12 *** phreeza has quit IRC (Ping timeout: 276 seconds) 2010-11-28T12:01:29 All players are splitted by leagues and by divisons 2010-11-28T12:02:08 lol another lamo map 2010-11-28T12:02:09 http://ai-contest.com/visualizer.php?game_id=8909412 2010-11-28T12:02:16 waited for an hour for this game 2010-11-28T12:02:18 bah 2010-11-28T12:02:22 fark 2010-11-28T12:02:26 so many crap maps 2010-11-28T12:02:48 GreenTea> it diamond ligue? 2010-11-28T12:02:51 i swear none of the example sets generated were this bad 2010-11-28T12:03:07 some maps are weird indeed. who wrote that generator?! ;-)) 2010-11-28T12:03:13 :P 2010-11-28T12:03:54 our man antimatroid 2010-11-28T12:04:01 Division Liberty Xi - platinum 2010-11-28T12:04:02 that was implied :P 2010-11-28T12:04:22 GreenTea> źīč8 ļīķčģąåų? 2010-11-28T12:04:31 GreenTea> koi8 2010-11-28T12:04:32 that doesn't make sense in any language 2010-11-28T12:05:22 *** perror has joined #aichallenge 2010-11-28T12:05:39 bx0r isn't doing very well, for him 2010-11-28T12:05:56 in fact, burny beat him 2010-11-28T12:06:08 *** amriedle has joined #aichallenge 2010-11-28T12:06:15 GreenTea> ĆŖĆ®ĆØ8 ĆÆƮƭĆØƬƠƄĆø? ?? 2010-11-28T12:06:23 crap, lost a third game in a row 2010-11-28T12:06:56 GreenTea> nea - ne mogu chiratx takoe 2010-11-28T12:07:04 GreenTea> utf8 ? 2010-11-28T12:07:32 ne znau , firifox ne otobrazaet normalno 2010-11-28T12:07:42 *** Ken69267 has quit IRC (Ping timeout: 276 seconds) 2010-11-28T12:08:12 i guess on that note i shall get some sleep, wake up and see how bad it manages to do in that time :) 2010-11-28T12:08:15 GreenTea> kodorovka koi8 2010-11-28T12:08:50 GreenTea> ili win1251 2010-11-28T12:08:52 dmj had a slow start, but now making his way back to the top 2010-11-28T12:09:27 eto wopros ili unverzdenie?) 2010-11-28T12:09:28 bocsi has almost made it to his place 2010-11-28T12:10:09 *** Ken69267 has joined #aichallenge 2010-11-28T12:10:30 GreenTea> utverzdenie- tm dolzna bit nastrojka 2010-11-28T12:10:41 GreenTea> utverzdenie- tam 2010-11-28T12:10:55 gde? 2010-11-28T12:11:10 GreenTea> v svoistah servera 2010-11-28T12:11:17 was there a webpage that shows the statistics for each map? (draw percentage etc) 2010-11-28T12:11:40 *** McLeopold has joined #aichallenge 2010-11-28T12:12:05 hi everybody 2010-11-28T12:12:10 translating for GreenTea :) "ŠŗŠ¾Šø8 ŠæŠ¾Š½ŠøŠ¼Š°ŠµŃˆ" 2010-11-28T12:12:28 McLeopold: hi 2010-11-28T12:12:31 tcp.zeroviz.us is running the final's map set right now, if you want to join 2010-11-28T12:12:32 you are doing reasonably well 2010-11-28T12:12:36 vot eto uvidel 2010-11-28T12:12:37 just started 2010-11-28T12:12:43 Š¼Š¾Šµ Š²ŠøŠ“Š½Š¾? 2010-11-28T12:12:48 уŠ³Ńƒ 2010-11-28T12:12:51 äą 2010-11-28T12:12:54 Top 10 players: protocolocon(3131), george(3126), bocsimacko(3120), medrimonia(3080), _iouri_(3071), _Astek_(3067), GreenTea(3049), goldman(3031), mogron(3031), jimrogerz(3022) 2010-11-28T12:13:02 ))) 2010-11-28T12:13:09 ņąéķą’ šóńńźą’ źīäčšīāźą 2010-11-28T12:13:10 Š Š¼Š½Šµ тŠ²Š¾Šµ Š½ŠµŃ‚) 2010-11-28T12:13:38 ia tvoi normalno mizu 2010-11-28T12:13:44 lol 2010-11-28T12:14:05 тŠ°Š¹Š½Š°Ń руссŠŗŠ°Ń ŠŗŠ¾Š“ŠøрŠ¾Š²ŠŗŠ° 2010-11-28T12:14:06 chrift biz ruskih bukv navernoe 2010-11-28T12:14:41 antimatroid1: I think I'm getting lucky right now. dmj111 is at 80 or so 2010-11-28T12:14:42 vidno vidno ;-) 2010-11-28T12:14:57 McLeopold: same, i'm just above 40 2010-11-28T12:15:07 ??????????, ???????? ?? ?????????? ????? ?? ?? ???????? ????? ?????. 2010-11-28T12:15:08 =] 2010-11-28T12:15:15 Kinda social that people started talking in a non-English language in a channel that seems to be mainly English... 2010-11-28T12:15:16 @rankings 2010-11-28T12:15:18 antimatroid1: Top 10 players: protocolocon(3131), george(3126), bocsimacko(3120), medrimonia(3080), _iouri_(3071), _Astek_(3067), GreenTea(3049), goldman(3031), mogron(3031), jimrogerz(3022) 2010-11-28T12:15:21 McLeopold: i'm gonna jump on your tcp. are you running final map set obly or other maps too? 2010-11-28T12:15:28 yes 2010-11-28T12:15:54 McLeopold: beat you :P 2010-11-28T12:15:56 ok great 2010-11-28T12:16:12 Where is specified that this channel is only english? 2010-11-28T12:16:25 wow, seven servers 2010-11-28T12:16:32 GreenTea: don't worry about it, keep conversing. 2010-11-28T12:16:36 mega1_: hey there 2010-11-28T12:16:40 ) 2010-11-28T12:16:44 Hi mega 2010-11-28T12:16:46 mega1_: looking good after a shaky start 2010-11-28T12:16:52 yeh 2010-11-28T12:16:58 must be the nerves 2010-11-28T12:17:05 Is burny around? I need to gloat... 2010-11-28T12:17:11 GreenTea: just an observation from a lurker who suddenly can't follow the conversation :-) 2010-11-28T12:17:39 turned out we can't converse normally - everybody has different encoding 2010-11-28T12:17:48 narnach: do you also study AI? 2010-11-28T12:17:50 :) 2010-11-28T12:17:54 nerves hehe 2010-11-28T12:17:55 McLeopold: how can you gloat about a bot made in under 2 weeks :P 2010-11-28T12:18:22 because my first one was made in under 2 weeks? 2010-11-28T12:18:41 Error323: as a hobbyist. went to university for a year, but decided academic life was not for me :-) 2010-11-28T12:19:00 ah 2010-11-28T12:19:12 1 gamin in ~ 15 minutes, cool.. 2010-11-28T12:19:14 at UvA? 2010-11-28T12:19:17 * game in 2010-11-28T12:19:26 GreenTea: more like 10 now 2010-11-28T12:19:45 Error323: aye, was back in 2004-2005 2010-11-28T12:19:56 oh that's when i started 2010-11-28T12:20:01 we should be in the same class then 2010-11-28T12:20:14 that's why your name looked familiar then :-) 2010-11-28T12:20:22 =] 2010-11-28T12:20:23 need more bots on mcleopold's tcp! join the after party ;-) 2010-11-28T12:20:26 antimatroid1 is on losing streak 2010-11-28T12:20:42 phew, i thought there was more 2010-11-28T12:20:45 i saw those :( 2010-11-28T12:20:56 i just beat mcleopold twice in a row on his server 2010-11-28T12:21:06 take that mcleopold :P 2010-11-28T12:21:20 ok ok i'll join aswell 2010-11-28T12:21:24 with iouri's code 2010-11-28T12:21:26 do you know what happens to someone who beats server's owner?! 2010-11-28T12:21:27 :P j/k 2010-11-28T12:21:34 Error323: heh 2010-11-28T12:21:48 asavis, bix0r and raschi are the negative surprises so far 2010-11-28T12:22:28 he got me this time i think 2010-11-28T12:22:47 mega1_: yeah i noticed those 2010-11-28T12:22:55 hm, I could put my horrific bot up on the tcp for lulz 2010-11-28T12:23:10 whoa, I just beat accoun? 2010-11-28T12:23:19 Ken69267: if that means I have a chance against someone on TCP, please join! :-) 2010-11-28T12:23:21 accoun just handed my rear to me 2010-11-28T12:23:23 especially bix0r. He's a reupload specialist but his bot was always around 2nd place. 2010-11-28T12:23:42 yeah bixor kinda screwed that up it seems... 2010-11-28T12:23:50 @rankings 2010-11-28T12:23:51 Error323: Top 10 players: bocsimacko(3170), protocolocon(3168), george(3159), medrimonia(3112), _Astek_(3097), GreenTea(3094), _iouri_(3082), mogron(3080), jimrogerz(3079), CorwinAlex(3063) 2010-11-28T12:23:54 there ya go 2010-11-28T12:23:58 a familiar sight ;) 2010-11-28T12:24:05 everything's back to normal :) 2010-11-28T12:24:06 I hope accoun stay away from me. My bot is most vulnerable to his. 2010-11-28T12:24:30 funny cuz I think out of all bots I beat him most 2010-11-28T12:24:38 out of all bots > m 2010-11-28T12:24:40 me* 2010-11-28T12:24:41 on tcp 2010-11-28T12:24:44 " asavis, bix0r and raschi are the negative surprises so far" dmj111 too 2010-11-28T12:24:46 Error323: i think you win on this map 2010-11-28T12:25:01 GreenTea: right, I missed that 2010-11-28T12:25:16 Account: I getting luck with the maps on my server. I'm pretty sure you have a better all around bot. 2010-11-28T12:25:20 phew, i broke my losing streak 2010-11-28T12:25:26 *** iFire` has joined #aichallenge 2010-11-28T12:25:41 leaderboard is still updated every 15 minutes? 2010-11-28T12:25:43 McLeopold: there are some really shit maps unfortunately 2010-11-28T12:25:47 but we're stuck with them 2010-11-28T12:25:47 http://ai-contest.com/profile_submissions.php?user_id=10464 - Raschi is submission world record holder (~4000 submissions) 2010-11-28T12:26:00 *** yasith has joined #aichallenge 2010-11-28T12:26:01 yeah serially 2010-11-28T12:26:30 thats insane 2010-11-28T12:26:40 must be a script 2010-11-28T12:26:41 wth is he thinking 2010-11-28T12:26:45 *** phreeza has joined #aichallenge 2010-11-28T12:26:49 perhaps that's the only reason he ever did well 2010-11-28T12:26:57 yeah 2010-11-28T12:27:02 just resubmit when losing 2010-11-28T12:27:06 hehe yeah 2010-11-28T12:27:23 quite ignorant :P 2010-11-28T12:27:27 :) 2010-11-28T12:27:28 i prefer my method 2010-11-28T12:27:31 preselected map set 2010-11-28T12:27:34 don't submit for the last month at all 2010-11-28T12:27:43 Accoun: the maps are too shit to have been pre selected 2010-11-28T12:27:46 *** bhasker has quit IRC (Quit: bhasker) 2010-11-28T12:27:55 Top 10 players: bocsimacko(3170), protocolocon(3168), george(3159), medrimonia(3112), _Astek_(3097), GreenTea(3094), _iouri_(3082), mogron(3080), jimrogerz(3079), CorwinAlex(3063) 2010-11-28T12:28:22 mega1_: could you release your sourcecode please? 2010-11-28T12:28:27 somebody made an error when pre-selecting ! 2010-11-28T12:28:41 Accoun: there we go, now I lose 2010-11-28T12:29:09 Error323: I'll clean it up before that. 2010-11-28T12:29:20 where is iouri's? 2010-11-28T12:29:25 mega1_: mkay 2010-11-28T12:29:26 in the code thread 2010-11-28T12:29:27 on github 2010-11-28T12:29:32 *** iFire has quit IRC (Ping timeout: 265 seconds) 2010-11-28T12:29:34 mega1_: So, are you going to reveal your secrets now? 2010-11-28T12:29:35 Error323: do you know lisp already ? :) 2010-11-28T12:29:38 http://zeroviz.us 2010-11-28T12:29:44 https://github.com/ikhramts/planet_wars/tree/simplified 2010-11-28T12:29:45 * 2010-11-28T12:29:46 tcp.zeroviz.us 2010-11-28T12:29:53 McLeopold: as I said I'll write a blog entry soonish 2010-11-28T12:29:56 krokokrusa: well i only wanna know his eval code :P 2010-11-28T12:30:17 and i'm familiar with its syntax 2010-11-28T12:30:22 so, basically, you did minimax for simultanious moves? 2010-11-28T12:30:24 never wrote in it though 2010-11-28T12:30:35 oh yay 2010-11-28T12:30:40 whaat 2010-11-28T12:30:40 he got first 2010-11-28T12:30:45 @rankings 2010-11-28T12:30:46 amstan: Top 10 players: bocsimacko(3220), protocolocon(3193), george(3160), _Astek_(3146), GreenTea(3132), medrimonia(3112), _iouri_(3103), mogron(3100), jimrogerz(3097), CorwinAlex(3084) 2010-11-28T12:30:48 mine crashed on tcp? O-o 2010-11-28T12:30:50 I tried to write in SBCL but can't resolve problem with launching bot locally.. 2010-11-28T12:30:52 antimatroid1: thanks 2010-11-28T12:31:00 invalid move Ken69267 2010-11-28T12:31:15 odd since its never crashed on the main server 2010-11-28T12:31:36 send 0 ships then 2010-11-28T12:31:43 i think tcp hates that while main server doesn;t? 2010-11-28T12:31:46 dang, I got a timeout in the finals. :( 2010-11-28T12:31:54 McLeopold: so did i 2010-11-28T12:31:55 McLeopold: ouch 2010-11-28T12:31:56 :) 2010-11-28T12:32:05 antimatroid1: ouch 2010-11-28T12:32:06 :P 2010-11-28T12:32:08 I beat Accoun ;) 2010-11-28T12:32:08 http://ai-contest.com/visualizer.php?game_id=8898177 2010-11-28T12:32:18 accoun is beating me, muggle! 2010-11-28T12:32:40 I beat iouri 22 out of 74 :P 2010-11-28T12:32:44 on official mapset 2010-11-28T12:33:04 i don't think my bots very good 2010-11-28T12:33:04 GreenTea: official or zeroviz? 2010-11-28T12:33:26 hm 2010-11-28T12:33:56 what you mean? official.. 2010-11-28T12:34:02 antimatroid1: ur bot is mostly extremely slow :P 2010-11-28T12:34:14 Error323: yep 2010-11-28T12:34:32 GreenTea: did you beat account in a game on my zeroviz server, or in the finals? You should post a link. 2010-11-28T12:34:39 Error323: is that so? wow 2010-11-28T12:34:41 oh you cant send ships = 0 on tcp? 2010-11-28T12:34:46 that would be why :p 2010-11-28T12:34:56 It's dhartmei's code... ask him. 2010-11-28T12:35:01 at least i win 1 batle) http://ai-contest.com/visualizer.php?game_id=8913743 2010-11-28T12:35:18 aww man, i almost had accoun on tcp 2010-11-28T12:35:31 mega1_: yes, just ran the test locally 2010-11-28T12:35:45 it preselected map set 2010-11-28T12:35:54 I can't find the code thread. 2010-11-28T12:35:57 http://zeroviz.us:8080/Accoun214_2_9.html 2010-11-28T12:36:22 mega1_: http://ai-contest.com/forum/viewtopic.php?f=3&t=1168 2010-11-28T12:36:23 Account: that's funny 2010-11-28T12:36:27 ah, it's in announcements 2010-11-28T12:36:30 im alweys loose McLeopold on it 2010-11-28T12:36:31 You only lose to me. :) 2010-11-28T12:36:34 Accoun: relax, man.. there is no conspiracy 2010-11-28T12:36:56 indeed, maps are just generated with the mapgen.py, its all fair. 2010-11-28T12:37:19 Account: I got you that time! 2010-11-28T12:37:22 and america landed on the moon 2010-11-28T12:37:23 =) 2010-11-28T12:37:25 damn, i did it too 2010-11-28T12:37:38 It's the maps that get drawn. My bot does very good if the best strategy is to wait and snipe. 2010-11-28T12:38:07 I do badly when you are far away and it's good to build fast, then I always get behind. 2010-11-28T12:38:08 McLeopold: sounds like you are me, don't make many actual moves each turn? 2010-11-28T12:38:15 most of my fleets are all show 2010-11-28T12:38:38 No, I make a lot of moves. Some of them are just to reserve fleets on a planet. 2010-11-28T12:38:42 this contest- playing agnust organizers 2010-11-28T12:39:04 aw, I've got a bunch of losses now.. still holding a high rank yet though 2010-11-28T12:39:16 antimatroid1: you just did 100 ships to the center planet... 2010-11-28T12:39:22 *** boegel has joined #aichallenge 2010-11-28T12:39:23 McLeopold: so what does your bot do? i might have missed if you already mentioned 2010-11-28T12:39:51 McLeopold: yep :) 2010-11-28T12:39:54 and i'd do it a gain 2010-11-28T12:40:01 are pairs of bots picked in a special way during the final round< 2010-11-28T12:40:03 ? 2010-11-28T12:40:18 cause my win/loss streaks are kind of striking... http://ai-contest.com/profile.php?user_id=7846 2010-11-28T12:40:24 Basically, I predict the future and look for holes in the enemy defense. 2010-11-28T12:40:27 although i ended up losing 2010-11-28T12:40:33 :( I'm gona have tons of losses cause of this crap. http://ai-contest.com/visualizer.php?game_id=8895479 2010-11-28T12:40:34 boegel, looks like the same match making algorithm is used.. i.e. normally close rankings are paired 2010-11-28T12:40:48 I have giant lists of numbers on max ships at every given planet at every turn. 2010-11-28T12:40:51 *** Gudradain has joined #aichallenge 2010-11-28T12:41:04 woohoo contest over 2010-11-28T12:41:11 not quite 2010-11-28T12:41:18 I then do some crazy ordering of every planet to pick the best target. 2010-11-28T12:41:36 I'm using ROI, so distance + num_ships/growth_rate 2010-11-28T12:42:00 Then I add some extra distance to return to the center so I value the center more. 2010-11-28T12:42:26 I sort enemyplanets on weakness, and attack the planet that has is the hardest to defend 2010-11-28T12:42:28 McLeopold: i'd say our bots are about equal 2010-11-28T12:42:33 but its not always optimal 2010-11-28T12:42:36 My bot doesn't try to expand too fast, it only tries to be a little growth over the opponent. 2010-11-28T12:42:56 Top 10 players: bocsimacko(3241), protocolocon(3222), _Astek_(3167), GreenTea(3148), george(3142), jimrogerz(3138), medrimonia(3132), CorwinAlex(3111), Accoun(3100), _iouri_(3093) 2010-11-28T12:43:05 Yeah, target selection is a big weakness in my bot. 2010-11-28T12:43:18 Almost in the top 150 now! 2010-11-28T12:43:25 my weakness is sending ships from planets i shouldn't and not making enough proper moves each turn 2010-11-28T12:43:30 McLeopold: i see. interesting. some aspects are similar to mine 2010-11-28T12:43:41 davaidjliu: how were you even not in the top 150? 