2011-06-22T00:01:02 contestbot: later tell McLeopold something is still fishy with with the pairing game #31096 and #31118 look to be the same matchup played repeated 2011-06-22T00:01:02 janzert: Aye, aye, sir 2011-06-22T00:13:58 *** acieroid has quit IRC (Read error: Connection reset by peer) 2011-06-22T00:14:14 *** acieroid has joined #aichallenge 2011-06-22T00:31:42 *** Accout has quit IRC (Read error: Connection reset by peer) 2011-06-22T00:37:52 *** Accoun has joined #aichallenge 2011-06-22T00:40:11 *** okay is now known as okayzed 2011-06-22T00:44:29 *** Mathnerd314 has quit IRC (Ping timeout: 240 seconds) 2011-06-22T00:48:17 *** okayzed is now known as okay 2011-06-22T00:53:54 *** jmcarthur has quit IRC (Ping timeout: 252 seconds) 2011-06-22T01:02:21 *** Accoun has quit IRC () 2011-06-22T01:04:40 *** george1914 has quit IRC (K-Lined) 2011-06-22T01:04:40 *** antimatroid-uni has quit IRC (K-Lined) 2011-06-22T01:04:40 *** astevenson has quit IRC (K-Lined) 2011-06-22T01:05:58 *** okay is now known as okayzed 2011-06-22T01:07:59 *** jmcarthur has joined #aichallenge 2011-06-22T01:19:14 *** Accoun has joined #aichallenge 2011-06-22T01:44:18 *** jmcarthur has quit IRC (Ping timeout: 255 seconds) 2011-06-22T02:02:07 *** mcstar has joined #aichallenge 2011-06-22T02:06:45 *** jmcarthur has joined #aichallenge 2011-06-22T02:17:12 *** stocha has joined #aichallenge 2011-06-22T02:23:42 *** stocha has quit IRC (Ping timeout: 252 seconds) 2011-06-22T02:26:34 *** sigh has quit IRC (Quit: This computer has gone to sleep) 2011-06-22T02:31:52 *** Cyndre has quit IRC (Quit: Leaving) 2011-06-22T02:34:24 *** gobang has joined #aichallenge 2011-06-22T02:42:33 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-22T02:44:53 *** stocha has joined #aichallenge 2011-06-22T02:45:07 even antimatroid wasn't opposed to a few 1vs1 games .... 2011-06-22T02:45:24 *** gobang has joined #aichallenge 2011-06-22T02:50:44 *** Palmik has joined #aichallenge 2011-06-22T03:01:43 *** stocha has quit IRC (Quit: Page closed) 2011-06-22T03:10:27 *** carlop has joined #aichallenge 2011-06-22T03:20:28 *** onensora has joined #aichallenge 2011-06-22T03:23:06 *** ltriant has quit IRC (Quit: Computer has gone to sleep) 2011-06-22T03:24:32 *** onensora has quit IRC (Ping timeout: 240 seconds) 2011-06-22T03:32:14 *** Cyndre has joined #aichallenge 2011-06-22T03:33:17 *** gobang has quit IRC (Read error: Connection reset by peer) 2011-06-22T03:47:15 *** boegel has joined #aichallenge 2011-06-22T04:09:27 *** Naktibalda has joined #aichallenge 2011-06-22T04:20:15 *** onensora has joined #aichallenge 2011-06-22T04:58:30 *** _0x47 <_0x47!~slash@p5B12F978.dip.t-dialin.net> has joined #aichallenge 2011-06-22T05:00:31 *** aerique has joined #aichallenge 2011-06-22T05:16:31 is the new firefox 5 out? 2011-06-22T05:16:46 new firefox again? 2011-06-22T05:17:13 yes no? is the canvas image shadow bug fixed? 2011-06-22T05:17:35 has noone upgraded? is everyone using other browsers? 2011-06-22T05:17:42 chrome 2011-06-22T05:18:03 * Naktibalda has 4 browsers openned :) 2011-06-22T05:18:17 err no, safari is closed, I don't need it 2011-06-22T05:18:55 I got an email from my free email service, advertising firefox 5 - roughly translated to english: "smart as a fox, fast as firefox 5" 2011-06-22T05:20:06 with special emphasize on the speed improvement due to "special modifications on the so-called JavaScript and canvas elements" 2011-06-22T05:20:24 ie team sent them a cake 2011-06-22T05:20:56 now if they bought that speedup by making the canvas unusable it makes me mad ... a bit 2011-06-22T05:21:21 ask them? 2011-06-22T05:21:25 surely they have an irc channel 2011-06-22T05:22:01 that they do 2011-06-22T05:22:25 no I'll download that Â§$&"&)" and try myself :p 2011-06-22T05:23:38 *doh* they released with the bug 2011-06-22T05:25:17 the cake is a lie 2011-06-22T05:25:27 mleise: how nice of them :( 2011-06-22T05:28:32 ok now that the visualizer is barely usable in FF 5 what do you think I should do? 2011-06-22T05:29:13 provide a work-around for ff5 until the bug is fixed or just wait for the