2011-02-08T00:12:53 aichallenge: McLeopold epsilon * r033a33a / ants/ant_map.py : Added render text map. Added fog of war stuff. - http://bit.ly/gAfgHc 2011-02-08T00:16:15 jmcarthur: http://i.imgur.com/BNPo7.png 2011-02-08T00:17:09 i really like how you can make it hard for them to invase a slightly wider passage, but if they come in right, they'll get a better in ratio 2011-02-08T00:17:24 kill* ratio 2011-02-08T00:26:17 hmm, why are you using the euclidean distance metric when you move in a manhattan? 2011-02-08T00:26:23 or am I reading it wrong? 2011-02-08T00:37:26 mcleopold wanted that 2011-02-08T00:37:51 just think like attacks can go in any direction 2011-02-08T00:38:07 with a max dist of 2 2011-02-08T00:38:41 antimatroid: still none where option 1 and the iterative version are different? 2011-02-08T00:38:58 hmm, then wouldn't the Chebyshev metric be better? 2011-02-08T00:39:01 http://en.wikipedia.org/wiki/Chebyshev_distance 2011-02-08T00:39:18 jmcarthur: 1 and 3? 2011-02-08T00:39:20 yeah 2011-02-08T00:39:22 i'm just about to try and find one 2011-02-08T00:39:31 i found a cool other one, hang on 2011-02-08T00:40:18 dist = max(|x2 - x1|, |y2 - y1|) 2011-02-08T00:40:50 not with euclidean 2011-02-08T00:40:57 that's manhatten 2011-02-08T00:41:06 no it is Chebyshev 2011-02-08T00:41:08 ^^ 2011-02-08T00:41:19 whatevs :P 2011-02-08T00:41:22 in manhatan you add delta x and delta y 2011-02-08T00:41:32 oh, i see 2011-02-08T00:41:37 in Chebyshev it is like moving the king on a chess board 2011-02-08T00:41:40 manhatten has a distance of 2 for the diagonals 2011-02-08T00:41:44 jmcarthur: http://i.imgur.com/RYUYS.png bottom left 2011-02-08T00:42:14 i quite like that distance metric for fights 2011-02-08T00:42:23 i still don't like option 2 though 2011-02-08T00:42:28 but i'll play around with that 2011-02-08T00:43:06 i'm coding up options 1 and 3 so i can just brute force them to find a case where they are different 2011-02-08T00:43:16 well, haven't started yet, but i'm about to 2011-02-08T00:43:37 with bottom left and option 2 it doesn't make sense for reds extra ant to cause him to lose both 2011-02-08T00:44:13 yeah, that's like the ab.c..d...e....f situation 2011-02-08T00:44:23 yep, but with the 2 fight region 2011-02-08T00:44:27 i made it off that example 2011-02-08T00:45:11 so you are using euclidean distance of 2 in these? 2011-02-08T00:45:25 yeah 2011-02-08T00:45:27 k 2011-02-08T00:45:32 making sure i use the right one 2011-02-08T00:45:35 i like what chris_0076 suggested 2011-02-08T00:45:50 diagonals = 1 away for fighting 2011-02-08T00:45:56 so it forms a square? 2011-02-08T00:46:29 well, you have the same fighting squares, but for option 2, bottom left, the red at distance 2 from green would survive 2011-02-08T00:46:29 i've suggested many times to do it that way :P 2011-02-08T00:46:42 oh i see 2011-02-08T00:46:44 because green is then 1 from blue 2011-02-08T00:46:48 easiest way to think of it is probably as a chess kings move distance 2011-02-08T00:46:59 but green could still kill red if blue wasn't there 2011-02-08T00:47:17 well i'll just run with euclidean for now 2011-02-08T00:49:56 *** AlliedEnvy has quit IRC (Read error: Connection reset by peer) 2011-02-08T00:51:05 if we had a working engine this would be a lot easier :P 2011-02-08T00:51:14 *** AlliedEnvy has joined #aichallenge 2011-02-08T00:51:51 jmcarthur: check second bottom right 2011-02-08T00:52:15 if blue comes in with anything but that, i think red is safe, but if blue does that, they take out 4 of reds with 3 of theirs 2011-02-08T00:52:17 that's cool 2011-02-08T00:52:51 this is for option 1, with option 2 blue can just walts down the middle with 1 and destroy you 2011-02-08T00:55:46 jmcarthur: i'm not convinced dimkadimons isn't just option 1 atm 2011-02-08T00:55:54 i think it's not the same 2011-02-08T00:56:10 but we'll see 2011-02-08T00:56:13 i'm trying to come up with an example that comes out differently atm 2011-02-08T00:56:25 so am i, once i get these actually implemented 2011-02-08T00:56:38 if after brute forcing it i can't find anything, then we can safely declare them the same 2011-02-08T00:56:47 but i kind of doubt it 2011-02-08T00:57:57 i might have one 2011-02-08T00:58:00 give me a minute 2011-02-08T01:02:18 yeah, i got one 2011-02-08T01:03:04 well fine, make my work for nothing :P 2011-02-08T01:03:22 :P 2011-02-08T01:03:47 http://i.imgur.com/HkMmT.png 2011-02-08T01:03:52 bottom left 2011-02-08T01:04:40 i see 2011-02-08T01:04:40 is his not more expensive computationally? 2011-02-08T01:04:46 much more 2011-02-08T01:04:48 i don't think we ever really supported scheme 2011-02-08T01:04:57 i think 1 or 2 2011-02-08T01:05:04 i think 2 with chris's distance metric 2011-02-08T01:05:07 i should play with that 2011-02-08T01:05:07 i can find references to three different interpreters... 2011-02-08T01:06:41 antimatroid: honestly, even if we were to go with this slower version, i bet option 1 would make a decent enough approximation for reasoning about most situations 2011-02-08T01:07:01 jmcarthur: yeah, or even if you just ran it on what you can see, i expect it to work most often 2011-02-08T01:07:19 but i don't see how it's any better? it's slower, almost always comes out the same, and does make it possible that you don't resolve correctly 2011-02-08T01:07:25 surely option 1 dominates option 3 2011-02-08T01:07:29 so then it's 1 or 2? 2011-02-08T01:08:02 option 1 will tend to kill more ants than option 3 is the main difference i guess 2011-02-08T01:08:26 very rarely, but sure, i don't think option 3 can kill more 2011-02-08T01:08:29 option 2 kills *way* more than either of them though 2011-02-08T01:08:32 Zannick: look at the languages.sql file to see a dump of the database table with the languages actually activated 2011-02-08T01:08:34 but isn't the point of option 2 that it kills more? 2011-02-08T01:08:36 http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/languages.sql 2011-02-08T01:08:38 well, one of jeff's points 2011-02-08T01:08:51 janzert: i know 2011-02-08T01:09:02 i think the point of jeff liking option 2 is that it makes good micro less powerful 2011-02-08T01:09:04 i wasn't sure how old that was, but it was what i was using 2011-02-08T01:09:23 any other languages in the compiler and such weren't actually being used 2011-02-08T01:09:28 because jeff thinks that micro is somehow restricted to awesome programmers 2011-02-08T01:09:28 yeah 2011-02-08T01:09:34 good micro being powerful is one of my biggest interests 2011-02-08T01:09:38 me too 2011-02-08T01:09:42 he and i are conflicted there 2011-02-08T01:10:03 *** dsockwell has joined #aichallenge 2011-02-08T01:10:10 he doesn't care if it interested you because you're one of the elites rather than a newb 2011-02-08T01:10:20 and that's going to be his approach to all of us 2011-02-08T01:10:36 i think the game should be attractive to the more experienced people 2011-02-08T01:10:48 as do i 2011-02-08T01:11:05 in fact, if i was a newb, i'm pretty sure i would be less attracted to something that doesn't attract the elite 2011-02-08T01:11:15 sure it needs to be simple enough for someone to get started, but the longer the path for improvement the better imo 2011-02-08T01:11:38 we're just circlejerking now :) 2011-02-08T01:12:00 yeah, not very productive 2011-02-08T01:13:43 are there are two-player cases where options 1 and 3 are different? 2011-02-08T01:13:55 i might finish this code just to test that if we don't come up with something before i'm done 2011-02-08T01:14:14 the two player cases are going to be the very common ones 2011-02-08T01:14:20 Zannick: that languages.sql is sligthly outdated it looks like the only one it's missing though is Scala 2011-02-08T01:14:34 ah 2011-02-08T01:14:56 is the command to run it "scala MyBot"? that's all i need 2011-02-08T01:15:07 yep 2011-02-08T01:15:14 cool, that's what i have 2011-02-08T01:17:01 here's a current dump http://pastebin.ca/2044450 2011-02-08T01:18:09 okay i got option1 implemented 2011-02-08T01:18:24 jmcarthur: get it in the repo 2011-02-08T01:18:31 amstan_: it's in haskell... 2011-02-08T01:19:20 janzert: thanks 2011-02-08T01:20:05 i made another post in the ant kill thread 2011-02-08T01:20:44 jmcarthur: i'll try to work out a 2 player instance of it 2011-02-08T01:21:08 i think you might be able to just chante the green to red and blue 2011-02-08T01:23:13 jmcarthur: yeah that example can be turned into 2 player and still work, 2011-02-08T01:23:20 ah okay 2011-02-08T01:23:29 that might be better to include in the graphic then. less misleading 2011-02-08T01:23:36 i will put both 2011-02-08T01:23:41 mmkay 2011-02-08T01:25:38 http://i.imgur.com/NQoRu.png 2011-02-08T01:25:40 there you go 2011-02-08T01:27:39 okay cool 2011-02-08T01:28:06 honestly, my intuition would say that that blue one is dead meat anyway 2011-02-08T01:28:29 whereas with option 3 *none* of the blue ants die 2011-02-08T01:29:03 then again... there actually appears to be more chances for micro with option 3 because of this sort of thing... assuming you can reason about it effectively 2011-02-08T01:29:16 big assumption i guess 2011-02-08T01:30:21 i very much think 3 is inferior to 1 with their closeness, i will play around with 2 with chebyshev, but i still love 1 2011-02-08T01:32:34 jmcarthur: i think i missed one that's different with 3 to 1 2011-02-08T01:32:43 you know the one you corrected me on yesterday? 2011-02-08T01:32:46 go take a look at that 2011-02-08T01:33:01 the zig zaggy one? 2011-02-08T01:33:05 i think blue loses 2 more of its ants 2011-02-08T01:33:06 yeah 2011-02-08T01:33:36 i think you are right 2011-02-08T01:33:37 which means it can kill more than option 1 2011-02-08T01:33:57 interesting 2011-02-08T01:34:00 kind of weird. i guess my intution about it was wrong 2011-02-08T01:34:47 when implementing it, you'd just decrement a killed ants neighbours enemy count if it's not to die as well 2011-02-08T01:35:01 still expensive 2011-02-08T01:35:15 well, the one i'm doing right now is going to be the naive, inefficient way 2011-02-08T01:35:25 because i don't feel like tweaking it forever 2011-02-08T01:35:29 yep 2011-02-08T01:36:10 it looks like the naive version of option 3 is roughly twice the code as a fairly naive implementation of option 1 2011-02-08T01:36:57 at least based on my guess of how much more code it will take by the time i'm done 2011-02-08T01:37:31 yeah i would expect that 2011-02-08T01:37:38 i said in the forum post it's harder to deal with the algorithm 2011-02-08T01:38:12 sometimes the more efficient version is actually shorter than the naive version, for some algorithms, though 2011-02-08T01:38:20 i just don't want to bother, is all 2011-02-08T01:40:28 yeah, shorter isn't always easiest 2011-02-08T01:40:40 well, looks like i got a bug in it 2011-02-08T01:40:49 it's actually not quite twice as long 2011-02-08T01:42:17 oh, haha. i implemented option2 by mistake 2011-02-08T01:42:19 amstan: have you looked at the new ant pics? 2011-02-08T01:42:22 hahaha 2011-02-08T01:42:24 that's hilarious 2011-02-08T01:42:37 okay, option3 next :P 2011-02-08T01:42:43 i mean... i meant to do that! 2011-02-08T01:47:46 *** sigh has quit IRC (Ping timeout: 240 seconds) 2011-02-08T01:48:01 *** McLeopold1 has joined #aichallenge 2011-02-08T01:48:39 amstan_: I'm looking at the aichallenge code on github, and noticed that your name produces a link, but mine doesn't. 2011-02-08T01:48:43 Any idea why? 2011-02-08T01:49:21 McLeopold1: make sure you have your git config setup properly 2011-02-08T01:49:35 there's an email address which you insert into git.. 2011-02-08T01:49:42 it has to match the one github has of you 2011-02-08T01:50:02 has to be a valid email too 2011-02-08T01:50:10 don't try to attach a non valid email to your account 2011-02-08T01:50:10 that's the "git config --global user.name" stuff? 2011-02-08T01:50:10 McLeopold1: there are more sample images in the forum thread now, also with stuff for dimkadimons option 2011-02-08T01:50:21 yes 2011-02-08T01:50:26 and user.email? i think 2011-02-08T01:50:47 antimatroid: I'm going to ignore the battle resolution discussions for a few days. I want to focus on getting the engine done and a tcp server up. 2011-02-08T01:50:56 Yeah, I've set both of those. 2011-02-08T01:51:12 is the some other option? 2011-02-08T01:51:24 McLeopold1: do they match the one on github? 