2010-11-28T12:43:41 davidjliu: I had a crash too :) 2010-11-28T12:43:47 oh 2010-11-28T12:43:51 I had a crash on one 2010-11-28T12:43:58 davidjliu: did you have a slow start? 2010-11-28T12:44:02 and my bot tends to always lose on some maps 2010-11-28T12:44:07 meh 2010-11-28T12:44:10 tcp hates me 2010-11-28T12:44:11 don't they all :P 2010-11-28T12:44:17 I lost to one guy around 400, timed out against one around 300 2010-11-28T12:44:28 ouch 2010-11-28T12:44:28 but I still have a 88% winnig percentage 2010-11-28T12:44:31 heh 2010-11-28T12:44:33 I wonder if its the map 2010-11-28T12:44:46 i've had a few losses on the main server lately 2010-11-28T12:44:46 Ken69267:are you not connecting right, or just losing? 2010-11-28T12:45:20 McLeopold: is it only official mapset atm? 2010-11-28T12:45:24 on zeroviz? 2010-11-28T12:45:25 yes 2010-11-28T12:45:29 good 2010-11-28T12:45:31 McLeopold: it just returns to a terminal, and the site says (I) 2010-11-28T12:46:10 McLeopold: is your bot as slow as mine? 2010-11-28T12:46:14 oh, invalid orders, I see 2010-11-28T12:46:17 someone else must be using their time up 2010-11-28T12:46:30 so is the top 10 now, looking just about right? 2010-11-28T12:46:31 I think I'm running with the 1 second limit 2010-11-28T12:46:41 is there anyway to tell why it craps out on tcp 2010-11-28T12:46:44 not really 2010-11-28T12:46:49 yeah, hold on 2010-11-28T12:47:32 @tcp 2010-11-28T12:47:33 Accoun: tcp = (#1) http://www.benzedrine.cx/planetwars/, or (#2) http://72.44.46.68/, or (#3) 98.247.248.39:995 (zeroviz.us), or (#4) http://tcp.zeroviz.us/, or (#5) http://tcp2.zeroviz.us/ (crazy!) 2010-11-28T12:47:55 burny: no i don't think top10 is settled at this point 2010-11-28T12:47:55 protocolocon #2? woah... 2010-11-28T12:47:59 http://tcp.zeroviz.us/ <- final map set 2010-11-28T12:48:00 Ken69267: It only records an invalid flag, I don't see what the order is. 2010-11-28T12:48:07 *** aizu has quit IRC (Read error: Connection reset by peer) 2010-11-28T12:48:36 *** aizu has joined #aichallenge 2010-11-28T12:49:09 McLeopold: hm, weird 2010-11-28T12:49:11 bah, i think my bots going downhill from here on the main site 2010-11-28T12:49:18 it's not defensive enough at all 2010-11-28T12:49:19 I think I could allow 0 ships... 2010-11-28T12:49:34 well I changed it to where it shouldn't send 0 ships 2010-11-28T12:50:12 CorwinAlex looks pretty good 2010-11-28T12:51:09 yes 2010-11-28T12:51:18 the ukrainians are pwning =) 2010-11-28T12:52:27 Error323: woah our bots got stuck on the 2nd page somehow :-( 2010-11-28T12:52:42 yeah 2010-11-28T12:52:50 temporarily i hope... 2010-11-28T12:53:30 McLeopold: what's up with timeouts on your tcp? i never had timeouts until today. it's not my connection for sure 2010-11-28T12:53:45 ? 2010-11-28T12:53:53 *** BaronTrozo has joined #aichallenge 2010-11-28T12:54:10 only 7 connections, it shouldn't be an issue 2010-11-28T12:54:35 just weird i had three time-outs in a row. all against accoun hmm.. never timed out before. is timeout 5 seconds on your end? 2010-11-28T12:54:46 yeah, 5 2010-11-28T12:55:15 cool 2010-11-28T12:55:36 have anybody a bot which holds some state between moves? 2010-11-28T12:55:48 nope 2010-11-28T12:55:54 * krokokrusa <-- 2010-11-28T12:55:57 GreenTea: how do you mean? 2010-11-28T12:56:01 Error323: yeah. i noticed it takes very long time to get to decent ranking once you have a slow start like we did :( 2010-11-28T12:56:10 I mean 2010-11-28T12:56:15 GreenTea: mine does 2010-11-28T12:56:23 rebelxt_: indeed. 2010-11-28T12:56:30 rebelxt_: have you published ur code yet? 2010-11-28T12:56:36 if so i wanna run a local test :) 2010-11-28T12:56:37 for exmample save on turn 10 some info witch will be used on turn 11? 2010-11-28T12:56:46 well if its c++ or java/python or sth 2010-11-28T12:56:57 GreenTea: scheduled defensive and offensive moves which are parts of compound moves 2010-11-28T12:57:11 interesting 2010-11-28T12:57:17 I have no scheduled moves.. 2010-11-28T12:57:21 *** testing has joined #aichallenge 2010-11-28T12:57:32 testing: testing? well.. it works 2010-11-28T12:57:36 not hard to debug? 2010-11-28T12:57:46 GreenTea: i had that, but it sucked balls 2010-11-28T12:57:47 *** testing is now known as Guest33036 2010-11-28T12:57:52 another win against accoun on tcp :) 2010-11-28T12:57:56 Top 10 players: bocsimacko(3261), protocolocon(3258), GreenTea(3188), _Astek_(3185), jimrogerz(3159), george(3159), Accoun(3146), medrimonia(3117), CorwinAlex(3114), _iouri_(3108) 2010-11-28T12:58:03 *** Guest33036 has quit IRC (Client Quit) 2010-11-28T12:58:19 will bcsimacko hold 1st place? 2010-11-28T12:58:25 i predict yes 2010-11-28T12:58:31 protocolocon is pretty close 2010-11-28T12:58:40 there's days left yet :P 2010-11-28T12:58:51 how lang will games be played? 2010-11-28T12:58:55 *long 2010-11-28T12:59:02 Ken69267: I see a 0 fleet order. 2010-11-28T12:59:04 *** rebelxt_ has quit IRC (Quit: Page closed) 2010-11-28T12:59:16 I'm at 453 now, I hope I can stay in the top 500 2010-11-28T12:59:22 *** rebelxt_ has joined #aichallenge 2010-11-28T12:59:28 that would be cool for a first-time contestant, no? :) 2010-11-28T12:59:29 10 12 0 2010-11-28T12:59:34 rebelxt_: could you publish ur code? 2010-11-28T13:00:17 Error323: sure i will tomorrow 2010-11-28T13:00:21 *** narnach has quit IRC (Ping timeout: 276 seconds) 2010-11-28T13:00:27 wow, i'm 3-3 against accoun on tcp 2010-11-28T13:00:27 sigh.. this loss demonstrates a second piece of fail in my bot: http://ai-contest.com/visualizer.php?game_id=8910339 2010-11-28T13:00:47 boegel: protocolon was pretty low before closing: http://ai-contest.com/final-open-ratings.txt 2010-11-28T13:00:57 rebelxt_: https://github.com/Error323/googleai 2010-11-28T13:00:59 amstan: he's been up there before 2010-11-28T13:01:02 @learn closing as http://ai-contest.com/final-open-ratings.txt 2010-11-28T13:01:02 amstan: Ready to serve, my lord. 2010-11-28T13:01:02 and was strong on tcp 2010-11-28T13:01:15 Error323: i was 93/1/7 against mega1, was is about same for you? 2010-11-28T13:01:30 rebelxt_: iirc i won 8 2010-11-28T13:01:31 Error323: awesome. tnx for the link 2010-11-28T13:02:01 this is my alphabeta attempt: https://github.com/Error323/googleai/tree/alphabeta 2010-11-28T13:02:12 if only i had a good eval function >_< 2010-11-28T13:02:20 *** davidjliu has quit IRC (Ping timeout: 265 seconds) 2010-11-28T13:02:50 Error323: nice. was this what you used to calculate optimum expansion? 2010-11-28T13:03:50 optimum everytihng given the actionset 2010-11-28T13:04:00 but i couldn't get it to work and started on it too late :/ 2010-11-28T13:04:17 alphabeta does work correctly though 2010-11-28T13:04:21 and is quite efficient 2010-11-28T13:04:34 it doesn't work for extending a tree 2010-11-28T13:05:20 man.. watching the games my bot is losing.. 2010-11-28T13:05:21 any word on the next contest btw? 2010-11-28T13:05:24 *** narnach has joined #aichallenge 2010-11-28T13:05:28 timeframe for example? 2010-11-28T13:05:33 if I had an extra few hours, I coulda possibly been top10 2010-11-28T13:05:33 boegel: next term 2010-11-28T13:05:44 Error323: very nice.. what I tried was simulated annealing to figure out best order of planet captures 2010-11-28T13:05:45 boegel: i.e. Jan-Feb? 2010-11-28T13:06:32 *** Ice_Harley has joined #aichallenge 2010-11-28T13:06:39 Error323: but it slowed down my bot to 0.75s+ per turn to get good results so local testing was becoming a nightmare 2010-11-28T13:06:49 yeah that too 2010-11-28T13:06:58 i sometimes had like 3500 actions 2010-11-28T13:07:02 to evaluate 2010-11-28T13:07:13 on a single branch 2010-11-28T13:07:29 and went 4 ply 2010-11-28T13:07:32 xD 2010-11-28T13:09:00 My first bot also have only set of moves and evaluation function to chose the best. But it is really hart to write good evaluation function.. 2010-11-28T13:09:04 so, who's narnach? 2010-11-28T13:09:08 if i'm not in top 100 tomorrow i'll cry 2010-11-28T13:09:09 :P 2010-11-28T13:09:11 c++ helps.. when i ported my early bot from python to c++, i got 35x speed improvement.. 2010-11-28T13:09:17 McLeopold: i think i might take top spot on your tcp in a minute :P 2010-11-28T13:09:23 McLeopold: me! :-) 2010-11-28T13:09:25 yeah c++ is sweet 2010-11-28T13:09:27 nearly every one of my losses against top100 bots, is because I don't re-enforce the front lines properly 2010-11-28T13:09:48 c++ <3 2010-11-28T13:09:59 I lost to rebelxt_, as soon as he rises up, so will I 2010-11-28T13:10:21 ah crap, i might lose this game to accoun 2010-11-28T13:10:38 Error323: you will get there today! i just did, and your bot is beating me on tcp at the moment 2010-11-28T13:10:39 no, i think i have it again 2010-11-28T13:10:55 when i ported my early bot from perl to C, i got 101x speed improvement.. 2010-11-28T13:11:17 rebelxt_: heh well it doesn't say eveerything but ye i do think i'll end up in #100 2010-11-28T13:12:06 http://zeroviz.us:8080/antimatroid.html :D 2010-11-28T13:12:25 I really gotta take down the 8080... 2010-11-28T13:12:31 *** Azrathud has quit IRC (Quit: Leaving) 2010-11-28T13:12:57 Top 10 players: bocsimacko(3329), protocolocon(3285), GreenTea(3250), george(3242), _Astek_(3235), Accoun(3225), jimrogerz(3213), _iouri_(3178), mogron(3157), CorwinAlex(3148) 2010-11-28T13:13:09 damn, i got him again i think 2010-11-28T13:13:56 no, he tore me a new one that time 2010-11-28T13:14:47 *** Azrathud has joined #aichallenge 2010-11-28T13:15:38 Error323: you just gave me a game on a draw map :) 2010-11-28T13:15:53 ohw 2010-11-28T13:15:59 *** Accoun has quit IRC () 2010-11-28T13:16:00 only on tcp though 2010-11-28T13:17:18 McLeopold: i finally updated my bookmark and script. no more 8080 ;-) 2010-11-28T13:18:28 wow, this bot is better than i thought 2010-11-28T13:18:46 *** margoth has quit IRC (Ping timeout: 265 seconds) 2010-11-28T13:18:54 *** Gudradain has quit IRC (Quit: Page closed) 2010-11-28T13:19:19 well i have no idea hwy my bot is imploding on tcp 2010-11-28T13:19:28 ok i'm getting pwned by timeouts on tcp.. gonna start bot from amazon EC2 2010-11-28T13:19:35 I'm finding the same thing for me anti... it makes me cry, how simple the things I'm losing to are 2010-11-28T13:19:39 if my bot properly saved itself from stupid moves it'd probably be a hell of a lot better 2010-11-28T13:19:41 accoun's 6th 2010-11-28T13:19:53 burny: http://zeroviz.us:8080/antimatroid.html 2010-11-28T13:19:57 lol anti, same.. 2010-11-28T13:20:05 amstan: look at my record against him there^^ 2010-11-28T13:20:07 like, turn 1, decide that attacking some planet is a good idea.. 2010-11-28T13:20:19 and then turn 4, decide it can't defend it(even though it can) 2010-11-28T13:20:58 a number of games against top20 bots that I lost like that, that would have otherwise won 2010-11-28T13:21:40 that, and deciding "if I attack that planet, I can defend it".. and attacking it.. and then losing everything expcet that planet in the next few turns 2010-11-28T13:21:49 Ken69267: It's 0 fleet orders 2010-11-28T13:21:57 burny: i have the exact same problems 2010-11-28T13:22:10 wait... already moving: go 2010-11-28T13:22:13 often in the middle of a battle i'll just send everything i have pretty much across the whole map 2010-11-28T13:23:14 Ken69267: you're sending 2 "go" commands in a row 2010-11-28T13:23:30 go go gadget! 2010-11-28T13:23:34 wat 2010-11-28T13:23:46 when you finish making your turn, you need to output go 2010-11-28T13:23:51 you're doing it twice 2010-11-28T13:23:59 nice try there mcleopold 2010-11-28T13:24:03 but that shit ain't gonna gly 2010-11-28T13:24:05 fly* 2010-11-28T13:24:21 * Ken69267 tries something 2010-11-28T13:24:28 http://zeroviz.us:8080/canvas.cgi?game_id=32611 2010-11-28T13:24:35 mcleopolds sneaky attempt at the end 2010-11-28T13:25:18 I need more time 2010-11-28T13:25:20 :) 2010-11-28T13:25:50 you've got me this time 2010-11-28T13:26:06 http://zeroviz.us:8080/ratings.html 2010-11-28T13:26:08 yeah, first :) 2010-11-28T13:26:57 Ken69267: successful game... 2010-11-28T13:27:05 well, you finished :) 2010-11-28T13:27:12 hehe 2010-11-28T13:27:55 despite that i'm doing terribly on the main site 2010-11-28T13:27:58 Top 10 players: bocsimacko(3378), protocolocon(3310), george(3284), GreenTea(3276), _Astek_(3254), jimrogerz(3253), Accoun(3249), _iouri_(3223), mogron(3187), CorwinAlex(3171) 2010-11-28T13:28:01 sitting around 50th and losing a bit 2010-11-28T13:28:34 I've been holding fairly steady around 30th 2010-11-28T13:28:43 Ken69267: that was a timeout 2010-11-28T13:28:53 in your favor 2010-11-28T13:29:20 collect $200 2010-11-28T13:29:36 http://ai-contest.com/visualizer.php?game_id=8921943 2010-11-28T13:29:48 why to send 100 ships to far isolated planet? 2010-11-28T13:30:19 rebelxt_: you're bouncing your fleets all over my back line :) 2010-11-28T13:30:30 oh aha, now that i wouldn't do again, it's because the planets are at an equal distance and it "thinks" (stupidly) that they need to both all out battle for that planet 2010-11-28T13:30:44 http://tcp.zeroviz.us/canvas.cgi?game_id=32623 2010-11-28T13:31:34 i think returning if ships=0 fixed it 2010-11-28T13:32:25 *** Accoun has joined #aichallenge 2010-11-28T13:32:32 *** boegel has quit IRC (Quit: Leaving) 2010-11-28T13:32:48 McLeopold: haha. let me watch 2010-11-28T13:33:01 *** medrimonia2 has joined #aichallenge 2010-11-28T13:33:15 *** medrimonia2 is now known as medrimonia 2010-11-28T13:33:29 *** amstan_ has joined #aichallenge 2010-11-28T13:33:29 *** ChanServ sets mode: +o amstan_ 2010-11-28T13:33:30 Hi everyone 2010-11-28T13:33:37 hey medrimonia 2010-11-28T13:33:38 *** amstan_ has quit IRC (Remote host closed the connection) 2010-11-28T13:33:38 hi 2010-11-28T13:33:45 Hi 2010-11-28T13:34:31 How are you? 2010-11-28T13:34:47 McLeopold: nice comeback by your bot: http://tcp.zeroviz.us/canvas.cgi?game_id=32623. i just don't re-inforce my attacks properly 2010-11-28T13:35:27 Ken69267: you seem to be stable now? 2010-11-28T13:35:34 Error323: running a local tourney against your final bot. will let you know soon 2010-11-28T13:35:40 McLeopold: yeah 2010-11-28T13:35:41 rebelxt_: cool 2010-11-28T13:35:49 rebelxt_: on official mapset i presume 2010-11-28T13:35:50 * Ken69267 prolly has the worst bot on tcp ;D 2010-11-28T13:36:01 Error323: yep final mapset 2010-11-28T13:36:15 *** Migi32 has quit IRC (Quit: www.ai-contest.com) 2010-11-28T13:36:31 on the official site im doing much worse than my unofficial rank was :/ 2010-11-28T13:36:33 hmmm i lose to smloh 39/35 2010-11-28T13:36:43 he has a good bot 2010-11-28T13:36:46 that isn't a bad thing 2010-11-28T13:37:15 antimatroid1: finally i'm winning a game against you on zeroviz :P 2010-11-28T13:38:00 BaronTrozo joined the after party too ;-) 2010-11-28T13:38:15 bixOr4ever is only #27? 2010-11-28T13:38:43 yeah, and i'm triying to convice the rest of the rufes band 2010-11-28T13:38:57 http://tcp.zeroviz.us/canvas.cgi?game_id=32649 2010-11-28T13:39:05 BaronTrozo: could you publish ur code please? 