2010-11-12T00:00:01 it seems to me like you're trying to prevent coordinates from being close to integer values 2010-11-12T00:00:28 sigh: :P 2010-11-12T00:00:45 Zannick: i was told to explicitly exclude integer coordinates 2010-11-12T00:01:01 not integer coordinates 2010-11-12T00:01:03 what?? 2010-11-12T00:01:08 ? 2010-11-12T00:01:10 near integer distances 2010-11-12T00:01:15 Zannick: yes, to prevent possible rounding errors 2010-11-12T00:01:17 oh okay, i need to change that thena 2010-11-12T00:01:33 and it wasn't checking it for home planets, will fix that up too 2010-11-12T00:01:51 hm 2010-11-12T00:02:01 i guess that's kind of legit 2010-11-12T00:02:07 for distances, at least 2010-11-12T00:02:08 errr... it doesn't matter if the coordinates are near integers 2010-11-12T00:02:10 only distances 2010-11-12T00:02:53 yeah, i see that now :P 2010-11-12T00:03:24 i think i said last night my life philosophy is do first think later 2010-11-12T00:03:36 code should check abs(distance) < epsilon or abs(distance) - 1 > epsilon 2010-11-12T00:03:53 err.. ignore that 2010-11-12T00:03:56 :P 2010-11-12T00:05:12 meh, you are smart enough to figure it out 2010-11-12T00:06:25 except you want to use the nonrounded form of distance for that 2010-11-12T00:06:34 yup 2010-11-12T00:07:25 *** acieroid has quit IRC (Read error: Connection reset by peer) 2010-11-12T00:07:38 *** acieroid has joined #aichallenge 2010-11-12T00:09:07 only other comments i have are on formatting 2010-11-12T00:09:10 *** hellman has joined #aichallenge 2010-11-12T00:09:49 it's missing that threshold thing based on growth rate. is that important? 2010-11-12T00:10:31 that was a way of establishing different minimum distances between planets 2010-11-12T00:11:02 threshold based on growth rate? 2010-11-12T00:11:04 it would be useful to keep that, though 2010-11-12T00:11:14 sure, but that was used on top of the <= 1 requirement 2010-11-12T00:11:20 *** hellman has quit IRC (Remote host closed the connection) 2010-11-12T00:11:33 *** davidjliu has quit IRC (Quit: Page closed) 2010-11-12T00:11:35 no 2010-11-12T00:11:42 *** hellman has joined #aichallenge 2010-11-12T00:12:05 let's see, 0.4 * the sum of the square roots of the growth rates 2010-11-12T00:12:10 that was a "what the fuck are you talking about" question mark 2010-11-12T00:12:23 was the minimum distance between two planets 2010-11-12T00:12:31 line 36 in the original script: if dist < threshold or dist <= 1: 2010-11-12T00:12:35 and that could be 0.8 for some cases 2010-11-12T00:13:12 wait, haha 2010-11-12T00:13:37 the biggest that would be is 0.4 * (sqrt(5) + sqrt(5)) 2010-11-12T00:13:41 which is less than 2010-11-12T00:13:43 which is less than 2 2010-11-12T00:13:57 hahahaha 2010-11-12T00:14:04 antimatroid: no, you're perfectly fine without that 2010-11-12T00:14:21 since you basically are using a bigger minimum distance 2010-11-12T00:14:44 *** delt0r has joined #aichallenge 2010-11-12T00:15:06 *** delt0r___ has quit IRC (Ping timeout: 245 seconds) 2010-11-12T00:15:06 i can't believe i didn't notice that when i edited that file last 2010-11-12T00:16:38 :) 2010-11-12T00:16:41 Top 10 players: bocsimacko(4416), george(4229), asavis(4184), shangas(4177), Raschi(4171), GreenTea(4126), oldman(4111), dmj111(4094), _Astek_(4086), _iouri_(4048) 2010-11-12T00:17:16 http://codepad.org/noN1rFFp take two 2010-11-12T00:17:26 i reduced min starting distance to 5 2010-11-12T00:17:43 *** codefun64 has quit IRC (Ping timeout: 265 seconds) 2010-11-12T00:18:28 *** iris1 has quit IRC (Ping timeout: 260 seconds) 2010-11-12T00:19:45 how hard would it be to instead ensure there is a neutral with cost less than 5 * home world distance 2010-11-12T00:20:13 make it the first neutral generated 2010-11-12T00:20:29 at least I think that is the correct formula, someone who is actually playing and thinking about these things feel free to chime in on that :) 2010-11-12T00:20:47 that's enough strict enough 2010-11-12T00:20:47 janzert: easy, i'll just make sure the first two neutrals added meet that 2010-11-12T00:21:05 actually 2010-11-12T00:21:08 wait, home world distance...as in dist(p1, p2)? 2010-11-12T00:21:22 hm, i see what that entials 2010-11-12T00:21:25 entails 2010-11-12T00:21:28 right distance between the two 2010-11-12T00:21:33 two home worlds 2010-11-12T00:21:33 hold on, i'll do it then let you guys look 2010-11-12T00:21:46 actually... that formula is fine 2010-11-12T00:22:34 I think it is better to just not make the home planets too close 2010-11-12T00:22:45 wait, i need to restrict the number of ships to 25 too? 2010-11-12T00:23:00 the neutral doesn't even need to be within that distance 2010-11-12T00:23:08 there needs to be a neutral with planets < 25? 2010-11-12T00:23:14 with ships* 2010-11-12T00:23:16 antimatroid the only restriction on the neutral is number of ships 2010-11-12T00:23:21 the position of the planet doesn't matter, only cost 2010-11-12T00:23:39 okay, so first neutrals added have ships in [1, 25)? 2010-11-12T00:23:53 5*distance(p1,p2) 2010-11-12T00:24:02 yes der :P 2010-11-12T00:24:05 okay, hold up 2010-11-12T00:24:13 [1, 5*distance(p1,p2)) 2010-11-12T00:24:26 you can make the range inclusive 2010-11-12T00:24:31 if you want 2010-11-12T00:24:41 i think it's better if it's not 2010-11-12T00:25:05 and you could make the home world distance minimum down lower then 2010-11-12T00:25:44 it still doesn't stop draws though... map 30 obeys this condition and is still a boring map 2010-11-12T00:26:20 any chance you have a handy playback showing map 30 2010-11-12T00:27:25 nm found it on tcp 2010-11-12T00:27:50 http://72.44.46.68/canvas?game_id=976319 for everyone else 2010-11-12T00:28:03 map 30 doesn't follow that rule does it? 2010-11-12T00:28:19 it is possible to aquire the 20 in the corner 2010-11-12T00:29:01 how far are the home planets, 4? 2010-11-12T00:29:06 or only 3 2010-11-12T00:29:20 I think 5 2010-11-12T00:29:44 startDist is 5 2010-11-12T00:29:59 janzert: i was just thinking, i'll remove starting distance being greater than 2 2010-11-12T00:30:03 i think it's fine now 2010-11-12T00:30:12 does it obey the condition ignoring the center planet? 2010-11-12T00:30:22 hmm, ok so you can expand to the 20 planet but no farther basically 2010-11-12T00:30:41 unless opponent is stupid: http://72.44.46.68/canvas?game_id=978085 2010-11-12T00:30:48 if home distances are 2, just means there exists neutrals for both players that have ships less than 10 2010-11-12T00:31:02 antimatroid: right 2010-11-12T00:31:02 if both bots suck, that's their fault 2010-11-12T00:31:10 :( 2010-11-12T00:31:50 if your opponent is not ragebot you can get planet 8 and 21 (starting at 2) 2010-11-12T00:32:56 *** Accoun has quit IRC () 2010-11-12T00:33:10 how about this will the top 100 bots always at least draw map 30? 2010-11-12T00:33:34 if so then we probably need to start worrying about disallowing it 2010-11-12T00:33:40 janzert: I think so 2010-11-12T00:33:47 i would think a really good bot should be able to win. 2010-11-12T00:33:49 can you be in the top 100 and be beaten by ragebot? 2010-11-12T00:33:49 http://codepad.org/JuoNVCba take 3 2010-11-12T00:34:00 Zannick: prove it :) 2010-11-12T00:34:27 but it will probably still be easier to do it manually than fixing the generator to not create them in the first place 2010-11-12T00:34:27 it was mentioned that one could take a 20-planet 2010-11-12T00:34:30 eburnette_: easily 2010-11-12T00:35:00 if that's true, then the distance from 1 to 2 is at least 5 2010-11-12T00:35:03 that might be an idea though... force everyone to play ragebot in main tournament 2010-11-12T00:35:08 sigh: but you think they all win or draw map 30? 2010-11-12T00:35:09 :P 2010-11-12T00:35:21 janzert: draw against other top bots 2010-11-12T00:35:25 and doing that ought to be enough to put a player ahead 2010-11-12T00:35:34 i was just thinking, with map generation in final tournament, would it be possible to generate a new map every time a pairing has played on all maps already generated? 