2011-02-08T01:51:46 McLeopold1: i see you're appearing both as scott hamilton and mcleopold 2011-02-08T01:51:48 antimatroid: okay, option 3 actually took only slightly more code than option 1 2011-02-08T01:51:52 and none of them use the right email 2011-02-08T01:52:06 and i implemented it pretty poorly 2011-02-08T01:52:43 My first commits were as "Scott Hamilton", but I changed it to try and get the links to work. 2011-02-08T01:53:12 Right now i'm "McLeopold" and "mcleopold@..." 2011-02-08T01:53:13 yes.. make sure you have mcl...@gmail.com in github's email addresses 2011-02-08T01:53:19 I do 2011-02-08T01:53:32 is it verified? 2011-02-08T01:53:58 McLeopold1: if so.. idk then, try submitting to github support or ask in #github 2011-02-08T01:54:12 whoops, I didn't...it was just for the ssh stuff 2011-02-08T01:54:13 arg 2011-02-08T01:55:26 jmcarthur: another reason for me to prefer 1 to 2 :P 2011-02-08T01:55:41 jbroman: python should be compiled as well, 2 reasons: speed at startup, sytax errors are emailed back to the user 2011-02-08T01:55:59 there's a way to make pyc files from py, without actually executing them 2011-02-08T01:58:01 antimatroid: option 2 has been the all around worst possible scenario in pretty much every way, so far 2011-02-08T01:58:08 IMO 2011-02-08T01:58:30 i agree :\ 2011-02-08T01:58:35 we just need to convince jeff of that 2011-02-08T02:00:44 amstan_: what OS do you use? 2011-02-08T02:00:51 ubuntu 2011-02-08T02:01:07 I was thinking I might try that. 2011-02-08T02:01:37 Just to get use to the environment we will run the servers on. 2011-02-08T02:01:52 probably not much different from fedora or gentoo... 2011-02-08T02:04:06 * antimatroid needs to switch to linux some time 2011-02-08T02:07:43 antimatroid: what do you run? 2011-02-08T02:08:04 xp :P 2011-02-08T02:08:12 it's done me well 2011-02-08T02:08:15 wow, at least i'm on win7 2011-02-08T02:08:22 i like xp 2011-02-08T02:08:45 I do too, but I'm pleased with win7 2011-02-08T02:09:09 i prefer the extra performance 2011-02-08T02:09:10 I just wish I could run make files and have real command line tools 2011-02-08T02:09:18 as a sanity check i just verifed that there are no 1 vs 2 scenarios where options 1 and 3 are different. i'm now checking for two teams of two each. i doubt there will be any differences here either 2011-02-08T02:09:29 yeah, i'm sick of things being harder on windows 2011-02-08T02:09:29 this is on a 10x10 grid 2011-02-08T02:09:39 hmm, 11x11 actually, because i suck 2011-02-08T02:09:54 -1 to 9? 2011-02-08T02:09:56 haha 2011-02-08T02:09:58 0..10 2011-02-08T02:10:10 um, what do you mean no 1vs2 scenarios? 2011-02-08T02:10:18 A vs. B B 2011-02-08T02:11:57 okay i shrunk it down because i got tired of waiting 2011-02-08T02:12:26 on a 8x8 grid, there are no scenarios where A A vs. B B end up being different when comparing option 1 to option 3 2011-02-08T02:12:53 try aaaavbb? 2011-02-08T02:13:00 actually, there's lots of those 2011-02-08T02:13:05 because mine can be varied heaps 2011-02-08T02:13:16 i'm trying to find the smallest case 2011-02-08T02:13:25 i'll try aaa vs bb 2011-02-08T02:14:15 okay, it found many differences for aaa vs. bb 2011-02-08T02:14:25 let me pick one of these out and diagram it 2011-02-08T02:15:01 yeah i was about to say, 3x2 is easy 2011-02-08T02:15:51 err, i made a mistake actually 2011-02-08T02:16:21 yeah, this is an error in my test case generator 2011-02-08T02:16:29 not filtering out an invalid scenario 2011-02-08T02:16:33 i'm less sure about my example 2011-02-08T02:17:12 a.b.a 2011-02-08T02:17:12 ..a.b? 2011-02-08T02:17:23 a . row between* 2011-02-08T02:17:35 might not be needed actually 2011-02-08T02:18:03 aichallenge: McLeopold epsilon * r47dbdf3 / ants/McMaps.py : Trival change to test username link. - http://bit.ly/g5eMjq 2011-02-08T02:18:12 i think that works 2011-02-08T02:18:38 you have 2011-02-08T02:18:38 1.3.2 2011-02-08T02:18:38 ..2.2 2011-02-08T02:18:46 i suck at life. i was generating only invalid scenarios instead of only valid ones 2011-02-08T02:18:53 hehehe 2011-02-08T02:18:54 i'm gonna try 2v2 again 2011-02-08T02:19:03 *** McLeopold1 has left #aichallenge 2011-02-08T02:19:07 i don't think 2v2 will happen 2011-02-08T02:19:21 i think i just got one with the smallest possible set :P 2011-02-08T02:20:50 okay, no 2v2 2011-02-08T02:21:10 *tons* of 3v2 2011-02-08T02:22:54 http://i.imgur.com/UhJdw.png 2011-02-08T02:22:57 bottom left again 2011-02-08T02:23:22 mine is much more compact ;) 2011-02-08T02:23:34 i could make that more compact 2011-02-08T02:23:36 abba;a 2011-02-08T02:23:45 that's with euclidean 2011-02-08T02:23:48 2 2011-02-08T02:24:00 i haven't worked it out yet. it's just been reported to me 2011-02-08T02:24:06 yeah that looks right 2011-02-08T02:24:18 i'll draw it up for you 2011-02-08T02:24:57 i can do even more compact 2011-02-08T02:25:39 that situtation above can happen too 2011-02-08T02:25:54 it'd possible to come up with situations that can't actually happen 2011-02-08T02:26:15 .a. 2011-02-08T02:26:15 .b. 2011-02-08T02:26:15 c.d 2011-02-08T02:26:15 is such an example 2011-02-08T02:30:06 http://i.imgur.com/6K7OC.png 2011-02-08T02:30:10 bottom right is your example 2011-02-08T02:31:08 no 3 vs 1 examples, btw 2011-02-08T02:31:16 duh 2011-02-08T02:31:21 :) 2011-02-08T02:31:28 *** delt0r_ has quit IRC (Read error: Operation timed out) 2011-02-08T02:31:36 *** delt0r_ has joined #aichallenge 2011-02-08T02:34:20 just over half of all scenarios in a 2x3 grid with 3 vs. 2 are different between option 1 and option 3, but of course not all those scenarios can actually happen due to the range thing 2011-02-08T02:34:40 still, that's interesting i think 2011-02-08T02:34:54 wait, waht about 2011-02-08T02:34:54 bab? 2011-02-08T02:35:02 that proportion should shrink with larger grids 2011-02-08T02:35:05 i think 2011-02-08T02:35:12 about about bab? 2011-02-08T02:35:21 *what about 2011-02-08T02:35:36 don't worry, i'm being silly 2011-02-08T02:36:20 just trying to get a feel for *how* different they are, that's all 2011-02-08T02:36:30 because until now we've felt they were very close to the same 2011-02-08T02:36:37 it's more different than i thought before 2011-02-08T02:43:27 i think the sample bots that come with the starter packages should have somewhat varying levels of difficulty this time 2011-02-08T02:43:36 so people have a local challenge to begin with as well 2011-02-08T02:43:48 not so difficult that a beginner putting in a decent effort can't exceed the quality though 2011-02-08T02:44:02 yeah, a path to improvement that's easy to see without even uploading 2011-02-08T02:44:09 let's face it, more experienced people aren't going to really care about those anyway 2011-02-08T02:44:48 downside to having a tough sample bot is that everybody will be uploading that one, so there will be a tougher curve at the lower end on the server 2011-02-08T02:45:00 no, you don't release the source for them? 2011-02-08T02:45:03 that was my intention 2011-02-08T02:45:22 huh? 2011-02-08T02:45:39 just release binary files for some sample bots 2011-02-08T02:45:49 of 6,375,600 valid positions on a 5x5 grid for 3 vs. 2, 335844 of them have different results between option 1 and option 3 2011-02-08T02:46:16 that's quite a lot 2011-02-08T02:46:26 about 5% 2011-02-08T02:46:31 a lot 2011-02-08T02:46:39 which is definitely a lot considering how many of those positions are probably not even fights 2011-02-08T02:46:57 not big ones involving all players at least 2011-02-08T02:47:07 s/players/ants/ 2011-02-08T02:47:08 can you count what proportion had 1 kill more than 3? 2011-02-08T02:47:11 *** sigh has joined #aichallenge 2011-02-08T02:47:31 would be some work 2011-02-08T02:47:42 not really important 2011-02-08T02:47:43 you mean where 1 ant killed more than 3? 2011-02-08T02:47:54 i'm more interested in what behaviour each rule encourages than overall deaths 2011-02-08T02:47:58 ah 2011-02-08T02:48:08 where option 1 killed more than option 3/vice versa 2011-02-08T02:48:14 oh 2011-02-08T02:48:24 i can do that pretty easily 2011-02-08T02:48:51 there are now 4 options :\ 2011-02-08T02:48:57 oh? 2011-02-08T02:49:00 if we count the two metrics for option 2 2011-02-08T02:49:04 ah 2011-02-08T02:49:27 i think pick between 1 and 3 and each of the two's, then pick between those 2011-02-08T02:49:32 how to pick, bleh 2011-02-08T02:50:09 actually, i have no idea how the second 2 option workds yet 2011-02-08T02:52:14 in 198576 of those scenarios, option 1 killed more than option 3 2011-02-08T02:52:23 oh 2011-02-08T02:52:25 wrong 2011-02-08T02:52:28 backward 2011-02-08T02:52:37 option 3 killed more than option 1 in that many 2011-02-08T02:53:36 interesting 2011-02-08T02:54:18 yeah, didn't expect that 2011-02-08T02:54:35 that means that option 3 actually killed more in about 60% of the cases where they were different 2011-02-08T02:54:58 137268 where option 1 killed more 2011-02-08T02:55:43 so option 3 can cause big changes in kill ratio? 2011-02-08T02:55:51 that means there were no cases where option 1 killed ant A but not ant B but option 3 killed ant B but not ant A 2011-02-08T02:56:12 their kills always overlapped, but one is still not a subset of the other in every case 2011-02-08T02:56:48 largest kill ratio! 2011-02-08T02:56:50 :P 2011-02-08T02:56:51 go 2011-02-08T02:57:05 maybe even save the example 2011-02-08T02:57:07 i'd be keen to see it 2011-02-08T02:58:04 hmm 2011-02-08T02:58:36 largest absolute kill ratio or largest difference between option 1 and option 3? 2011-02-08T03:00:53 both? :P 2011-02-08T03:01:44 i think i got another one wrong 2011-02-08T03:02:12 oh shit, i think you can block the 3 passage with 3 2011-02-08T03:02:25 but not with 1? 2011-02-08T03:02:40 not with 1 2011-02-08T03:02:48 go to 3rd row from bottom, right side 2011-02-08T03:02:51 it's wrong atm 2011-02-08T03:02:57 i think red keeps all their ants with option 3? 2011-02-08T03:03:08 you remove the blue 4, then each red has 1 blue enemy? 2011-02-08T03:03:34 which distance metric? 2011-02-08T03:03:53 what do you mean? it's the 2 diamond 2011-02-08T03:03:54 they all are 2011-02-08T03:04:05 i thought you said you were using euclidean 2011-02-08T03:04:10 that's for option 2 2011-02-08T03:04:13 oh 2011-02-08T03:04:15 hence the rt(2)'s 2011-02-08T03:04:17 then i'm doing all these wrong 2011-02-08T03:04:35 there shouldn't be a difference with manhatten and euclidean for 1 and 3? 2011-02-08T03:04:44 not sure 2011-02-08T03:04:45 the fight region is the same, and that's all it determines? 2011-02-08T03:05:10 okay, none of these scenarios varied by more than 2 kills, and 137268 of them varies by exactly 2 kills 2011-02-08T03:05:14 *varied 2011-02-08T03:05:23 i guess so 2011-02-08T03:05:31 i think you're safe on that one 2011-02-08T03:05:33 it's a diamond with 2 euclidean? 2011-02-08T03:05:38 yeah 2011-02-08T03:05:38 i hadn't thought about it 2011-02-08T03:05:40 okay 2011-02-08T03:05:46 i realised that with snowman game 2011-02-08T03:05:52 if you go larger it isn't 2011-02-08T03:06:29 but.. is that bad that they can block that passage? 2011-02-08T03:06:49 okay, so the number of cases where the difference was exactly 2 kills is exactly the same as the number of cases where option 1 had more kills than option 3. i doubt that's a coincidence 2011-02-08T03:07:20 that's a pretty big passage to be blocking 2011-02-08T03:07:28 are you sure that would be impenetrable? 