2010-11-28T13:39:50 Id want to see more java code :( 2010-11-28T13:40:07 smloh has javacode 2010-11-28T13:40:12 ; 2010-11-28T13:40:12 ; 2010-11-28T13:40:23 Error323: :P 2010-11-28T13:40:37 antimatroid1: shush 2010-11-28T13:40:40 :P 2010-11-28T13:40:53 jeez just kill me already :P 2010-11-28T13:40:53 McLeopold: 2-2 ;-) tie breaker coming up? 2010-11-28T13:41:14 i can't believe my bot is doing to well on tcp 2010-11-28T13:41:40 well pairing isn't exactly good 2010-11-28T13:41:46 but ye ur bot is sweet 2010-11-28T13:41:51 i'll run it locally against mine 2010-11-28T13:42:01 you'll be there for a year :) 2010-11-28T13:42:15 antimatroid1: man, you have some mad last minute coding skills! i honestly expected your bot to be buggy since you reverted back to tree idea so late ;) 2010-11-28T13:42:16 hehe 2010-11-28T13:42:25 agree 2010-11-28T13:42:27 nice work 2010-11-28T13:42:28 :) 2010-11-28T13:42:48 rebelxt_: i have been on holidays for 3 weeks and hardly slept :) 2010-11-28T13:42:51 I've already beaten #2 #3 and #4 =P But I'm still loosing too much on the other side 2010-11-28T13:42:57 antimatroid1: is your code open source now? i wanna run a local tourney if it is 2010-11-28T13:42:58 and had quite a bit written before that 2010-11-28T13:42:58 Top 10 players: bocsimacko(3416), protocolocon(3313), george(3312), GreenTea(3283), jimrogerz(3280), _iouri_(3271), _Astek_(3259), mogron(3237), Accoun(3237), CorwinAlex(3221) 2010-11-28T13:43:02 it is 2010-11-28T13:43:03 Mistmanov: http://ai-contest.com/visualizer.php?game_id=8923232 2010-11-28T13:43:11 of all games 2010-11-28T13:43:13 https://github.com/amstan/antimatroid 2010-11-28T13:43:15 a draw game -_- 2010-11-28T13:43:28 the tree code doesn't make sense for what it's actually doing, just roll with it :) 2010-11-28T13:43:41 rebelxt_: how is our pairing doing locally? 2010-11-28T13:43:47 Error323: :) 2010-11-28T13:43:59 what is "rufes band"? 2010-11-28T13:44:00 Error323: 57 out of 75.. 5 more minutes 2010-11-28T13:44:07 they're a group of friends 2010-11-28T13:44:18 rebelxt_: and am i ahead? :P 2010-11-28T13:44:18 working at the same company they don't like, hence don't want to give credit to 2010-11-28T13:44:45 Error323: your bot is fast. mine is ... python ;-) 2010-11-28T13:44:57 heh yeah 2010-11-28T13:45:04 Error23: Publish my code?? I have no problem in publishing it, but it's so dirty, and so triky 2010-11-28T13:45:07 Error323: you are coming from behind now... very close 2010-11-28T13:45:11 GreenTea: note, i've been asking that for a month, only just found out today when boran came on here, like you, they decided to stear clear the entire contest :\ 2010-11-28T13:45:30 baron* 2010-11-28T13:45:34 BaronTrozo: we don't mind dirty code. i think that's the case for everyone ;) 2010-11-28T13:45:38 BaronTrozo: doesn't matter, mine is here: https://github.com/Error323/googleai also check the alphabeta branch 2010-11-28T13:45:41 indeed. 2010-11-28T13:45:47 almost oall of the code is done at 1a.m. - 2a.m. 2010-11-28T13:45:55 my code is not safe for anyones eyes 2010-11-28T13:46:34 with variables name some in sapnish, some in english and some variables with part in spanish and part on english 2010-11-28T13:46:46 BaronTrozo: ouch :P 2010-11-28T13:46:48 whyyyyy. 2010-11-28T13:47:07 god damnit, i want the greek alphabet and common maths relations etc. on my keyboard and usable in programs 2010-11-28T13:47:10 all of my variables are a,b,c,...z for ease of reading 2010-11-28T13:47:11 becasue it was done at 1 a.m. 2010-11-28T13:47:13 j/k 2010-11-28T13:47:32 BaronTrozo: that's no excuse to go spanish on us :P 2010-11-28T13:48:03 okey, but first i will clean the code before publishing it 2010-11-28T13:48:24 BaronTrozo: sure :) 2010-11-28T13:48:44 http://zeroviz.us:8080/canvas.cgi?game_id=32678 2010-11-28T13:48:51 the 'ol switch eroo 2010-11-28T13:49:28 Error323: 44-2-29 pretty close! games were fun to watch since only four games were 200-turn standoffs 2010-11-28T13:49:31 BaronTrozo: Me too, I've a mix of french and english in my variables 2010-11-28T13:50:02 rebelxt_: 44-2-29 you mean you win 44 times of me? 2010-11-28T13:50:16 cuz that's a lot :/ 2010-11-28T13:50:19 and also the english part, are writting in my own english, that of xourse is diferent than cambrige one 2010-11-28T13:51:22 Error323: yep mine was ahead.. but i don't think that's a big difference. let me run on a bigger 500-map set 2010-11-28T13:52:09 mm 2010-11-28T13:52:11 i doubt 75 maps is a good test.. i was normally doing 250 maps in local tests to get a decent picture which bot is stronger 2010-11-28T13:52:13 *** medrimonia1 has joined #aichallenge 2010-11-28T13:52:21 yeah 200 here 2010-11-28T13:52:29 *** medrimonia1 has left #aichallenge 2010-11-28T13:52:41 lemme add antima's bot to the tourney 2010-11-28T13:52:47 rebelxt_: don't 2010-11-28T13:52:53 rebelxt it's slow 2010-11-28T13:52:53 *** wh1teside has joined #aichallenge 2010-11-28T13:52:54 his bot is reaaaaaaaaaaally slow 2010-11-28T13:52:58 it uses it's full second 2010-11-28T13:53:00 1 sec a turn 2010-11-28T13:53:15 and it's not that simple to just adjust either 2010-11-28T13:53:18 I have 5 map sets for local testing. 100, 100, 200, and 2x200 generated by new generator 2010-11-28T13:53:33 ok, i will make my seconds shorter to speed it up ;) 2010-11-28T13:53:58 rebelxt_: it'll just crash on you :) 2010-11-28T13:54:25 will it cause blue screen of death on ubuntu? and permanent damage to hard drive? 2010-11-28T13:54:37 no, but it wont give accurate results 2010-11-28T13:54:57 *** wh1teside_ has quit IRC (Ping timeout: 276 seconds) 2010-11-28T13:55:21 McLeopold: tie breaker finally. you got this! 2010-11-28T13:55:58 well bix0r isn't going to get second i don't think 2010-11-28T13:57:55 http://ai-contest.com/visualizer.php?game_id=8926489 2010-11-28T13:57:59 Top 10 players: bocsimacko(3431), protocolocon(3330), GreenTea(3309), george(3303), jimrogerz(3301), _iouri_(3298), _Astek_(3276), Accoun(3258), CorwinAlex(3255), goldman(3233) 2010-11-28T13:58:15 one open question: who was ObligatoryShampoo in the tcp server? 2010-11-28T13:58:44 bah, i found that out the other day 2010-11-28T13:58:48 i hated that bot 2010-11-28T13:58:54 me too 2010-11-28T13:58:54 it was the bane of my existence 2010-11-28T13:59:00 everyone did 2010-11-28T13:59:05 bhasker: was it you? 2010-11-28T13:59:05 meneky: flagcapper 2010-11-28T13:59:10 ah, that's right 2010-11-28T13:59:15 or flag 2010-11-28T13:59:20 he's flagcapper @ official 2010-11-28T13:59:41 i think i will put that down as my most annoying opponent 2010-11-28T14:00:13 ok, thanks :), he crushed me so many times ...... 2010-11-28T14:00:25 .. and me :( 2010-11-28T14:01:57 not me. hahaha! 2010-11-28T14:02:03 lol 2010-11-28T14:03:16 flagcapper capped my flag so often ;) 2010-11-28T14:03:18 pathetic no dutch player in #100 yet :( 2010-11-28T14:03:18 @tcp 2010-11-28T14:03:19 Mistmanov: tcp = (#1) http://www.benzedrine.cx/planetwars/, or (#2) http://72.44.46.68/, or (#3) 98.247.248.39:995 (zeroviz.us), or (#4) http://tcp.zeroviz.us/, or (#5) http://tcp2.zeroviz.us/ (crazy!) 2010-11-28T14:03:45 Gosh lost another one... good beginning for me but now it's really loosing a lot 2010-11-28T14:04:09 medrimonia: you're right near the top, i wouldn't worry too much 2010-11-28T14:04:12 you're playing the best 2010-11-28T14:04:29 i keep losing to people and i'm in 44th place :( 2010-11-28T14:04:43 I think the canadians are chiting to keep the dutch people out 2010-11-28T14:04:52 generated maps that we suck on 2010-11-28T14:04:57 :) 2010-11-28T14:05:00 hahah 2010-11-28T14:05:05 yeah that must be it xD 2010-11-28T14:05:19 antimatroid1: locally we are 6 - 6 here 2010-11-28T14:05:19 janzert spent hours with the submitted bots finding ones you guys suck on 2010-11-28T14:05:29 started the games an hour ago 2010-11-28T14:05:30 uhuh 2010-11-28T14:05:32 soooooo slow 2010-11-28T14:05:39 Mistmanov: this applies to every nation except for canadian 2010-11-28T14:05:42 you're welcome :) 2010-11-28T14:05:49 good possibility 2010-11-28T14:06:02 they had all our source code, and all the time in the world to hand craft maps =) 2010-11-28T14:06:09 Mistmanov: lol another draw @ tc 2010-11-28T14:06:11 tcp 2010-11-28T14:06:23 " janzert spent hours with the submitted bots finding ones you guys suck on" ))) 2010-11-28T14:06:35 GreenTea: it was a joke :) 2010-11-28T14:06:43 antimatroid1: wow, your bot is making a great heater out of my CPU! ;-) 2010-11-28T14:06:51 come on server, i want one more game before going to sleep 2010-11-28T14:06:59 it's 6am and i had 2 hours of sleep last night 2010-11-28T14:07:10 well, the night before i guess 2010-11-28T14:07:11 I understand sarcasm :) 2010-11-28T14:07:19 lol 6am 2010-11-28T14:07:22 8 pm here 2010-11-28T14:07:43 rebelxt_: don't say you weren't warned :P 2010-11-28T14:08:04 Australia is anti everywhere else 2010-11-28T14:08:06 i quite like it 2010-11-28T14:08:10 hehe.. 75 maps are gonna take a couple of hours 2010-11-28T14:08:12 especially as it's now summer 2010-11-28T14:08:48 7 to 8 now for you antimatroid1 2010-11-28T14:08:56 in wins 2010-11-28T14:09:28 if i could only stop stupid moves from happening i would have been a lot stronger 2010-11-28T14:09:36 same here :P 2010-11-28T14:09:40 same everywhere :P 2010-11-28T14:09:43 my bot falls to pieces once there are a lot of takeable planets though 2010-11-28T14:10:31 i only take neutrals when losing 2010-11-28T14:10:40 and doing it poorly aswell sometimes :/ 2010-11-28T14:11:36 olman got me on the main site, bah, bed time 2010-11-28T14:12:17 whoot 8 servers up 2010-11-28T14:12:21 buurn baby buuuurn 2010-11-28T14:12:59 Top 10 players: bocsimacko(3431), GreenTea(3368), _iouri_(3358), protocolocon(3347), george(3337), jimrogerz(3319), _Astek_(3319), CorwinAlex(3296), Accoun(3294), shangas(3278) 2010-11-28T14:13:33 159 games a minute 2010-11-28T14:13:40 then every once in a while i play :P 2010-11-28T14:14:02 Error323: My code is online now 2010-11-28T14:14:16 medrimonia: where 2010-11-28T14:14:28 http://ai-contest.com/forum/viewtopic.php?f=19&t=1168&p=7858#p7858 2010-11-28T14:15:47 medrimonia: is it fast? 2010-11-28T14:17:13 *** krokokrusa has quit IRC (Quit: leaving) 2010-11-28T14:17:42 hmm quite fast 2010-11-28T14:17:53 I don't think more than 0.4sec usual 2010-11-28T14:18:07 how do you not find things to do in a second? 2010-11-28T14:18:22 I'm not doing any minmax or all 2010-11-28T14:19:20 It's a sequential bot, so if I make too much time on one part, the other won't happend 2010-11-28T14:20:38 *** McLeopold has quit IRC (Quit: Page closed) 2010-11-28T14:22:32 *** Xayide has joined #aichallenge 2010-11-28T14:22:40 *** aizu has quit IRC (Read error: Connection reset by peer) 2010-11-28T14:26:12 *** davidjliu has joined #aichallenge 2010-11-28T14:28:00 Top 10 players: bocsimacko(3415), GreenTea(3381), jimrogerz(3354), george(3346), _iouri_(3341), protocolocon(3340), Accoun(3322), _Astek_(3320), CorwinAlex(3319), shangas(3307) 2010-11-28T14:28:57 *** davidjliu_ has joined #aichallenge 2010-11-28T14:29:09 *** Migi32 has joined #aichallenge 2010-11-28T14:30:47 *** davidjliu has quit IRC (Ping timeout: 265 seconds) 2010-11-28T14:34:08 *** Migi32 has quit IRC (Quit: www.ai-contest.com) 2010-11-28T14:37:08 *** bhasker has joined #aichallenge 2010-11-28T14:37:44 so how is it going 2010-11-28T14:38:03 *** JeroenDeDauw has joined #aichallenge 2010-11-28T14:38:55 " if i could only stop stupid moves from happening i would have been a lot stronger" one of last improvements of my bot is to filter captures of neutral planet which alltow to enemy to capture beter neutral planet (for example in the middle of the map) 2010-11-28T14:39:22 * allow 2010-11-28T14:40:00 my bot seems to be stuck in the 70s-80s 2010-11-28T14:40:23 .. this is primitive look 1 step forward.. 2010-11-28T14:40:44 *** Xayide has quit IRC (Quit: Page closed) 2010-11-28T14:40:50 134 now 2010-11-28T14:40:52 all dutch people are stuck at 100+ -_- gameserver is refusing to give us games against higher ranked people -_- 2010-11-28T14:41:04 yeah :/ 2010-11-28T14:41:08 just wait 2010-11-28T14:41:12 we're still climbing the ranks 2010-11-28T14:41:14 I know, patience patience 2010-11-28T14:41:31 the dutch will RAAAAAAGE 2010-11-28T14:41:31 xD 2010-11-28T14:41:37 but it's just... I know that I'd get the same win/loss ratio when playing against ~50 as I get now 2010-11-28T14:41:39 *** politik has joined #aichallenge 2010-11-28T14:41:46 yea me too 2010-11-28T14:41:51 since I simply lose whenever I get a map that's just not suited to my strategy :p 2010-11-28T14:41:57 indd :P 2010-11-28T14:42:18 heh 2010-11-28T14:42:24 started up malazan on tcp 2010-11-28T14:42:27 almost into the top 100! 2010-11-28T14:42:28 yay 2010-11-28T14:42:33 something strange with bix0r4ever.. 2010-11-28T14:42:36 hi all. is there any talk of another contest? I found out about this one too late 2010-11-28T14:42:46 i just seem to be getting games against whiteside 2010-11-28T14:43:01 Top 10 players: bocsimacko(3458), GreenTea(3402), _iouri_(3384), jimrogerz(3358), _Astek_(3355), Accoun(3353), george(3348), protocolocon(3344), CorwinAlex(3307), shangas(3304) 2010-11-28T14:43:04 grats david i guess the reset is throwing people around a lot 2010-11-28T14:43:15 people from the top 50 are still outside the top 100 2010-11-28T14:43:34 amazing how you are 1 game better than me bhasker 2010-11-28T14:43:35 =) 2010-11-28T14:43:53 heh against you not necessarily against others though 2010-11-28T14:44:06 mja dunno 2010-11-28T14:44:07 you may be stronger against other opponents for all you know 2010-11-28T14:44:12 but it'll be close i think 2010-11-28T14:44:16 barontrozo just broke my winning streak on tcp 2010-11-28T14:44:21 we'll dance around eachother =) 2010-11-28T14:44:22 !@#! 2010-11-28T14:44:25 hehe hopefully 2010-11-28T14:44:30 i think rebelxt will be close too 2010-11-28T14:44:36 his bot is also of similar strength 2010-11-28T14:44:52 did anyone here end up doing minimax? 2010-11-28T14:45:01 quite a few people 2010-11-28T14:45:03 i think antimatroid 2010-11-28T14:45:06 E323 gave it a shot 2010-11-28T14:45:08 i got the code 2010-11-28T14:45:11 *** BaronTrozo has quit IRC (Quit: Page closed) 2010-11-28T14:45:11 dmj does a hybrid 2010-11-28T14:45:13 but phailed 2010-11-28T14:45:30 Error323: failed how? 2010-11-28T14:45:33 politik: https://github.com/Error323/googleai/tree/alphabeta 2010-11-28T14:45:36 just not enough time in 1s? 2010-11-28T14:45:41 poor evaluation function 2010-11-28T14:45:52 baron trozo is killing me:(* 2010-11-28T14:46:00 ah, what rank was it? 2010-11-28T14:46:16 I would think even very dumb eval would be alright, compared to most bots 2010-11-28T14:46:27 the problem is not really just the eval though 2010-11-28T14:46:30 http://ai-contest.com/profile.php?user_id=9309 OMG SUA... 2010-11-28T14:46:31 its also the move generation 2010-11-28T14:46:45 he was in top 50 2010-11-28T14:46:46 I did a half-ply with just ship count as eval, and it's not good but it's not awful 2010-11-28T14:46:49 yea i remember 2010-11-28T14:47:00 how does one do a half ply 2010-11-28T14:47:06 isnt a ply one move by one player 2010-11-28T14:47:11 Mistmanov: yeah it seems we'll go really really slow 2010-11-28T14:47:12 bah 2010-11-28T14:47:19 I think a ply is a full move by both players 2010-11-28T14:47:24 thats how it is in chess anyway 2010-11-28T14:47:38 I think... googling.. 2010-11-28T14:47:45 ply is one move 2010-11-28T14:47:48 I hope they don't do the cut-off to soon :p 2010-11-28T14:47:49 by one player 2010-11-28T14:47:50 ply is one move iirc 2010-11-28T14:47:52 ye 2010-11-28T14:47:58 ah, you're right 2010-11-28T14:48:03 like that tomorrow they're already going to stop playing everyone besides top100 -_- 2010-11-28T14:48:25 ? 2010-11-28T14:48:41 I hope they don't do that :p 2010-11-28T14:48:45 the probably won't 2010-11-28T14:49:03 wow, seems sudden to go to top 100 2010-11-28T14:49:07 maybe top 300 2010-11-28T14:49:35 i think top 100 will be at least a day off 2010-11-28T14:49:40 they should cut off the bottom ~4000 anyway, so many people just submitted ragebot or something 2010-11-28T14:49:53 i thought it was totally 4000 odd bots 2010-11-28T14:50:12 4619 2010-11-28T14:50:15 imo playing top 500 is good enough right? 