bugfix 2011-06-22T05:29:33 (the bug is in their bugzilla already) 2011-06-22T05:30:25 unless the work-around won't take much time i'd just wait for the bugfix 2011-06-22T05:32:58 mleise: with this reasoning ("i cant use my latest software, doh") many of could say, that the whole contest is f'ed, cause it runs on out-of-date ubuntu 2011-06-22T05:33:31 us left us out 2011-06-22T05:34:29 mcstar: there is a different between what stuff runs on the server and what people can do locally 2011-06-22T05:34:47 mleise: i dunno, i'm amazed they would release something with a known bug 2011-06-22T05:34:56 is it likely to screw many developers over? 2011-06-22T05:35:14 you have to write your program so that it runs the same on the server, languages and impementations change 2011-06-22T05:35:29 it isnt different 2011-06-22T05:36:32 antimatroid: the bug is in a feature neither ie nor chrome support, so it is quite possible that few developers make use of it. 2011-06-22T05:36:48 take a look at the ifpc contest or what the f ever that was called: they supported only a specific version of debian with strict package selection 2011-06-22T05:36:48 mleise: it's still really poor on their part imo 2011-06-22T05:37:03 they're knowingly causing a fair bit of grief to anyone who does use it and needs to care about what their users think 2011-06-22T05:37:12 antimatroid: yes, especially since it is a regression that an automated test case would have catched 2011-06-22T05:38:17 i would question them about that decision personally, but i don't really know what the bug is? 2011-06-22T05:38:22 then again it is an open-source project so i don't want to be that harsh 2011-06-22T05:39:05 the still hold a fairly large market share in the browser market, they should care somewhat about the people who develop stuff to work with their browser 2011-06-22T05:39:13 antimatroid: The bug is, if you render an image into the canvas with active drop shadow it now colorizes and blurs the whole image 2011-06-22T05:39:49 so with the medium gray shadow I use the images look gray and blurry, like through dirty glass 2011-06-22T05:40:14 is there an easy way to do it a different way? 2011-06-22T05:41:52 i would have to create a shadow in some gfx software and change my button class to load two images. it's more work than I want to put into that bug. Disabling shadows altogether would be another option 2011-06-22T05:43:23 *** okayzed is now known as okay 2011-06-22T05:43:59 mleise: i couldn't help myself and asked about it in #firefox, and now i have a question for you :P 2011-06-22T05:44:00 "does the image come from cross domain?" 2011-06-22T05:46:14 of course not :p 2011-06-22T05:46:54 they ask the basic questions, that's ok. but once I find the bug in the bugzilla I tell you what the status is 2011-06-22T05:47:28 cross-domain would be if i didn't load it from aichallengebeta.hypertriangle.com 2011-06-22T05:47:49 but all the files are local to that server 2011-06-22T05:55:14 The bug status is RESOLVED FIXED, but I'm not sure in which version 2011-06-22T05:55:23 ah ff6 2011-06-22T05:56:14 lol 2011-06-22T05:57:52 http://ai-contest.com/ down áëÿòü 2011-06-22T05:57:59 please.. 2011-06-22T05:58:01 i wonder what they do if there is a bug that allows hackers to take control of your computer 2011-06-22T05:58:47 ÃControl stack exhausted (no more space for function call frames). 2011-06-22T05:59:29 Accoun: yeah it might not be up for a while, the beta servers is still up 2011-06-22T05:59:40 http://aichallengebeta.hypertriangle.com/index.php 2011-06-22T05:59:50 ohh f i left out a 0-check 2011-06-22T06:00:03 antimatroid: dont feed the troll 2011-06-22T06:00:21 mcstar: he's a good competitor 2011-06-22T06:00:24 who says he is trolling this time? maybe he was away for ling 2011-06-22T06:00:30 *long 2011-06-22T06:00:33 now the challenge has really started 2011-06-22T06:00:41 Rabidus: pls explain to them im not in the mood 2011-06-22T06:00:49 he's not in the mood 2011-06-22T06:00:51 :D 2011-06-22T06:00:53 lol 2011-06-22T06:01:12 did someone open a window? its getting cold 2011-06-22T06:01:22 yeah, few days back he repeatedly said that site is down, and we told him that we know it but he kept repeating. 