2011-02-08T03:07:32 look at it 2011-02-08T03:07:37 you can't send ants in there to break it 2011-02-08T03:07:44 any other variation and they all die straight up 2011-02-08T03:07:52 it would be an awesome trap with option 1 2011-02-08T03:08:12 yeah i think you're right 2011-02-08T03:08:49 with option 1 it's breakable and with option 3 it's not 2011-02-08T03:10:07 i think you can even flow your ants through that hole and remain stable 2011-02-08T03:10:23 it's like a one way portal 2011-02-08T03:10:27 that's actually kind of awesome 2011-02-08T03:10:46 2 way if you're a friend 2011-02-08T03:12:14 it's neat, but probably bad for the contest anyway :( 2011-02-08T03:12:30 the existence of one that large hints that there may be an even larger one 2011-02-08T03:12:40 i don't think so 2011-02-08T03:12:46 i could be wrong 2011-02-08T03:12:52 i have been over and over again 2011-02-08T03:16:15 *** Palmik has joined #aichallenge 2011-02-08T03:17:31 simple case where option 3 kills more than option 1: AB-AB;-----;-A--- 2011-02-08T03:17:50 option 1: A---B;-----;-A--- 2011-02-08T03:18:00 option 3: A----;-----;-A--- 2011-02-08T03:18:15 acutally, i think they can block the 4 passage too 2011-02-08T03:18:21 rule 3 is definitely out 2011-02-08T03:18:25 i guess we already had an even smaller case for that 2011-02-08T03:18:36 oh, maybe they can't, hang on 2011-02-08T03:18:56 i'm not sure. i thought about the blocking a bit more and i don't think it should be possible to block any larger 2011-02-08T03:19:30 ah, no we didn't have a simpler case already for option 3 killing more than option 1 2011-02-08T03:19:57 so there's a simpler one than the one we had already (which was a 5v5) 2011-02-08T03:20:24 yeah, pretty sure they can block 4 2011-02-08T03:20:28 i guess the corridor made a 3v4 one 2011-02-08T03:20:31 dang 2011-02-08T03:21:08 i should go to bed soon 2011-02-08T03:21:11 WWb..bWW 2011-02-08T03:21:11 Wa.bb.aW 2011-02-08T03:21:11 W.aaaa.W 2011-02-08T03:21:18 i'm pretty convinced against option 3 now 2011-02-08T03:21:53 oh wait, they could take out some of your things there 2011-02-08T03:21:57 it's not trivial to block it 2011-02-08T03:22:59 bleh, i am going around in circles, move middle a squares down and it may be blocked 2011-02-08T03:23:19 ugh 2011-02-08T03:23:53 even if it takes active moves to do it, I would guess if it can be done on a 4 wide corridor it can be done for any width corridor? 2011-02-08T03:24:47 http://i.imgur.com/5X6jq.png 2011-02-08T03:24:50 i think that's blocked 2011-02-08T03:24:53 or linear space, the only question to form an impenetrable fortress then is to find a way to turn a corner 2011-02-08T03:25:11 janzert: you might be right 2011-02-08T03:26:18 antimatroid: stick one blue ant to the immediate right of the leftmost red ant and stick another blue ant above that one 2011-02-08T03:26:40 jmcarthur: an ant couldn't get that close 2011-02-08T03:26:58 okay, get rid of the lower blue ant then 2011-02-08T03:27:02 ah i see 2011-02-08T03:27:04 nevermind 2011-02-08T03:27:13 yeah, that's blocked 2011-02-08T03:27:23 i think you can do that for any size given space 2011-02-08T03:27:32 it could turn into a game of blocking off large regions 2011-02-08T03:27:35 option 3 is out 2011-02-08T03:28:46 are we sure option 1 is fine for this one? 2011-02-08T03:28:51 although it would be fun to hole someone into a corridor from both ends 2011-02-08T03:29:00 i might just be tired, but i'm not thinking of how to attackit 2011-02-08T03:29:03 *attack it 2011-02-08T03:29:07 i really should head to bead 2011-02-08T03:29:09 *bed 2011-02-08T03:29:11 gah 2011-02-08T03:29:20 hmm, i'll find something that breaks it 2011-02-08T03:30:12 jmcarthur: you can take out red's edge pieces 2011-02-08T03:30:24 put blue ones diagonally above each 2011-02-08T03:30:42 so they can block it pretty well, but it's breakable 2011-02-08T03:30:51 i'm not seeing it 2011-02-08T03:30:57 hang on, i'll do the numbers 2011-02-08T03:31:11 ah i see one 2011-02-08T03:32:12 it takes two ants to kill each side ant 2011-02-08T03:32:21 then you go in with more to take the rest 2011-02-08T03:32:28 oh, maybe 2011-02-08T03:32:46 i think that's blocked 2011-02-08T03:32:48 crud 2011-02-08T03:32:51 no no 2011-02-08T03:33:02 bah, i'll just run it through my algorithm and be sure :P 2011-02-08T03:33:07 aichallenge: Benjamin S Wolf zeta * rc2bcc70 / (worker/language.py worker/submission.py worker/compiler.py): 2011-02-08T03:33:07 aichallenge: Another overhaul of compilation system. Moved compiler.py into 2011-02-08T03:33:07 aichallenge: language.py and did a major refactor of the compilers dict 2011-02-08T03:33:07 aichallenge: into a languages dict. 2011-02-08T03:33:07 aichallenge: Now language maps to compiler rather than other way around, 2011-02-08T03:33:07 aichallenge: and the run commands are listed in the same place. 2011-02-08T03:33:08 aichallenge: Credit for the form of the languages dict goes to jbroman. - http://bit.ly/fvVQ7l 2011-02-08T03:33:09 aichallenge: Benjamin S Wolf zeta * re4acb86 / worker/README.md : Modify README to reflect the refactor. - http://bit.ly/hdZHBQ 2011-02-08T03:33:12 omg 2011-02-08T03:33:42 wow did that take too long 2011-02-08T03:33:51 yeah, okay, those red corners do die under 1 2011-02-08T03:35:22 http://i.imgur.com/0zBs2.png 2011-02-08T03:35:23 i'm really going to bed now 2011-02-08T03:35:25 easy to break 2011-02-08T03:35:46 yeah, i need to sleep 2011-02-08T03:35:49 night 2011-02-08T03:35:51 wow, took out more than i thought it would 2011-02-08T03:35:58 ah i see why 2011-02-08T03:36:04 cool 2011-02-08T03:36:20 okay, option 1 is still my definitive favorite of the bunch 2011-02-08T03:36:24 gnite 2011-02-08T03:36:33 night 2011-02-08T03:42:56 *** mega1 has joined #aichallenge 2011-02-08T03:47:20 *** antimatroid has quit IRC (Ping timeout: 265 seconds) 2011-02-08T04:01:01 *** antimatroid has joined #aichallenge 2011-02-08T04:04:05 damnit mcleopold i have a question for him 2011-02-08T04:04:17 i need to make sure that if 3 friend ants walk into the same square, he kills all 3 2011-02-08T04:10:45 janzert: http://i.imgur.com/CMwVE.png 2011-02-08T04:12:02 with option 3, i think you basically want to attack with an inverse duck thing 2011-02-08T04:20:36 *** Naktibalda has joined #aichallenge 2011-02-08T04:33:41 *** p4p4p4 has joined #aichallenge 2011-02-08T04:42:44 *** amstan_ has quit IRC (Remote host closed the connection) 2011-02-08T04:57:51 *** boegel has joined #aichallenge 2011-02-08T05:12:29 there is a way to break through with 3 though? 2011-02-08T05:14:11 otherwise I think 3 is dead 2011-02-08T05:15:13 *** boegel has quit IRC (Remote host closed the connection) 2011-02-08T05:15:45 janzert: with 3 what? 2011-02-08T05:15:45 *** boegel has joined #aichallenge 2011-02-08T05:15:48 i think 3 is dead 2011-02-08T05:15:52 option 3 2011-02-08T05:15:55 yeah 2011-02-08T05:16:13 i still think 1 would be way cooler than 2 2011-02-08T05:16:40 either option is going to require caring about the micro side, but 1 makes it better for people who do it well 2011-02-08T05:16:57 as opposed to those that don't just suck 2011-02-08T05:21:11 did anyone take a serious look at one with distance just the 8 adjacent squares 2011-02-08T05:21:19 option 1 that is 2011-02-08T05:23:16 i liked the idea that enemy ants can never land in the same square 2011-02-08T05:23:25 i haven't really looked at it though 2011-02-08T05:23:41 the above wouldn't allow them to would it? 2011-02-08T05:23:46 i like chris_0076's idea to use chebyshev metric for option 2 2011-02-08T05:24:01 oops, I confused option 2 and 1 2011-02-08T05:24:21 if you do the 8, you can have a.b and them walk into each other 2011-02-08T05:24:22 I meant option 2 with chebyshev distance 1 above 2011-02-08T05:24:30 ah okay 2011-02-08T05:24:31 ahh, yeah 2011-02-08T05:24:36 i plan to play around with that 2011-02-08T05:25:27 hmm, so distance 1 is too small and distance 2 starts to seem rather large 2011-02-08T05:25:58 maybe it would still work out though 2011-02-08T05:26:33 i don't think it's too bad, if you consider manhatten metric with option 1 2011-02-08T05:26:37 it's all just the 2 circle 2011-02-08T05:26:47 it's the same with euclidean metric too 2011-02-08T05:26:50 right, I meant the chebyshev 2011-02-08T05:27:07 yeah, i wouldn't necessarily set d=2 2011-02-08T05:27:17 give the diamon, but use chebyshev as distance metric 2011-02-08T05:27:39 it just changes ajcant to d=1 instead of rt2 2011-02-08T05:28:03 ahh, so they only interact within the diamond but distance measurement is the chebyshev 2011-02-08T05:28:09 yeah, that might work nice 2011-02-08T05:28:22 that way you can fight around you or long range if necessary 2011-02-08T05:28:31 but only long range in your movement direction 2011-02-08T05:28:37 it sort of comes out alright explanation wise 2011-02-08T05:28:45 yeah 2011-02-08T05:30:09 i'm implementing an iterateState function in my simple bot now so i can just feed it a game state to iterate for me 2011-02-08T05:30:20 i'm doing option 1, but i will do 2 after 2011-02-08T05:41:43 janzert: another issue the extra fighting region presents is fighting through walls 2011-02-08T05:41:56 although diagonal also had the issue of fighting through a crack 2011-02-08T05:42:22 one "explanation" is to say walls aren't high enough to block fighting, only movement 2011-02-08T05:42:25 yeah, I thought of that but I don't think it'll be a big problem 2011-02-08T05:42:36 it might make it cool if you can fight through the walls 2011-02-08T05:42:39 or just say, hey it's a game :P 2011-02-08T05:42:57 that's what i would do, but i've been told an explanation is important 2011-02-08T05:43:13 i'd be happy saying it's a game on a graph with blah blah blah :P 2011-02-08T05:43:35 I guess I think the whole ants thing is pretty thin at this point anyway so... 2011-02-08T05:43:55 nah, ant colonies fighting for land area and resources for growing 2011-02-08T05:44:04 that makes perfect sense :\ 2011-02-08T05:44:26 i especially like the idea of not knowing how many other colonies there are 2011-02-08T05:44:33 with no hole and a central brain and.. :P 2011-02-08T05:44:35 but that may be unavoidable 2011-02-08T05:44:58 well.. the starting square could be an ants hole, and if another ant lands on it, kills that plays ants, all of them 2011-02-08T05:45:03 but i don't think people will like that 2011-02-08T05:45:20 actually, manages to kill the hole with the same rules? 2011-02-08T05:45:24 and the hole can't move 2011-02-08T05:45:34 but what advantage does the hole give? 2011-02-08T05:45:41 your survival? 2011-02-08T05:45:54 ants could spawn there on a turn that you don't have an ant on it 2011-02-08T05:46:07 and you get a counter for amount of food squares you claimed 2011-02-08T05:46:12 but it's going back to my game then 2011-02-08T05:46:22 that would certainly change the nature of the game yeah 2011-02-08T05:46:38 i think this game as is could be quite cool 2011-02-08T05:46:49 i just think it needs option 1 to make the micro more awesome 2011-02-08T05:46:51 not that it would necessarily be bad, but it's a lot different than what people are looking at now 2011-02-08T05:47:10 and i've not seen so much agreement over the general game until now 2011-02-08T05:47:17 it might be best to go with where we're at 2011-02-08T05:48:04 probably 2011-02-08T05:50:10 i'm pretty iffy about the non-symmetric maps idea 2011-02-08T05:50:41 i like that at least with the maps i have, if two players come across each other at the start, it's not going to be on symmetrical terms 2011-02-08T05:50:48 well, very unlikely 2011-02-08T05:50:48 yeah, I really like the idea in theory. But I don't think it's going to work in practice like Jeff thinks it will 2011-02-08T05:51:27 if he can actually pull it off though then great 2011-02-08T05:51:51 i think both should be done initially and results compared 2011-02-08T05:51:59 and generally look at how it goes 2011-02-08T05:52:01 I think the noise and feedback between map rating and player rating is going to be too much though 2011-02-08T05:52:11 that and the number of games needed 2011-02-08T05:52:18 yep 2011-02-08T05:52:36 tcp could help the games thing maybe 2011-02-08T05:53:23 possibly, but you know even less about player strengths there 2011-02-08T05:53:58 although I don't think he was planning to try and take player strength into account in the calculation at all 2011-02-08T05:54:12 and just hope that it averaged out I guess 2011-02-08T05:55:43 i don't think it'd work 2011-02-08T05:55:47 but i'm happy to try 2011-02-08T05:55:51 oh well 2011-02-08T05:56:25 * janzert nods 2011-02-08T05:57:56 *** aerique has joined #aichallenge 2011-02-08T06:09:23 wtf dabino 2011-02-08T06:09:29 calculate kills before moves, that is stupid 2011-02-08T06:10:18 and no different, just makes you iterate the state yourself before making moves 2011-02-08T06:24:01 *** boegel has quit IRC (Quit: This computer has gone to sleep) 2011-02-08T06:29:40 *** RudyRain has joined #aichallenge 2011-02-08T06:46:28 I added possible visualiser input to etherpad if anyone wants to look 2011-02-08T06:46:31 http://ietherpad.com/TR0zbEqvx1 2011-02-08T06:50:57 *** boegel has joined #aichallenge 2011-02-08T07:14:31 *** chris_0076 has quit IRC (Ping timeout: 240 seconds) 2011-02-08T07:14:39 *** p4p4p4 has quit IRC (Read error: Connection timed out) 2011-02-08T07:15:26 *** p4p4p4 has joined #aichallenge 2011-02-08T07:33:21 *** sigh has quit IRC (Remote host closed the connection) 2011-02-08T07:40:36 * antimatroid also made the mistake of killing off friend ants near each other 2011-02-08T07:46:05 *** boegel has quit IRC (Quit: This computer has gone to sleep) 2011-02-08T07:51:34 *** p4p4p4 has quit IRC (Read error: Connection timed out) 2011-02-08T07:57:50 *** boegel has joined #aichallenge 2011-02-08T08:41:37 *** Error323 has joined #aichallenge 2011-02-08T08:43:34 Error323: morning 2011-02-08T08:43:42 http://www.ai-contest.com/forum/viewtopic.php?f=21&t=1313&start=10 2011-02-08T08:43:47 antimatroid: hooooooooowdy you tasmanian devil 2011-02-08T08:43:50 have you looked at the possible battle resolutions? 