2010-11-28T14:50:26 at 200 gpm 2010-11-28T14:50:57 I would think it would be sort of adaptive, where if your rating is moving, you get more games 2010-11-28T14:51:00 how is it that on zeroviz 2010-11-28T14:51:07 but it seems to be totally spread evenly 2010-11-28T14:51:11 there is e323-final at 249 elo 2010-11-28T14:51:16 and the next bot is mine at -55 elo 2010-11-28T14:51:22 shouldn't there be something in the middle 2010-11-28T14:51:28 or a more gradual fall off in elo 2010-11-28T14:51:33 it only shows active bots bhasker 2010-11-28T14:51:39 oooh nm 2010-11-28T14:51:40 doh 2010-11-28T14:51:41 but there's a lot in between 2010-11-28T14:51:50 http://tcp.zeroviz.us/ratings.html 2010-11-28T14:51:56 too late --- 2010-11-28T14:52:47 http://ai-contest.com/profile.php?user_id=14232 7 submissions, and the last one doesn't even play 2010-11-28T14:52:50 looks like mistmanov is beating you e323 2010-11-28T14:52:51 poor guy 2010-11-28T14:53:04 bhasker: YEAH :( :(( :( 2010-11-28T14:53:09 I think he'll win 2010-11-28T14:53:22 unless the random generator is on my side :P 2010-11-28T14:53:24 assuming the server is ever going to give me some good games :p 2010-11-28T14:53:32 I think that in theory 2010-11-28T14:53:45 i'm almost #1 of NL 2010-11-28T14:53:51 it's perfectly possible to just get a crap map every time that I get a good opponent (like every 20 games) 2010-11-28T14:53:51 with Mistmanov on my tail 2010-11-28T14:53:52 hehe 2010-11-28T14:54:27 who is Ken and narnach on tcp and whiteside? 2010-11-28T14:54:42 as of now those seem to be the only bots i can beat ^ 2010-11-28T14:54:49 accoun and barontrozo both thrashed my bot 2010-11-28T14:55:14 bhasker: narnach is me! 2010-11-28T14:55:22 oh k 2010-11-28T14:55:29 is that your name on official too 2010-11-28T14:57:05 it is 2010-11-28T14:57:12 hes dutc 2010-11-28T14:57:12 accoun stop killing my bot 2010-11-28T14:58:01 Top 10 players: bocsimacko(3486), GreenTea(3418), _iouri_(3403), Accoun(3387), protocolocon(3372), george(3362), jimrogerz(3357), _Astek_(3355), CorwinAlex(3322), shangas(3317) 2010-11-28T14:58:02 *** mceier has joined #aichallenge 2010-11-28T14:58:28 the elo ratings are still climbing 2010-11-28T14:58:33 bhasker: oh man ur 89 @ official :/ 2010-11-28T14:58:39 yea dropping 2010-11-28T14:58:40 :-\ 2010-11-28T14:58:44 lost a bunch of games in a row 2010-11-28T14:58:48 I see 2010-11-28T14:59:07 right now in a cafe and looks like my bot is timing out 2010-11-28T14:59:08 every now and then 2010-11-28T14:59:11 !! 2010-11-28T14:59:19 just lost to narnach:p 2010-11-28T14:59:22 :P 2010-11-28T14:59:27 timeout4 tl 2010-11-28T14:59:43 timed out again 2010-11-28T14:59:43 wtf 2010-11-28T15:00:20 looks like my bot is about to drop out of the top 100 2010-11-28T15:00:23 just lost a 138 2010-11-28T15:00:29 just lost to a #138 2010-11-28T15:00:45 #98 now 2010-11-28T15:00:47 \o/ #1 @ NL :P 2010-11-28T15:00:48 bhasker: i dropped out as well. weird stuff.. hope we find a way to get back in 2010-11-28T15:00:56 yea lets see :-\ 2010-11-28T15:01:02 congrats 2010-11-28T15:01:03 *** sigh has joined #aichallenge 2010-11-28T15:01:17 now lets keep it that way Mistmanov xD 2010-11-28T15:01:21 grats E323 2010-11-28T15:01:45 ty 2010-11-28T15:02:05 bhasker: i was also having timeouts on zeroviz.. had to start my bot from amazon EC2 2010-11-28T15:02:20 ah i am not spending money for running a bot! 2010-11-28T15:02:24 free wifi @ peets for now 2010-11-28T15:02:43 waiting for my wife to finish one of her classes 2010-11-28T15:02:49 have to go pick her up so tinkering on free wifi 2010-11-28T15:02:59 *** Azrathud has quit IRC (Ping timeout: 245 seconds) 2010-11-28T15:03:26 hehehe well, this EC2 server is my other side project so i'm not spending any extra money.. 3 cents per hour is not too bad 2010-11-28T15:04:03 bhasker: nooooooooo don't go out of 100 2010-11-28T15:04:12 still at #98! 2010-11-28T15:04:13 that prolly means the same for me :S :S :S 2010-11-28T15:04:15 holding on 2010-11-28T15:04:15 anyone do random playouts, i.e. Monte Carlo? I considered that since it seems to be great for Go AI 2010-11-28T15:04:29 I think that running your CPU at full speed in order to run the bot locally comes close to costing 3 cents per hour in energy consumption.. 2010-11-28T15:04:49 hmm probably does 2010-11-28T15:05:23 bhasker, Error323, Mistmanov, all or us are not where we are used to be but i feel ratings are too volatile at the moment, so let's keep our hopes up! go, bots! 2010-11-28T15:05:23 a kwh is something like 12 cents, so if your cpu takes 250w (it doesn't, but eh..), it would cost 3 cents per hour 2010-11-28T15:05:42 rebelxt_: indeed :D 2010-11-28T15:05:47 lol yea rebel... lets cheer for our bots! 2010-11-28T15:05:48 yeah, I'm certainly expecting to still breach the top 100 :p 2010-11-28T15:05:49 in russian lowest power price in world 2010-11-28T15:05:52 *** Azrathud has joined #aichallenge 2010-11-28T15:06:02 in russian lowest electric power price in world 2010-11-28T15:06:03 worst case i blame antimatroid for writing a crappy map generator 2010-11-28T15:06:13 well, my low voltage CPU only takes 35 watts at full power, but still i like to use it for other things ;) 2010-11-28T15:06:21 *** Palmik has quit IRC (Remote host closed the connection) 2010-11-28T15:06:33 ah yes, I see my bot is doing excellently horrible on tcp :p 2010-11-28T15:06:37 my pc is also very nrg efficient 2010-11-28T15:06:38 i don't knwo how much i am paying to run my laptop its just a core2duo 2010-11-28T15:06:41 antimatroid1 is safe to blame for anything that happens ;)) 2010-11-28T15:06:53 oh i beat ruilov 2010-11-28T15:06:56 so going back up a bit now 2010-11-28T15:07:07 back to 80s 2010-11-28T15:07:14 =] 2010-11-28T15:07:24 yay i had another win asweel 2010-11-28T15:07:30 from #124 2010-11-28T15:07:45 woohoo 2010-11-28T15:07:53 sigh i am having random timeouts 2010-11-28T15:08:02 on tcp i.e 2010-11-28T15:09:13 hmm weird i won but my rating/elo didnt move:-\ 2010-11-28T15:09:58 bhasker: give it time 2010-11-28T15:10:01 antimatroid1's bot is very anti-green ;) local tourney has been running for close to two hours and still going 2010-11-28T15:10:04 it updates every now and then 2010-11-28T15:10:17 hehe 2010-11-28T15:10:22 rebelxt_: i just finished pllaying anti 2010-11-28T15:10:28 but he beats me 29/19 2010-11-28T15:10:30 are you running all the open sourced bots? 2010-11-28T15:10:38 GreenTea, are you here? 2010-11-28T15:10:43 though he's experiencing timeouts @ tournament 2010-11-28T15:10:47 so that's good :P 2010-11-28T15:10:48 yea 2010-11-28T15:10:48 bhasker: yeah i'm running all open sourced ones 2010-11-28T15:10:48 ah back to #89 2010-11-28T15:11:09 #119 now 2010-11-28T15:11:18 how is mine doing 2010-11-28T15:11:20 gonna run medrimonia and Mistmanov next 2010-11-28T15:11:28 oh you are running them in serial 2010-11-28T15:11:33 mm, i am finding private in this IRC client. One second 2010-11-28T15:11:35 i'll run Mistmanov aswell 2010-11-28T15:11:43 hurray! 2010-11-28T15:11:51 see if i can get it to work 2010-11-28T15:11:59 I wonder.. do you guys really check the entire code to see if it isn't going to blow up your computer? 2010-11-28T15:12:03 yay beat accoun finally! 2010-11-28T15:12:12 i don't think anyone is doing that mist lol 2010-11-28T15:12:20 Mistmanov: i run on EC2 so it's a JAIL 2010-11-28T15:12:21 (not that my code blows up your computer, but that's one of the things that makes me wary of running other people's code :p) 2010-11-28T15:12:35 i didnt run other's code either 2010-11-28T15:12:38 thats why i rely on tcp 2010-11-28T15:12:41 to do all my testing 2010-11-28T15:12:49 except for antima's which is running locally so he might blow something up hehe 2010-11-28T15:12:59 rebel he is not even a programmer 2010-11-28T15:13:02 Top 10 players: bocsimacko(3508), _iouri_(3434), protocolocon(3408), Accoun(3406), GreenTea(3403), jimrogerz(3382), george(3371), _Astek_(3369), shangas(3352), felixcoto(3352) 2010-11-28T15:13:05 so be careful! 2010-11-28T15:13:13 though maybe if he was one you have something to worry 2010-11-28T15:13:23 hah http://zeroviz.us:8080/canvas.cgi?game_id=33028 2010-11-28T15:13:47 all my losses so far are on maps with low planet counts 2010-11-28T15:13:50 yeah, he's just pretending to not be a programmer so you won't search for the keylogger that he cunningly hid in some future state calculations ;) 2010-11-28T15:13:50 looking at antima's code.. maybe all my passwords are gone now 2010-11-28T15:14:36 mega1 your bot will this tourney 2010-11-28T15:14:39 i think that is decided 2010-11-28T15:14:44 now its for rank 2-100 2010-11-28T15:14:50 you can go rest 2010-11-28T15:15:20 Error323: timing out :o 2010-11-28T15:15:28 on tcp 2010-11-28T15:16:00 bhasker: I'm still uncertain about that. Maybe if we had 100+ games at this point. 2010-11-28T15:16:08 *** cybertron_ has joined #aichallenge 2010-11-28T15:16:23 mega1_: you wrote bocsimacko, I guess? 2010-11-28T15:16:26 i don't see why you are uncertain, i don't think any bot came out in last few days that gave you any significant amount of trouble 2010-11-28T15:16:28 *** Neverstu has joined #aichallenge 2010-11-28T15:16:37 politik: yes 2010-11-28T15:16:39 most bots beat you maybe on 5% of the maps 2010-11-28T15:16:43 thats about it 2010-11-28T15:16:46 mega1_: looks like we will be over 100 games each in 24 hours 2010-11-28T15:16:56 it's pretty quick 2010-11-28T15:16:58 *** cybertron_ has quit IRC (Client Quit) 2010-11-28T15:17:04 Lisp was an awesome choice... hope you're open sourcing that one 2010-11-28T15:17:05 *** Error323 has quit IRC (Read error: Connection reset by peer) 2010-11-28T15:17:08 and iouri is quite strong 2010-11-28T15:17:22 I cannot help but feel sorry for bix0r 2010-11-28T15:17:26 next best Lisp bot is #293 2010-11-28T15:17:34 i got a fluke win, i feel like all my time is validated :P 2010-11-28T15:17:39 *** Error323_ has joined #aichallenge 2010-11-28T15:17:39 although I hated that bot with vengeance. 2010-11-28T15:17:42 what's the best quick 30-min tutorial on lisp online? if anyone knows 2010-11-28T15:17:50 oops i crashed my system 2010-11-28T15:18:00 *** Error323_ is now known as Error323 2010-11-28T15:18:26 rebelxt_: you've heard of SICP probably? definitely not a 30 min tutorial though 2010-11-28T15:18:26 Error323: malicious open source bots? ;-) 2010-11-28T15:18:39 rebelxt_: I'll get some recommendations for you. 2010-11-28T15:19:08 *** medrimonia2 has joined #aichallenge 2010-11-28T15:19:21 *** bhasker has quit IRC (Quit: bhasker) 2010-11-28T15:19:38 I think I wont be able to stop improving my bot now that people released sources :| I cant help it but this game is an awesome way to learn some new aspects in prediction and whatnot for ai's 2010-11-28T15:19:44 politik: will check it out 2010-11-28T15:19:46 kinda sad...isnt it? 2010-11-28T15:19:52 whoot another win 2010-11-28T15:20:53 :/ I haven't had a game in ages 2010-11-28T15:20:55 i did not really expect top100 to be so competitive. awesome ;-) 2010-11-28T15:21:02 well, half an hour 2010-11-28T15:21:14 *** sigh has quit IRC (Ping timeout: 242 seconds) 2010-11-28T15:21:15 Mistmanov: ok, let me schedule a game for you 2010-11-28T15:21:16 ;-) 2010-11-28T15:21:40 mega1_: is ur src released yet? 2010-11-28T15:21:40 just give it give it give it :D :D :D 2010-11-28T15:21:40 put it on github 2010-11-28T15:21:40 *** Ken69267 has quit IRC (Ping timeout: 240 seconds) 2010-11-28T15:21:43 quite amazing how well shangas's bot competes considering he hasnt updated it for a month 2010-11-28T15:21:54 *** nann has joined #aichallenge 2010-11-28T15:22:04 *** medrimonia has quit IRC (Ping timeout: 245 seconds) 2010-11-28T15:22:05 *** Apophis_ has quit IRC (Read error: Connection reset by peer) 2010-11-28T15:22:05 *** dmj111 has quit IRC (Ping timeout: 245 seconds) 2010-11-28T15:22:06 *** sigh_ has joined #aichallenge 2010-11-28T15:22:08 *** Ken69267 has joined #aichallenge 2010-11-28T15:22:08 i had a 30-min period between games too earlier today. match making is not perfect 2010-11-28T15:22:12 *** javagamer__ has quit IRC (Ping timeout: 246 seconds) 2010-11-28T15:22:13 Error323: not yet, you'll have to wait a couple of days 2010-11-28T15:22:21 :( 2010-11-28T15:22:41 *** javagamer__ has joined #aichallenge 2010-11-28T15:22:46 mega1 is still making it stronger so that noone can best it for 2-3 more years 2010-11-28T15:22:48 the lack of the "so many games left until your next game" means that I'm refreshing the page the entire time instead of doing something useful 2010-11-28T15:22:49 *** Appleman1234 has quit IRC (Ping timeout: 264 seconds) 2010-11-28T15:22:51 I'm hopeless -_- 2010-11-28T15:23:18 rebelxt_: I'm not touching it after making it a somewhat acceptable read. 2010-11-28T15:23:19 yeah it's quite addictive 2010-11-28T15:23:19 *** Appleman1234 has joined #aichallenge 2010-11-28T15:23:39 *** sigh has joined #aichallenge 2010-11-28T15:23:51 and then you finally get a game, and it's against ~1000 and you lose because he does something ridiculous that triggers a bug that you've never encountered 2010-11-28T15:23:52 just wait 2010-11-28T15:23:54 *** Rubicon-|-Cross has joined #aichallenge 2010-11-28T15:23:58 lol Mistmanov 2010-11-28T15:24:01 *** Rubicon-|-Cross has quit IRC (Read error: Connection reset by peer) 2010-11-28T15:24:08 Mistmanov: lol exactly 2010-11-28T15:24:09 *** jaspervdj has joined #aichallenge 2010-11-28T15:24:43 so many servers! this is exciting! 2010-11-28T15:24:51 ye its awesome 2010-11-28T15:24:53 8 servers 2010-11-28T15:25:04 oh 9 now O_O 2010-11-28T15:25:11 oh no w8 2010-11-28T15:25:17 * Error323 needs to learn howto count 2010-11-28T15:25:36 clearly math is not required for this contest 2010-11-28T15:25:42 :P 2010-11-28T15:25:48 mega1, deep boltzmann machines = yummy.. revisiting some of your blog posts for fun 2010-11-28T15:26:26 those are my favourite. Nothing like that in the code. 2010-11-28T15:26:44 where is his blog 2010-11-28T15:26:52 *** perror has quit IRC (Quit: Bye all !) 2010-11-28T15:27:01 *** Zepp_ is now known as Zepp 2010-11-28T15:27:04 mega1_: can't you just push to github? :X I don't care if code is messy 2010-11-28T15:27:10 * Error323 wants to reaaaaaaaaaad 2010-11-28T15:27:31 mega1_: i'm expecting to see your code do something like: (fold-spacetime gametree planets) 2010-11-28T15:27:32 Error323: what if it's 55K lines of code? ;) 2010-11-28T15:27:36 *** Zepp has left #aichallenge 2010-11-28T15:27:43 rebelxt_: its lisp, i highly doubt it :P 2010-11-28T15:27:48 hehe 2010-11-28T15:27:57 i'll compile lisp 2010-11-28T15:28:03 Top 10 players: bocsimacko(3525), _iouri_(3440), GreenTea(3419), protocolocon(3415), Accoun(3413), jimrogerz(3412), george(3378), shangas(3366), _Astek_(3361), felixcoto(3359) 2010-11-28T15:28:13 as I said, don't get your hopes up. 2010-11-28T15:28:20 yeah, probably, 40 LOC so that it could fit on a screen ;) 2010-11-28T15:28:27 lol 2010-11-28T15:28:37 he'd be godlike then xD 2010-11-28T15:28:46 *** Rubicon-|-Cross has joined #aichallenge 2010-11-28T15:29:01 after eeking out a win over bocsimacko, my bot is spent 2010-11-28T15:29:01 let's see how space.invaders is doing 2010-11-28T15:29:05 he will now lose all future matches 2010-11-28T15:29:15 SLOC Directory SLOC-by-Language (Sorted) 2010-11-28T15:29:15 1748 top_dir cpp=1748 2010-11-28T15:29:15 Totals grouped by language (dominant language first): 2010-11-28T15:29:15 cpp: 1748 (100.00%) 2010-11-28T15:29:32 space.invaders at 271 pretty good 2010-11-28T15:29:35 nice 2010-11-28T15:30:09 http://ai-contest.com/visualizer.php?game_id=8941220 2010-11-28T15:30:13 Mistmanov: another loss -.