2011-06-22T06:01:29 it's winter, it's damn freezing! 2011-06-22T06:01:35 winter? lol 2011-06-22T06:01:38 mid-summer here :D 2011-06-22T06:01:52 next saturday we'll have a midsummer festival 2011-06-22T06:02:21 we had fÃªte de la musique here yesterday. there were 37 locations were local bands and djs played music. 2011-06-22T06:03:21 summer rocks \_/ 2011-06-22T06:03:31 ah well, we get summer over christmas and new years, which is awesome 2011-06-22T06:03:41 where are your fingers? \m/ 2011-06-22T06:03:55 sry, im new to pictograms 2011-06-22T06:04:09 christmas must have snow 2011-06-22T06:04:43 yea, snow and santa clause. pines. candles... ah 2011-06-22T06:05:03 we have a music festival from 29 Dec - 1 January, it's great 2011-06-22T06:05:16 christmas is a market strategy 2011-06-22T06:05:53 the three kings were business men in reality, they were on the way to sell things 2011-06-22T06:06:20 then their goods were stolen? 2011-06-22T06:06:26 yhyy... i want to australia 2011-06-22T06:06:41 :< 2011-06-22T06:06:41 poor jesus didnt have anything and they gave product samples to him 2011-06-22T06:07:56 sorry, for ruining your image of the main christian holiday 2011-06-22T06:08:05 :D 2011-06-22T06:11:24 *** Zaphus has joined #aichallenge 2011-06-22T06:15:43 *** okay is now known as okayzed 2011-06-22T06:21:34 tell me a joke 2011-06-22T06:21:52 ant joke 2011-06-22T06:23:11 *** george1914 has joined #aichallenge 2011-06-22T06:25:32 tcp server is down due cpu usage, george 2011-06-22T06:30:08 i've got one, wait a sec... 2011-06-22T06:31:01 http://woodcote.files.wordpress.com/2011/03/miracle.jpg 2011-06-22T06:31:42 i knew this wo ant 2011-06-22T06:31:43 s 2011-06-22T06:32:12 isnt this on xkcd? 2011-06-22T06:34:28 http://avionod.files.wordpress.com/2009/04/then-a-miracle-happens.gif?w=300&h=364 2011-06-22T06:34:34 not on xkcd 2011-06-22T06:40:27 *** antimatroid has quit IRC (Ping timeout: 258 seconds) 2011-06-22T06:41:12 ff 4.0.1-1 -> 5.0-1 2011-06-22T06:41:27 i dont care so ill upgrade 2011-06-22T06:41:51 did this mornings fix for the broken match scheduler get broken again ? 2011-06-22T06:41:56 My local python says this syntax is illegal, is that correct?: https://github.com/aichallenge/aichallenge/blob/epsilon/ants/ants.py#L89 2011-06-22T06:42:57 works on 2.7.1 2011-06-22T06:43:20 right, i'm using 2.6.6. what will the official server use? 2011-06-22T06:43:36 already 2.7 i guess 2011-06-22T06:44:48 aerique: you are among the organisers, right? 2011-06-22T06:45:07 mcstar: no 2011-06-22T06:45:27 mcstar: i just made the cl starter pack :) 2011-06-22T06:46:01 your name is the first below organization members on github/aichellenge thats why i thought 2011-06-22T06:46:25 mcstar: amstan got tired of my pull requests 2011-06-22T06:52:06 hmm - some country codes don't quite work: one says "Saint BarthÃï¿½Â©lemy" :-) 2011-06-22T06:53:12 Rabidus: hi. it anyway seemed yesterday that some more things needed doing to make the game engine run correctly ... so i guess we need to wait a while before we have fully functioning tcp server ? 2011-06-22T06:58:00 yes 2011-06-22T06:58:17 it would be nice if i just could leave it running and forget it like that :) 2011-06-22T06:58:35 now everything else starts lagging because it takes everything it can 2011-06-22T06:58:59 hm - sounds like you are doing valuable testing though :/ 2011-06-22T06:59:30 *** olexs has joined #aichallenge 2011-06-22T07:00:03 just running others code :p 2011-06-22T07:01:45 aye. well i'll stick to writing my bot for the time being, and will keep checking for when tcp code is good ... 2011-06-22T07:01:57 thanks for running it though. 