2011-02-08T08:43:58 I have not 2011-02-08T08:44:40 in fact I haven't looked in such a long time I need to read up massively 2011-02-08T08:46:07 But am not gonna do that now =) 2011-02-08T08:46:13 :) 2011-02-08T08:46:30 i've been implementing said battle resolutions methods 2011-02-08T08:46:42 i'm just doing it straight into some code for a bot 2011-02-08T08:46:48 then passing mock game states to that 2011-02-08T08:48:31 cool 2011-02-08T08:48:54 is it already certain that it will be the ant game 2011-02-08T08:49:11 i've not seen any more agreement than for the current form of this 2011-02-08T08:49:25 i would be very surprised if there is a completely different game used 2011-02-08T08:49:36 I see 2011-02-08T08:49:41 as in people are in the process of making official stuff for this 2011-02-08T08:49:44 would be a waste of a lot of work indeed. 2011-02-08T08:52:35 i hated the game until we moved away from 1:1 battle resolution 2011-02-08T08:52:46 now i think it will be quite cool 2011-02-08T08:53:30 is there already an engine and webbased visualizer? 2011-02-08T08:54:03 there's an old engine j3camero wrote (i think, maybe amstan), but mcleopold is working on one 2011-02-08T08:54:12 i think he plans to get a tcp server going soon for testing on 2011-02-08T09:53:19 *** mceier has joined #aichallenge 2011-02-08T10:01:01 *** RudyRain has quit IRC (Ping timeout: 240 seconds) 2011-02-08T10:05:54 hmmm well i have very naive implementations of each battle resolution 2011-02-08T10:05:57 must be time for bed 2011-02-08T10:32:08 *** delt0r___ has joined #aichallenge 2011-02-08T10:33:15 *** delt0r_ has quit IRC (Ping timeout: 240 seconds) 2011-02-08T11:21:48 *** aerique has quit IRC (Quit: ...) 2011-02-08T11:36:14 *** McLeopold has joined #aichallenge 2011-02-08T11:38:55 *** rebelxt has joined #aichallenge 2011-02-08T12:01:42 *** Meatkat has quit IRC (Ping timeout: 276 seconds) 2011-02-08T12:05:54 *** boegel has quit IRC (Quit: Leaving) 2011-02-08T12:37:29 *** bhasker has quit IRC (Quit: bhasker) 2011-02-08T12:42:39 *** amstan has joined #aichallenge 2011-02-08T12:42:39 *** ChanServ sets mode: +o amstan 2011-02-08T12:59:00 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2011-02-08T13:03:39 *** Meatkat has joined #aichallenge 2011-02-08T13:09:50 *** bhasker has joined #aichallenge 2011-02-08T13:21:30 i'll just leave this here: http://www.secretgeek.net/fuv_intro.asp 2011-02-08T13:27:18 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-02-08T13:31:57 looks useful 2011-02-08T13:48:56 *** mceier has quit IRC (Quit: leaving) 2011-02-08T14:05:42 *** Meatkat has quit IRC (Ping timeout: 240 seconds) 2011-02-08T14:23:37 *** boegel has joined #aichallenge 2011-02-08T15:25:25 *** Accoun has quit IRC () 2011-02-08T15:36:28 *** Accoun has joined #aichallenge 2011-02-08T15:38:52 *** sigh has joined #aichallenge 2011-02-08T16:01:19 *** mega1 has quit IRC (Ping timeout: 240 seconds) 2011-02-08T16:08:52 *** boegel has quit IRC (Quit: Leaving) 2011-02-08T16:09:51 *** sigh has quit IRC (Remote host closed the connection) 2011-02-08T16:18:32 *** McLeopold1 has joined #aichallenge 2011-02-08T16:19:01 janzert: I thought of something we could do for the scoring. 2011-02-08T16:19:28 It would ensure the last bot standing gets the win, or a least make no incentive for keeping the game going. 2011-02-08T16:20:02 If the amount of food to be spawned is known, then when there is 1 bot left, they get that amount added to their score. 2011-02-08T16:20:36 At 1000 turns, if you win at turn 750, you get a 250 point bonus, assuming 1 food spawned per turn. 2011-02-08T16:21:12 Letting someone live in a corner will only take away points if food spawns near them. 2011-02-08T16:21:43 So there should be no incentive to not end the game for anyone. 2011-02-08T16:22:11 And dumb bots with survival luck don't get credit. 2011-02-08T16:28:41 *** Meatkat has joined #aichallenge 2011-02-08T16:32:12 jmcarthur: another battle resolution idea: ants only die when they are on the same square as other ants (including self) 2011-02-08T16:32:43 McLeopold1: so just require them to run into each other? 2011-02-08T16:32:47 yep 2011-02-08T16:32:53 a bot bot can ensure 1:1 2011-02-08T16:32:58 what about overshooting? 2011-02-08T16:33:06 ab, a moves right and b moves left 2011-02-08T16:33:15 kill em 2011-02-08T16:33:29 alright, so like my stratego rules 2011-02-08T16:34:01 that would certainly be 1:1 and fair, but i can't say it sounds very... compelling :) 2011-02-08T16:35:42 it makes blocking impossible too, obviously 2011-02-08T16:36:05 there's pretty much no real micro there except for the occasional game of chase 2011-02-08T16:36:12 yeah, now let's say an ant is really a 2x2 square 2011-02-08T16:36:21 or, a 3x3 2011-02-08T16:36:33 *** Meatkat has quit IRC (Ping timeout: 250 seconds) 2011-02-08T16:36:52 then, we are back to, if in range, kill ant, and not trying to maintain 1:1 with shorter distance requirements 2011-02-08T16:37:02 a player can still ensure 1:1 2011-02-08T16:37:06 the shorter distance requirements were always silly 2011-02-08T16:37:09 just keep yourself spaced 2011-02-08T16:37:11 this is basically just kill all ants in range 2011-02-08T16:37:23 yes, but without serial numbers 2011-02-08T16:37:33 well right, because it's not actually 1:1 2011-02-08T16:37:45 no, but a bot can ensure 1:1 2011-02-08T16:37:53 given that they have the space to move about 2011-02-08T16:37:56 without a "CS degree" 2011-02-08T16:38:09 if they are cornered it's still not going to be 1:1 necessarily 2011-02-08T16:38:31 no 2011-02-08T16:38:40 locally deterministic 2011-02-08T16:38:52 If jeff will accept it 2011-02-08T16:39:05 i'm pretty sure we had mentioned it before 2011-02-08T16:39:09 maybe not in front of jeff 2011-02-08T16:39:29 he was against it at first 2011-02-08T16:39:43 anyway, i'm not a fan just because i really want micro, but it beats the distance sorting thing 2011-02-08T16:39:50 But, I think he made a statement where he thought it was a good idea. 2011-02-08T16:40:08 You get some micro, just not formations 2011-02-08T16:40:51 well, you get basically the same kind of micro as you get with the distance one: keep your ants spaced out, attack with one at a time, play chicken 2011-02-08T16:40:52 He also wants range attacks to discourage chasing and breaking through behind lines. 2011-02-08T16:41:48 But, I think those are actually fun *macro* strategies. 2011-02-08T16:41:50 how do range attacks discourage those? 2011-02-08T16:42:16 yes, i agree, but i don't want the game to be solely macro 2011-02-08T16:42:16 you can't sneak past an ant in a corridor if he has a range 3 attack. 2011-02-08T16:42:39 less ants to "block" a chokepoint 2011-02-08T16:42:56 err 2011-02-08T16:43:01 also, read above for my scoring idea 2011-02-08T16:43:13 last bot standing gets food bonus 2011-02-08T16:43:21 less ants to get *a* kill in a chokepoint, sure 2011-02-08T16:44:05 the scoring idea sounds good at first glance 2011-02-08T16:57:14 *** delt0r_ has joined #aichallenge 2011-02-08T16:58:36 *** delt0r___ has quit IRC (Ping timeout: 240 seconds) 2011-02-08T17:00:12 *** delt0r___ has joined #aichallenge 2011-02-08T17:01:34 *** delt0r_ has quit IRC (Ping timeout: 246 seconds) 2011-02-08T17:03:11 *** McLeopold1 has left #aichallenge 2011-02-08T17:10:18 *** Palmik has quit IRC (Remote host closed the connection) 2011-02-08T17:11:22 *** amstan has joined #aichallenge 2011-02-08T17:11:22 *** ChanServ sets mode: +o amstan 2011-02-08T17:17:13 *** sigh has joined #aichallenge 2011-02-08T17:33:46 *** delt0r_ has joined #aichallenge 2011-02-08T17:35:25 *** delt0r___ has quit IRC (Ping timeout: 255 seconds) 2011-02-08T18:11:27 I'm beginning to think that understanding the sample bot and game engine should be super-simple, not the rules... 2011-02-08T18:11:44 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-02-08T18:12:05 yeah 2011-02-08T18:12:22 the whole point of the rules, to me, is to create some emergent complexity anyway 2011-02-08T18:12:33 otherwise it's not really an AI competition at all 2011-02-08T18:12:38 even with our "simple" ants, I can imagine a lot of questions that will be hard to answer. 2011-02-08T18:14:15 McLeopold: such as? 2011-02-08T18:14:42 i don't like that "colliding ants die" rule, i'm with jmcarthur, i want interesting micro, and i think formations would be awesome to watch 2011-02-08T18:15:14 i've got all 4 options implemented (fairly naively) in my bot code, so I can throw input at it and get the result of iterating the state 2011-02-08T18:23:13 antimatroid: I think it would be to hard to explain recursive collision resolution to the newbs. It's easier to kill their ants and slap their hands. 2011-02-08T18:23:47 With range attacks, enemy ants can't collide. 2011-02-08T18:23:53 McLeopold: that's one of the reasons i like option 1 to 2 2011-02-08T18:24:35 i actually found each algorithm was pretty easy to do naively 2011-02-08T18:24:42 option 2 code might have even been the shortest 2011-02-08T18:24:59 honestly, I think we have "the one and only option" and "wish #1" and "wish #2" 2011-02-08T18:25:22 what's "the one and only option"? 2011-02-08T18:25:41 range attacks, attempt 1:1 ratio 2011-02-08T18:25:51 i think that's shit personally 2011-02-08T18:26:14 i only got interested in this game once we moved away from 1:1 and introduced "attack" in my ants version so it wasn't also 2011-02-08T18:26:36 without micro being interesting, it's boring and doesn't leave interesting battles 2011-02-08T18:27:24 If we can trivially break the game, then jeff will listen, but probably not until then 2011-02-08T18:28:13 so, write an unstoppable bot with range attacks 2011-02-08T18:28:59 the visualiser is easy as to make with ranged attacks 2011-02-08T18:29:14 did you see my very rough proposal for passing info to it? 2011-02-08T18:30:08 atm i'm imagining "pixels" sliding between squares each turn, then fading kills/spawns/food removal 2011-02-08T18:30:22 if ants are going into the same square, that gets trickier and less visually appealing 2011-02-08T18:30:38 *** Mathnerd314 has quit IRC (Ping timeout: 240 seconds) 2011-02-08T18:30:39 and it's complicated for killing ants that are passing through the same edge 2011-02-08T18:31:02 ideally you'd want to visualise that as the two ants sliding off their square to die in a given direction 2011-02-08T18:31:28 I too want to see dead ants. 2011-02-08T18:31:39 on the visualiser? 2011-02-08T18:31:49 isn't it enough to see them fade? 2011-02-08T18:31:55 ?!%*.abcdefghijklmnopqrstuvwxyz 2011-02-08T18:32:11 and what if there's more than one ant landing in a square? then how do you show a "dead" ant? there's many 2011-02-08T18:32:13 That's my current set of characters for the map matrix 2011-02-08T18:32:33 ?=unseen, !=dead ant, %=wall/water, *=food, .=open land 2011-02-08T18:32:40 no more hills 2011-02-08T18:32:45 i think allowing attack range ==2 with manhatten makes it a lot easier, and blocking passageways etc. is more complicated 2011-02-08T18:33:15 So, as for sliding ants, that would be nice, but it is a lot of data to try and save for the visualizer. 2011-02-08T18:33:39 McLeopold: check the input i proposed, i don't think it is 2011-02-08T18:33:41 we'd have to echo all ant moves to every player 2011-02-08T18:33:49 where do I check? 2011-02-08T18:33:52 etherpad? 2011-02-08T18:33:53 etherpad 2011-02-08T18:34:36 you basically echo the moves to the visualiser, along with information at the end about spawning/killing squares, which is done with fading 2011-02-08T18:35:45 surely it'd be less info than sent for planet wars? 2011-02-08T18:36:17 McLeopold: can i replace the io at the top? 2011-02-08T18:36:27 i'll change X to %, i wanted to do something like that anyway 2011-02-08T18:36:27 I think it might be more 2011-02-08T18:36:33 yeah 2011-02-08T18:36:35 me too 2011-02-08T18:36:41 keep letters for players 2011-02-08T18:36:45 yep 2011-02-08T18:36:51 you can make a moved players value capital too 2011-02-08T18:36:54 etc. etc. 2011-02-08T18:37:04 you need to give info about revealed spots too 2011-02-08T18:37:13 not to the visualiser? 