- 2010-11-28T15:30:16 * Mistmanov pleegt zelfmoord 2010-11-28T15:30:28 hahaha 2010-11-28T15:30:45 *** Mistmanov is now known as MistmanovsGhost 2010-11-28T15:31:35 dont think i have won a single game on tcp now :p 2010-11-28T15:32:56 My "can I attack this neutral without making myself vulnerable" code is borked 2010-11-28T15:33:03 which is a bit of a disadvantage on many maps -_- 2010-11-28T15:33:16 ur beating me locally here MistmanovsGhost 2010-11-28T15:33:27 which hurts 2010-11-28T15:33:29 ;_; 2010-11-28T15:33:52 well yeah but.. 2010-11-28T15:33:53 .. 2010-11-28T15:33:58 ;_;x2 2010-11-28T15:34:19 I already played our bots together on my pc -_- 2010-11-28T15:34:23 didn't you trust my results :p 2010-11-28T15:34:26 * Error323 casts his trapped in local optima spell on MistmanovsGhost 2010-11-28T15:34:46 well just curious =) 2010-11-28T15:34:52 if it was the same here :P 2010-11-28T15:35:03 ;-) 2010-11-28T15:35:04 *** Ice_Harley has quit IRC (Ping timeout: 265 seconds) 2010-11-28T15:35:05 playing you against iouri now 2010-11-28T15:35:19 i win 22 out of 74 of iouri 2010-11-28T15:35:22 1 draw 2010-11-28T15:35:29 *** Neverstu has quit IRC (Quit: Leaving) 2010-11-28T15:35:33 quite happy since he's second 2010-11-28T15:35:52 i think i'm just really good on some maps 2010-11-28T15:35:57 and very bad on others 2010-11-28T15:36:15 yeah that's probably quite common 2010-11-28T15:36:29 for me it's because some of my heuristics are pretty good 2010-11-28T15:36:39 and some are really bad (e.g., the "can I take a neutral?") 2010-11-28T15:36:54 and some maps just invoke one set of heuristics, and other maps another set 2010-11-28T15:37:40 maps where you would have two clear front planets would be a disaster for me 2010-11-28T15:37:46 since it would require coordinated attacks 2010-11-28T15:37:54 and my bot simply doesn't have coordinated attacks ^^ 2010-11-28T15:37:59 my attacking system is very very different 2010-11-28T15:38:24 what, I beat narnach on tcp! 2010-11-28T15:38:36 *** Ice_Harley has joined #aichallenge 2010-11-28T15:38:49 iouri's code is sloooooooow 2010-11-28T15:38:52 but good 2010-11-28T15:38:56 he writes rlly nice code 2010-11-28T15:39:09 * Error323 will eat it all 2010-11-28T15:39:21 is iouri's minimax or heuristic? 2010-11-28T15:39:34 its maximax 2010-11-28T15:39:47 evaluation of actions 2010-11-28T15:39:50 across time 2010-11-28T15:40:29 ahh see his forum post 2010-11-28T15:41:31 *** foucist has quit IRC (Ping timeout: 245 seconds) 2010-11-28T15:43:05 MistmanovsGhost: your bot is pretty quick, isn't it? 2010-11-28T15:43:05 Top 10 players: bocsimacko(3564), _iouri_(3477), Accoun(3468), protocolocon(3445), jimrogerz(3445), GreenTea(3443), _Astek_(3399), shangas(3390), felixcoto(3374), george(3369) 2010-11-28T15:43:17 yeah 2010-11-28T15:43:30 here it takes ~20 ms per turn 2010-11-28T15:43:30 max 2010-11-28T15:43:37 cool then gonna run it next. antima's is killing me ;) 2010-11-28T15:44:06 (wait, that's a lie, I'm seeing a lot of 32 ms turns :p) 2010-11-28T15:44:10 2 minutes per game, i wonder how he managed debugging... 2010-11-28T15:44:21 *** nann is now known as Apophis 2010-11-28T15:44:31 3ms here 2010-11-28T15:44:41 :o 2010-11-28T15:44:53 wow.. i'm under 100ms python so would probably be 3-4ms c++ 2010-11-28T15:45:00 goddamn when will I face rankings < me 2010-11-28T15:45:03 =.= 2010-11-28T15:45:12 Error323: 3ms is so awesome for volume testing! 2010-11-28T15:45:16 yeah I think it has somethign to do with memory management or something.. I often get a 0 ms turn, and 16, and 32.. but not really the stuff in between 2010-11-28T15:45:26 or maybe my time thing is just off, I dunno 2010-11-28T15:45:40 rebelxt_: yeah especially with this: https://github.com/albertz/planet_wars-cpp 2010-11-28T15:46:39 yeah awesome engine and viewer. i used it too 2010-11-28T15:46:57 and extremely fast 2010-11-28T15:47:08 no java vm startup crap 2010-11-28T15:47:38 *** davidjliu_ has quit IRC (Ping timeout: 265 seconds) 2010-11-28T15:47:54 yeah.. well, on my laptop chrome starts instantly (html5 visualizer) mainly due to using solid state drive 2010-11-28T15:48:10 oh man, there was a better viewer out there? sweet jesus, that would have saved me some time 2010-11-28T15:48:26 politik: zomg, you did not know? ;-) 2010-11-28T15:49:01 *** foucist has joined #aichallenge 2010-11-28T15:49:23 http://ai-contest.com/profile.php?user_id=10154 I like green 2010-11-28T15:49:34 this seems like a lot of work this dude undertook just for a better viewer 2010-11-28T15:49:39 pretty amazing 2010-11-28T15:49:47 nice GREEN streak 2010-11-28T15:50:02 if i were to start again i would created my own visualizer with clickable debugging etc 2010-11-28T15:50:11 but i just thought ohh how hard can this be xD 2010-11-28T15:50:15 well he wrote engine in c++, also viewer, and ported all example bots. pretty amazing 2010-11-28T15:50:18 * Error323 <-- fail 2010-11-28T15:50:19 xD 2010-11-28T15:50:56 MistmanovsGhost: duuuuuude wtf are you doing 2010-11-28T15:51:03 where? 2010-11-28T15:51:11 klosing, probably 2010-11-28T15:51:49 ye 2010-11-28T15:51:52 rebelxt vs antimatroid 35-0-40 gg's (i think he went to sleep) 2010-11-28T15:52:18 #103 2010-11-28T15:52:21 grrrrrrr 2010-11-28T15:52:30 just.a.few.moar... 2010-11-28T15:52:42 you were running me against iouri right? 2010-11-28T15:52:49 results in yet? 2010-11-28T15:53:01 no forgot to flush the resultsbuffer 2010-11-28T15:53:04 so rerunning 2010-11-28T15:53:11 owk 2010-11-28T15:53:16 ah they are now 2010-11-28T15:53:18 25 wins 2010-11-28T15:53:20 48 losses 2010-11-28T15:53:22 3 more than me 2010-11-28T15:53:25 bastard 2010-11-28T15:53:27 =) 2010-11-28T15:53:39 Error323: you just passed me! wait up ;-) 2010-11-28T15:53:48 rebelxt_: haha ok 2010-11-28T15:54:10 *** foucist has quit IRC (Ping timeout: 255 seconds) 2010-11-28T15:54:24 just kidding, if you go higher, so do i ;-) let's aim for top 10! 2010-11-28T15:54:33 =] oh ye 2010-11-28T15:54:49 *** sigh_ has quit IRC (Remote host closed the connection) 2010-11-28T15:54:54 if the official server just gave me 10 games against iouri, I'm sure it would do wonders for my rank -_- 2010-11-28T15:54:57 though I think i'll end up @ 70-90 somewhere :/ 2010-11-28T15:55:05 *** foucist has joined #aichallenge 2010-11-28T15:55:05 MistmanovsGhost: i love your bot. it's the fastest i've tested against so far except Error323's probably 2010-11-28T15:55:18 where is the official map set? 2010-11-28T15:55:18 I'm not sure whether that's a good thing :p 2010-11-28T15:55:47 MistmanovsGhost: 50-50 so far.. will let you know the results 2010-11-28T15:56:41 lol i'm gathering attack actions and defend actions 2010-11-28T15:56:46 and sort them on growthrate XD 2010-11-28T15:56:47 mega1_: http://ai-contest.com/forum/download/file.php?id=190 2010-11-28T15:56:48 that's it 2010-11-28T15:56:58 *** p4p4 has joined #aichallenge 2010-11-28T15:57:02 tried a lot of other crap 2010-11-28T15:57:07 but this seems to work best 2010-11-28T15:57:07 thanks 2010-11-28T15:57:17 now I cannot clone iouri's repo 2010-11-28T15:57:26 ? 2010-11-28T15:57:32 shouldn't be hard 2010-11-28T15:57:40 mega1_: could you clone mine and tell results? 2010-11-28T15:58:04 Top 10 players: bocsimacko(3548), _iouri_(3488), Accoun(3479), GreenTea(3463), protocolocon(3458), jimrogerz(3441), _Astek_(3419), shangas(3402), Slin-.-(3391), felixcoto(3385) 2010-11-28T15:58:05 I got it 2010-11-28T15:58:11 ah cool 2010-11-28T15:58:21 url for yours? 2010-11-28T15:58:27 *** antimatroid1 has quit IRC (Ping timeout: 276 seconds) 2010-11-28T15:58:49 *** McLeopold has joined #aichallenge 2010-11-28T15:58:55 mega1_: https://github.com/Error323/googleai 2010-11-28T15:59:00 git@github.com:Error323/googleai.git 2010-11-28T16:00:02 *** bhasker has joined #aichallenge 2010-11-28T16:00:19 *** winexception has quit IRC (Quit: Bye) 2010-11-28T16:00:36 *** winexception has joined #aichallenge 2010-11-28T16:01:56 bhasker: last game was not exciting to watch :( 2010-11-28T16:02:30 *** asdfghv has quit IRC (Quit: Page closed) 2010-11-28T16:02:31 Error323: did you compile iouri's bot with -DIS_SUBMISSION? 2010-11-28T16:02:36 Interesting to look code of wagstaff bot (in Haskell) :) 2010-11-28T16:02:54 mega1_: yes 2010-11-28T16:03:05 GreenTea: could you give me the url for it? 2010-11-28T16:03:09 just add in makefile 2010-11-28T16:03:37 *** insectores has joined #aichallenge 2010-11-28T16:03:59 I have no url, just say it will be interesting to see) 2010-11-28T16:04:15 oh, oops. *will* be :) 2010-11-28T16:04:21 *** Migi32 has joined #aichallenge 2010-11-28T16:04:22 never ever seen haskell code 2010-11-28T16:04:24 in ##flash 2010-11-28T16:04:28 sorry 2010-11-28T16:04:42 I typed /join ##flash and it apparently missed the /jo 2010-11-28T16:05:02 *** Ice_Harley has quit IRC (Ping timeout: 265 seconds) 2010-11-28T16:05:16 yay another win 2010-11-28T16:05:25 with the next rank update i'll be in #100 in think 2010-11-28T16:05:28 i hope -.- 2010-11-28T16:05:47 accoun is #3 2010-11-28T16:05:50 nice 2010-11-28T16:06:13 Accoun: wanna share ur code? 2010-11-28T16:06:24 or is it russian 2010-11-28T16:06:47 i think i'm going to stick with c++ for the next contest.. just for efficient local testing even though code is not going to be pretty 2010-11-28T16:07:13 i'm gonna do things very differently next contest 2010-11-28T16:07:42 like research the game itself first 2010-11-28T16:07:48 and see what strategies are good 2010-11-28T16:07:54 ridiculous 2010-11-28T16:07:56 i just started hacking right away this time 2010-11-28T16:08:02 *** mceier has quit IRC (Quit: leaving) 2010-11-28T16:08:03 we had some great engines and visualizers this time, but no testing framework. that would be awesome.. 2010-11-28T16:08:34 I might look into an actual development envionment next time.. 2010-11-28T16:08:45 *** sanddbox has joined #aichallenge 2010-11-28T16:08:48 I did everything with notepad2 (basically.. notepad with syntax highlightning) 2010-11-28T16:08:49 vim ftw 2010-11-28T16:08:56 not the most efficient way to work I think 2010-11-28T16:09:08 Error323: i took a 5-week break in the middle and read AI book last month. it was a fun read but did not really help me with this contest ;-) 2010-11-28T16:09:09 *** Migi32 has left #aichallenge 2010-11-28T16:09:34 I wasted sooooo much time on perfecting alphabeta :( 2010-11-28T16:09:59 only to find out my eval function was crap and i couldn't find out a good one 2010-11-28T16:10:56 YEEEEEEEEEEEEEEES 2010-11-28T16:10:59 #95 2010-11-28T16:11:07 :) 2010-11-28T16:11:14 here I come :P 2010-11-28T16:11:18 Error323: me too back in 100, awesome! ;) 2010-11-28T16:11:24 sweeeeeeet 2010-11-28T16:11:36 thanks, Meatkat ;) 2010-11-28T16:11:46 thanks oddshrimp 2010-11-28T16:11:48 xD 2010-11-28T16:12:03 MistmanovsGhost: i have results. interested? 2010-11-28T16:12:06 i'm on quite a winningstreak 2010-11-28T16:12:09 http://ai-contest.com/profile.php?user_id=10154 2010-11-28T16:12:10 always 2010-11-28T16:12:13 as long as they aren't depressing 2010-11-28T16:12:25 nah close actually 2010-11-28T16:12:33 *** davidjliu has joined #aichallenge 2010-11-28T16:12:33 39-3-33 2010-11-28T16:12:48 33 for me? that's pretty good =) 2010-11-28T16:12:58 yep 33 wins for you, 3 draws 2010-11-28T16:13:05 Top 10 players: bocsimacko(3578), _iouri_(3525), Accoun(3522), protocolocon(3501), GreenTea(3495), jimrogerz(3471), _Astek_(3452), george(3428), Slin-.-(3415), CorwinAlex(3412) 2010-11-28T16:13:39 what's interesting is that you win very fast.. avg turns for you to win only 117, for me 151 2010-11-28T16:13:57 yeah I either win fast or I lose ^^ 2010-11-28T16:14:04 hehe 2010-11-28T16:14:10 same here 2010-11-28T16:14:12 depending on mapset 2010-11-28T16:14:16 maps with a lot of direct fighting often go may way 2010-11-28T16:14:33 maps with a lot of expanding and waiting for ships to replenish go horribly 2010-11-28T16:14:43 ok, medrimonia2 up next. python vs python. this is gonna be super painful for my bot 2010-11-28T16:14:53 I was losing a lot at first, then I figured ok once binary knapsack is done, i'll remove the center planets and save my ships... then BOOM top 40 :P 2010-11-28T16:17:31 mega1_: are you running ur bot yet against iouri's or mine? 2010-11-28T16:17:44 against iouri 2010-11-28T16:17:51 how is it doing 2010-11-28T16:18:07 29/1/7 so far for me 2010-11-28T16:18:13 nice 2010-11-28T16:18:18 what's everyone's horizon? i.e. number of turns you use when calculating expected value of planet captured.. if you used such a metric in eval function 2010-11-28T16:18:46 well sometimes i simulate to the end 2010-11-28T16:18:49 without moves though 2010-11-28T16:18:52 just to see who'd win 2010-11-28T16:19:02 if its the enemy i'll cap neutrals 2010-11-28T16:19:19 do you simulate to 200 even on first turn? 2010-11-28T16:19:25 yes 2010-11-28T16:19:28 its rlly fast 2010-11-28T16:19:49 https://github.com/Error323/googleai/blob/heuristics/Simulator.cc 2010-11-28T16:19:57 interesting. i had to cap it to ~60 turns or so to prevent from capturing expensive planets 2010-11-28T16:20:15 oh i never tried that :$ 2010-11-28T16:20:25 I go to 20 turns, just because thats what I guessed last night when I got it working :) 2010-11-28T16:20:31 lol 2010-11-28T16:20:44 *** antimatroid has joined #aichallenge 2010-11-28T16:20:47 yeah, the more turns, the more those 4/5 growth neutrals will look better than they are 2010-11-28T16:20:53 antimatroid: lol, goto bed 2010-11-28T16:20:54 =) 2010-11-28T16:21:04 *** dmj111 has joined #aichallenge 2010-11-28T16:21:15 damnit so far for my winning streak -.- 2010-11-28T16:21:32 medrimonia2's bot is fast.. and very strong too 2010-11-28T16:21:57 *** davidjliu has quit IRC (Ping timeout: 265 seconds) 2010-11-28T16:23:03 how did Accoun rise to second? 2010-11-28T16:23:14 wowa 2010-11-28T16:23:19 They must have cheated to favour him. 2010-11-28T16:23:19 dunno 2010-11-28T16:23:23 hahaha 2010-11-28T16:23:36 biased mapset 2010-11-28T16:23:37 rebelxt_: my horizon is 33 turns) 2010-11-28T16:23:45 lol Error323 2010-11-28T16:24:30 canadians are known to cheat to favour russians in AI contests.. 2010-11-28T16:25:02 yes 2010-11-28T16:25:12 XD 2010-11-28T16:25:17 canadians very cheaty 2010-11-28T16:25:20 Accoun: wanna release ur sourcecode? 2010-11-28T16:25:37 Error323> dont wanna 2010-11-28T16:25:45 lol ok 2010-11-28T16:26:03 one thing I thought of today that I wish I tried... in the eval of the game at the N turn horizon, *divide* my ship count by the enemy ship count, rather than subtracting. then even exchanges in attacks are a win, would make my bot more aggressive 2010-11-28T16:26:51 with subtracting, the bot ends up being conservative, so it doesn't use it's advantage 2010-11-28T16:27:02 mega1_: was accon the one with highest win rate against you? 2010-11-28T16:27:13 "*divide* my ship count by the enemy ship count" i do such thing.. 2010-11-28T16:27:55 aw man, I could have been in the top 10... ;) 2010-11-28T16:28:06 Top 10 players: bocsimacko(3584), Accoun(3533), protocolocon(3512), _iouri_(3505), GreenTea(3485), jimrogerz(3477), _Astek_(3457), george(3432), Slin-.