2010-11-12T00:35:43 I mean win or draw against perfect play :) 2010-11-12T00:35:46 if it takes this much thought then it should be in there 2010-11-12T00:35:48 but there is no way to prevent bots from doing the exact same thing 2010-11-12T00:35:57 and start with 1 map, then generate as needed? 2010-11-12T00:35:58 i.e. against anyone 2010-11-12T00:36:10 *** Accoun has joined #aichallenge 2010-11-12T00:36:32 antimatroid: I'd rather not add that complexity 2010-11-12T00:36:34 janzert: I'm 90% sure it will draw against perfect play 2010-11-12T00:36:38 99% 2010-11-12T00:36:54 then we're going to need to work out a number of planets to be generated for the final contest 2010-11-12T00:37:05 generate 100,000 maps and you won't have to worry about it antimatroid 2010-11-12T00:37:09 i'd say too many will make results more unstable, but too few results in repeats 2010-11-12T00:37:34 just round-robin n times on n maps 2010-11-12T00:37:40 you can't round robin 2010-11-12T00:37:47 it's too inefficient 2010-11-12T00:38:06 :P 2010-11-12T00:38:17 yeah, round-robin is a terribly inefficient use of games 2010-11-12T00:38:49 i'd say 20-30 maps max 2010-11-12T00:38:56 hmm, yeah I guess ideally we want as many maps as the largest number of pairings 2010-11-12T00:39:06 repeat pairings 2010-11-12T00:39:17 repeat pairings on a map are pointless 2010-11-12T00:39:26 right 2010-11-12T00:39:33 maybe 2010-11-12T00:39:36 although can repeat starting positions i think 2010-11-12T00:39:46 repeat with them switched* 2010-11-12T00:39:50 yeah... I think you should actually work out estimates on paper to figure out the number of maps 2010-11-12T00:39:59 there is a random factor due to timing 2010-11-12T00:40:07 or use of random() 2010-11-12T00:40:09 antimatroid: yeah, if necessary not as ideal as a different map though 2010-11-12T00:40:34 would it be very hard for it to generate a new map every time a pairing requires it? 2010-11-12T00:40:41 i think that'd be ideal 2010-11-12T00:41:04 I would have to think about how hard that might be 2010-11-12T00:41:14 you'd have to keep track of the number of times a pairing occurs 2010-11-12T00:41:21 remember new maps would have to be distributed to all the workers 2010-11-12T00:41:23 already happening pretty sure 2010-11-12T00:41:37 yeah, i have no idea how everything works :) 2010-11-12T00:41:40 Zannick: the new pairing code does that 2010-11-12T00:41:49 janzert: ah, excellent 2010-11-12T00:41:57 only probl is you could never repeat a battle (for understanding why the bot made a certain move) 2010-11-12T00:42:14 i think that map generator is solid now 2010-11-12T00:42:19 at least by map number anyway 2010-11-12T00:42:55 why not have first pairing on map 1, second on map 2, etc? 2010-11-12T00:43:04 or would that be less fair? 2010-11-12T00:43:39 that would heavily favor map 1 in initial rankings 2010-11-12T00:43:46 antimatroid: mind making up a zip of say 20 maps from it and post it on the forums to try and get some feedback from players on them 2010-11-12T00:43:47 sigh: that would do it 2010-11-12T00:43:53 generate 200 maps, then do what sigh said 2010-11-12T00:44:03 nah ,I think Zannick is right 2010-11-12T00:44:20 and you'd probably not get to all 200 2010-11-12T00:44:39 yeah, you really need to randomly choose within the set of maps being used 2010-11-12T00:44:39 yeah, i read that after 2010-11-12T00:45:08 in that case, why do we want to restrict the maps to a small set? 2010-11-12T00:45:10 which really means you want the initial set to be about the right size to start with and only grow if needed 2010-11-12T00:45:12 why not 1000? 2010-11-12T00:45:21 janzert: that sounds good though, increment usable maps when necessary, and pick between unused maps when paired 2010-11-12T00:45:24 what is the advantage of a small set? 2010-11-12T00:45:26 yet in that case, you need to remember which maps you've used 2010-11-12T00:45:31 start with like 20 2010-11-12T00:46:03 sigh: i think it's better if lots of different bots play on the same map 2010-11-12T00:47:09 playing devil's advocate, why is that better? 2010-11-12T00:48:10 better confidence in rankings 2010-11-12T00:48:29 if you have a beats b beats c it's probably more likely a beats c if they both played the same map(s) 2010-11-12T00:48:47 janzert: do you want me to post a link to the code in the forum post too? 2010-11-12T00:48:57 hmm..fair enough 2010-11-12T00:48:58 sure, then people can play with it 2010-11-12T00:49:37 when you make the sample maps, give them new numbers, >1000 2010-11-12T00:50:06 map2k1.txt 2010-11-12T00:50:14 eburnette_: what? 2010-11-12T00:50:14 heh 2010-11-12T00:50:19 i've already made them, i'm uploading them now 2010-11-12T00:50:22 no, numbers are better, actually 2010-11-12T00:50:48 I don't want them to be confused with the maps that already exist 2010-11-12T00:51:07 wait, no i'm not 2010-11-12T00:51:12 call 'em antimap.txt :) 2010-11-12T00:51:21 antimaproid 2010-11-12T00:51:29 that works too 2010-11-12T00:51:43 just don't callem map1 map2 etc. 2010-11-12T00:54:51 they're testmap1-20 2010-11-12T00:56:17 http://tinyurl.com/testmaps 2010-11-12T00:57:42 *** virdo has quit IRC (Ping timeout: 240 seconds) 2010-11-12T00:59:52 blech I hate spending all night on improvements that don't improve things 2010-11-12T01:00:03 night all 2010-11-12T01:00:14 *** eburnette_ has quit IRC (Quit: Page closed) 2010-11-12T01:01:27 *** kincody has joined #aichallenge 2010-11-12T01:04:13 *** hellman has quit IRC (Remote host closed the connection) 2010-11-12T01:08:59 btw, if everyone here could test those maps and report back how your bot handles them (crashes, plays stupidly or plays normal) that would be great 2010-11-12T01:09:38 http://ai-contest.com/forum/viewtopic.php?f=19&t=1087 2010-11-12T01:10:23 oh, i'll add something about brand new maps being generated for contest too 2010-11-12T01:10:53 janzert: are you considering using that map generator for the finals? 2010-11-12T01:10:53 *** kincody has quit IRC (Ping timeout: 265 seconds) 2010-11-12T01:11:12 antimatroid: you raise good points, i agree, we should have those maps in 2010-11-12T01:11:15 like.. now 2010-11-12T01:11:17 lol 2010-11-12T01:11:17 mega1: yes 2010-11-12T01:11:27 *** perestre1ka has quit IRC (Ping timeout: 255 seconds) 2010-11-12T01:11:28 mega1: I believe j3camero was very much in favor of it 2010-11-12T01:11:31 janzert: let's just add them 2010-11-12T01:11:36 *** yasith has joined #aichallenge 2010-11-12T01:11:36 amstan: patience 2010-11-12T01:11:48 I'm personally slightly worried about doing it at this stage 2010-11-12T01:11:50 if they're valid maps i don't see why not 2010-11-12T01:12:08 we constantly added maps last contest 2010-11-12T01:12:16 eburnette_ for example had something in his code that depends on the old maps 2010-11-12T01:12:25 amstan: because we are sort of changing the definition of valid here 2010-11-12T01:12:26 but i think people shouldn't have assumed that anyway 2010-11-12T01:12:37 I asked two weeks ago to pin down the generator 2010-11-12T01:12:43 janzert: the only thing is different is the symmetry 2010-11-12T01:13:01 mega1: i'm trying to pin it down now, do you not like these types of maps? any particular reason? 2010-11-12T01:13:12 no, I like them 2010-11-12T01:13:34 janzert: if there is a time to add maps to test things it's not 2010-11-12T01:13:35 now* 2010-11-12T01:13:43 what I don't much like is changing the generator two weeks before the end of a three month contest 2010-11-12T01:13:53 hmm, I think it was 2 weeks ago :/ 2010-11-12T01:14:03 *** perestrelka has joined #aichallenge 2010-11-12T01:14:07 but yeah, I certainly don't want to just throw them into the final 2010-11-12T01:14:19 janzert: here.. 2010-11-12T01:14:25 janzert: we can add just one 2010-11-12T01:14:33 how many maps are there in total 2010-11-12T01:14:37 janzert: what about adding some of those 20 now, and seeing what happens? 