2011-02-08T18:37:30 wait, what? 2011-02-08T18:37:33 but, to the players 2011-02-08T18:37:47 fog of war 2011-02-08T18:38:06 Just catching up on the IRC. I would support "kill all ants that are within range of an enemy ant" rule. It would produce some interesting micro where you try to kill two or three of their ants at a time using only one of your ants. It would be pretty hard to exploit that death rule to consistently get odds that are much better than 1:1. 2011-02-08T18:38:30 I think that was a suggestion just above, not sure if I read that right. 2011-02-08T18:38:49 j3camero: I thought you were against micro that allowed 3:1 deaths? 2011-02-08T18:38:52 hmmm, it still encourages fighting with just 1 ant 2011-02-08T18:39:03 *** Apophis_ has joined #aichallenge 2011-02-08T18:39:10 i think it's cooler if you have to try and coordinate a good fight with numerous ants to be effective 2011-02-08T18:39:30 j3camero: what's not awesome about watching ants try to move themselves into formations for offense? 2011-02-08T18:39:34 j3camero: so you are okay with offensive micro, but not defensive micro? 2011-02-08T18:39:54 it'd be a little bit like watching game of life 2011-02-08T18:39:56 McLeopold: 3:1 deaths are technically possible with the suicide bomber death rule, but would be very rare. Especially with simultaneous moves, it would be pretty hard to do consistently. In any typical battle it would be hard to do much better than 1:1. 2011-02-08T18:40:22 j3camero: we're at complete odds about whether deviating from 1:1 is good i think 2011-02-08T18:40:42 j3camero: also, I'm in the middle of massive changes to the engine I wrote. 2011-02-08T18:40:57 McLeopold: just make yourself the official map gen person already :P 2011-02-08T18:41:00 I think I'm going to get a bot that processes stdin by today 2011-02-08T18:41:28 *** Mathnerd314 has joined #aichallenge 2011-02-08T18:41:35 antimatroid: I plan to compete with you for map gen guru. You won the last round. ;) 2011-02-08T18:41:44 McLeopold: bring it :P 2011-02-08T18:41:51 Sounds god guys. 2011-02-08T18:41:57 i'm hoping lots of people will play around with maps and early 2011-02-08T18:42:02 interesting typo... 2011-02-08T18:42:08 I think there will be room for several map gens this time around. Why not? 2011-02-08T18:42:08 *** Apophis has quit IRC (Ping timeout: 240 seconds) 2011-02-08T18:42:17 j3camero: i'd say go with one at the end 2011-02-08T18:42:28 i think it's important to keeps things clear and simple at the end 2011-02-08T18:42:33 j3camero: I think your right. Pure random will be too noisy. There are so many options. 2011-02-08T18:43:02 i had an idea for a map gen, just send tron like snakes randomly around the map until you have filled enough walls? 2011-02-08T18:43:16 would encourage "walls" more i think? 2011-02-08T18:43:34 did you see my first draft at one? 2011-02-08T18:43:40 nope 2011-02-08T18:44:07 start with all walls, the carve out symmetric "rooms". ensure a path to all starting ants 2011-02-08T18:44:25 j3camero: have you seen the pictures of resolutions in the forums? 2011-02-08T18:44:31 I have seen some of antimatroid's maps. They are cool I think people will like them. They are very mazelike. Perhaps we can also generate some that are like islands joined by narrow land bridges. 2011-02-08T18:44:32 Although it crashes 50% of the time. 2011-02-08T18:44:59 my engine? was that just on codepad? that times out, otherwise it shouldn't 2011-02-08T18:45:01 j3camero: that's what my first attempt was trying to do 2011-02-08T18:45:04 but it was a very rough draft anyway 2011-02-08T18:45:25 j3camero: while mine are symmetric, coming across another player wont be on symmetrical terms, i like that and think you will? 2011-02-08T18:45:55 You played much Starcraft? You know The Hunters? That's sort of my concept for maps. http://www.teamliquid.net/tlpd/images/maps/173_Hunters.jpg 2011-02-08T18:46:02 never played it 2011-02-08T18:46:06 I *hate* that map... 2011-02-08T18:46:23 but yeah 2011-02-08T18:46:26 antimatroid: I think we should explore non-symmetric maps, but there's no reason why we should make a rule against them. Of course we should use your map gen. 2011-02-08T18:46:36 Actually, we could use the planet wars generator for this. 2011-02-08T18:46:37 i think it'd be better to focus on maps once an engine is running 2011-02-08T18:47:02 Sorry, there is no reason we should make a rule against symmetric maps is what I meant. 2011-02-08T18:47:09 Just make the planets a little bigger, draw some paths if they are close enough, convert to png. 2011-02-08T18:47:20 j3camero: i'm happy to try the non-symmetric maps idea, it sounds cool, but i don't see it working 2011-02-08T18:47:33 are we still using pngs as maps? :( 2011-02-08T18:47:39 Zannick: no 2011-02-08T18:47:42 i don't think anyway 2011-02-08T18:47:42 I'm supporting both right now. 2011-02-08T18:47:46 ah 2011-02-08T18:47:51 that's good 2011-02-08T18:47:51 We will get lots of feedback from the contestants on maps once we launch, so there will be time to zero in on what people are finding most interesting. 2011-02-08T18:48:00 It's easier to "draw" a png just to test stuff out. 2011-02-08T18:48:08 i'm all about battle resolution atm 2011-02-08T18:48:12 I think the community would enjoy that functionality too. 2011-02-08T18:48:19 Yeah PNGs are a terrible way to store maps. I just did that because I was lazy when I was writing a throw-away ants simulator. 2011-02-08T18:48:21 and i'm still very much in favour of option 1, but i'd prefer 2 to the new thing mcleopold mentione 2011-02-08T18:48:28 or what j3camero just mentioned 2011-02-08T18:49:17 someone needs to make an "ascii paint" program, where you can pick the character to draw with and make different shapes with it on a lattice canvas 2011-02-08T18:49:22 that'd be sweet for making maps in 2011-02-08T18:49:45 Wow why doesn't that exist? 2011-02-08T18:49:48 Or does it? 2011-02-08T18:49:50 ant_map can be used as a png2txt converter right now 2011-02-08T18:49:50 it probably does 2011-02-08T18:50:01 I can make it go the other way too. 2011-02-08T18:50:08 yeah, other way would be good 2011-02-08T18:50:45 Okay I have to drill through some work that I promised I would do for people, so I am going into "ignore distracting IRC mode". I'll be back in a while to make some headway on engine stuff. 2011-02-08T18:50:46 http://asciipaint.com/ 2011-02-08T18:51:03 lol well there you go 2011-02-08T18:51:17 that's so cool 2011-02-08T18:51:20 i drew boobies 2011-02-08T18:51:36 yes i'm 2 2011-02-08T18:51:38 1* 2011-02-08T18:51:40 12** 2011-02-08T18:51:46 120**? 2011-02-08T18:51:49 FTFY 2011-02-08T18:52:34 i love how it "fills" with an animation 2011-02-08T18:52:38 flood fill yay 2011-02-08T18:52:59 i broke it 2011-02-08T18:53:29 i was randomly filling a region with different characters and it ended up missing some 2011-02-08T18:53:36 now it's jumbled inside the rectangle 2011-02-08T18:54:11 McLeopold: so normal scoring is 1 point per food eaten with 1 food spawned per move and last player gets all remaining food in map + future spawns at end of game? 2011-02-08T18:54:21 yes 2011-02-08T18:54:25 on first blush, seems like it could work 2011-02-08T18:54:30 no incentive to make it last long 2011-02-08T18:54:34 with my maps you really ought to spawn n food squares per turn 2011-02-08T18:54:50 as long as the rate is known, we can give the bonus 2011-02-08T18:55:07 yeah, as long as there is a fixed rate it will work 2011-02-08T18:55:43 McLeopold: has your engine got fog of war? 2011-02-08T18:55:46 we could have any type of rate, diminishing or increasing, and we could still calc the bonus 2011-02-08T18:55:49 yes 2011-02-08T18:56:01 I can produce fogged maps. 2011-02-08T18:56:15 I just putting it all together so I can get a sample movie going. 2011-02-08T18:56:24 oh, do you want the visualiser to show squares that can't be seen by anyone as unseen? 2011-02-08T18:56:27 that could be cool 2011-02-08T18:56:37 yes, show all, or show perspective 2011-02-08T18:56:37 and spawn points need to be random not predetermined 2011-02-08T18:56:47 or just do that a darker "shade" 2011-02-08T18:56:51 that might depend on the map 2011-02-08T18:56:59 * janzert has no idea how it is now 2011-02-08T18:57:10 janzert: but with the option of using a seed so you can test your bot on maps 2011-02-08T18:57:11 food spawn that is not player spawn 2011-02-08T18:57:22 oh, yeah 2011-02-08T18:57:37 antimatroid: yeah, that would be good too 2011-02-08T18:57:44 we could just blast the whole map every 40 turns 2011-02-08T18:57:58 ants would spawn like a wild fire 2011-02-08T18:57:59 McLeopold: i think it'd be easiest if you could just provide a bool parameter to the engine as to whether the seed should be set or not? 2011-02-08T18:58:23 if you don't supply a seed, then just generate one :) 2011-02-08T18:58:26 What would happen if we spawned 1 food every time an ant dies? How about every time a food is eaten? 2011-02-08T18:58:40 McLeopold: that's another issue, should food squares ever be within range of other ones? and do you get to spawn them like rapid fire in the one move? or over a few turns? 2011-02-08T18:58:48 These two schemes mean that the number of ants on the map stays more or less constant. 2011-02-08T18:58:56 j3camero: i like that 2011-02-08T18:59:08 j3camero: former gives constant population, latter constant food density 2011-02-08T18:59:09 but, there goes my bonus idea 2011-02-08T18:59:29 that would upset the scoring mechanism 2011-02-08T18:59:30 Well I'm just thinking what incentive it would produce to prolong the game. 2011-02-08T18:59:36 yeah, and the McLeopold scoring system doesn't work anymore 2011-02-08T18:59:46 yeah, go with a known rate 2011-02-08T19:00:04 And I can't emphasize enough how *BAD* ranking bots by surviving turn is. 2011-02-08T19:00:23 Yeah that would produce camping. 2011-02-08T19:00:30 no 2011-02-08T19:00:34 j3camero: we don't want prolonged games... 2011-02-08T19:00:35 McLeopold: what's the roadmap to a testing tcp server like? 2011-02-08T19:00:46 it's hard to tell too much about the resolutions until then i think 2011-02-08T19:00:58 step 1: finish engine, step 2: ?, step 3: TCP! 2011-02-08T19:01:23 yeah, step 2? is that lots of work? or just a bit of changing? 2011-02-08T19:01:30 work 2011-02-08T19:01:44 any camping bot will be wiped out by an expanding bot, so camping only works if everyone camps 2011-02-08T19:01:56 okay, got to ignore irc, must ...finish...engine... 2011-02-08T19:01:57 i don't think camping would work well in any case 2011-02-08T19:02:01 Constant ants seems like it would encourage camping. Whenever you decide to fight you are giving away 1 ant in return for a 1/turf chance of getting a food back. 2011-02-08T19:02:20 janzert: my concern isn't camping or not as a strategy, my concert is proper ranking of bots 2011-02-08T19:02:21 yeah, I think constant ants is probably bad 2011-02-08T19:03:24 j3camero: with ranged attacks, it might be more like 30/turf 2011-02-08T19:03:35 I still think it would have some bad orderings but would work out fine overall 2011-02-08T19:04:06 i.e. a bad ant may end up higher than it should occasionally but won't consistently 2011-02-08T19:04:10 can we talk fight range actually? 2011-02-08T19:04:16 as that changes up each rule as well 2011-02-08T19:04:32 anyway like antimatroid said it will really need testing at this point to tell 2011-02-08T19:04:50 *** rebelxt1 has joined #aichallenge 2011-02-08T19:05:13 *** rebelxt has quit IRC (Read error: Connection reset by peer) 2011-02-08T19:05:48 the biggest problem i see with allowing ants to land in the same square is complicating the visualiser 2011-02-08T19:06:11 and i think the smaller region makes it easier to block/defend with option 1? 2011-02-08T19:06:19 if we do 1 food per spawn/death, then that would allow hoarding 2011-02-08T19:06:51 it no longer becomes necessary to hold onto land 2011-02-08T19:07:01 Oh true. 2011-02-08T19:07:06 McLeopold: i think the other way is better anyway, if people camp, eventually they're all going to grow so large that they have to fight 2011-02-08T19:07:42 Camping is a bad idea if you can't figure out whether other people are fighting. 