-(3418), felixcoto(3409) 2010-11-28T16:28:09 *** delt0r has joined #aichallenge 2010-11-28T16:28:23 rebelxt_: I'm not sure how high because he ran so many bots at the same time on tcp, but certainly big percentage of my losses came from him. 2010-11-28T16:29:00 *** delt0r___ has quit IRC (Ping timeout: 276 seconds) 2010-11-28T16:29:03 63/3/9 against iouri 2010-11-28T16:29:06 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-11-28T16:29:07 *** rebelxt_ has quit IRC (Quit: Page closed) 2010-11-28T16:29:11 i found it to be very dangerous to run several versions of my bot on tcp 2010-11-28T16:29:24 mega1_: nice 2010-11-28T16:29:25 *** rebelxt_ has joined #aichallenge 2010-11-28T16:29:51 what's really surprising is how many strong bots there are which i've never seen testing on tcp... how did they not get stuck in local optimum? or maybe they tested secretly using aliases 2010-11-28T16:29:53 MistmanovsGhost: oh man you keep losing o_O 2010-11-28T16:30:05 yeah I've just given up :p 2010-11-28T16:30:44 how do you guys compute the frontline 2010-11-28T16:31:16 my method is: if i'm the closest planet to an enemy planet i'm frontline. 2010-11-28T16:31:17 mega1_: woah, so iuri only gets two more wins that i did. that's weird since he appears way stronger than my bot. 2010-11-28T16:31:31 rebelxt_: different map set 2010-11-28T16:31:39 ahh right 2010-11-28T16:31:54 different number of maps too 2010-11-28T16:32:15 mega1_: could you run mine now? :D 2010-11-28T16:32:18 makes sense.. i'll run his code next 2010-11-28T16:32:23 plz 2010-11-28T16:32:38 mega1_: do you know what your bot does that is unique, or is it a mystery to you why it is so awesome? :) 2010-11-28T16:32:51 Error323: the "if I'm closest to an enemy planet I'm front line" is also at the basis of my front calculation yes, but I also try to look into the future.. 2010-11-28T16:33:07 MistmanovsGhost: ah yes so do i 2010-11-28T16:33:11 for the feeding part 2010-11-28T16:33:18 like.. if planet X is going to send to ships to Y, Y needs to be the front when the ships arrive 2010-11-28T16:33:31 *** fawek has joined #aichallenge 2010-11-28T16:33:57 *** Appleman1234 has joined #aichallenge 2010-11-28T16:33:58 but my calculation is a bit borked. Works most of the time, but doesn't do what I want it to in some special situations 2010-11-28T16:34:07 same here 2010-11-28T16:34:13 sometimes it very wrong 2010-11-28T16:34:27 Error: i not have any calculations of "front line" just send support fleets to planets that's can be captured if enemy send all forces to to it.. 2010-11-28T16:34:31 on sparse maps 2010-11-28T16:34:31 its horror 2010-11-28T16:34:44 I see 2010-11-28T16:34:58 clearly a better strategy :P 2010-11-28T16:35:54 ratings are ceep growing. it will be 4000 soon.. 2010-11-28T16:36:10 what does this mean 2010-11-28T16:36:49 medrimonia2: probably sleeping? you beat me 48-2-25. nice 2010-11-28T16:36:56 I think, this meens that difference in strength of bots is very big. 2010-11-28T16:37:20 mm yes ur right 2010-11-28T16:37:35 *** fawek has quit IRC (Ping timeout: 245 seconds) 2010-11-28T16:37:43 still at 95 2010-11-28T16:37:51 scary low :/ 2010-11-28T16:40:26 in case anyone is interested :D https://github.com/phreeza/planetwars the sum of my dabblings 2010-11-28T16:42:14 fuuuuuuk 2010-11-28T16:42:24 if i look ahead less it performs better indeed 2010-11-28T16:42:32 yeah :) 2010-11-28T16:42:48 you didn't notice it attacking expensive neutrals? 2010-11-28T16:43:06 mega1: did you have to tweak iouri's code to compile on linux? 2010-11-28T16:43:07 Top 10 players: bocsimacko(3614), Accoun(3562), protocolocon(3542), _iouri_(3535), GreenTea(3513), jimrogerz(3512), _Astek_(3484), george(3463), Slin-.-(3445), shangas(3444) 2010-11-28T16:43:08 I would think you'd be getting sniped all the time 2010-11-28T16:43:09 yes but didn't figure this caused it >_< 2010-11-28T16:43:40 rebelxt_: -DIS_SUBMISSION 2010-11-28T16:43:44 and it works 2010-11-28T16:43:48 rebelxt_: add -DIS_SUBMISSION after CC=g++ 2010-11-28T16:43:50 removes timer crap 2010-11-28T16:44:28 worked, tnx both of you 2010-11-28T16:44:57 MistmanovsGhost: oh man I hope you get out of it 2010-11-28T16:45:02 this is really scary 2010-11-28T16:45:27 eh. 2010-11-28T16:45:31 what can you do about it :p 2010-11-28T16:45:36 may a bot that sucks less, maybe 2010-11-28T16:45:37 :p 2010-11-28T16:45:41 make 2010-11-28T16:46:04 mja 2010-11-28T16:46:11 i could've 2010-11-28T16:46:19 with my too extreme lookahead lol 2010-11-28T16:46:22 fuk 2010-11-28T16:46:24 xD 2010-11-28T16:46:40 why didn't i think of that 2010-11-28T16:46:54 I dunno, thats like the only thing I *did* think of :) 2010-11-28T16:47:08 *** choas has quit IRC (Quit: leaving) 2010-11-28T16:47:23 the last big change I made was to change how far I looked ahead based on the closest distance between my planets and the enemy planets. 2010-11-28T16:47:53 mja i should've done that too 2010-11-28T16:47:58 Error323: well, it's probably a 10% improvement if you optimize horizon. not dramatic.. it seems to matter the most on the first turn 2010-11-28T16:48:01 cuz this is just >_< frustrating 2010-11-28T16:48:25 dmj111: will you publish code? 2010-11-28T16:48:34 dmj111: hey, curious if your bot is heuristic as well like most? 2010-11-28T16:48:45 yeah, I will. I am too depressed at the moment. 2010-11-28T16:48:50 :) 2010-11-28T16:49:03 dmj111: whyh? 2010-11-28T16:49:06 where u at 2010-11-28T16:49:12 28 2010-11-28T16:49:13 mm 2010-11-28T16:49:21 u'll climb 2010-11-28T16:49:22 dmj111: you had a bad start (imho). climbing back up 2010-11-28T16:49:24 just takes more time 2010-11-28T16:49:27 I am rising. I had my hopes up after the last few weeks :) 2010-11-28T16:49:40 Even though I told myself not to. 2010-11-28T16:50:05 I learned to _always_ trust statistical data 2010-11-28T16:50:05 rebelxt_: It is mostly heuristic. I do a 2-ply lookahead, against my own heuristics. 2010-11-28T16:50:10 and not look at first losses :P 2010-11-28T16:50:11 bix0r4ever is #27, I don't understand that either. I thought he was top 5 consistently 2010-11-28T16:50:23 though its psychologically a hit on the head haha 2010-11-28T16:50:25 *** yasith has quit IRC (Ping timeout: 264 seconds) 2010-11-28T16:50:35 dmj111: nice. i often watched your games and tried reverse engineering some of the heuristics ;) 2010-11-28T16:50:54 Error323: 67/1/7 against you 2010-11-28T16:51:13 mega1_: cool thnx 2010-11-28T16:51:19 I'm off to bed. 2010-11-28T16:51:45 yeah me too in a bit 2010-11-28T16:51:53 going crazy over the refresh button its insane 2010-11-28T16:53:02 OMFG I WON A GAME 2010-11-28T16:53:11 against #472 -_- 2010-11-28T16:53:13 )) 2010-11-28T16:53:17 BRING OUT THE CLOWNS!!!!!!!!! 2010-11-28T16:53:34 BALOONDROP 2010-11-28T16:53:35 ))) 2010-11-28T16:53:39 ugh I lost to medved 2010-11-28T16:53:42 CURSES 2010-11-28T16:53:48 41 games per bot 2010-11-28T16:55:54 just when i thought i was beating iouri 10/10, i realized i was running his c++ bot with the following: "python $file" where file is his binary.. python did not complain really and i just kept winning games ;-) 2010-11-28T16:56:30 Error323: haha 2010-11-28T16:56:32 er 2010-11-28T16:56:51 does reducing the lookahead help much? 2010-11-28T16:57:07 politik: tbh I don't wanna know anymore :P 2010-11-28T16:58:07 Top 10 players: bocsimacko(3631), Accoun(3581), protocolocon(3554), jimrogerz(3529), _iouri_(3525), GreenTea(3525), _Astek_(3498), george(3457), felixcoto(3455), Slin-.-(3455) 2010-11-28T16:58:35 politik: it does.. i burnt lots of cpu cycles to find optimal value 2010-11-28T16:58:52 I don't blame youdid you say you did 60? 2010-11-28T16:59:00 jesus I keep half-finishing things 2010-11-28T16:59:06 mine is 40 turns, i tested increasing it, but it didnĀ“t work 2010-11-28T16:59:45 politik: yes 60 2010-11-28T16:59:48 it was better 2010-11-28T16:59:55 I think dmj111's approach of making it proportional to distance to enemy is the best... that's the measure of how uncertain you are about how quickly the enemy could change the state 2010-11-28T16:59:58 i did something weird.. which is 40 + (max_distance_between_planets - distance), then multiply by growth and subtract cost if it's neutral 2010-11-28T16:59:59 yep 2010-11-28T17:00:09 I did 50, for what it's worth 2010-11-28T17:00:18 rebelxt_: that is weird indeed :P 2010-11-28T17:00:35 rebelxt_: wow 2010-11-28T17:00:53 what is distance, distance of the move you're considering? 2010-11-28T17:01:01 yes 2010-11-28T17:01:15 )#$(%)#$(*%)#($%@ 102 2010-11-28T17:01:16 rebelxt_: i tried something similar, but it didn't make mine better. 2010-11-28T17:02:00 if max distance is 30, then neutral's value is essentially (70 - distance) * growth - cost 2010-11-28T17:02:25 *** yasith has joined #aichallenge 2010-11-28T17:02:26 if it's enemy, then multiply by two since enemy loses production, and you gain, so the score is twice as big 2010-11-28T17:02:41 hm so that is sort of valuing closer planets higher... makes sense 2010-11-28T17:02:54 dmj111: curious what worked the best for you? 2010-11-28T17:02:56 sounds semi-hard to tune though 2010-11-28T17:03:12 I used: growth^2 / (ships * distance + epsilon) 2010-11-28T17:03:14 still seems like distance-to-enemy should be in there somewhere 2010-11-28T17:03:52 maybe distance-to-enemy - distance? but that will give quite low numbers 2010-11-28T17:04:02 distance-to-enemy being the distance of the target planet to the enemy 2010-11-28T17:04:14 politik: distance to enemy is taken into account since i only attack as long as i can actually capture the target.. so i can get more ships to the planet than enemy can to be able to hold it until profit 2010-11-28T17:04:48 I calc rating to neutral this way: getGrowthRate() / (getShipsCount() + distanceTo(x, y) * getGrowthRate());, where x y is center of mass of my planets.. 2010-11-28T17:04:59 ok, that's based on the assumption that the enemy will attack it with everything? makes sense 2010-11-28T17:05:16 siple and effective formula 2010-11-28T17:05:17 politik: yeah, a worst case scenario 2010-11-28T17:05:39 these new maps are destroying my bot :( 2010-11-28T17:05:46 GreenTea: 2010-11-28T17:05:50 public int rankPlanetAttackability(Planet home, Planet expansion) { 2010-11-28T17:05:50 return ((this.Distance(home.PlanetID(), expansion.PlanetID())) - (2 * expansion.GrowthRate()) + (expansion.NumShips() / 10)); 2010-11-28T17:05:50 } 2010-11-28T17:05:55 ugly formatting 2010-11-28T17:05:56 xD 2010-11-28T17:06:10 the lower the score the better 2010-11-28T17:06:33 see magic constants) 2010-11-28T17:06:40 10 - why not 15.. 2010-11-28T17:06:54 so many ways to skin a cat ;) 2010-11-28T17:07:07 =) 2010-11-28T17:07:10 rebelxt_: I'm not sure I can explain a little, instead of a lot. I'll work on commenting my code, and try to get it out. 2010-11-28T17:07:18 GreenTea, i use a constant 2010-11-28T17:07:22 i really hate magic constants though. some of those were impossible to eliminate by HUNDRED of lines of code 2010-11-28T17:07:23 i fear magic constants :P 2010-11-28T17:07:24 i just put 10 so i didn't have to tell you the value of the constant 2010-11-28T17:07:59 dmj111: ok. looking forward to it 2010-11-28T17:09:06 i tried to use as less constants as possible - because this is synthetic thing.. 2010-11-28T17:09:21 "as less constants" 2010-11-28T17:09:29 i try to use as much proper english as possible 2010-11-28T17:09:33 the most magic constant i have is 140.. that thing is still there /sigh 2010-11-28T17:09:36 as few constants is correct 2010-11-28T17:09:58 yes, thank you) 2010-11-28T17:10:01 * Error323 trying GreenTea's eval function 2010-11-28T17:10:01 few 2010-11-28T17:10:03 how much tact do you try to use 2010-11-28T17:10:19 not a lot 2010-11-28T17:10:38 yay i won from rank #1200 -.- 2010-11-28T17:11:15 * Error323 has never been challenged by a rating < 90 :( 2010-11-28T17:12:46 experimental safe strategy not to capture neutrals if my growth rate is better seems work http://ai-contest.com/visualizer.php?game_id=8956546 2010-11-28T17:13:08 Top 10 players: bocsimacko(3616), Accoun(3605), protocolocon(3553), GreenTea(3546), jimrogerz(3545), _Astek_(3520), _iouri_(3518), george(3477), shangas(3471), felixcoto(3468) 2010-11-28T17:13:12 boring but effective ;) 2010-11-28T17:13:35 GreenTea: i do the same yes 2010-11-28T17:13:43 GreenTea, same 2010-11-28T17:13:44 hm even from turn 1? 2010-11-28T17:13:45 i did that from the start 2010-11-28T17:13:50 well 2010-11-28T17:13:56 figuring out when to capture neutrals and which is imo hardest part of this contest 2010-11-28T17:13:56 GreenTea: do you really like to juggle? ;) 2010-11-28T17:14:04 in turn 1 growth ratie is same (5) 2010-11-28T17:14:09 i expand if my growth is <= to the opponent 2010-11-28T17:14:18 sanddbox: so do i 2010-11-28T17:14:22 rebelxt_: yes) 2010-11-28T17:14:25 in the future* 2010-11-28T17:14:38 4 balls - max =) 2010-11-28T17:14:49 GreenTea: cool, haha that's what i'm doing at the moment ;-) 2010-11-28T17:15:03 programmers are weird ^_^ 2010-11-28T17:15:05 really?) 2010-11-28T17:15:21 cool.. 2010-11-28T17:15:22 3 balls is all i have though ;) 2010-11-28T17:15:33 ok, this probably sounded funny 2010-11-28T17:16:12 well guys its been fun 2010-11-28T17:16:17 but i'm gonna go ZzzZZ 2010-11-28T17:16:28 cya tomorrow, hf bashing F5 ;) 2010-11-28T17:16:39 night Error323! 2010-11-28T17:16:41 GreenTea: i also wanted to put a check in place to prevent expansion when not needed.. was losing some games due to this. ran out of time though 2010-11-28T17:16:50 ur submission is going strong 2010-11-28T17:17:00 night Error323 2010-11-28T17:17:06 bye 2010-11-28T17:17:20 good nigth 2010-11-28T17:17:51 *** p4p4 has quit IRC (Quit: ChatZilla 0.9.84 [SeaMonkey 2.0a3/20090223135443]) 2010-11-28T17:19:24 *** davidjliu has joined #aichallenge 2010-11-28T17:19:48 iouri vs rebelxt 43-0-32.. i'm surprised tbh 2010-11-28T17:21:52 *** winexception has quit IRC (Quit: Bye) 2010-11-28T17:22:08 *** winexception has joined #aichallenge 2010-11-28T17:22:18 *** MistmanovsGhost has quit IRC (Quit: Page closed) 2010-11-28T17:22:56 *** winexception has quit IRC (Client Quit) 2010-11-28T17:23:13 *** winexception has joined #aichallenge 2010-11-28T17:25:08 *** winexception has quit IRC (Client Quit) 2010-11-28T17:28:09 Top 10 players: bocsimacko(3640), Accoun(3592), jimrogerz(3562), GreenTea(3559), protocolocon(3549), _Astek_(3541), _iouri_(3540), felixcoto(3488), george(3473), Slin-.-(3472) 2010-11-28T17:31:35 *** rebelxt_ has quit IRC (Quit: Page closed) 2010-11-28T17:33:54 does anyone know how set an alert for the next context? 2010-11-28T17:35:40 meneky: don't the organizers have our e-mail so they can e-mail us when the next contest is announced? 2010-11-28T17:36:16 the question then is, will they? 2010-11-28T17:37:13 I hope next contest will be next year in the same months 2010-11-28T17:38:04 not sure, i think last contest finished on feb/mar 2010-11-28T17:38:21 so twice a year seems likely if they continue doing this 2010-11-28T17:38:27 which I hope they do :-) 2010-11-28T17:38:55 make it more frequent than once a year, I don't know.. 2010-11-28T17:39:11 it take to much time) 2010-11-28T17:39:13 my mental health may not toleate it, but i'd like it 2010-11-28T17:40:09 how much time you guys spend on this contest? 2010-11-28T17:41:27 50 hours or so, I think 2010-11-28T17:43:11 Top 10 players: bocsimacko(3667), Accoun(3596), GreenTea(3594), protocolocon(3580), jimrogerz(3574), _iouri_(3565), _Astek_(3557), felixcoto(3523), shangas(3510), Slin-.-(3508) 2010-11-28T17:43:12 i think from 10 october to 27 november - maybe 3-4 hour a day 2010-11-28T17:43:30 * 10 september 2010-11-28T17:43:48 hehe, that's why you're 350 ranks above me :-) 2010-11-28T17:43:55 =) 2010-11-28T17:44:04 i started working seriously on it from 1 Oct, almost all my free time dedicated to it (most of it thinking), too much for the result i finally expect 2010-11-28T17:44:36 maybe 3 hours a day, weekend plenty more 2010-11-28T17:44:59 yes, on weekens ~ 10 hours 2010-11-28T17:44:59 i followed too many dead ends 2010-11-28T17:45:17 and 90% of time is testing.. 2010-11-28T17:45:42 against all older versions on different mapsets.. 