2011-06-22T07:01:59 it really shouldnt use much CPU if it's coded right - it can sleep waiting for the bots to respond 2011-06-22T07:02:07 not that I could write one though :-) 2011-06-22T07:03:31 tcp is almost the important part of the contest (apart from final results) imo. the place you really test your bot against others, and improve/debug it more easily than on official server 2011-06-22T07:05:00 mcleopold said its not hard to mod the official engine to add tcp support to the sandbox, berak might just do it 2011-06-22T07:11:10 hope so - would be great to have it 2011-06-22T07:11:39 *** FireFly has joined #aichallenge 2011-06-22T07:12:02 mcstar: you were telling me about lisp yesterday? 2011-06-22T07:12:22 yes? 2011-06-22T07:12:48 well mcleopold is pro 2011-06-22T07:13:01 "you just dip dap dappaduu, and it's ready" 2011-06-22T07:13:03 ;) 2011-06-22T07:13:06 george1914: sorry if i upset you 2011-06-22T07:13:10 lisp is/was something i have looked at a few times, wondering about using/learning it ... 2011-06-22T07:13:14 :) 2011-06-22T07:13:33 though i have never encountered a lisp user 2011-06-22T07:13:36 wondering is bad, its wasted time: do or do not 2011-06-22T07:13:47 and i say do :D 2011-06-22T07:13:51 no, no. that leads to lots of wasted time 2011-06-22T07:13:56 planning is important 2011-06-22T07:14:07 planning is for the weak 2011-06-22T07:14:15 no im kidding 2011-06-22T07:14:16 you should see the wasted efforts i did in the last 2 contests, on clever plans that did not work 2011-06-22T07:14:34 im sure youre much wiser now 2011-06-22T07:14:35 and maybe should not have started 2011-06-22T07:14:36 planning takes just a week 2011-06-22T07:14:44 :) 2011-06-22T07:14:55 fuu, back to work 2011-06-22T07:14:55 -< 2011-06-22T07:15:53 mcstar: so you have used lisp for the clever stuff? - modifying own code during running, etc? 2011-06-22T07:16:19 why would i want that? 2011-06-22T07:16:41 or maybe we arent talking about the same thing 2011-06-22T07:16:48 you mean the REPL or simply eval? 2011-06-22T07:17:05 (i had a brief look at bocsimako's code from last contest - it seemed he used it just because he knows the language - he could have done the same thing in fortran, c++, or whatever) 2011-06-22T07:17:31 well, i guess the question is how comfortably 2011-06-22T07:17:36 well, i thought the big deal in lisp was that you could manipule lisp code with lisp code (easily) 2011-06-22T07:17:59 you can know the popular languages well, some things nevertheless will always look ugly 2011-06-22T07:18:07 so potentially very powerful for AI 2011-06-22T07:18:08 lisp helps with those usually 2011-06-22T07:18:31 thats true, and its not just ai that it is powerful in 2011-06-22T07:18:40 but you should elaborate on your question 2011-06-22T07:18:45 the first one 2011-06-22T07:18:53 modifying own code during running 2011-06-22T07:19:47 *** stocha has joined #aichallenge 2011-06-22T07:19:59 still there? 2011-06-22T07:20:08 hey guys i had this new crazy idea bout how to make games shorter. 2011-06-22T07:20:16 stocha: very nice 2011-06-22T07:20:22 yeah 2011-06-22T07:20:31 (but i havent even told it yet!) 2011-06-22T07:20:35 ok, go ahead 2011-06-22T07:20:39 current top bot seems to start with peacefullness in mind 2011-06-22T07:20:51 but in 1vs1 it would probably be bad ... 2011-06-22T07:20:51 ok, simple stuff: doing genetic programming. i imagine it is easy for the master 'evolving program' to mutate the 'candidates' in lisp, than it would be in c++ (where eg i might need to compile each 'candidate' on the run) 2011-06-22T07:21:07 So make 10% of the game be 1vs1, and then people will have to tune their aggressivity 2011-06-22T07:21:20 ha ha ha. okay it's not funy. sorry :p 2011-06-22T07:22:00 george1914: does that kind of "genetic" programming actually work ? 2011-06-22T07:22:08 george1914: in that example, as i beleive aerique can confirm, yes its easier, you can directly compile as in assemble code that you can execute, without resorting to writing a full minilanguge on top of C/whatever 2011-06-22T07:22:37 and the whole thing would fit into lisp very well 2011-06-22T07:22:38 hm, does it work (so books tell me - i have not done it). i believe it to be so, for some problems 2011-06-22T07:22:48 but thats not all 2011-06-22T07:23:21 for example i had in mind two different interpretations of your first question 2011-06-22T07:23:41 the first one relates to the repl 2011-06-22T07:24:06 