2010-11-12T01:14:47 i say yes 2010-11-12T01:14:50 can we wait till tomorrow and see what the feedback is on the forums 2010-11-12T01:14:57 yeah, that's probably good 2010-11-12T01:15:06 removing them again is much harder than adding them 2010-11-12T01:15:24 i don't think it's really changing the generator, when no generator had ever been set in stone 2010-11-12T01:15:28 janzert: is it? 2010-11-12T01:15:31 it's always been said that the old one wasn't intended to be used 2010-11-12T01:15:45 amstan: rankings depend on the results 2010-11-12T01:15:56 antimatroid: was it said? 2010-11-12T01:15:57 yes, although adding them won't be completely trivial either :} 2010-11-12T01:15:59 we need dhartmei to add some to tcp 2010-11-12T01:16:00 janzert: the old games will still be there, yes, but taking maps out so no further people use them 2010-11-12T01:16:17 mega1: j3camero has said it numerous times on here, but that's not really official i will admit 2010-11-12T01:16:40 Top 10 players: bocsimacko(4447), george(4277), Raschi(4198), shangas(4194), asavis(4184), GreenTea(4155), dmj111(4125), oldman(4124), dimkadimon(4109), _Astek_(4093) 2010-11-12T01:16:52 and tables in the database refer to them 2010-11-12T01:16:55 j3camero is pretty spontaneous, he won't have much objections to this either 2010-11-12T01:17:10 best not to really delete them at all but add the code to disable them really 2010-11-12T01:17:16 i'm in favour of waiting for forum responses 2010-11-12T01:18:08 if someone does decide to add them, remember to get them on all the workers and in svn for new workers before adding them to the database 2010-11-12T01:18:16 otherwise it's going to be a real mess 2010-11-12T01:19:07 *** sigh has quit IRC () 2010-11-12T01:19:51 fyi, to answer a question I saw go by earlier, there are currently 100 maps in use 2010-11-12T01:22:39 is it still just the 100 that came with starter packs? 2010-11-12T01:24:37 I believe so 2010-11-12T01:24:44 it's the 100 in svn 2010-11-12T01:25:01 the first responded in the thread is under the impression that currently those 100 are intended to be the ones used 2010-11-12T01:25:20 that would be the worst imo, it's meant to be an ai contest, not who can solve these 100 maps 2010-11-12T01:25:51 at least i think that's what they meant 2010-11-12T01:25:53 yeah, definitely the maps will be regenerated for the final 2010-11-12T01:26:44 i think it's important all these decisions are made before the contest next time, which includes working out how to make maps properly 2010-11-12T01:27:44 the only problem, well one anyway, is that a lot of these problems don't become evident until lots of people start pounding on them 2010-11-12T01:28:37 yep, neutral resolution was a good one for that 2010-11-12T01:30:12 *** codefun has joined #aichallenge 2010-11-12T01:30:28 * codefun pokes the chat and wonders if anything's up 2010-11-12T01:33:24 codefun: http://ai-contest.com/forum/viewtopic.php?f=19&t=1087&p=7127#p7127 2010-11-12T01:34:14 antimatroid: i would be interested to know what he meant by " People (such as myself) have written bots that are specifically geared to perform well on the current maps." 2010-11-12T01:34:27 does he actually rely on the middle planet thing? 2010-11-12T01:34:36 eburnette said he does 2010-11-12T01:34:42 but not majorly 2010-11-12T01:34:57 imo it's never been said that maps will meet that requirement 2010-11-12T01:35:04 relying on specific settings = not good AI. 2010-11-12T01:35:05 one common thing is simple code for splitting the map in two may break I suppose 2010-11-12T01:35:22 you can still split the map in half though 2010-11-12T01:35:24 and that seems to be a fairly common thing to be mentioned as first turn code on the forums 2010-11-12T01:35:36 antimatroid: yeah, just the current code might break 2010-11-12T01:35:42 the way i split it is unaffected by the change 2010-11-12T01:35:46 but it shouldn't be a big deal 2010-11-12T01:35:47 codefun: well.. no... it is good AI, like... you can only go so far.. 2010-11-12T01:38:18 antimatroid: point symmetric maps might be in the theme too, they do look like galaxies 2010-11-12T01:38:40 point symmetric? 2010-11-12T01:38:54 point, symmetric... 2010-11-12T01:40:37 antimatroid: how they are now, i mean 2010-11-12T01:40:39 Has anyone else thought of the fact the "radially symmetric" maps are just inversely bilaterally symmetric? Like you make one side, copy it, turn it upside down, then put it on the other side..? 2010-11-12T01:41:19 codefun: i don't think you can do radially symmetric maps unless players are at angles 180 degrees apart 2010-11-12T01:42:02 i was going to have the generator randomly switch between them but it doesn't work 2010-11-12T01:43:04 codefun: yeah.. i'm not sure.. i'm split now 2010-11-12T01:43:26 i was fine with introducing the linearly symmetric maps 2010-11-12T01:43:51 till i saw that people were relying on the center planet thing: "players home planets were required to start in a position such that the line joining them intersected the center neutral planet." 2010-11-12T01:44:19 we have spent two and a half months [unknowingly] optimizing for the old map generator 2010-11-12T01:44:22 mega1: are you doing this? 2010-11-12T01:44:39 do you really think it's a good idea to change it now? 2010-11-12T01:44:47 amstan: no 2010-11-12T01:45:00 mega1: well.. you said unknowingly 2010-11-12T01:45:01 How can you unknowingly optimize for the exact same map configurations to be used? 2010-11-12T01:45:17 for the _distribution_ of maps 2010-11-12T01:45:47 So.. the rules do not state that we may not change maps, we could do it at any time. The problems with that is that we'll make a lot of people unhappy, and to what end? 2010-11-12T01:45:49 I'm not sure I follow. 2010-11-12T01:46:05 Well, I, for one, think switching up maps is great. 2010-11-12T01:46:22 codefun: once you have different map types, you'll surely find some bugs that manifests only on the linear symmetric maps 2010-11-12T01:46:22 codefun: do you have tunable parameters in your code? 2010-11-12T01:46:27 It gives people a chance to see how their AI fares in the big bad world of uncertainties. 2010-11-12T01:47:23 amstan: jeff gave me the idea of these maps 2010-11-12T01:47:24 codefun: did you throw away certain changes or approaches because they didn't perform well enough? 2010-11-12T01:47:40 antimatroid: oh man.. he'll definitely put them in then 2010-11-12T01:47:59 well, the maps that you tested on were the current ones 2010-11-12T01:48:27 everyone has burned weeks of cpu power testing ideas 2010-11-12T01:48:49 changing the maps late invalidates much of that effort. 2010-11-12T01:49:19 mega1: starting maps have always just been examples 2010-11-12T01:49:29 I the generator is changed I have to put trying new developments on hold and repeat testing. 2010-11-12T01:49:31 i have always been under that understanding, both for this and tron 2010-11-12T01:49:39 tron has significantly different maps added at the end 2010-11-12T01:49:48 antimatroid: just in terms of size 2010-11-12T01:50:00 amstan: and obstables 2010-11-12T01:50:01 antimatroid: that's was unfair for the same reason 2010-11-12T01:50:15 antimatroid: we didn't have those change, just size 2010-11-12T01:50:40 i think a size change is worse 2010-11-12T01:50:55 antimatroid: for an optimization task moving the target is a strange thing to do 2010-11-12T01:50:58 size change would make battles a bit... drawn out and strange 2010-11-12T01:51:24 and the target is a distribution of maps (i.e. what the generator produces) 2010-11-12T01:51:46 i prefer the greater level of generality 2010-11-12T01:51:55 larger maps means more calculations which means that actually could cause our bots to crash due to time constraints 2010-11-12T01:52:00 and i don't think anyones code is going to break in such a way that can't easily be fixed 2010-11-12T01:52:03 the 900 hundred sample maps that I used for testing should be a pretty good sample 2010-11-12T01:52:27 antimatroid: i see mega's pov, he spent tons of time testing and optimizing his code 2010-11-12T01:52:38 on an assumption he made 2010-11-12T01:52:51 anti, I think changing map design/configuration/symmetry is fine, personally 2010-11-12T01:52:53 of course, I have the most invested in the current status quo 2010-11-12T01:53:03 antimatroid: how would you of done it then? 2010-11-12T01:53:05 If my AI fails because of map configuration I must've failed in making an intelligent machine 2010-11-12T01:53:14 i'd be surprised of your bot wasn't still far superior on these maps 2010-11-12T01:53:15 but we all have optimized for that generator 2010-11-12T01:53:34 antimatroid: that's not really the point 2010-11-12T01:53:35 mega1: k.. so like.. 2010-11-12T01:53:37 try it! 2010-11-12T01:53:50 codefun: not entirely true, what happens if i thow your both in a 10-dimensional 4 player gmae? 2010-11-12T01:54:14 amstan: i write my bot to be as general as possible inside teh rules that have been given 2010-11-12T01:54:33 antimatroid: ok.. sounds good\ 2010-11-12T01:54:40 antimatroid: does it perform equally well for every map generator within those rules? 