2011-02-08T19:08:11 j3camero: exactly, it's tough to know if you should even take on a tough battle because of that 2011-02-08T19:08:12 if all bots camp, then the game will go long 2011-02-08T19:08:30 McLeopold: that will just be a max turn game, but i find it very unlikely 2011-02-08T19:08:31 If only one bot attempts to seek and destroy, they will take out the campers 2011-02-08T19:08:58 i don't think people will find it a very lucrative strategy to camp 2011-02-08T19:09:14 there really isn't "camping" in the sense of FPS games, since you can easily find others 2011-02-08T19:09:19 Well and you can get away with demolishing one neighbor as long as you can feign continued strength to you other neighbors. There's an incentive to give it to your neighbor if you think you'll win. 2011-02-08T19:09:58 even if you can push a battle line back, you still should get a few more ants in return over time 2011-02-08T19:10:19 What if score the games thus: when you kill a player's last ant, you get a point. 2011-02-08T19:10:40 Produces an incentive to play interesting games and take risks. 2011-02-08T19:10:40 j3camero: it's not always clear who killed who i don't think 2011-02-08T19:10:47 so, I'm still for constant food, or at least something not geared off of deaths/births 2011-02-08T19:11:02 McLeopold: i think everyone here is in agreement on that 2011-02-08T19:11:22 Yeah I would say having a steady flow of new food being spawned is a good idea. 2011-02-08T19:11:29 j3camero: that makes cherry picking lucrative, I don't like it 2011-02-08T19:12:15 Cherry picking would be pretty tough I think, since you can't see your neighbors' other battle fronts. 2011-02-08T19:12:50 2 stupid bots that kill each other off with the first ant? 2011-02-08T19:12:58 good strategy? 2011-02-08T19:13:03 It encourages bots to sick on a neighbor that looks like he might be having trouble on another front, or may be generally weak. You'll get players that suddenly collapse, with their two neighbors sprinting inwards towards each other looking for the kill. 2011-02-08T19:13:45 or, you do an expert job taking out an opponent, then he runs his last ant over to someone else 2011-02-08T19:14:05 dying bots get to be king maker 2011-02-08T19:14:44 1 point per kill (roughly) with food bonus for last bot is still best 2011-02-08T19:15:03 You still want to rush in and kill 2011-02-08T19:15:19 because you get the land 2011-02-08T19:17:01 basically, I don't think you can make a single few events in the game decide the winners. The game should be a marathon and favor those that did well through the entire game. 2011-02-08T19:17:58 Yeah I still don't know what to do about that. I'm afraid that the food bonuses may be so huge that they'll screw up the rankings. 2011-02-08T19:18:04 But it's best I know so far. 2011-02-08T19:18:17 rankings are only a rank, score doesn't matter, for trueskill 2011-02-08T19:18:32 so, you get 1st, 2nd, 3rd and so on 2011-02-08T19:18:44 winning by 1 point is the same as winning by 1000 2011-02-08T19:18:49 What if we score up until the last turn when the winning bot had less than 50% of the ants on the map? 2011-02-08T19:18:51 yeah, score from a game is just the ordering that is important 2011-02-08T19:19:05 so, the food bonus is just to ensure the last bot standing should be in 1st 2011-02-08T19:19:18 I see. 2011-02-08T19:19:54 Scoring by rank seems a bit bad to me. Would that not encourage camping? I know that poker players play ridiculously tight when there are six players left and 5 prizes left. 2011-02-08T19:20:23 If a bot guesses that there are about 3 players left it might clam up and avoid conflict. 2011-02-08T19:20:31 that's why I want I want performance, not how long you survive 2011-02-08T19:20:50 the bot that avoids conflict does not score much 2011-02-08T19:21:44 so, even if you make it to the last 2, if you didn't do anything, you won't be ranked 2nd 2011-02-08T19:23:02 We could call the game when 1 bot owns 75% or more of the ants? 2011-02-08T19:23:43 no, that would encourage non-conflict 2011-02-08T19:24:07 although, the bot that stopped the game could still be last... 2011-02-08T19:27:13 How about this. As the game runs, in the background, assign a color to each square on the map. They start off blank. When an ant moves around, you "paint" all the squares within a radius of 5 their color. As the ants move around, you repaint the squares. What we're going for here is some way to guess roughly who "controls" a piece of territory. We stop scoring after the last turn when every bot controlled less than 50% of the territory. 2011-02-08T19:27:31 So there's no point prolonging the game to delay an inevitable win. 2011-02-08T19:29:06 so, the first bot that claims 50% wins? 2011-02-08T19:29:11 so you want to gain control of 51% of the map and then camp right? 2011-02-08T19:29:19 If a bot is cocky and want to pump up its score it could just chill on 45% of the turf, but that would be risky. 2011-02-08T19:29:55 what about this: you get a point for every square you can access first? 2011-02-08T19:29:56 so, the painting is only for stoping the game, not scoring? 2011-02-08T19:30:00 It's not first to 50% wins. It's stop scoring after a bot passes the 50% mark for the *last time*. 2011-02-08T19:30:22 including your currently occupied squares (so a point for every live ant) 2011-02-08T19:30:57 You still want to win as fast and as strongly as possible, but there's no point delaying a certain win. 2011-02-08T19:31:14 any variation seems pretty complicated to calculate 2011-02-08T19:31:15 j3camero: that would encourage delaying the game if you came from behind 2011-02-08T19:31:29 The only perverse incentive left is to chill at 49% for as long as possible. But even that is risky. 2011-02-08T19:31:39 so hold 50% for a long time, then give up 50% to your opponent? 2011-02-08T19:31:44 the food bonus will get rid of any incentive to delay 2011-02-08T19:32:10 a lot less confusing for beginners to understand 2011-02-08T19:32:20 actually we really want not just no incentive to delay, but an incentive to end early 2011-02-08T19:32:27 yes 2011-02-08T19:32:34 Oh yeah you could still game the system by chilling at 90% for a while then retreating and letting the opponent demolish you. 2011-02-08T19:33:16 j3camero: remember, the score doesn't really matter except to rank bots 2011-02-08T19:33:22 True yes. 2011-02-08T19:33:36 so, that wouldn't really be gaming anything, but the server resources 2011-02-08T19:33:42 Yeah the food bonus is fine. 2011-02-08T19:33:50 consider it written 2011-02-08T19:33:53 ... 2011-02-08T19:34:06 what are you doing? 2011-02-08T19:34:07 Now the issue that is shaky to me is having these really sharp discontinuities in the payoff function. 2011-02-08T19:34:23 ? 2011-02-08T19:34:31 "consider it written" 2011-02-08T19:34:32 which thing? 2011-02-08T19:34:37 food bonus 2011-02-08T19:34:58 If you get the idea that there are only three bots left, you really just want to avoid conflict. 2011-02-08T19:35:17 maybe the end food bonus should better than the 1 point for 1 food during the game? say 1.5x or something 2011-02-08T19:35:31 Are you sure? If you can wipe them out, you get the win. 2011-02-08T19:35:51 j3camero: i think it'll be hard to work that out 2011-02-08T19:36:00 plus, conflict gives you points 2011-02-08T19:36:23 yeah, an outright win will give you more in ranking than a 3 way tie 2011-02-08T19:36:30 if you are fighting 3 bots, and they are only fighting you, you should have about 3 times the score, which makes sense 2011-02-08T19:37:21 we're hiding players scores yes? 2011-02-08T19:37:23 the scoring system will ensure very few ties 2011-02-08T19:37:25 McLeopold: only if you're eating 3 times the food though right? 2011-02-08T19:37:32 otherwise you might work out who is dead 2011-02-08T19:37:34 yeah 2011-02-08T19:37:39 even if there are 3 bots left, they probably won't have the same score 2011-02-08T19:37:47 that's true 2011-02-08T19:37:56 but if a bots score isn't changing, it's most likely dead 2011-02-08T19:38:07 janzert: yes, I'm assuming if you can hold off 3 bots, you have roughly the same amount of food as them combined 2011-02-08T19:38:11 you'll want to expand your territory whenever you think you can 2011-02-08T19:38:20 ahh, ok 2011-02-08T19:38:58 antimatroid: I think giving the bots ongoing scores during a game is probably bad 2011-02-08T19:39:16 at least I don't see any real benefit and possible problems 2011-02-08T19:39:18 are we going to try and hide the number of starting players? 2011-02-08T19:39:22 if we are hiding player numbers, we can't give scores 2011-02-08T19:39:34 McLeopold: there are other issues with hiding player numbers 2011-02-08T19:39:37 like time limits 2011-02-08T19:39:42 if doing moves simultaneously 2011-02-08T19:39:49 I think hiding player numbers would be nice 2011-02-08T19:39:51 Players can try an guess at the total points given the map size and turn limit, so they can try for some type of goal. 2011-02-08T19:40:03 I came up with a solution to that. Have the same time limit regardless of #players. 2011-02-08T19:40:14 j3camero: so like 5 seconds? 2011-02-08T19:40:23 Something like that yeah. 2011-02-08T19:40:27 fyi, my bot will probably always use whatever you give it 2011-02-08T19:40:33 j3camero: yes, that's what I was trying to say the other day :) 2011-02-08T19:40:39 j3camero: can we have a time bank? 200 seconds for the game, use at your discretion? 2011-02-08T19:40:55 That would work better than you expect, especially if the majority of games have between 5 and 8 players or something. 2011-02-08T19:41:01 McLeopold: hard to do when people aren't making their moves sequentially isn't it? 2011-02-08T19:41:14 McLeopold: I thought a time bank would be nice, but how do you enforce that? 2011-02-08T19:41:18 j3camero: i was thinking 4-16? 2011-02-08T19:41:24 a time bank really needs to still have a per turn time as well 2011-02-08T19:41:27 That's your problem. :) 2011-02-08T19:41:58 e.g. the time bank can be enough to average out to 4 seconds per turn and the per turn limit is set at 10 seconds 2011-02-08T19:41:59 You can stop a player's clock when they send back their move to the engine, but what's to stop them from pondering the next move after that? 2011-02-08T19:42:08 bank = 200 seconds, can use up to x of it per turn 2011-02-08T19:42:08 turn limit = normal + min(x, bank) 2011-02-08T19:42:21 j3camero: you "pause" processes 2011-02-08T19:42:27 j3camero: pondering is fine, they already committed to the current information they have 2011-02-08T19:42:38 let 'em ponder 2011-02-08T19:42:43 yeah, or that 2011-02-08T19:43:01 or go into "suck up server resources to crash other bots" mode 2011-02-08T19:43:01 with tron that would have been bad, but for this i don't think you could do that much 2011-02-08T19:43:08 other than waste other peoples time 2011-02-08T19:43:14 ^^ yeah that 2011-02-08T19:43:20 even with tron pondering didn't do much 2011-02-08T19:43:34 at least at 1 second per turn and the top bot's using pretty much all of it 2011-02-08T19:43:36 Haha you get more food if you think faster. 2011-02-08T19:43:45 janzert: yes it did, we recycled our tree from the move they picked, we could have kept it growing 2011-02-08T19:43:56 yep, I did that 2011-02-08T19:43:59 you're growing stuff you don't want as well, but you'd also grow some of what you do want 2011-02-08T19:44:13 you can prune the move you made at least, that's a large chunk 2011-02-08T19:44:16 antimatroid: but there was very little pondering in practice 2011-02-08T19:44:22 1 random ant dies per second of time you think 2011-02-08T19:44:30 no limits. 