its the interface to the language, when running interactively 2011-06-22T07:24:12 what kind of problem does it work ? (when you tweak actual instruction flow, and not just some parameters) 2011-06-22T07:24:53 you can change basically a running program rededfine functions, classes, everything updates accordingly 2011-06-22T07:25:25 and related to this lisp has got a very powerful error handling capability 2011-06-22T07:25:35 @later tell janzert Current top bots seems to be peacefull at the start. Think having 10% of game 1vs1 might incit more aggressiv behaviors (hence shorter games) 2011-06-22T07:25:35 stocha: Aye, aye, sir 2011-06-22T07:25:54 you can basically recover from any error(except segfaults that happen in the core(i beleive)) 2011-06-22T07:26:19 and the stack is not unwound as they say, that is all the nested calls remain in place 2011-06-22T07:26:25 so you say that you can develop/debug/test faster in lisp than in compiled language? 2011-06-22T07:26:34 stocha: make the games really short, first one to find some food wins ! :-) 2011-06-22T07:26:37 lisp is compiled 2011-06-22T07:26:50 i can develop/debug faster in java then i can in lisp 2011-06-22T07:26:58 (it depends on implementation, but for example sbcl always compiles) 2011-06-22T07:27:10 Zaphus: bleh 2011-06-22T07:27:17 :-) 2011-06-22T07:27:22 i want there to be a few 1vs1 games 2011-06-22T07:27:29 the nearest i know to this is eg mathematica or matlab, where i can run each function from command line, and inspect/modify variables from the command line 2011-06-22T07:27:42 thats not the same 2011-06-22T07:28:11 the other interpretation is that you can literally make up code as your program runs 2011-06-22T07:28:21 ok. (i found that useful in mathematica). but lisp has more than that? 2011-06-22T07:28:23 stocha: I'd like some of the other things fixed first, the change to the matching algorithm sucks now. But I don't see any reason we can't have 2-bot maps 2011-06-22T07:28:31 there is this tiny tiny map, i wouldn't have test my bot 1vs1 on. And yet they manage to crank 5 bots in there :p I wonder what the ranking for this particular map looks like 2011-06-22T07:28:42 george1914: here are my first steps into genetic programming using lisp: http://aerique.blogspot.com/2011/01/baby-steps-into-genetic-programming.html 2011-06-22T07:28:47 but you dont have to use strings or something else to represent it, you can use the structure that lisp uses otherwise: linked lists 2011-06-22T07:29:01 Zaphus: what's wrong with the current matching ? 2011-06-22T07:29:16 stocha: it's playing the same sets over and over, with very little variety 2011-06-22T07:29:18 i have seen some games between close strength bot. I'm happy with that 2011-06-22T07:29:21 george1914: ant there is another possibility which is related : macro expansion 2011-06-22T07:29:27 t->d 2011-06-22T07:29:33 george1914: best thing is to use lisp for a while and see if you like it 2011-06-22T07:29:41 Zaphus: it only play close games ? 2011-06-22T07:29:50 Zaphus: or worst, some do never play ? 2011-06-22T07:29:55 george1914: and bug mcstar and me (and perhaps people on #lisp) for help 2011-06-22T07:30:19 george1914: it's a bit of a love-hate language 2011-06-22T07:30:28 stocha: only close games - and that means you play the same opponents over and over until someone manages to move up or down a spot. - no variety any more 2011-06-22T07:30:39 aerique - thanks for that. i have looked at the GP field guide also. 2011-06-22T07:30:40 i guess the people who really hate it just didnt try hard enough 2011-06-22T07:30:44 Although variety is of course a cool addition, i would be happy whith anthing from 10% to 90% games being "close" games 2011-06-22T07:31:06 george1914: i'm working on a video for McLeopold for using CL in this challenge 2011-06-22T07:31:10 well, since we're probably at 90%, are you happy ? 2011-06-22T07:31:13 *screencast 2011-06-22T07:31:14 Zaphus: it seems fair though. fairer than pure random i would say 2011-06-22T07:31:17 george1914: let me give you a nice example 2011-06-22T07:31:21 ill look it up 2011-06-22T07:31:30 i say that i write (encapsulated) fortran in any/every language. if i were to learn lisp, it would have to be because it allowed me to do more than that ... 