2010-11-12T01:55:09 i would expect it to be as random as the maps generated under either generator 2010-11-12T01:55:21 have you tested and discarded changes during the last two months? 2010-11-12T01:55:23 i have nothing about forcing that symmetry 2010-11-12T01:55:41 and on the same side, i could argue that i would have spent more time thinking under that restriction if it were in stone 2010-11-12T01:55:51 so i'm at a disadvantage to those who made an assumption 2010-11-12T01:55:58 oh man.. this is exactly what happened last year with accoun 2010-11-12T01:56:00 although i probably wouldn't have 2010-11-12T01:56:06 he just died when he heard we'll have smaller maps 2010-11-12T01:56:55 so arguing one is disadvantaged from an assumption they made is an invalid argument imo, because the same holds for those who didn't make the assumption and then have that forced on them 2010-11-12T01:57:11 Too much bureaucracy! Make a command decision and stick with it. 2010-11-12T01:57:20 antimatroid: it can easily be that among the versions you discarded as bad, there was a true gem 2010-11-12T01:57:32 There will be some who like it and some who don't, no matter what you do, and they'll both have the same arguments. 2010-11-12T01:57:40 so the thing is... these different maps make the games more interesting, it's under the rules, and if you optimized your bot for the first map generator you made a gamble 2010-11-12T01:58:25 i don't think many people really will have optimised with that restriction anyway 2010-11-12T01:58:33 right, and you can also make bigger changes if you want to shuffle ranking around ... which I don't :-) 2010-11-12T01:58:47 antimatroid: they probably all do, because they used the default map generator 2010-11-12T01:59:12 i don't think anyone modified the map generator to produce other maps such as different number of planets or changes to the symmetry 2010-11-12T01:59:21 given the current rankings are based off the first 100 maps atm, i'd say the rankings are terrible atm 2010-11-12T01:59:32 because a lot of peoples bots can just be running strategies they already created on there 2010-11-12T01:59:39 although i doubt many are 2010-11-12T01:59:48 maybe the guy on top using lisp is 2010-11-12T01:59:51 *** bhasker has quit IRC (Quit: bhasker) 2010-11-12T01:59:57 codefun: guy on top is mega1 2010-11-12T02:00:29 I'm not relying explicitly on the map generator being as it is. 2010-11-12T02:00:44 dimka in the forum says basically the same thing. 2010-11-12T02:01:08 antimatroid: hold on, there's 2 threads, can i merge them? 2010-11-12T02:01:25 amstan: remove the old one if anything 2010-11-12T02:02:02 antimatroid: oh! make it a poll 2010-11-12T02:02:18 *** yasith has quit IRC (Ping timeout: 260 seconds) 2010-11-12T02:02:50 amstan: i'm not sure i can alter it now 2010-11-12T02:03:00 antimatroid: i can do it 2010-11-12T02:05:16 antimatroid: is there any rules about the map designs? 2010-11-12T02:07:10 read the comment post 2010-11-12T02:07:13 *** asdffasf has joined #aichallenge 2010-11-12T02:07:46 amstan: isn't line symmetric the old one? 2010-11-12T02:07:58 shouldn't people be able to vote for that exclusively if you are going to do a poll? 2010-11-12T02:09:04 antimatroid: point symmetric was the old one 2010-11-12T02:10:27 okay, cool 2010-11-12T02:10:41 the generator i wrote will need to be altered to do both kinds of maps 2010-11-12T02:11:02 antimatroid: should be fairly easy 2010-11-12T02:11:07 antimatroid: can i have a link to the code? 2010-11-12T02:11:08 http://codepad.org/JuoNVCba 2010-11-12T02:11:16 bam, i was just on my way to provide it :P 2010-11-12T02:11:45 i had never done anything in python before that, so don't be too harsh on the coding style 2010-11-12T02:12:06 i didn't even realise i could just run the file without compiling, mind blown 2010-11-12T02:12:48 you must learn the ways of scripting... ;) 2010-11-12T02:12:52 antimatroid: where's the symmetry stuff 2010-11-12T02:13:35 amstan: it picks radius and rotation angles randomly then generates coordinates based off of those 2010-11-12T02:13:48 antimatroid: ugh.. yeah.. you need a total redesign then 2010-11-12T02:14:06 why? 2010-11-12T02:14:06 antimatroid: what i did for tron was.. i generated the top half, picked a starting point 2010-11-12T02:14:25 then.. for point symmetry i would copy that top part, mirror X and Y and put it under 2010-11-12T02:14:32 for line symmetry just mirror X 2010-11-12T02:15:01 *** yasith has joined #aichallenge 2010-11-12T02:15:03 oh man! 2010-11-12T02:15:06 epic kind of maps 2010-11-12T02:15:09 make 2 galaxies 2010-11-12T02:15:12 far from eachother 2010-11-12T02:15:19 put a player in each galaxy 2010-11-12T02:15:23 that's just a race to expand 2010-11-12T02:15:28 make the galaxies so far that yeah... 2010-11-12T02:15:30 that's hardly even a game 2010-11-12T02:15:36 so.. the winner will be the one that expands the best 2010-11-12T02:15:52 that's more of who is best at a decision problem than a game 2010-11-12T02:15:57 i did a test with those maps for tron, they were isolated from eachother, it was who survives the most in the maze 2010-11-12T02:16:02 in that my decisions don't affect the other player 2010-11-12T02:16:21 like mario ai, that's not a game 2010-11-12T02:16:25 i see 2010-11-12T02:16:25 it's a decision problem 2010-11-12T02:16:26 ok 2010-11-12T02:16:40 Top 10 players: bocsimacko(4463), george(4276), shangas(4228), asavis(4210), Raschi(4197), GreenTea(4153), oldman(4134), dmj111(4127), Doraj8(4109), _Astek_(4096) 2010-11-12T02:17:20 i wanna know what mcleopold and a1k0n think about this 2010-11-12T02:17:27 *** Tobu has quit IRC (Ping timeout: 255 seconds) 2010-11-12T02:17:49 so do i, Mcleopold hasn't been in the channel all day though 2010-11-12T02:18:27 #later tell McLeopold see this, ping us when done: http://ai-contest.com/forum/viewtopic.php?f=19&t=1087 2010-11-12T02:18:30 @later tell McLeopold see this, ping us when done: http://ai-contest.com/forum/viewtopic.php?f=19&t=1087 2010-11-12T02:18:30 amstan: I think that worked... 2010-11-12T02:18:38 @later tell a1k0n see this, ping us when done: http://ai-contest.com/forum/viewtopic.php?f=19&t=1087 2010-11-12T02:18:38 amstan: OK 2010-11-12T02:20:12 janzert: oh yeah.. there's a guy complaining about clojure support 2010-11-12T02:23:09 hmm 2010-11-12T02:24:31 anyway, i'm going to buy some dinner 2010-11-12T02:25:53 janzert: what do you think of "Number of planets is not stated in the rules, and it might change, (assume 100 if you really want a top limit)." 