2011-02-08T19:44:35 janzert: it gave you extra time to grow the tree, that was our strategy 2011-02-08T19:45:33 because you thought for 0.95 seconds or whatever, even if the opponent thought for 0.99 seconds you only got about .05 seconds of pondering, about 1/3 of that was wasted 2011-02-08T19:46:49 yes, but if players got to make their decisions sequentially and your opponent barely used their time, you basically got double the time yeah? 2011-02-08T19:47:33 if it had been sequential it would have most likely have been useful yes 2011-02-08T19:48:11 way better than pondering in chess even :) 2011-02-08T19:48:33 yeah, extending the depths of the tree can be futile, but it's still better than not 2011-02-08T19:48:48 tree's will be the rage for this with fighting 2011-02-08T19:50:15 i just had an idea for a sample bot 2011-02-08T19:50:19 scardybot 2011-02-08T19:50:23 runs away from enemies 2011-02-08T19:50:53 passive agressive bot, it takes you food 2011-02-08T19:51:02 hehe 2011-02-08T19:51:17 actually, scaredybot would be a terrible example for quick games 2011-02-08T19:51:36 it's too bad there's going to be fog of war 2011-02-08T19:51:48 otherwise i would write a copybot for the starter packages 2011-02-08T19:52:10 it'd be a good test bot though, run away from any enemy else run toward any food else random move 2011-02-08T19:52:30 Zannick: while that'd be nice, fog of war is too much better :P 2011-02-08T19:52:37 i agree 2011-02-08T19:52:48 janzert: you can do better than move randomly 2011-02-08T19:53:05 if you can't go towards a food square, move towards the closest unseen square or something 2011-02-08T19:53:18 or check how recently something was seen etc. 2011-02-08T19:54:46 it's suppose to be a simple test bot though :) 2011-02-08T19:55:17 for inclusion in the start package I meant 2011-02-08T19:56:22 "it'd be a good test bot though" 2011-02-08T19:56:30 i thought you meant as an actual test bot 2011-02-08T19:56:34 not a starter bot, :P 2011-02-08T19:57:38 McLeopold: what battle resolution are you currently using for the engine? 2011-02-08T19:57:40 hmm, to me starter bot means the bot for people to build off of in the starter package. The other bots in the starter package I think of as test bots for people to test against 2011-02-08T19:58:06 janzert: i am in favour of having a few harder bots included there 2011-02-08T19:58:06 oh, yes. 2011-02-08T19:58:16 so people imediately have a range of difficulties to test with locally 2011-02-08T19:58:26 and just don't release the code for those bots 2011-02-08T19:58:27 antimatroid: the problem is someone needs to come up with those ;) 2011-02-08T19:58:35 Zannick: I and I'm sure others will 2011-02-08T19:58:35 yeah, I wouldn't mind that. I would even include the code 2011-02-08T19:58:46 although.. I can't do that in c++ can i? 2011-02-08T19:58:52 as in my binary file wont run on any computer? 2011-02-08T19:58:52 janzert: I am having a really stupid n00b problem. I just set up a new domain for someone but every time I visit their index.php with my browser, I download the .php file instead of displaying it. Do you know what is causing this? I remember having this problem at some point long ago but I forgot what the problem was. 2011-02-08T19:59:01 I was disappointed how closed off people were from sharing ideas in planetwars, tron was much better in that respect 2011-02-08T19:59:03 binary files won't run on every computer 2011-02-08T19:59:17 janzert: i pretty much shared every idea i had 2011-02-08T19:59:28 pw's test bots were in jars, but the source was given, too 2011-02-08T19:59:44 janzert: I was explaining a lot of stuff in the channel 2011-02-08T19:59:52 j3camero: sounds like apache isn't setup to server php correctly 2011-02-08T19:59:59 there weren't any good tutorials like with tron 2011-02-08T20:00:05 i had no idea where to start without those 2011-02-08T20:00:23 i'll do a forum tutorial at some point for this one with pretty .gifs 2011-02-08T20:00:24 some people made some good write ups in the forums 2011-02-08T20:00:27 I'm mostly thinking of the guy that posted his bot source and got soundly beat up for it 2011-02-08T20:00:44 i don't think it was beating him up, their points were very vali 2011-02-08T20:00:46 janzert: that's the weird thing. It's the same server as the contest. I have set up like a dozen domains on there and never seen this problem. It's just a dummy hello world php file. 2011-02-08T20:01:00 it sort of invalidates the work of people below that level if anyone can then upload a bot of better quality 2011-02-08T20:01:15 hmm, it's been a long time since I've setup a new domain myself :} 2011-02-08T20:01:27 that's like saying that you shouldn't participate unless you expect to get #1 2011-02-08T20:01:39 I still think it's a great way for the people below to learn 2011-02-08T20:01:44 sigh: not really, but it gives you less of an idea about how you've gone relative to other people 2011-02-08T20:01:54 sure some of them will just copy and submit but... 2011-02-08T20:01:56 as there might be quite a few copies of a better bot 2011-02-08T20:02:13 janzert: they can particupate themselves and learn from shared source at the end 2011-02-08T20:02:28 like tackling any problem, i think it's best to do it yourself, then look at how others did it 2011-02-08T20:02:38 When people sign up, they should choose from one of three options (1) never publish my code (2) publish my code after a delay of n days (3) publish my code after the submission deadline. 2011-02-08T20:02:50 j3camero: skip option 2 2011-02-08T20:03:08 and i would make it after results are released, just in case 2011-02-08T20:03:12 janzert: yeah but I remember having this conversation in the channel... in the end my opinion was that it would have a negative effect *because* people didn't like it 2011-02-08T20:03:23 not for any logical reason :) 2011-02-08T20:03:29 (1) never (2) after deadline (3) after results 2011-02-08T20:03:37 i think their reasons are fine, even though i don't think it would affect me 2011-02-08T20:03:48 why distinguish 2 and 3? 2011-02-08T20:04:00 * Zannick shrug 2011-02-08T20:04:16 yeah, I suppose. I just liked the much more open feel the tron contest community ended up with. not sure there is any way to encourage it and there is probably a limit in size for it to happen at all 2011-02-08T20:04:22 although unlikely, should something bad happen during the running of the contest and code is shared, it's ruined 2011-02-08T20:04:30 something bad as in a server loses all bot data 2011-02-08T20:04:41 janzert, I was't in tron but I like a more open atmosphere as well 2011-02-08T20:04:53 when there is months of work going into some bots, i think it's better to be safe than sorry 2011-02-08T20:05:25 i like general sharing of strategy/algorithms rather than actual raw code 2011-02-08T20:05:30 let them go and code it up themselves 2011-02-08T20:05:51 yeah, there was a fair bit of that though 2011-02-08T20:06:05 yeah, and i thought that was sufficient 2011-02-08T20:06:21 in fact i'd rather converse with people about their strategy rather than trudge through people's source 2011-02-08T20:06:22 but still could have been better :) 2011-02-08T20:07:36 what's the deal with google having MOVE_H_ instead of MOVE_H for #ifdef's etc.? 2011-02-08T20:07:44 it's a redundant character as far as i'm concerned 2011-02-08T20:08:16 makes it clear that it is a special variable *shrug* 2011-02-08T20:08:51 trailing/leading underscores are pretty standard for marking variables/whatever like that 2011-02-08T20:09:18 bleh, i can't think of a single situation where it'd be unclear 2011-02-08T20:09:21 *** McLeopold has left #aichallenge 2011-02-08T20:09:27 i'm a minimalist 2011-02-08T20:09:50 provided it's clear 2011-02-08T20:10:00 i don't use 1 name variables for everything, that would be painful 2011-02-08T20:10:02 but you were against "ur" :P 2011-02-08T20:10:09 and not ugly 2011-02-08T20:10:30 bah, mathematicians love one letter variables 2011-02-08T20:11:19 "Ur" is the name of a programming language i've been looking into today 2011-02-08T20:12:10 i also think that it's better to share strategies and high level algorithms than to share straight up source code 2011-02-08T20:12:10 'The signature of the standard library is such that well-typed Ur/Web programs "don't go wrong" ' 2011-02-08T20:12:11 haha 2011-02-08T20:12:33 sigh: that's actually a common thing to say about most statically typed, pure languages 2011-02-08T20:12:54 :P 2011-02-08T20:12:54 but yeah, that's the gist 2011-02-08T20:13:20 you can screw up, but it is guaranteed not to crash, error out, or become a security problem 2011-02-08T20:13:32 and the compiler will catch most screwups anyway 2011-02-08T20:14:05 fair enough 2011-02-08T20:14:35 sigh: i use a lot of 1 letter variables, but not everything 2011-02-08T20:14:55 the thing about 1-letter variables is that sometimes there's really just not anything better to use 2011-02-08T20:15:06 probably because you are forced to, I remember you complaining about being restricted to ascii 2011-02-08T20:15:07 especially if you are writing very polymorphic code (no idea what the actual type is) 2011-02-08T20:15:17 usually just for iterators in for loops etc. a = ant, d = direction, r,c = row,col etc. 2011-02-08T20:15:30 well, for example, generic list operations 2011-02-08T20:15:37 no idea what to name the elements. x is fine 2011-02-08T20:15:42 yeah I know, usally in maths everything is very genral 2011-02-08T20:15:47 exactly 2011-02-08T20:15:51 *** Mathnerd314 has quit IRC (Ping timeout: 250 seconds) 2011-02-08T20:15:56 i hope mcleopold has switched to using row col instead of x y 2011-02-08T20:16:03 i refuse to back down on that 2011-02-08T20:16:08 somebody did i think 2011-02-08T20:16:20 i saw a big commit message related to that somewhere 2011-02-08T20:16:34 i think everyone should mack row col and x y equivalent 2011-02-08T20:16:42 although it's tricky when drawing 3d graphs 2011-02-08T20:17:02 i have no idea how i got mack out of make 2011-02-08T20:17:10 row col isn't *always* better 2011-02-08T20:17:17 when isn't it? 2011-02-08T20:17:25 e.g. one of the operands in matrix multiplication 2011-02-08T20:17:50 what for traversing the right multipliees columns? 2011-02-08T20:17:55 when you know that multiplication is coming up, might as well orient it ideally in the first place 2011-02-08T20:18:24 i've never cared to optimise that far 2011-02-08T20:18:39 *** Frontier has quit IRC (Quit: Ex-Chat) 2011-02-08T20:18:41 a decent language and library combination can do it on your behalf 2011-02-08T20:19:04 i'm a fan of writing my own tools when i don't need "the fastest", it's good learning 2011-02-08T20:19:05 or at least make it very easy 2011-02-08T20:19:52 well it's easier to write the actual multiplication that way too 2011-02-08T20:20:05 you can make it a pretty simple parallel loop 2011-02-08T20:20:17 it's just a matter of how hard/easy it is to arrange for that simplicity in the first place 2011-02-08T20:21:30 matrix multiplication algorithm is so easy i love it 2011-02-08T20:21:31 i sure think about optimization a lot for somebody who doesn't code in Fortran... 2011-02-08T20:21:43 it's like 4-5 lines 2011-02-08T20:21:56 4-5 lines? 2011-02-08T20:22:16 if dot product is defined it's like 1 line 2011-02-08T20:22:25 and dot product is like one line, too 2011-02-08T20:22:32 yeah i just use for loops 2011-02-08T20:34:54 aichallenge: McLeopold epsilon * r5a4cab0 / (4 files in 2 dirs): Started switch to bots running on stdin. Added common functionality to ants that should exist for all games. - http://bit.ly/g0jiA5 2011-02-08T20:34:55 aichallenge: McLeopold epsilon * r3978692 / (ants/bots/MyBot.py ants/bots/ants.py ants/playgame.py): Beginning of python starter pack bot. - http://bit.ly/eTyQdw 2011-02-08T20:37:59 *** Mathnerd314 has joined #aichallenge 2011-02-08T20:52:06 *** bhasker has quit IRC (Ping timeout: 255 seconds) 2011-02-08T20:59:06 *** amstan has joined #aichallenge 2011-02-08T20:59:06 *** ChanServ sets mode: +o amstan 2011-02-08T21:05:55 *** Chris_0076 has joined #aichallenge 2011-02-08T21:14:51 McLeopold is engine.py in epsilon yours? 2011-02-08T21:15:59 j3camero: he's not here 2011-02-08T21:16:17 j3camero: git blame engine.py 2011-02-08T21:17:00 Wait who is Scott Hamilton? 2011-02-08T21:17:09 mcleopold? 2011-02-08T21:17:24 I see both names in the blame file. 2011-02-08T21:17:30 McLeopold == Scott Hamilton 2011-02-08T21:17:33 Just wondering if it's a dude I knew in high school. 2011-02-08T21:17:42 i think he's older than you 2011-02-08T21:17:42 You know roughly where he lives? 2011-02-08T21:17:50 us somewhere 2011-02-08T21:17:58 Hmm probably just a coincidence. 2011-02-08T21:18:03 he has a kid 2011-02-08T21:18:09 i imagine mcleopold in his thirties 2011-02-08T21:18:18 whois says Olympia, WA 2011-02-08T21:18:58 So probably a different Scott Hamilton. 2011-02-08T21:22:14 *** janzert has quit IRC (Read error: Connection reset by peer) 2011-02-08T21:23:46 stalking time: http://www.myspace.com/mcleopold 2011-02-08T21:24:02 *** janzert has joined #aichallenge 2011-02-08T21:24:14 MySpace still exists? 2011-02-08T21:24:23 (And doesn't look horribly ugly anymore, apparently?) 