2011-06-22T07:31:34 http://aichallengebeta.hypertriangle.com/profile.php?user=93 this guy wouldnt think so (ignore the fact that all those bots suck) 2011-06-22T07:31:44 Zaphus: i'm happier with how it is now, that what it used to be. Then maybe 50% / 50% would be okay 2011-06-22T07:32:11 or, 4/5 bots close, one outsider dropped in - something like that perhaps ? 2011-06-22T07:32:28 look at his site under examples: http://common-lisp.net/project/parenscript/tutorial.html 2011-06-22T07:32:44 you can do whatever you want with the remaining 50% 2011-06-22T07:32:52 from pure random, to anything in between :p 2011-06-22T07:33:07 it show the power of lisp: you can embed html/javascript in it, and serve it on a webserver directly 2011-06-22T07:33:10 s 2011-06-22T07:33:29 i guess pure random as it merits. or you split 25% 25% and go for 25% with on or two random outsiders, and 25 with pure random 2011-06-22T07:33:49 people in the middle are better off, they will get matches against people above and below - but the person at the top is playing the same 9 opponents every game now it seems (or a subset of them) 2011-06-22T07:34:12 Zaphus: your outsider idea is good i guess. 2011-06-22T07:34:24 aerique: are you familiar with racket? 2011-06-22T07:34:53 mcstar: very little, i'm doing The Little Schemer in it currently 2011-06-22T07:35:06 is that a tutorial? 2011-06-22T07:35:18 Zaphus: let's say we go for 25% close game. Then 25% one outsider. and then 50% for other pairing schemes :p 2011-06-22T07:35:26 Zaphus: have a deal ? 2011-06-22T07:35:29 mcstar: no, a nice short book about recursion 2011-06-22T07:35:44 Zaphus: (yeah i realize we have no authoritie, but it's nice to agree on something still :p ) 2011-06-22T07:35:51 stocha: :-) yeah 2011-06-22T07:36:25 mcstar: http://junix-linux-config.googlecode.com/files/The.Little.Schemer.4th.2.pdf 2011-06-22T07:36:28 Zaphus: send a word to people with power maybe :p 2011-06-22T07:36:28 mcstar: wooops 2011-06-22T07:36:33 aerique: i look at your github again, and there was that auxesia stuff, which language was the easier to implement it? 2011-06-22T07:36:48 mcstar: http://www.amazon.com/Little-Schemer-Daniel-P-Friedman/dp/0262560992 2011-06-22T07:37:03 mcstar: aucesia? 2011-06-22T07:37:07 mcstar: eh, what?> 2011-06-22T07:37:13 auxesia 2011-06-22T07:37:18 i probably should be back at having a start with this new bot i want to test. 2011-06-22T07:37:39 why is it so not-fun to implement ideas ? 2011-06-22T07:37:44 the GAHElloWorld stuff 2011-06-22T07:37:50 stocha: A mix of methods would be good, not sure how you make sure that all bots get involved in all mixes, leave that to someone smart 2011-06-22T07:37:53 you used that as a namespace 2011-06-22T07:38:23 Zaphus: still i'll be very happy with a large number of game being close one :) 2011-06-22T07:38:29 mcstar: i did? 2011-06-22T07:38:46 i think i had way more information about how the consest might go on, with those late games , than in the whole previous weeks. 2011-06-22T07:38:53 mcstar: ah right, i only did the CL part of that 2011-06-22T07:39:01 ah, ok i didnt know 2011-06-22T07:39:02 stocha: how are the games your bots get at the moment ? going the way you'd expect ? 2011-06-22T07:39:15 Zaphus: ? 2011-06-22T07:39:21 i was expecting a bit of a comparison :) 2011-06-22T07:39:24 are you getting close games ? 2011-06-22T07:39:25 i don't know. i don't have a bot really at the moment. 2011-06-22T07:39:26 mcstar: sorry :) 2011-06-22T07:39:36 Zaphus: i think i do yeah. 2011-06-22T07:39:50 aerique: so did you try to run the engine) 2011-06-22T07:39:51 didn't check to hard for my bots. But it seems close game are frequent now. 2011-06-22T07:39:56 ? 