2010-11-12T02:26:34 i'm against changing it from 23 2010-11-12T02:26:39 although i could personally handle it 2010-11-12T02:26:40 antimatroid: why? 2010-11-12T02:27:00 if you make it much past 30 you will cripple a significant portion of the bots 2010-11-12T02:27:08 and a fix wont be easy without completely changing strategies 2010-11-12T02:27:37 antimatroid: but again.. this is exactly the same problem as the linear symmetry 2010-11-12T02:28:06 it is, like i said before, these kind of things need to be worked out before the contest even starts if possible 2010-11-12T02:28:18 varying around below 30 would be fine 2010-11-12T02:28:35 but if you go above that, you're going to really render a lot of the top bots useless 2010-11-12T02:28:42 antimatroid: that sounds good 2010-11-12T02:28:46 so.. s/100/30 2010-11-12T02:28:59 i just think it's really stupid to have a constant number of planets 2010-11-12T02:29:10 i don't think it really matters 2010-11-12T02:29:16 strategy wise 2010-11-12T02:30:31 one option is to randomly decide whether to include the centre neutral planet or not 2010-11-12T02:30:57 amstan: yeah moving it up is likely to break any bots using bitmasks for planets 2010-11-12T02:31:08 i think the number of planets should be in the [15:30] range 2010-11-12T02:31:23 i wouldn't be opposed to [3,30] 2010-11-12T02:31:28 sure.. 2010-11-12T02:31:38 but a lot of people would 2010-11-12T02:31:47 i always wanted 3 planets, but i think i was the only one 2010-11-12T02:31:59 a lot of people wouldn't**** 2010-11-12T02:32:50 hang on, i have a good idea for even more general maps, i'll draw a pic 2010-11-12T02:35:02 http://imgur.com/N2LzL.png 2010-11-12T02:35:16 antimatroid: that's line symmetric 2010-11-12T02:35:17 you can pick any number of neutrals to go along that yellow line, and only have to put one of them in 2010-11-12T02:35:40 yes, but it's adding more planets starting equal distance between players 2010-11-12T02:36:00 won't cause RPS? 2010-11-12T02:36:12 the idea is to have the other ones as well 2010-11-12T02:36:13 that would be good :) 2010-11-12T02:36:28 janzert: do we still have the priority column for the maps? 2010-11-12T02:36:39 yes, but it's currently unused 2010-11-12T02:36:42 *** wh1teside_ has quit IRC (Remote host closed the connection) 2010-11-12T02:36:42 generate players starting positions, generate random number of single neutrals, then fill the rest with two symmetrically placed neutrals 2010-11-12T02:36:51 janzert: as in.. the code ignores it? 2010-11-12T02:37:06 yeah, the code doesn't reference it anywhere that I have seen 2010-11-12T02:37:14 that's not cool 2010-11-12T02:37:19 that's how we disabled maps last term 2010-11-12T02:37:23 and certainly the rewrite of pairing code I just did doesn't 2010-11-12T02:37:23 set priority to 0 2010-11-12T02:37:29 yep 2010-11-12T02:38:19 after I remembered seeing that the column was there adding it's use back in shouldn't be too hard 2010-11-12T02:38:56 in fact I think now that's how we should disable the current maps for the final 2010-11-12T02:39:03 alright, i'm going to change it to randomly pick number of planets and number of middle neutrals 2010-11-12T02:39:09 just set all of them to 0 then generate the new maps 2010-11-12T02:39:10 janzert: k.. so we have that solved then 2010-11-12T02:39:23 well, theoretically :) 2010-11-12T02:39:32 still need to write the code of course 2010-11-12T02:39:32 antimatroid: how about starting a github repo for the new version of the map generators? 2010-11-12T02:39:43 antimatroid: i'll help you on there 2010-11-12T02:40:20 i wouldn't have a clue how to do that sorry 2010-11-12T02:41:15 i'm not willing to do anything or get any access to anything that renders me inelligible of competing though 2010-11-12T02:41:16 antimatroid: do you know git? 2010-11-12T02:41:24 nope 2010-11-12T02:41:28 antimatroid: oh man 2010-11-12T02:41:39 antimatroid: yeah.. dont' worry about the inegibility, this should be fine 2010-11-12T02:41:53 antimatroid: do you know svn? 2010-11-12T02:41:56 nope :P 2010-11-12T02:42:12 you should learn 2010-11-12T02:42:17 i should learn a lot of things 2010-11-12T02:42:43 *** eburnette has quit IRC (Ping timeout: 265 seconds) 2010-11-12T02:42:47 janzert: help me on this one 2010-11-12T02:42:54 ;) 2010-11-12T02:42:55 http://imgur.com/N2LzL.png peace? 2010-11-12T02:43:10 i'd recommend just using the current svn repo so as not to get divided all over the place 2010-11-12T02:43:42 oh man.. that's even more complicated, since i don't know it myself, therefore i can't help him on it 2010-11-12T02:43:51 hmm? 2010-11-12T02:44:13 what do you want to do? 2010-11-12T02:45:40 oh maybe you want, "learning a to use a version control system is crucial for your development as a programmer and all around person, a distributed version control system is a particularly good way" :) 2010-11-12T02:45:48 *** tapwater has quit IRC (Quit: tapwater) 2010-11-12T02:46:20 learning svn is good, too 2010-11-12T02:46:24 although I like bzr > hg > git myself which is about the opposite of the rest of the world :) 2010-11-12T02:46:36 makes you feel so much better when you step up to git or hg 2010-11-12T02:46:48 hey, i like hg over git :P 2010-11-12T02:46:52 haven't used bzr, though 2010-11-12T02:47:13 janzert: yay! that! 2010-11-12T02:47:58 now, if one was to use both symmetries, what sort of split would you like the maps to be? 2010-11-12T02:47:59 50 50? 2010-11-12T02:48:33 i think it'd be best to just make the one generator generate all maps 2010-11-12T02:48:51 50/50 is good, make it even chance to go either way 2010-11-12T02:50:53 *** fawek has joined #aichallenge 2010-11-12T02:51:14 antimatroid: what's the point of translate_planets? 2010-11-12T02:52:01 amstan: positive coordinates 2010-11-12T02:52:07 apparently it's needed for viewers 2010-11-12T02:52:20 antimatroid: oh.. decent, k 2010-11-12T02:53:25 *** jaspervdj has joined #aichallenge 2010-11-12T03:00:33 how does one do else if in python? 2010-11-12T03:00:40 antimatroid: elif 2010-11-12T03:07:05 antimatroid: what's a realistic mapsize? 2010-11-12T03:07:18 as in the biggest horizontal or vertical distance between planets 2010-11-12T03:11:32 i think people are happy with a map radius of 12 2010-11-12T03:11:48 oh.. 2010-11-12T03:11:50 i have it switching between symmetries 2010-11-12T03:12:00 so.. 10,20? 2010-11-12T03:12:02 how's that? 2010-11-12T03:12:02 just going to add variable number of planets and "central" planets 2010-11-12T03:12:12 10,20? 2010-11-12T03:12:16 10-20 2010-11-12T03:12:16 map radius? 2010-11-12T03:12:19 could be variable 2010-11-12T03:12:22 it's not exactly radius 2010-11-12T03:12:24 it's size 2010-11-12T03:12:39 i say just leave it at 12 max, then it doesn't necessarily reach that distance anyway 2010-11-12T03:12:44 so max distance is 24 2010-11-12T03:13:25 antimatroid: i'm creating X and Y values based on random.uniform(10,15) 2010-11-12T03:13:47 i'm geneating radius and theta values randomly, then converting to cartesian coordinates 2010-11-12T03:13:53 so it's made inside a cirlce 2010-11-12T03:14:03 i'm making another map gen 2010-11-12T03:14:06 it makes doing the new symmetry much easier 2010-11-12T03:14:09 why? 2010-11-12T03:14:16 for linear maps 2010-11-12T03:14:23 linear symmetry* 2010-11-12T03:14:23 linear? 2010-11-12T03:14:32 1d maps? 2010-11-12T03:15:39 ^^ 2010-11-12T03:16:28 *** codefun has quit IRC (Quit: Page closed) 2010-11-12T03:16:41 Top 10 players: bocsimacko(4457), george(4275), asavis(4258), shangas(4225), Raschi(4180), GreenTea(4148), dmj111(4122), _Astek_(4100), Doraj8(4088), _iouri_(4086) 2010-11-12T03:16:57 *** aerique has joined #aichallenge 2010-11-12T03:26:23 *** perror has joined #aichallenge 2010-11-12T03:44:04 antimatroid: http://codepad.org/ZhgQWDtM 2010-11-12T03:45:01 antimatroid: the only problem is that it produces even number of planets all the time 2010-11-12T03:45:19 amstan: home planets are to always start with 100 ships and 5 growth 2010-11-12T03:45:28 that's easy 2010-11-12T03:46:26 there's a whole bunch of checks etc. people want as well 2010-11-12T03:46:38 like? 2010-11-12T03:46:54 like distances aren't allowed to be withing epsilon of an integer, players must be able to take at least one neutral on the first move without losing their home planet to rage bot 2010-11-12T03:48:09 oh man 2010-11-12T03:48:11 hold on 2010-11-12T03:48:13 i have a problem 2010-11-12T03:48:31 ? 