2010-11-28T17:45:49 wow, I've just been working on it since thursday, and I thought I put in a lot of time... 2010-11-28T17:46:16 ^_^ 2010-11-28T17:46:26 i'm quite obsesive when it is about programming :P 2010-11-28T17:46:27 :-) 2010-11-28T17:46:52 *** insectores has quit IRC (Quit: Page closed) 2010-11-28T17:46:53 took me a couple days to give up on hacking the starter package and rewrite from scratch, too 2010-11-28T17:47:25 I'm afraid that whoever wrote the C++ starter package did some things the hard way... 2010-11-28T17:47:54 testing and restesing minor changes has consumed most of the time 2010-11-28T17:49:14 I think if I was in top 50, I would have declared success and done something else :) 2010-11-28T17:49:15 *** Appleman1234 has quit IRC (Remote host closed the connection) 2010-11-28T17:50:05 *** Azrathud has quit IRC (Ping timeout: 245 seconds) 2010-11-28T17:50:37 *** davidjliu_ has joined #aichallenge 2010-11-28T17:51:22 *** davidjliu has quit IRC (Ping timeout: 265 seconds) 2010-11-28T17:53:24 but yeah I am looking forward to the next one, then I can be obsessed with it from the beginning like you guys :) 2010-11-28T17:54:14 might be hard to do something as cool as this challenge though... I like the game 2010-11-28T17:56:58 me too, it is worth the time and effort, and i agree, the game is very very cool, and real time rankings make it very addictive 2010-11-28T17:58:10 Top 10 players: bocsimacko(3678), Accoun(3606), GreenTea(3605), protocolocon(3590), _iouri_(3580), jimrogerz(3565), _Astek_(3565), felixcoto(3537), Slin-.-(3526), shangas(3517) 2010-11-28T18:02:44 *** albert has joined #aichallenge 2010-11-28T18:13:11 Top 10 players: bocsimacko(3701), Accoun(3627), protocolocon(3613), _iouri_(3608), GreenTea(3604), jimrogerz(3590), _Astek_(3569), felixcoto(3562), Slin-.-(3554), shangas(3546) 2010-11-28T18:15:34 the rankings seem higher than previously.. that make any sense? 2010-11-28T18:22:06 burny: higher? 2010-11-28T18:22:18 highest was 3400-3500 2010-11-28T18:25:10 *** Rubicon-|-Cross has quit IRC (Ping timeout: 250 seconds) 2010-11-28T18:26:48 *** Rubicon-|-Cross has joined #aichallenge 2010-11-28T18:27:09 holy crap 2010-11-28T18:27:12 accoun is 2??????????????????????? 2010-11-28T18:28:16 Top 10 players: bocsimacko(3713), Accoun(3646), protocolocon(3623), _iouri_(3615), GreenTea(3611), jimrogerz(3601), felixcoto(3577), _Astek_(3562), shangas(3559), Slin-.-(3553) 2010-11-28T18:28:58 why? 2010-11-28T18:29:13 it good 2010-11-28T18:32:19 *** phreeza has quit IRC (Read error: Connection reset by peer) 2010-11-28T18:33:01 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-11-28T18:38:02 woulda been nice to see how this game ends :\ http://ai-contest.com/visualizer.php?game_id=8970075 2010-11-28T18:38:14 *** Azrathud has joined #aichallenge 2010-11-28T18:43:13 Top 10 players: bocsimacko(3707), Accoun(3635), jimrogerz(3619), protocolocon(3617), GreenTea(3611), _iouri_(3593), _Astek_(3578), felixcoto(3572), Slin-.-(3570), shangas(3556) 2010-11-28T18:45:43 Account: http://tcp.zeroviz.us/canvas.cgi?game_id=33687 2010-11-28T18:45:50 Accoun: 2010-11-28T18:46:00 I keep misspelling your name, sry 2010-11-28T18:47:10 *** rebelxt_ has joined #aichallenge 2010-11-28T18:48:37 @rankings 18 2010-11-28T18:48:38 McLeopold: Top 18 players: bocsimacko(3707), Accoun(3635), jimrogerz(3619), protocolocon(3617), GreenTea(3611), _iouri_(3593), _Astek_(3578), felixcoto(3572), Slin-.-(3570), shangas(3556), george(3531), Neverstu(3496), CorwinAlex(3486), wagstaff(3481), joakim_mjardner(3478), jambachili(3477), Manwe56(3470), McLeopold(3466) 2010-11-28T18:48:46 whoohoo ^^ 2010-11-28T18:49:06 nice :) 2010-11-28T18:49:20 I think I'm just getting really lucky right now. 2010-11-28T18:49:49 I hear ya. Its fun anyways... 2010-11-28T18:50:41 I'm amazed at how quick the games are coming now. 2010-11-28T18:51:02 dmj111: how come your not top 10? I thought you maintained it the whole time? 2010-11-28T18:51:41 Could be many reasons, first of which that it was a fluke I was in the top 10. 2010-11-28T18:52:01 Since I resubmitted rarely, maybe that gave me an artificial inflation? 2010-11-28T18:52:19 I don't play defense, and maybe the new maps demand it... 2010-11-28T18:52:21 I thought that too, but janzert said they only used 3 days of results 2010-11-28T18:52:32 I do play defense... 2010-11-28T18:52:41 It comes first it my ordering 2010-11-28T18:52:51 *** medrimonia has joined #aichallenge 2010-11-28T18:53:12 when I say defense, I mean against possible attacks, not attacks already on the way. Is that what you mean? 2010-11-28T18:53:16 Maybe a high ranking get's you games with better people, so you don't lose to people way down in the rankings? 2010-11-28T18:53:42 No, I have defense fleets and security reserves. 2010-11-28T18:53:44 I think that may be it. I am suseptible to losing to anyone, but can manage to beat the higher bots too. 2010-11-28T18:53:55 Ah, I never got reserves to work. 2010-11-28T18:54:02 heh, I have successfully secured myself in the top 100 2010-11-28T18:54:05 I've gotten some big wins against the top ten so far. 2010-11-28T18:54:26 pythonistas, unite! 2010-11-28T18:54:35 Each order I have has a set of 3 numbers, fleets, reserves and boost. 2010-11-28T18:54:57 I match that against num_ships, potential_ships and security levels. 2010-11-28T18:55:16 *** JeroenDeDauw has quit IRC (Quit: Leaving.) 2010-11-28T18:55:19 wow... now I am more convinced my bot is a hack job. 2010-11-28T18:55:40 Even when someone gets more growth than me, it takes them a long time to take me down unless the do one planet at a time. 2010-11-28T18:55:43 give to dmj111 some points ;) 2010-11-28T18:56:07 GreenTea: thanks. we'll see after a few days though :) 2010-11-28T18:56:21 if not, I'll take what I can get. 2010-11-28T18:56:27 The problem is though, this game need positional analysis, and my bot just doesn't do that. 2010-11-28T18:56:57 My bot has lots of bad assumptions that made it hard to improve it much. 2010-11-28T18:57:17 davidjliu: do not rely on the position, check the elo difference instead 2010-11-28T18:57:18 *** sanddbox has quit IRC () 2010-11-28T18:57:20 I wish I found out about this think like a month earlier :\ 2010-11-28T18:57:54 *** sanddbox has joined #aichallenge 2010-11-28T18:58:17 Top 10 players: bocsimacko(3702), Accoun(3629), jimrogerz(3617), protocolocon(3611), GreenTea(3591), _iouri_(3587), _Astek_(3579), Slin-.-(3573), felixcoto(3572), shangas(3541) 2010-11-28T18:58:19 from me (69) to 100 there is a difference of 80 ELO 2010-11-28T18:58:35 it is a meaningless difference 2010-11-28T18:58:48 meneky: yeah, I think my bot still hasn't converged to where it will finish. hopefully at least :) 2010-11-28T18:58:58 but my objective of #1 in the US is looking less and less plausible 2010-11-28T18:59:09 i think my bots been sufficiently destroyed on tcp :> 2010-11-28T18:59:38 is anyone else plesantly surprised at how well their last-minute bot changes are performing? 2010-11-28T19:01:40 I was hoping for #2 python 2010-11-28T19:01:54 behind dmj11 O_o 2010-11-28T19:06:10 @gamestats 2010-11-28T19:06:11 McLeopold: gamestats = http://ai-contest.com/game_stats.php 2010-11-28T19:06:23 *** Rubicon-|-Cross has joined #aichallenge 2010-11-28T19:07:03 did the game stats start over for the finals? 2010-11-28T19:07:30 ye 2010-11-28T19:07:58 *** albert has quit IRC (Quit: Page closed) 2010-11-28T19:09:17 *** dmj111 has quit IRC (Remote host closed the connection) 2010-11-28T19:10:17 *** dmj111 has joined #aichallenge 2010-11-28T19:10:38 if 2 bots are basically identical(always tie against each other, and always win against people below them, and lose to people above them).. if one of them has more games against lower ranks(wins), would they be higher, same or lower than their counterpart? 2010-11-28T19:10:55 higher 2010-11-28T19:13:14 Top 10 players: bocsimacko(3714), Accoun(3640), jimrogerz(3608), protocolocon(3604), GreenTea(3599), _Astek_(3590), felixcoto(3579), _iouri_(3579), Slin-.-(3562), shangas(3552) 2010-11-28T19:15:39 =) Win again my main opponent: Manwe56 2010-11-28T19:17:45 Am I the only to win/loose by streak? 2010-11-28T19:18:53 timeout against #13, and I was winning. :( 2010-11-28T19:20:30 :( 2010-11-28T19:20:40 time outs are a pooy way to lose 2010-11-28T19:21:00 it'd be nice if timeout would just mean that all moves for that turn, aren't executed 2010-11-28T19:21:51 that would be fair I'd think 2010-11-28T19:21:52 I've never had any time out with my officials bots 2010-11-28T19:24:58 Ohhh! bocsimacko has lost against #400 2010-11-28T19:25:55 ;o 2010-11-28T19:26:10 bet that's the number 1 watched game of the next 5minutes :P 2010-11-28T19:27:02 Yes I think so^^ 2010-11-28T19:28:13 Top 10 players: bocsimacko(3697), Accoun(3644), protocolocon(3610), GreenTea(3604), jimrogerz(3600), _Astek_(3595), _iouri_(3580), felixcoto(3571), Slin-.-(3567), shangas(3561) 2010-11-28T19:28:23 that's gotta hurt though 2010-11-28T19:29:26 Normally, at the end, winning five or six games in a streak won't change too much the rankings right? 2010-11-28T19:32:15 correct 2010-11-28T19:32:24 " did the game stats start over for the finals?" finals? what is format of finals? 2010-11-28T19:32:29 the order you win in, doesn't matter.. only the relative goodness of the opponents you beat 2010-11-28T19:33:09 Ok good because the rankings are still changing a lot. 2010-11-28T19:34:53 from main page "Wednesday Dec 1, 2010: official results and final rankings will be released." nothing about some special final round. just final rankings.. 2010-11-28T19:36:46 green... new maps, and resetting of the ELO/rankings 2010-11-28T19:37:03 *** yasith has quit IRC (Ping timeout: 240 seconds) 2010-11-28T19:38:42 it'd be interesting to see how the results for the top100 would be, with the pairing method beign used, vs 100% random pairing 2010-11-28T19:40:24 they mentioned pairing cutoff will be lowered gradually. things should get more interesting 2010-11-28T19:40:35 oh. it must be I dont undertend the question is about past or about future) my bad english =) 2010-11-28T19:41:01 rebel, what's that mean? 2010-11-28T19:41:22 go more and more towards 100% random? or less random? 2010-11-28T19:41:42 freezing the last players right? 2010-11-28T19:41:45 no. just stop playing game for rankings over 1000, 500, 200, 100, etc 2010-11-28T19:41:51 what is "pairing cutoff"? 2010-11-28T19:41:51 medrimonia: yep 2010-11-28T19:42:07 bah, that doesn't help me.. 2010-11-28T19:42:31 I need for both me, and the other top 100s.. to play against more of the sub-100 bots 2010-11-28T19:42:37 pairing cutoff is a nice way to eliminate all the dual and rage bots and let higher ranks get more games 2010-11-28T19:43:04 pairing cutoff will be shown at http://ai-contest.com/game_stats.php (currently is None) 2010-11-28T19:43:15 Top 10 players: bocsimacko(3721), Accoun(3674), GreenTea(3631), _Astek_(3624), protocolocon(3623), jimrogerz(3622), _iouri_(3606), Slin-.-(3593), shangas(3586), felixcoto(3577) 2010-11-28T19:43:17 e.i. disabling low rated bots? 2010-11-28T19:43:40 yep 2010-11-28T19:43:46 correct. just not letting them play after a certain time 2010-11-28T19:43:50 Why no explanation on offical forum... 2010-11-28T19:44:03 there is one. let me find a link. 2010-11-28T19:44:16 janzert explained.. 2010-11-28T19:44:59 http://ai-contest.com/forum/viewtopic.php?f=20&t=1131#p7471 2010-11-28T19:46:42 thank you 2010-11-28T19:47:35 image crazy number of games we will be getting when pairing cutoff is eventually in place 2010-11-28T19:48:07 yep something like a game every minute 2010-11-28T19:48:12 when they drop it to top100 for instance, everyone would be getting two games per minute it seems 2010-11-28T19:48:40 more like 3 per min 2010-11-28T19:48:41 dont think so, top 100 bots are more likely to consume the entire second 2010-11-28T19:48:47 per turn 2010-11-28T19:48:53 err meant "imagine" 2010-11-28T19:49:23 *** yasith has joined #aichallenge 2010-11-28T19:49:30 meneky: good point. but still likely a game per 1-2 minutes. would be nice to watch nearly real time ;) 2010-11-28T19:49:40 yep :) 2010-11-28T19:55:45 the more i see my games... the more I think that I made a mistake in how to choose my neutrals after turn 10 or something like that 2010-11-28T19:56:27 the more I see my games, the more I think /cry, stop making me play against top50 bots so much 2010-11-28T19:57:55 medrimonia: fyi you beat my bot 48-2-25 on final map set. nice 2010-11-28T19:58:20 Top 10 players: bocsimacko(3733), Accoun(3685), GreenTea(3646), protocolocon(3638), _Astek_(3635), jimrogerz(3633), _iouri_(3616), Slin-.-(3613), shangas(3585), felixcoto(3576) 2010-11-28T19:58:25 =) thanks rebelxt_ 2010-11-28T19:58:35 medrimonia: for instance, iouri beats me 43-0-32. so you are actually the one i lose the most from all open sourced bots 2010-11-28T19:58:56 really? very good! 2010-11-28T19:59:22 open sourced? 2010-11-28T19:59:46 still bouncing around it looks like 2010-11-28T19:59:56 Yep a lot of people have open sourced their bots 2010-11-28T20:00:03 bots open sourced today http://ai-contest.com/forum/viewtopic.php?f=19&t=1168 2010-11-28T20:00:13 oh.. you mean just: shared the code, since final? 2010-11-28T20:00:26 today was lucky day.. 2010-11-28T20:00:28 yes 2010-11-28T20:00:44 rebel how did mine in the local tourney 2010-11-28T20:01:09 bhasker: running right now. did you end up submitting 0.23 or minmax? 2010-11-28T20:01:15 go to sleep. bye-bye! 2010-11-28T20:01:15 max 2010-11-28T20:01:23 didnt want to resubmit 0.23 2010-11-28T20:01:31 though it may have done better probably 2010-11-28T20:01:37 it was holding 40-50 much better than this one 2010-11-28T20:01:40 good night GreenTea 2010-11-28T20:01:52 ok test should be done in 5 minutes 2010-11-28T20:02:01 *** GreenTea has quit IRC (Quit: Page closed) 2010-11-28T20:02:06 cool 2010-11-28T20:02:57 sanddbox: ? 2010-11-28T20:03:05 http://www.ai-contest.com/visualizer.php?game_id=8984385 2010-11-28T20:03:09 i have no idea why my bot did that 2010-11-28T20:03:11 !! 2010-11-28T20:03:27 suddenly decided it needed to send 182 ships to another planet 2010-11-28T20:03:28 :-\ 2010-11-28T20:03:57 though e323 http://www.ai-contest.com/visualizer.php?game_id=8983436 2010-11-28T20:04:40 nice 2010-11-28T20:05:06 weird my bot decided to turn a draw into a loss lol 2010-11-28T20:05:10 :-\ 2010-11-28T20:06:15 it's just experimenting with different strategies ;-) 2010-11-28T20:06:19 hehe i wish 2010-11-28T20:06:20 it probably just thought that sending the 112, would soak up some of their 200, and then your other 200 can capture the side 2010-11-28T20:06:34 i don't have any such code, i think its a bug lol 2010-11-28T20:06:54 oh i thin i know what happened 2010-11-28T20:07:04 it sent the 112 which changed my staging planet 2010-11-28T20:07:05 to the right 2010-11-28T20:07:11 and it sent the ships to the staging 2010-11-28T20:07:12 bhasker: 40-31-4 2010-11-28T20:07:16 not bad 2010-11-28T20:07:48 i thought it would do much worse rebel 2010-11-28T20:07:58 wow upto 190 games a minute 2010-11-28T20:08:10 thats a game every 10 minutes? 