2011-06-22T07:40:00 mcstar: i did some GP stuff in C++ recently, but i can't compare that very well either since it's been 10 years since I did c++ 2011-06-22T07:40:12 mcstar: no, not yet 2011-06-22T07:40:22 top 10 seems pretty similar to before, but the order has changed a bit 2011-06-22T07:40:34 Zaphus: then i see that new submission have no touble reaching their rank. Like flag capper, or my new (old) FastRef bot (stoha02 last update) 2011-06-22T07:40:38 george1914: what do you think about the link i gave you? 2011-06-22T07:40:52 mcstar: except optimizing it was as much as a drag as it is for cl 2011-06-22T07:41:06 stocha: yeah, I think I was wrong that it might take a while to settle, it seems to get there fast right now 2011-06-22T07:41:13 Zaphus: the order is sensible to the bot population. Now aggressive bots (like my reference bot) get crush pretty quick in high strength games. 2011-06-22T07:41:50 mcstar: hm - i want to do AI, am less interested in web apps... 2011-06-22T07:42:07 i guess you are showing me that you can do mainstream things in lisp? 2011-06-22T07:42:15 no 2011-06-22T07:42:26 i wanted to point at its fexibility 2011-06-22T07:42:29 but i would learn it for the things it can do that other languages cannot 2011-06-22T07:42:30 with l 2011-06-22T07:42:46 george1914: http://www.gigamonkeys.com/book/ 2011-06-22T07:42:47 hm 2011-06-22T07:42:50 i guess you dont appreciate it cause you dont know what is going on there... 2011-06-22T07:42:51 Zaphus: for example davidjlui is now a top bot. Although i don't think it was in the past :p 2011-06-22T07:42:52 my fault 2011-06-22T07:43:10 mcstar: maybe so - if so, my apologies 2011-06-22T07:43:16 np 2011-06-22T07:43:39 stocha: yes, it's interesting what wins now... especially that my dumber older bot does better than my current one that I thought fought better 2011-06-22T07:43:42 siebel's book is good, loop for blackbelts is a recurring chapter for me :D 2011-06-22T07:44:00 george1914: most languages can do what others can, question is how hard do they make it for you 2011-06-22T07:44:07 Zaphus: i think i'm pretty happy with the ranking right now. I could have enough data that i can try to build a bot that will aim for top rank. If i had patience for that :p 2011-06-22T07:44:45 Zaphus and Zaphus test are same ranking aren't they ? 2011-06-22T07:45:02 Zaphus: what is your older bot, what it the newest one ? 2011-06-22T07:45:10 close, yes - but the test one has no 'attack' code - it's older 2011-06-22T07:45:22 Zaphus: the top game, a very noisy. It's hard to understand what's going on, during the melting phase. 2011-06-22T07:45:39 Zaphus: they are same strength it appears :p 2011-06-22T07:46:25 stocha: yes, very noisy games.. those two are probably the same strength because the attack opportunities dont come up as much as I'd hoped (the way I have coded it) 2011-06-22T07:46:44 the current difference between stocha and Sto03ReplirCrush is that they run the same code, but after 600 turns, Sto03 transforms itself into the refGreedyBot. They have same strength. That give me confidence in the ratings. 2011-06-22T07:46:49 aerique: in c++ i need to call compiler if i chage code in a function - i could do that i guess, but would be -very- difficult/flaky etc 2011-06-22T07:47:06 C-c C-c 2011-06-22T07:47:10 sure one can say 'turing complete' or something, but in practical terms ... 2011-06-22T07:47:14 Zaphus: adding 1vs1 games would make the challenge more interesting though, don't you think ? 2011-06-22T07:47:38 george1914: turing complet has practical meaning. It will be hard to debut for example :) 2011-06-22T07:47:44 debug 2011-06-22T07:48:17 aye - i mean that i agree with aerique that one can do anything. but in practice one cannot 2011-06-22T07:48:21 stocha: perhaps, I've done local testing of that by creating a "sleeperbot" that just sits there, then I can effectively do 1v1 . I see no reason to prevent 1v1, although my bots tend to do better in larger battles for some reason 2011-06-22T07:48:23 george1914: turing complete is the same as saying that you can make program arbitrarily impossible to understand what they are doing, i believe. 2011-06-22T07:49:04 Zaphus: so, now i only have 99 more people to convince of that :p Who's next ? 