2010-11-12T03:52:55 @visualizer 2010-11-12T03:52:56 amstan: visualizer = (#1) http://github.com/DanielVF/Planet-Wars-Canvas-Visualizer, or (#2) http://csclub.uwaterloo.ca/~amstan/aichallenge/canvas-visualizer/ 2010-11-12T03:55:27 antimatroid: http://codepad.org/KkdSp33e 2010-11-12T03:55:40 antimatroid: how would you not make ragebot win? 2010-11-12T04:06:11 *** hellman has joined #aichallenge 2010-11-12T04:07:52 amstan: first neutral added has ships < 5*dist(players home planets) 2010-11-12T04:08:00 more importantly, http://codepad.org/1R9ILAWg 2010-11-12T04:08:17 that generator will do both symmetries, and variable numbers of "centre planets" and varying number of planets 2010-11-12T04:08:31 it's messy as fuck though 2010-11-12T04:08:33 that's a lot of code 2010-11-12T04:08:39 ^^ :) 2010-11-12T04:08:45 i've never programmed in python before 2010-11-12T04:09:01 i can't believe it doesn't have operator overloading :( 2010-11-12T04:09:16 operator overloading? like what? 2010-11-12T04:09:28 change behavoir of a+b? 2010-11-12T04:09:30 sure it does 2010-11-12T04:09:35 *** hellman has quit IRC (Remote host closed the connection) 2010-11-12T04:09:42 you need to make a Planet class 2010-11-12T04:09:49 and then you have 2010-11-12T04:09:55 def __add__ (self, other): 2010-11-12T04:10:00 return what it'll return 2010-11-12T04:11:01 i wanted not_valid(p1, p2) and not_valid(p1) but it didn't seem to like that 2010-11-12T04:11:11 that's function overloading 2010-11-12T04:11:13 my bad 2010-11-12T04:11:14 you mean function overloading 2010-11-12T04:11:16 oh.. 2010-11-12T04:11:18 that's not pythonic 2010-11-12T04:11:25 who math is connection limit for TCP server? 2010-11-12T04:11:41 unandestending problems i have 2010-11-12T04:11:41 Accoun: we are considering changes the types of maps generated 2010-11-12T04:11:51 your opinion on the idea would be great 2010-11-12T04:11:57 antimatroid> stuped idia 2010-11-12T04:12:07 they're better maps 2010-11-12T04:12:28 stuped idea change reles in last week 2010-11-12T04:12:34 stuped idea change rules in last week 2010-11-12T04:12:36 it's not the last week 2010-11-12T04:12:49 in last 1.5 weeks 2010-11-12T04:12:55 Accoun: it's not a rule 2010-11-12T04:12:55 Accoun: please vote in the forum 2010-11-12T04:13:08 Accoun: http://ai-contest.com/forum/viewtopic.php?f=19&t=1087&start=10 2010-11-12T04:13:10 all canadia math less main mass 2010-11-12T04:13:17 there's more than 2 weeks left, and rules were never stated either way 2010-11-12T04:13:33 do all u want 2010-11-12T04:13:44 Accoun: vote in the poll 2010-11-12T04:13:45 Accoun: have you taken a look at the new maps? 2010-11-12T04:13:47 canadian organiser very chitty 2010-11-12T04:13:56 :( 2010-11-12T04:14:21 Accoun: don't look at me, it was suggested by antimatroid 2010-11-12T04:14:32 please install new maps in TCP cerver of stuped canadian relay cange rules 2010-11-12T04:14:35 i'm sure he's neither an organizer nor canadian 2010-11-12T04:14:46 if 2010-11-12T04:14:56 server 2010-11-12T04:15:00 i haven't seen dhartmei in a while, i will try to get some on there when i do 2010-11-12T04:15:07 his email address bounced yesterday 2010-11-12T04:15:11 hope u andesend me 2010-11-12T04:15:22 we can make out what you mean 2010-11-12T04:15:49 Accoun: you can download 20 such maps and a generator from that link above 2010-11-12T04:15:50 and i was enjoying coding this.. 2010-11-12T04:16:02 now it's all ruined 2010-11-12T04:16:13 some tiames tcp.exe dont finish bot and logs is ceep bloked 2010-11-12T04:16:17 sthange bug 2010-11-12T04:16:41 Top 10 players: bocsimacko(4447), bix0r4ever(4285), george(4254), asavis(4253), shangas(4225), Raschi(4166), GreenTea(4143), dmj111(4124), davidjliu(4105), _iouri_(4097) 2010-11-12T04:18:20 @later tell dmj111 see this, ping us when done: http://ai-contest.com/forum/viewtopic.php?f=19&t=1087 2010-11-12T04:18:20 amstan: Aye, aye, sir 2010-11-12T04:26:05 is it possible to change votes? 2010-11-12T04:26:37 mega1: yes, i just did 2010-11-12T04:26:48 *** Utkarsh has joined #aichallenge 2010-11-12T04:26:54 didn't know that 2010-11-12T04:27:05 mega1: ya, i made it changable 2010-11-12T04:27:26 i take it that was from seeing my change and not a change of heart? :P 2010-11-12T04:27:45 antimatroid: what was? 2010-11-12T04:27:46 not of mine :-) 2010-11-12T04:27:59 amstan: i changed my vote from both to wild 2010-11-12T04:28:07 mega1: and you? 2010-11-12T04:28:13 saw my change 2010-11-12T04:28:17 and wondered how poll number dropped 2010-11-12T04:28:22 yup 2010-11-12T04:28:34 mega1: what did you vote for? 2010-11-12T04:28:54 amstan: the "stay where you are" option 2010-11-12T04:28:58 ok 2010-11-12T04:29:42 but line symmetry is so much cooler than point symmetry! 2010-11-12T04:30:03 mega1 made assumptions about maps being like that and "threw out ideas based on that" 2010-11-12T04:30:36 like.. why would you throw out an idea just because current maps are like that? 2010-11-12T04:30:52 the least you can do is ask 2010-11-12T04:31:01 he did 2010-11-12T04:31:01 wait.. maybe you did 2010-11-12T04:31:04 darn 2010-11-12T04:31:07 I made the assumption that the strength a bot displays is closely related to its playing strength in the final tournament 2010-11-12T04:31:22 i don't think the rankings will change much at all 2010-11-12T04:31:28 mega1: how sure are you that changing it to line symmetric too will break your bot? 2010-11-12T04:31:41 antimatroid: i don't think either 2010-11-12T04:31:43 is it possible to run some of the top bots on new maps separately and see if the rankings remain much the same? 2010-11-12T04:31:56 my intentions are not to mix up the rankings at all 2010-11-12T04:32:04 antimatroid: not really, code's kinda crappy, so it's inflexible at doing these things 2010-11-12T04:32:17 I don't know if this change will change anything which is precisely the reason for not doing it. 2010-11-12T04:32:18 antimatroid: and running them without a sandbox is not a cool idea either 2010-11-12T04:32:36 what we need is some of the maps on tcp 2010-11-12T04:32:59 I have played about 20000 games locally so far. 2010-11-12T04:33:27 testing countless changes 2010-11-12T04:33:43 anyone know grep? 2010-11-12T04:33:58 nvm 2010-11-12T04:34:09 mega1: you were welcome to create your own test maps 2010-11-12T04:34:21 the same was said for tron, don't assume maps are going to be like those provided 2010-11-12T04:34:54 antimatroid: I'm still not talking about the maps but the generator. 2010-11-12T04:35:32 mega1: you're welcome to create your own generator 2010-11-12T04:35:47 are or were? 2010-11-12T04:35:54 still are 2010-11-12T04:35:58 mega1: that's my point, there has never been a specified "official generator for the contest" 2010-11-12T04:36:00 and it's just too late 2010-11-12T04:36:15 an example genreator and example genrated maps were provided 2010-11-12T04:36:17 guys, the lawyer stance is just not useful 2010-11-12T04:36:27 mega1: the point is that you should have thought about this, other people did 2010-11-12T04:36:42 I did and asked on the mailing list. 2010-11-12T04:37:03 k, hold on 2010-11-12T04:37:17 mega1: what email did you use as? 2010-11-12T04:37:25 mega@retes.hu 2010-11-12T04:37:31 darn 2010-11-12T04:37:33 k 2010-11-12T04:38:17 mega1: wait, you sure? do you have a subject name, i can't find your actual email address 2010-11-12T04:38:32 yes, i'm looking at it now 2010-11-12T04:38:37 antimatroid: link? 2010-11-12T04:38:42 it's in my email :P 2010-11-12T04:38:49 antimatroid: subject line? 2010-11-12T04:38:58 DECISION TIME 2010-11-12T04:39:01 second last 2010-11-12T04:39:02 yes 2010-11-12T04:39:39 it's melis.