2011-02-08T21:24:25 no no... it's my_____ 2011-02-08T21:24:36 Oh dear, that's lame. 2011-02-08T21:24:54 :P 2011-02-08T21:24:55 yup 2011-02-08T21:25:13 it's still horrible 2011-02-08T21:28:24 ha, i can't delete my myspace page 2011-02-08T21:28:31 i don't have access to the email account 2011-02-08T21:29:26 hmm... now I need to find antimatroid's myspace page 2011-02-08T21:29:43 http://www.myspace.com/helloshittyass 2011-02-08T21:30:16 hmm... that's not your name 2011-02-08T21:30:20 nope :P 2011-02-08T21:30:28 :P 2011-02-08T21:31:26 mmm, that made my life easier 2011-02-08T21:31:54 i changed it to read input from an istream, so i can just give it a file stream rather than pasting stuff to the terminal when doing stuff 2011-02-08T21:32:59 antimatroid: ./myapp Any reason that wouldn't have worked? 2011-02-08T21:33:14 http://www.facebook.com/nick.ham 2011-02-08T21:33:19 you may be able to access that 2011-02-08T21:33:22 but i think it's private 2011-02-08T21:33:24 (Not that istreams aren't ultimately better.) 2011-02-08T21:34:44 it's private :P, but if you are just going to paste these things it's not as fun :D 2011-02-08T21:34:56 haha, :P 2011-02-08T22:06:32 Ugh just finished troubleshooting things for other people's websites. Now going to bed. Looks like I'm not working on the engine today. Sorry guys. 2011-02-08T22:14:10 If somebody else wants to yank that task from me go right ahead. I recommend using the sadbox python module from the planet_wars/backend directory. You use it to launch a bot and then you send and receive strings to/from it. 2011-02-08T22:21:39 wow, Ur rocks. we should just drop Django and use Ur for Zeta. :D 2011-02-08T22:31:34 j3camero: i'm hoping mcleopolds engine will be sufficient 2011-02-08T22:37:38 antimatroid: I don't think he has been building it to talk to sandboxed clients. Doing that is a bit different than talking with a subclass of the Strategy class. I may be mistaken though. 2011-02-08T22:38:56 okay, i kind of take that back. error messages are... cryptic, at best 2011-02-08T22:39:44 jmcarthur: i was going to say, that's a very early comment :P 2011-02-08T22:40:01 antimatroid: it was a joke :P 2011-02-08T22:40:02 fair enough, i'm more interested in a working engine for testing out rules/scoring atm 2011-02-08T22:52:12 *** Prillicy has joined #aichallenge 2011-02-08T22:52:22 from the sounds of things, the competition is nearly ready to go, am I right? 2011-02-08T22:53:32 you all are quite talkative 2011-02-08T22:53:42 Prillicy: That only means that we talk a lot. 2011-02-08T22:54:01 ? 2011-02-08T22:54:24 That we're talkative doesn't mean we're nearly ready to go. Progress has been made, but there is still much progress yet to be made. 2011-02-08T22:54:36 you misunderstand 2011-02-08T22:54:47 those were two entirely different statements 2011-02-08T22:55:00 the first is based on my reading on the forums 2011-02-08T22:55:10 the second was a sarcastic observation 2011-02-08T22:55:37 it's mostly a bunch of bikeshedding 2011-02-08T22:55:42 but I was more hoping it was close to being done more than anything 2011-02-08T22:57:05 Prillicy: It's still probably a few weeks out, but this is unpredictable. 2011-02-08T22:57:28 well, simplistic twitter clone in about 30 lines including html isn't too bad i guess 2011-02-08T22:58:14 no ajax though 2011-02-08T23:02:19 jmcarthur: can we talk food for a second? 2011-02-08T23:02:50 i had pizza tonight 2011-02-08T23:03:41 atm i think it's the case that any food square within range of more than 2 ants dies after killing ants? how about this idea: 2011-02-08T23:03:42 move ants, resolve spawning of new ants/killing food, kill ants/battle 2011-02-08T23:03:42 option 1: if you have more ants closer to it than anyone else, you get the square, otherwise it dies 2011-02-08T23:03:42 option 2: if you have the closest ant to the square, you get it, otherwise it dies 2011-02-08T23:03:52 if you do that after battle resolution, you have to resolve them yet again 2011-02-08T23:04:44 anyone else please comment :P 2011-02-08T23:06:30 i had pizza too 2011-02-08T23:06:37 * antimatroid needs to get lunch 2011-02-08T23:06:41 i shall ride to subway soonish 2011-02-08T23:06:41 I don't like the 1 ant kills many 2011-02-08T23:06:51 Prillicy: what do you mean? 2011-02-08T23:06:51 but that's not food oriented 2011-02-08T23:07:05 it adds complexity to the game in a good way 2011-02-08T23:07:11 from the post 2011-02-08T23:07:18 antimatroid: by option 1 do you mean if you have more ants *in range* of it than anyone else? 2011-02-08T23:08:19 one ant couldn't kill many 2011-02-08T23:08:19 besides, it takes away the benefit of keeping your ants grouped together 2011-02-08T23:08:19 in doing so, it encourages intentional separation and scavenging, removing strategy 2011-02-08T23:08:42 Prillicy: you are talking about option 2 in the forum post? 2011-02-08T23:08:43 jmcarthur: yeah 2011-02-08T23:09:07 Prillicy: jmcarthur and i are very much in favour of option 1 over 2 2011-02-08T23:09:07 I am not sure where it was in the post 2011-02-08T23:09:21 antimatroid: i'm not sure which i like better. the first one mimicks my favorite combat rule, but maybe food shouldn't be the same 2011-02-08T23:09:22 which encourages you to move around in packs to attack/defend better 2011-02-08T23:09:31 but you also need to separate out if you want any chance of winning 2011-02-08T23:09:36 not only in packs, but in *formations* ^^ :) 2011-02-08T23:09:52 precisely, creating strategy 2011-02-08T23:10:01 seems that we all agree then 2011-02-08T23:10:04 yes, that's what we want, but jeff thinks that will discourage beginners 2011-02-08T23:10:30 and unfortunately jeff is the only one whose opinion matters because he is the benevolent dictator 2011-02-08T23:10:44 checking what option 1 was 2011-02-08T23:10:57 you need to have someone make the final choices, i'd pick jeff over a lot of other people 2011-02-08T23:12:23 can anyone work out what the hell dimkadimons last example is doing? 2011-02-08T23:12:27 .A. -> ... 2011-02-08T23:12:27 BaC    ... 2011-02-08T23:12:46 * jmcarthur checks the forums for the first time all day 2011-02-08T23:14:20 just remind jeff that the option he creates removes all semblance of strategy and replaces it with, simply, if your bot unctions at all, you're equal 2011-02-08T23:14:26 i think the capitals are one team and the lowercase is another, but what he says along with it is simply wrong, if he's talking about option 1 2011-02-08T23:14:31 i have another example for why i like 1 over 2, give me a minute 2011-02-08T23:14:50 *** janzert1 has joined #aichallenge 2011-02-08T23:14:50 if he's talking about option 2, then yeah, that's right 2011-02-08T23:15:12 a.b.c 2011-02-08T23:15:12 ..... 2011-02-08T23:15:12 ..d.. 2011-02-08T23:15:14 in fact, if you refer to the need for formations to counter that, then you're seriously discouraging beginners 2011-02-08T23:15:22 surely it's better to have just b die there? 2011-02-08T23:15:51 *** janzert has quit IRC (Ping timeout: 276 seconds) 2011-02-08T23:16:51 i don't see why people think "ranged" attacks are hard to understand? 2011-02-08T23:16:54 it's pretty straight forward 2011-02-08T23:18:42 I guess I'm just preaching to the choir 2011-02-08T23:18:51 Prillicy: pretty much 2011-02-08T23:18:54 I'll go back to waiting patiently 2011-02-08T23:19:00 i think they see complicated diagrams and go "ow my bwain!" without taking a moment to think 2011-02-08T23:19:02 most people seem to want option 1 imo 2011-02-08T23:20:37 I wouldn't even suspect option 2 to be a possibility, it was odd to me when I saw it. If I had coded something like that, I'd call it a bug 2011-02-08T23:23:15 it's harder to calculate and less efficient 2011-02-08T23:23:21 ad requires full map information to ensure you get it right 2011-02-08T23:24:15 we haven't even really analysed situations where lots of players ants are in a region 2011-02-08T23:24:30 i think that would be awesome with option 1 2011-02-08T23:24:43 not fight region, a larger region working out where best to move 2011-02-08T23:24:49 it also sounds like it would greatly favor whoever goes first 2011-02-08T23:24:56 it's simultaneous 2011-02-08T23:25:06 oh, right, that was stated 2011-02-08T23:25:23 ants don't die until after 2011-02-08T23:25:36 yep, move ants, then "resolve state" 2011-02-08T23:25:48 it's got a very game of life appeal 2011-02-08T23:28:00 well, option 2 and 3 don't 2011-02-08T23:28:15 i'm all about option 1:P 2011-02-08T23:34:05 *** Error323 has quit IRC (*.net *.split) 2011-02-08T23:34:05 *** jmpespxoreax has quit IRC (*.net *.split) 2011-02-08T23:34:07 *** BtbN has quit IRC (*.net *.split) 2011-02-08T23:34:08 *** Accoun has quit IRC (*.net *.split) 2011-02-08T23:34:08 *** j3camero has quit IRC (*.net *.split) 2011-02-08T23:34:09 *** janzert1 has quit IRC (*.net *.split) 2011-02-08T23:34:09 *** Apophis_ has quit IRC (*.net *.split) 2011-02-08T23:34:09 *** delt0r_ has quit IRC (*.net *.split) 2011-02-08T23:34:09 *** smellyhippy has quit IRC (*.net *.split) 2011-02-08T23:34:10 *** Prillicy has quit IRC (*.net *.split) 2011-02-08T23:34:11 *** pgpaskar_ has quit IRC (*.net *.split) 2011-02-08T23:34:12 *** Chris_0076 has quit IRC (*.net *.split) 2011-02-08T23:34:12 *** amstan has quit IRC (*.net *.split) 2011-02-08T23:34:12 *** Mathnerd314 has quit IRC (*.net *.split) 2011-02-08T23:34:13 *** AlliedEnvy has quit IRC (*.net *.split) 2011-02-08T23:34:13 *** javagamer_ has quit IRC (*.net *.split) 2011-02-08T23:34:14 *** Zannick has quit IRC (*.net *.split) 2011-02-08T23:34:14 *** nullkuhl has quit IRC (*.net *.split) 2011-02-08T23:34:14 *** dr`away has quit IRC (*.net *.split) 2011-02-08T23:34:15 *** antimatroid has quit IRC (*.net *.split) 2011-02-08T23:34:16 *** JamesMG has quit IRC (*.net *.split) 2011-02-08T23:34:16 *** politik has quit IRC (*.net *.split) 2011-02-08T23:37:50 *** janzert1 has joined #aichallenge 2011-02-08T23:37:50 *** Prillicy has joined #aichallenge 2011-02-08T23:37:50 *** Chris_0076 has joined #aichallenge 2011-02-08T23:37:50 *** amstan has joined #aichallenge 2011-02-08T23:37:50 *** Mathnerd314 has joined #aichallenge 2011-02-08T23:37:50 *** Apophis_ has joined #aichallenge 2011-02-08T23:37:50 *** delt0r_ has joined #aichallenge 2011-02-08T23:37:50 *** Accoun has joined #aichallenge 2011-02-08T23:37:50 *** Error323 has joined #aichallenge 2011-02-08T23:37:50 *** antimatroid has joined #aichallenge 2011-02-08T23:37:50 *** AlliedEnvy has joined #aichallenge 2011-02-08T23:37:50 *** smellyhippy has joined #aichallenge 2011-02-08T23:37:50 *** javagamer_ has joined #aichallenge 2011-02-08T23:37:50 *** Zannick has joined #aichallenge 2011-02-08T23:37:50 *** jmpespxoreax has joined #aichallenge 2011-02-08T23:37:50 *** j3camero has joined #aichallenge 2011-02-08T23:37:50 *** nullkuhl has joined #aichallenge 2011-02-08T23:37:50 *** dr`away has joined #aichallenge 2011-02-08T23:37:50 *** JamesMG has joined #aichallenge 2011-02-08T23:37:50 *** pgpaskar_ has joined #aichallenge 2011-02-08T23:37:50 *** BtbN has joined #aichallenge 2011-02-08T23:37:50 *** politik has joined #aichallenge 2011-02-08T23:37:50 *** card.freenode.net sets mode: +o amstan 2011-02-08T23:38:00 *** janzert1 has quit IRC (*.net *.split) 2011-02-08T23:38:00 *** Apophis_ has quit IRC (*.net *.split) 2011-02-08T23:38:01 *** delt0r_ has quit IRC (*.net *.split) 2011-02-08T23:38:01 *** smellyhippy has quit IRC (*.net *.split) 2011-02-08T23:38:02 *** j3camero has quit IRC (Read error: Connection reset by peer) 2011-02-08T23:41:32 *** j3camero has joined #aichallenge 2011-02-08T23:41:32 *** janzert1 has joined #aichallenge 2011-02-08T23:41:32 *** Apophis_ has joined #aichallenge 2011-02-08T23:41:32 *** delt0r_ has joined #aichallenge 2011-02-08T23:41:32 *** smellyhippy has joined #aichallenge 2011-02-08T23:43:25 with option 2, you also want to end up attacking people from as far as possible don't you? least some idiot decided to attack from closer 2011-02-08T23:43:45 yeah 2011-02-08T23:43:51 i've described it as a game of chicken 2011-02-08T23:44:02 you keep everybody spread out and attack with one ant at a time 2011-02-08T23:44:41 there could be some micro in the sense that you can use "fear" to kind of herd an opponent's army around 2011-02-08T23:44:48 but that's... kind of a macro thing 2011-02-08T23:46:16 damnit, we must have option 1 2011-02-08T23:59:38 hello, is there anything confirmed yet regarding the start date of the next contest ? or even an estimate ?