2011-06-22T07:49:09 as i undertood it, it meant i could reproduce any other computer program in it 2011-06-22T07:49:27 Zaphus: would you send a word to mcleopold that you crave for 1vs1 to exist too ? :p 2011-06-22T07:49:37 Has someone added an "end the game early" rule ? - I don;t understand why this game ends when it does: http://aichallengebeta.hypertriangle.com/visualizer.php?game=31850&user=225 2011-06-22T07:49:52 Zaphus: yeah. 2011-06-22T07:50:04 turing complete essentially means, it is capable of doing any deterministic calculation 2011-06-22T07:50:08 Zaphus: a game end if 90% of the ants are owned by one player or by the food player. 2011-06-22T07:50:23 i wasn't trying to play the turing complete card :-| 2011-06-22T07:50:30 mcstar: why deterministic ? 2011-06-22T07:50:31 mcstar: aerique: thanks for your comments - nice to encounter some lisp people. i'll have at look at the links you sent. 2011-06-22T07:50:38 and they have to hold that 90% for 100 turns or something ? 2011-06-22T07:50:49 Zaphus: ah yeah maybe. 2011-06-22T07:51:06 just think of quantum computers: they are non-deterministic non-turing computing device 2011-06-22T07:51:08 s 2011-06-22T07:51:16 george1914: i'll post the screencast to the forums when it's complete 2011-06-22T07:51:26 stocha: so why does this end: http://aichallengebeta.hypertriangle.com/visualizer.php?game=31922&user=93 (other than the fact that it is boring) 2011-06-22T07:51:35 aerique: sure, i know. we are instead debating quite 'how hard is it to do it in this langauge'? 2011-06-22T07:51:36 all bots have the same number of ants in that 2011-06-22T07:52:16 Zaphus: food player won with 90% more ants than everyone else for 100 turns. 2011-06-22T07:52:25 aerique: i'll look out for it 2011-06-22T07:52:46 stocha: whats "food player" ? 2011-06-22T07:52:49 george1914: yes. i find lisp fights me the least (and bockimacko said the same in his blog i think) 2011-06-22T07:52:52 Zaphus: nice one that blooby 2011-06-22T07:52:59 Zaphus: it's the food . 2011-06-22T07:53:22 stocha: oh, right :-) lots of food = stop the game because nobody is really playing 2011-06-22T07:53:33 Zaphus: never saw a film with vandam ? Vandam hits the tree, and breaks his leg. The master come, and proclaim the tree is the winner. 2011-06-22T07:54:12 isnt that jackie chan? 2011-06-22T07:54:15 stocha: MrBlobby is another of my testing bots, totally different algorithm (barely enough code to register as an algorithm) - I'm trying to add some variety in there :-) 2011-06-22T07:54:15 Zaphus: is mister bloby one of yours ? 2011-06-22T07:54:28 quick answer if there was one :) 2011-06-22T07:54:41 nanotime lapse answer :p 2011-06-22T07:55:09 Zaphus: yeah i do that too. Though it's a fact i already had the bots ready 2011-06-22T07:55:39 I wanted to get a 'clumping' bot in there, rather than all the spread out ones... although I suspect it won't get too high on the ladder 2011-06-22T07:55:46 Zaphus: i wanted to do a blob bot. i always kinda of failed. This one looks nice. Although maybe you will need a better start (i didn't look at it) 2011-06-22T07:56:01 Zaphus: mine are like 20 in rank 2011-06-22T07:56:23 Sto03 is 20 and stocha is 19 2011-06-22T07:56:37 so that's a lesser bound of how high a blobber bot should aim. 2011-06-22T07:58:00 nice, MrBlobby is unaware of enemy ants so he won't get that high, but if we can get more blob ones in there then that may give smarter people ideas 2011-06-22T07:58:25 Zaphus: aren't we the smarter people in there ? 2011-06-22T07:58:49 i'm a strong believer in diversity. you probably are too :p 2011-06-22T07:58:59 :-) 2011-06-22T07:59:27 do you mind sharing the internals of the blobby part ? 2011-06-22T08:00:00 it seems distracted by food 2011-06-22T08:00:42 is it like pick up random blob point, and go to closest one, unless there is food ? 2011-06-22T08:00:45 it's pretty simple - for each ant it tries all 4 directions and scores the new position. food distracts it, but otherwise it likes to be next to it's friends, and it likes to go to places that are not currently visible 2011-06-22T08:01:07 oh it's a heavy heat map calculation then :p 2011-06-22T08:01:24