* 2010-11-12T04:40:32 mega1: you didn't get a reply though 2010-11-12T04:40:41 that's his point 2010-11-12T04:41:11 yes, no response doesn't mean the generator there is official, but then the point moves onto it being really late in the contest 2010-11-12T04:41:30 but i've been going under the assumption the whole time that maps could change, and my understanding of the rules was to expect that 2010-11-12T04:41:47 yes, the maps 2010-11-12T04:42:25 but if the generator too then you have to think and lay down some guidelines 2010-11-12T04:42:27 it's a situation where there is no solution that doesn't in a way disadvantage someone, that's why these decisions need to be made (where possible) before the contest even starts 2010-11-12T04:42:36 yes 2010-11-12T04:44:04 my point is that it shouldn't give any particular person an unfair advantage with the new maps, whether optimising on old ones or not, the new maps are way cooler and the rules always had this open to be done 2010-11-12T04:44:51 and it does give an advantage to people who made assumptions and have spent considerable time working under those assumptions to then have them enforced 2010-11-12T04:46:07 so.. 2010-11-12T04:46:23 i did say "just changing the seed": http://pastebin.com/r2nRuc9u 2010-11-12T04:46:51 amstan: "You should not train your bots on the current maps." should read "You should not train your bots exclusively on the current maps." 2010-11-12T04:47:27 there is nothing wrong with training bots on current maps, but they were only ever given as examples 2010-11-12T04:47:42 antimatroid: yeah, lol 2010-11-12T04:47:44 antimatroid: thanks 2010-11-12T04:48:14 i'm going to get back to work on the new bot anyway, it's not going to write itself 2010-11-12T04:50:01 I also asked about whether server #0 is going to be in the finals 2010-11-12T04:50:15 because it's way slower than the others 2010-11-12T04:50:33 and the amount of ram one can expect to have without being forced to swap and time out. 2010-11-12T04:50:34 mega1: jeff seemed to think the servers were going a lot better now 2010-11-12T04:50:40 it was getting 90 games a minute before 2010-11-12T04:50:46 but not sure about your specific question 2010-11-12T04:51:13 it isn't that good anymore, hmmm 2010-11-12T04:52:39 there's a new australian in the top 50 2010-11-12T05:05:41 *** Naktibalda has joined #aichallenge 2010-11-12T05:16:42 Top 10 players: bocsimacko(4451), bix0r4ever(4273), asavis(4262), george(4243), shangas(4227), Raschi(4208), GreenTea(4142), dmj111(4125), Doraj8(4114), _Astek_(4095) 2010-11-12T05:26:20 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-11-12T05:26:28 *** iNtERrUpT has joined #aichallenge 2010-11-12T05:26:50 *** nullkuhl has joined #aichallenge 2010-11-12T05:27:41 *** p4p4 has joined #aichallenge 2010-11-12T05:29:31 *** iNtERrUpT has quit IRC (Remote host closed the connection) 2010-11-12T05:37:15 *** Azrathud has quit IRC (Ping timeout: 255 seconds) 2010-11-12T05:39:31 haha, i generated a map that looks like a cock and balls 2010-11-12T05:40:30 you have to hope that your bot doesn't suck at it :) 2010-11-12T05:40:47 http://imgur.com/TfTBN.png 2010-11-12T05:41:01 that's what a map would look like with many "central" neutral planets 2010-11-12T05:41:04 and a cock and balls theme 2010-11-12T05:48:35 http://ai-contest.com/game_stats.php woooooooow it's blaaazing 2010-11-12T05:49:24 antimatroid: how is that symmetrical? 2010-11-12T05:49:55 Error323_: it is, it's just scaled annoyingly for some reason 2010-11-12T05:50:18 Error323_: it's symmetric about that line of planets 2010-11-12T05:50:29 yeah, it is symmetrical, i have a seprate file i can check symmetry of a map on and I have counted turns in jbotmanager 2010-11-12T05:50:30 ah ok 2010-11-12T05:50:46 http://codepad.org/5A5vJQuD 2010-11-12T05:50:51 try generating yourself some maps on that page 2010-11-12T05:51:04 it'll give you alternating symmetries and multiple "central" neutral planets 2010-11-12T05:53:20 oh, and varying number of planets 2010-11-12T05:54:16 ah 2010-11-12T05:54:25 those are excellent given my bot's workings :P 2010-11-12T05:54:57 Error323_: http://ai-contest.com/forum/viewtopic.php?f=19&t=1087&start=20 2010-11-12T05:55:08 there is a poll on peoples preferences about using new maps like these in the final contest 2010-11-12T05:58:24 Error323_: make sure you read the discussion too :-) 2010-11-12T05:58:28 wait, i'm not sure the game stays symmetric when radially symmetric if i put "centre" neutrals off dead centre 2010-11-12T06:00:54 mega1: yeah imo its too late aswell 2010-11-12T06:01:22 it's always been intended for them to change though 2010-11-12T06:01:34 people were told not to assume the starting maps are a good representation of all maps 2010-11-12T06:01:52 i think people missed that memo 2010-11-12T06:02:01 it was touted throughout tron 2010-11-12T06:02:05 and i've seen it said for this 2010-11-12T06:03:31 well 2010-11-12T06:03:58 even if it was said it would still be very misleading to let the unofficial ranking server never ever play those maps 2010-11-12T06:04:32 Error323_: have you voted yet? ;-) 2010-11-12T06:05:10 yes 2010-11-12T06:06:02 Error323_: i think it's unfair to let people who made assumptions get to have those maps exclusively 2010-11-12T06:06:13 anyone who didn't make any assumptions shouldn't be disadvantaged 2010-11-12T06:06:59 i haven't made any assumptions for that reason, cause it'd suck if things changed 2010-11-12T06:07:04 but it's always been said that they would 2010-11-12T06:07:16 it's unfair to then just leave it imo 2010-11-12T06:07:30 more unfair* 2010-11-12T06:08:17 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-11-12T06:09:44 *** Error323_ has quit IRC (Ping timeout: 272 seconds) 2010-11-12T06:11:52 *** Bobng has joined #aichallenge 2010-11-12T06:12:31 *** Error323 has joined #aichallenge 2010-11-12T06:12:38 @rankings 2010-11-12T06:12:39 Error323: Top 10 players: bix0r4ever(4301), shangas(4297), asavis(4273), Raschi(4260), george(4245), GreenTea(4197), Error323(4192), dmj111(4141), _Astek_(4111), Doraj8(4111) 2010-11-12T06:12:47 I just had to see that 2010-11-12T06:13:03 hehe 2010-11-12T06:13:03 Being temporarily in the top 10 :P 2010-11-12T06:13:16 oh, i thought you were referring to bocsimacko not being there 2010-11-12T06:13:19 i take it he resubmitted 2010-11-12T06:13:24 congrats :) 2010-11-12T06:13:28 oh 2010-11-12T06:13:31 hehe 2010-11-12T06:13:44 yeah, well its a fluke =) 2010-11-12T06:13:57 only played four games 2010-11-12T06:14:14 it'll fall down soon i think 2010-11-12T06:15:00 there is no bot for bocsimacko 2010-11-12T06:15:15 mega1: did you mean to take your bot down? 2010-11-12T06:16:03 wait, nevermind 2010-11-12T06:16:11 i don't know why it didn't show up under "lisp" bots 2010-11-12T06:16:43 Top 10 players: bix0r4ever(4302), shangas(4291), asavis(4268), Raschi(4255), george(4228), GreenTea(4197), Error323(4188), dmj111(4138), Doraj8(4118), _Astek_(4109) 2010-11-12T06:18:04 antimatroid: just reuploaded 2010-11-12T06:18:25 yep, i didn't think i'd have noticed in the time it took to show up 2010-11-12T06:18:48 Error323: is that the alpha beta bot? 2010-11-12T06:18:55 it got to 12th pretty quick 2010-11-12T06:18:57 one game 2010-11-12T06:19:05 mega1: nah, the alphabeta bot sux 2010-11-12T06:19:09 ouch 2010-11-12T06:19:14 yeeeeeep 2010-11-12T06:19:31 wasted about 1.5 weeks of codetime on that :/ 2010-11-12T06:19:46 i wasted so much time on minimax and zero sum simultaneous games 2010-11-12T06:19:48 I did play with alpha beta and nash too 2010-11-12T06:20:41 i'm in the process of writing a complete final revision to try and get rid of bugs and rid myself of terrible hacks i made previously 2010-11-12T06:21:00 i just don't know a good eval function 2010-11-12T06:21:28 thats not a very precisely tweaked heuristic thing 2010-11-12T06:21:40 which would take another 2 weeks to get right 2010-11-12T06:22:03 i want awesomeness in a linear function :P 2010-11-12T06:22:05 heh 2010-11-12T06:22:21 or through non manual tweaking anyway 2010-11-12T06:22:27 *** Tobu has joined #aichallenge 2010-11-12T06:22:29 can be nonlinear 2010-11-12T06:27:00 *** Utkarsh has joined #aichallenge 2010-11-12T06:33:13 mega1: )@(#*$)@#($*%)#$% http://ai-contest.com/visualizer.php?game_id=7031517 2010-11-12T06:33:27 its not even remotely fair how i don't stand a chance 2010-11-12T06:33:29 thank you :-) 2010-11-12T06:33:29 xD 2010-11-12T06:33:56 @rankings 2010-11-12T06:33:57 