2010-11-28T20:08:14 almost 2010-11-28T20:08:51 well, i did not do any tweaks last few days so this kinda indicates that new map set did not affect relative strength of our bots. good data point for both 2010-11-28T20:09:12 true 2010-11-28T20:09:16 neither did i tweak anything 2010-11-28T20:09:21 i kind of ran out of ideas 2010-11-28T20:09:39 short of a full rewrite 2010-11-28T20:09:43 to do a proper minmax 2010-11-28T20:10:03 hmm #games per minute is still climbing are we spinning up more servers 2010-11-28T20:10:06 upto 200 now 2010-11-28T20:10:45 mcleopold seems to be doing quite well 2010-11-28T20:10:46 #21 2010-11-28T20:10:54 i'm almost 50-50 with antimadroid on 200-map test... but significant difference in the rankings. indicates that local tests against other people's bots can still be very misleading ;-) 2010-11-28T20:11:06 he is doing great 2010-11-28T20:11:10 yea 2010-11-28T20:11:21 wow upto 16 now 2010-11-28T20:11:25 @rankings 20 2010-11-28T20:11:26 bhasker: Top 20 players: bocsimacko(3746), Accoun(3686), protocolocon(3654), GreenTea(3649), jimrogerz(3640), _Astek_(3634), Slin-.-(3633), _iouri_(3624), shangas(3586), felixcoto(3578), george(3578), goldman(3534), Manwe56(3524), wagstaff(3522), joakim_mjardner(3520), McLeopold(3513), oldman(3507), asavis(3507), medrimonia(3506), ChrisH(3504) 2010-11-28T20:12:04 e323 finally in the top 100 2010-11-28T20:12:05 #97 2010-11-28T20:12:06 bhasker: 200gpm - it's probably playing lowest rated bots now.. i.e. those which issue invalid orders on first turn ;-) 2010-11-28T20:12:15 rebel you are at 99! 2010-11-28T20:13:14 Top 10 players: bocsimacko(3746), Accoun(3686), protocolocon(3654), GreenTea(3649), jimrogerz(3640), _Astek_(3634), Slin-.-(3633), _iouri_(3624), shangas(3586), felixcoto(3578) 2010-11-28T20:13:18 the sad part though is that there are quite a few bots which haven't been updated in a month or more and are still in the top 50-60 2010-11-28T20:13:30 that shows that not much improvement happened in the last month:-\ 2010-11-28T20:13:31 bhasker: yeah but some terrible losses ;) i was stuck in 100-200 land for too long hehe 2010-11-28T20:13:35 *** Azrathud has quit IRC (Ping timeout: 260 seconds) 2010-11-28T20:14:08 ah k 2010-11-28T20:14:26 rebel you on east coast? 2010-11-28T20:14:42 central - austin 2010-11-28T20:14:46 ah 2010-11-28T20:14:52 anyone on the west coast here? 2010-11-28T20:14:55 in the bay area? 2010-11-28T20:15:27 if the number of losses to a lower-ranked bot effects the ratings much.. then higher ranked bots should all have to face lower bots, some fixed number of times(to balance it out, so some don't get screwed out of points from randomness) 2010-11-28T20:16:07 but the lower ranked bot is a moving target is it not? 2010-11-28T20:16:48 burny, i dunno if it's a good idea but yeah some good bots had a slow start due to getting stuck in the low rankings. i think when they start lowering pairing cutoff gradually, things will hopefully settle down eventually 2010-11-28T20:17:14 burny: it actually evens out in the end 2010-11-28T20:17:37 burny: if you are losing a lot of games to lower ranked bots then your bot isn't very good anyway 2010-11-28T20:17:41 ya, true.. if when it's only playing top 500 bots, it's more random, and less playing against equal rated bots.. it'll balance out fine 2010-11-28T20:17:52 burny: and if a higher ranked bot is able to hold its own in the higher ranks then it certainly deserves to be there 2010-11-28T20:18:05 jmc, lots of high ranked bots lose a few games to lowbies.. 2010-11-28T20:18:12 yeah i know it happens 2010-11-28T20:18:20 and also lots of high rankers, didn't have to play against any lowbies, due to randomness 2010-11-28T20:18:22 but nowhere near as often as the lower ranked bots 2010-11-28T20:18:38 those high rankers don't get to stay high rank for long 2010-11-28T20:20:58 looks like the low ranked bots are playing i don't see any movement in the top 100 right now 2010-11-28T20:21:26 asavis and bocs both lost to a 400ish ranked bot 2010-11-28T20:22:57 I'm sure a number of others have also 2010-11-28T20:23:55 ragebot test fail http://zeroviz.us:8080/canvas.cgi?game_id=33868 2010-11-28T20:24:14 the problem with my dual availability list 2010-11-28T20:24:20 screws up my neutral decision making 2010-11-28T20:24:30 i should have tested against ragebot lol 2010-11-28T20:25:41 *** iFire` has quit IRC (Read error: Connection reset by peer) 2010-11-28T20:26:03 *** iFire has joined #aichallenge 2010-11-28T20:26:58 heh i can't recall last time i tested against rage bot. getting 100% win ratio against all example bots felt challenging in the very beginning though 2010-11-28T20:27:10 *** Mathnerd314 has joined #aichallenge 2010-11-28T20:28:21 Top 10 players: bocsimacko(3765), Accoun(3703), protocolocon(3668), GreenTea(3661), _Astek_(3653), jimrogerz(3653), Slin-.-(3646), _iouri_(3640), shangas(3601), george(3590) 2010-11-28T20:35:22 horrible map? or is taking one of those 81's a good move? http://ai-contest.com/visualizer.php?game_id=8989329 2010-11-28T20:36:11 you should take a 81 I think 2010-11-28T20:36:24 hm, ya.. probably should 2010-11-28T20:36:47 the distance between you two is of more than eight and you have a growth of 12 2010-11-28T20:37:06 so you can take it 2010-11-28T20:37:15 it's not distance between bases that matters, it's base->target vs opponents base->target 2010-11-28T20:37:23 (But my bot wouldn't do it too I think) 2010-11-28T20:37:34 The two matters 2010-11-28T20:37:48 and the 81 isn't a 5grow 2010-11-28T20:38:06 No but you've already 12 of growth 2010-11-28T20:38:11 er, ya nm.. doesn't matter what it's grow is 2010-11-28T20:38:58 if the 81 is a distance of 5 vs 11 though.. means they can get there, before I get 80 extra units to it 2010-11-28T20:39:24 yep but I think there's more than eight of difference 2010-11-28T20:39:57 and at worst, you keep 82 ships in the little planet (2 growth) 2010-11-28T20:40:11 or even all your ships 2010-11-28T20:40:17 and here it has to be good 2010-11-28T20:40:39 But... that's human thinking, not bot thinking 2010-11-28T20:41:12 I don't think the differece is more than 8 2010-11-28T20:42:18 er, mayybe.. i estimate 6 vs 15, by appearance 2010-11-28T20:43:16 Top 10 players: bocsimacko(3767), Accoun(3714), protocolocon(3674), _Astek_(3657), GreenTea(3650), jimrogerz(3646), Slin-.-(3635), _iouri_(3631), george(3599), shangas(3596) 2010-11-28T20:43:50 I'll sleep, good luck all 2010-11-28T20:45:18 *** superflit has joined #aichallenge 2010-11-28T20:46:26 burny: and in any case, if you attack the 81, that means the enemy will be able to take the 68 in the middle, I think 2010-11-28T20:47:04 oh ya.. true 2010-11-28T20:47:08 weird game agains manwe: http://ai-contest.com/visualizer.php?game_id=8992224 2010-11-28T20:58:16 Top 10 players: bocsimacko(3785), Accoun(3727), protocolocon(3690), _Astek_(3671), GreenTea(3663), jimrogerz(3656), Slin-.-(3650), _iouri_(3642), george(3616), shangas(3606) 2010-11-28T21:02:36 lost about 16 of my last 20 games :( 2010-11-28T21:05:25 *** protocolocon has quit IRC (Quit: Leaving) 2010-11-28T21:07:12 *** yasith has quit IRC (Ping timeout: 276 seconds) 2010-11-28T21:11:00 *** Azrathud has joined #aichallenge 2010-11-28T21:12:38 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-11-28T21:13:16 Top 10 players: bocsimacko(3807), Accoun(3750), _Astek_(3696), protocolocon(3695), GreenTea(3681), jimrogerz(3679), Slin-.-(3678), _iouri_(3670), george(3643), shangas(3619) 2010-11-28T21:17:30 *** Azrathud has quit IRC (Ping timeout: 265 seconds) 2010-11-28T21:27:44 *** javagamer__ is now known as javagamer 2010-11-28T21:28:17 Top 10 players: bocsimacko(3825), Accoun(3767), protocolocon(3721), _Astek_(3713), GreenTea(3701), jimrogerz(3696), Slin-.-(3694), _iouri_(3686), george(3659), shangas(3625) 2010-11-28T21:28:29 *** javagamer has joined #aichallenge 2010-11-28T21:35:40 *** winexception has joined #aichallenge 2010-11-28T21:43:17 Top 10 players: bocsimacko(3828), Accoun(3771), protocolocon(3727), GreenTea(3705), _iouri_(3703), jimrogerz(3700), _Astek_(3700), Slin-.-(3694), george(3677), shangas(3642) 2010-11-28T21:58:17 Top 10 players: bocsimacko(3828), Accoun(3772), protocolocon(3716), _iouri_(3708), _Astek_(3706), jimrogerz(3705), Slin-.-(3703), GreenTea(3697), george(3683), shangas(3654) 2010-11-28T21:58:29 *** amstan_ has joined #aichallenge 2010-11-28T21:58:29 *** ChanServ sets mode: +o amstan_ 2010-11-28T22:00:05 *** amstan has quit IRC (Ping timeout: 245 seconds) 2010-11-28T22:05:07 *** rebelxt_ has quit IRC (Ping timeout: 265 seconds) 2010-11-28T22:07:04 *** Snowplan has joined #aichallenge 2010-11-28T22:10:22 watch out top 50, here I come... 2010-11-28T22:11:06 *** i_ito has joined #aichallenge 2010-11-28T22:13:22 Top 10 players: bocsimacko(3844), Accoun(3787), protocolocon(3731), jimrogerz(3724), _Astek_(3723), _iouri_(3722), Slin-.-(3713), george(3701), GreenTea(3697), shangas(3655) 2010-11-28T22:20:10 *** Accoun has quit IRC (Ping timeout: 260 seconds) 2010-11-28T22:21:00 *** Accoun has joined #aichallenge 2010-11-28T22:26:57 *** rebelxt_ has joined #aichallenge 2010-11-28T22:28:18 Top 10 players: bocsimacko(3848), Accoun(3798), protocolocon(3739), jimrogerz(3734), _Astek_(3727), _iouri_(3726), Slin-.-(3705), george(3705), GreenTea(3696), shangas(3653) 2010-11-28T22:36:04 *** Accoun has quit IRC (Ping timeout: 250 seconds) 2010-11-28T22:36:48 *** Accoun has joined #aichallenge 2010-11-28T22:42:43 *** tapwater has joined #aichallenge 2010-11-28T22:43:19 Top 10 players: bocsimacko(3861), Accoun(3797), jimrogerz(3750), _Astek_(3749), _iouri_(3747), protocolocon(3742), george(3728), Slin-.-(3723), GreenTea(3713), shangas(3660) 2010-11-28T22:44:07 maybe it's time to slow those announcements back down now that the rankings are starting to stabilize? 2010-11-28T22:45:52 *** rebelxt_ has quit IRC (Quit: Page closed) 2010-11-28T22:46:21 janzert: yeah, will do 2010-11-28T22:46:28 @scheduler remove rankings 2010-11-28T22:46:28 amstan_: Ready to serve. 2010-11-28T22:46:42 @scheduler add 2010-11-28T22:46:42 amstan_: (scheduler add ) -- Schedules the command string to run seconds in the future. For example, 'scheduler add [seconds 30m] "echo [cpu]"' will schedule the command "cpu" to be sent to the channel the schedule add command was given in (with no prefixed nick, a consequence of using echo). Do pay attention to the quotes in that example. 2010-11-28T22:47:48 @list Scheduler 2010-11-28T22:47:48 amstan_: add, list, remove, and repeat 2010-11-28T22:47:57 @scheduler repeat 2010-11-28T22:47:57 amstan_: (scheduler repeat ) -- Schedules the command to run every seconds, starting now (i.e., the command runs now, and every seconds thereafter). is a name by which the command can be unscheduled. 2010-11-28T22:48:13 @scheduler repeat rankings 3600 "[echo rankings]" 2010-11-28T22:48:14 Top 10 players: bocsimacko(3861), Accoun(3797), jimrogerz(3750), _Astek_(3749), _iouri_(3747), protocolocon(3742), george(3728), Slin-.-(3723), GreenTea(3713), shangas(3660) 2010-11-28T22:48:41 *** Snowplan has left #aichallenge 2010-11-28T22:53:59 Apparently I must have preselected maps to make Accoun win :P 2010-11-28T22:54:49 *** Appleman1234 has joined #aichallenge 2010-11-28T22:57:31 yes, you are making my hair grey 2010-11-28T22:57:38 what kind of announcements? 2010-11-28T22:58:38 i think just the contestbot top 10 announcements. 2010-11-28T23:00:05 I see 2010-11-28T23:00:45 yes 2010-11-28T23:01:16 I did just post in the forums on some more finals details people may be interested in 2010-11-28T23:05:38 cool, thank you 2010-11-28T23:10:31 new additional maps coming soon 2010-11-28T23:13:13 how many maps? 2010-11-28T23:13:25 another 75 2010-11-28T23:14:01 so it will be 150 total 2010-11-28T23:14:21 nice 2010-11-28T23:14:26 should make for interesting games 2010-11-28T23:14:35 btw can you add that counter back 2010-11-28T23:14:43 where it was showing how many games before you get the next game? 2010-11-28T23:14:48 that was kind of cool 2010-11-28T23:14:58 sorry, there isn't any easy way to make it work now 2010-11-28T23:15:01 oh 2010-11-28T23:15:02 :-( 2010-11-28T23:15:19 the pairing mechanism is not the same? 2010-11-28T23:15:31 *** Neverstu has joined #aichallenge 2010-11-28T23:15:54 pairing itself is but the way the first player is chosen changed, your next game isn't based on the time of the last game but how many games you have played 2010-11-28T23:16:10 *** i_ito has quit IRC (Ping timeout: 265 seconds) 2010-11-28T23:16:34 so one of the players with the least number of games is randomly chosen and on average that is somewhere a little under half of all the players 2010-11-28T23:17:20 I guess it could go by how many other players have the same or less number of games as you 2010-11-28T23:17:48 but that would be a pretty heavy query and I'm pretty sure too much load 2010-11-28T23:18:01 I'd rather try and keep the server playing as many games as possible 2010-11-28T23:18:22 ah k 2010-11-28T23:19:23 and rankings is updated every 10-15 minutes? 2010-11-28T23:20:10 currently 10 minutes, as the total number of games increases and it takes longer to calculate that will almost certainly change though 2010-11-28T23:24:28 ah k 2010-11-28T23:27:32 i beat iouri on official over night :) 2010-11-28T23:28:41 *** winexception1 has joined #aichallenge 2010-11-28T23:29:15 *** winexception has quit IRC (Ping timeout: 245 seconds) 2010-11-28T23:39:15 *** Neverstu has quit IRC (Quit: Leaving) 2010-11-28T23:41:27 gratz 2010-11-28T23:42:08 yeah, it's still making some really stupid moves 2010-11-28T23:42:31 so i don't think there's any chance of top 10, but it may hang around in the top 30 2010-11-28T23:45:59 i am kind of losing my inclination to write a proper minmax bot now that the competition is over:-\ 2010-11-28T23:47:15 hehe, the bot i uploaded isn't doing what i thought at all 2010-11-28T23:47:43 but it seems to work so meh aha 2010-11-28T23:47:43 nice 2010-11-28T23:47:51 burny: you beat me, you little bugger ! :P: 2010-11-28T23:47:56 :P 2010-11-28T23:48:06 my rankings is higher now though :) 2010-11-28T23:48:15 Top 10 players: bocsimacko(3885), Accoun(3804), _iouri_(3770), _Astek_(3761), protocolocon(3758), Slin-.-(3744), george(3730), jimrogerz(3727), GreenTea(3723), shangas(3673) 2010-11-28T23:48:18 ya, I keep getting matched with higher ranked bots 2010-11-28T23:48:26 and my bot sucks against higher ranked ones 2010-11-28T23:48:30 i have a win against iouri and a draw against accoun 2010-11-28T23:48:50 *** bhasker has quit IRC (Quit: bhasker) 2010-11-28T23:49:02 think my best still is a win against bix 2010-11-28T23:51:03 heh, still slowly rising 2010-11-28T23:51:11 davidjliu: :( 2010-11-28T23:51:14 you had a very rough start 2010-11-28T23:51:44 hopefully I finish moving up by the time the contest's over 2010-11-28T23:52:16 also, how do I get rid of the underscore after my name? 2010-11-28T23:52:30 type /nick newnick 2010-11-28T23:52:40 *** davidjliu_ is now known as davidjliu 2010-11-28T23:52:45 yay. danke 2010-11-28T23:52:48 :) 2010-11-28T23:55:31 I still have, like, an 80% winning percentage. It's pretty exciting 2010-11-28T23:55:52 that gets ruined once you're closer to the top :) 2010-11-28T23:55:59 http://ai-contest.com/profile_games.php?user_id=3945 2010-11-28T23:56:03 that's my current record :) 2010-11-28T23:56:17 i make too many stupid moves to get the top ten i think 2010-11-28T23:56:42 my bot has 2 flaws that's costing it about 90% of it's games :\ 2010-11-28T23:56:51 I lose too often against 400-ranked bots to get to the top ten I think 2010-11-28T23:56:53 as in twice 2010-11-28T23:56:53 heh 2010-11-28T23:57:03 even all kinds of top20 bots I've faced, i would have won, if not for those flaws 2010-11-28T23:57:31 dmj111: FUUUUUUUU, you beat me 2010-11-28T23:57:33 bocs and asavis both lost to some 400s 2010-11-28T23:58:48 *** medrimonia1 has joined #aichallenge 2010-11-28T23:59:09 *** medrimonia3 has joined #aichallenge 2010-11-28T23:59:09 antimatroid you've climbed upper than me ^^ 2010-11-28T23:59:32 *** medrimonia has quit IRC (Ping timeout: 240 seconds) 2010-11-28T23:59:41 :) 2010-11-28T23:59:49 take that!