Error323: Top 10 players: bix0r4ever(4308), shangas(4297), asavis(4274), Raschi(4261), george(4234), GreenTea(4203), Error323(4190), dmj111(4143), Doraj8(4124), _Astek_(4115) 2010-11-12T06:33:59 at least it's over quick :) 2010-11-12T06:34:03 ah still there 2010-11-12T06:34:07 hehe 2010-11-12T06:34:40 i'm so curious as to what your heuristic eval function is 2010-11-12T06:34:53 I think the rating calc is lagging a bit. 2010-11-12T06:35:19 Error323: that should be a good topic for a short blog post 2010-11-12T06:35:45 i do hope nobody beats mega1 (unless it's me), he's been on top so long now 2010-11-12T06:36:05 george worries me still 2010-11-12T06:36:12 hehe 2010-11-12T06:36:14 yeah, i saw his bot earlier 2010-11-12T06:36:19 he has an old bot on the official server 2010-11-12T06:36:28 that;s more worrying 2010-11-12T06:36:34 last time i had a running bot i beat him 2 for 1 2010-11-12T06:36:38 but that was probably pure luck 2010-11-12T06:36:42 and on tcp he was getting very near to the top when he took his bots down 2010-11-12T06:37:03 perhaps he had the eureka moment :P 2010-11-12T06:37:09 and isn't releasing until final tourny now 2010-11-12T06:37:57 mega1: yes blogging that would be really cool 2010-11-12T06:38:19 unfortunately I'll only have time to blog on the 28th :-) 2010-11-12T06:38:24 hehe 2010-11-12T06:38:51 mega1: how so? 2010-11-12T06:39:03 I'm so busy writing code 2010-11-12T06:39:25 still? 2010-11-12T06:39:25 Error323: tthat wasn't self explanatory? 2010-11-12T06:39:36 one could program for a few years on this problem 2010-11-12T06:39:52 no i mean why not 29th, 30th et 2010-11-12T06:39:54 etc 2010-11-12T06:39:55 +) 2010-11-12T06:40:11 because the contest ends on the 27th? 2010-11-12T06:40:39 i'm putting my final bot up a good day before deadline 2010-11-12T06:41:00 mega1: yeah i know, so you can blog afterwards right =) 2010-11-12T06:41:05 doesn't have to be just 28th 2010-11-12T06:41:27 give him a break, english is his second language 2010-11-12T06:41:51 hmm it seems i'm being misunderstood 2010-11-12T06:42:00 i understand 2010-11-12T06:42:12 he meant that he wont be blogging until at least then 2010-11-12T06:42:18 got nothing but respect for the man ;) 2010-11-12T06:42:53 i think everyone is in awe 2010-11-12T06:42:59 over a monthy uncontested on top 2010-11-12T06:43:05 nobody managed that last time 2010-11-12T06:45:29 hehe yes amazing 2010-11-12T06:46:01 lol, i managed to get trapped in free minecraft 2010-11-12T06:46:13 *** narnach has joined #aichallenge 2010-11-12T06:46:13 *** narnach is now known as narnach_mb 2010-11-12T06:46:17 i went down and hit lava and water, then it blocked me on all 6 sides 2010-11-12T06:53:09 who's the funny guy with metternich and nesselrode on tcp? 2010-11-12T06:54:47 all austrians are suspicious 2010-11-12T06:55:01 lol 2010-11-12T06:56:31 i always like the story about russian scienctists vs us scientists coming up with writing devices in space 2010-11-12T06:56:37 mega1: what's funny abou tit? 2010-11-12T06:56:37 i take it you've both heard that one? 2010-11-12T06:57:04 yeah the pencil vs million dollar pen 2010-11-12T06:57:06 or sth 2010-11-12T06:57:18 yeah aha 2010-11-12T06:57:27 i always write in pacer 2010-11-12T06:57:54 techniclick is my choice 2010-11-12T06:58:17 pink for some reason, not sure why, i've just always got the pink one 2010-11-12T06:58:43 Error323: they were involved in putting down the hungarian revolution (1844-1848) against the austrians 2010-11-12T06:58:43 mega1: lol, how do you test your bots when you win of everyone but yourself :P must be a real problem for you 2010-11-12T06:58:55 mega1: ahhh lol 2010-11-12T06:59:25 it takes many games to get info on tcp 2010-11-12T06:59:42 and even then I'm mostly counting number of losses 2010-11-12T06:59:50 to compare versions. Oh, and timeouts. 2010-11-12T06:59:50 yeah... 2010-11-12T06:59:57 still its very tricky 2010-11-12T07:00:15 i assume you first play against eachother locally 2010-11-12T07:00:52 yes, and a couple of times lost 100+ elos with a change that was quite a bit better against its old self 2010-11-12T07:01:16 yeah but you can't play them against eachother on tcp 2010-11-12T07:01:27 can, but don't want to 2010-11-12T07:01:30 that would seriously bias the results as they would be picked against eachother quite a lot 2010-11-12T07:01:35 yes 2010-11-12T07:02:09 that's the trouble Accoun has with running several versions at the same time 2010-11-12T07:03:37 i think your only good testbed is the official 2010-11-12T07:03:55 tcp is too unstable in terms of connections/players 2010-11-12T07:04:34 true, on the other hand you play a whole lot more games 2010-11-12T07:05:28 mega1: are you running linux by any chance? 2010-11-12T07:05:40 Error323: yes 2010-11-12T07:05:57 you don't want a binary, right? ;-) 2010-11-12T07:06:59 no :P 2010-11-12T07:07:13 i want you to have mine :P 2010-11-12T07:07:38 just curiuos :P 2010-11-12T07:14:25 remember official site still only has the original 100 maps 2010-11-12T07:14:42 i don't think that is a good indication of possible maps, even with the original generator 2010-11-12T07:16:43 Top 10 players: bocsimacko(4311), asavis(4291), shangas(4283), Raschi(4270), bix0r4ever(4257), george(4210), GreenTea(4193), dmj111(4146), _Astek_(4142), Doraj8(4134) 2010-11-12T07:17:06 mega1: that didn't take long :P 2010-11-12T07:17:25 lol 2010-11-12T07:17:53 is the servercode using old rank as startrank nowadays? 2010-11-12T07:18:22 i think it might be 2010-11-12T07:18:32 but only based off watching your bots go up the ranks 2010-11-12T07:20:47 I think it makes a lot of sense to bias opponent selection this way 2010-11-12T07:23:07 *** phreeza has joined #aichallenge 2010-11-12T07:23:34 hi everyone 2010-11-12T07:23:39 how are the bots going? 2010-11-12T07:23:59 some newcomers on the tcp i see 2010-11-12T07:24:10 new in the top10 al least 2010-11-12T07:24:45 mega1: it does 2010-11-12T07:25:02 phreeza: i haven't got a running bot atm, in the middle of a complete rewrite 2010-11-12T07:25:42 antimatroid: aha, i haven't got anything workble running either :( 2010-11-12T07:25:53 your heuristic bot that you are rewriting? 2010-11-12T07:28:07 yep, i've given up on modeling the other player too 2010-11-12T07:28:10 it's too expensive 2010-11-12T07:29:17 dmj111: ping? 2010-11-12T07:29:35 antimatroid: computionally? 2010-11-12T07:30:06 yes 2010-11-12T07:30:14 yea i guess thats the only thing that makes sense in this context :P 2010-11-12T07:30:29 have you taken a look at the new maps i generated? 2010-11-12T07:30:35 http://ai-contest.com/forum/viewtopic.php?f=19&t=1087&start=20 2010-11-12T07:30:35 no 2010-11-12T07:30:39 there is a discussion here 2010-11-12T07:30:41 and links to some 2010-11-12T07:30:51 ok i'll have a look 2010-11-12T07:32:12 *** mega1 has quit IRC (Ping timeout: 276 seconds) 2010-11-12T07:32:59 i think i agree with the first reply... interesting in principle, but a bit late to change them now 2010-11-12T07:33:41 it has always been said that they will change damnit 2010-11-12T07:33:53 hehe 2010-11-12T07:33:53 it's unfair not to change them 2010-11-12T07:34:04 you've probably heard this many times now? 2010-11-12T07:34:13 yes 2010-11-12T07:34:13 *** _Utkarsh <_Utkarsh!~admin@117.201.90.251> has joined #aichallenge 2010-11-12T07:34:22 but i've always been working under the assumption that maps will change 2010-11-12T07:34:29 and it was stated like that 2010-11-12T07:34:36 *** Rubicon-|-Cross has joined #aichallenge 2010-11-12T07:34:44 ah but what you are suggesting is keeping the generator rules but generating new maps? 2010-11-12T07:34:54 no, i have a new generator 2010-11-12T07:34:57 does more symmetrie 2010-11-12T07:35:01 and more general maps 2010-11-12T07:35:14 *** Utkarsh has quit IRC (Ping timeout: 272 seconds) 2010-11-12T07:35:26 ah i see 2010-11-12T07:35:30 it will always be the case that a fresh set of maps is generated for the final contest 2010-11-12T07:36:03 yea i'm cool with that 2010-11-12T07:36:17 i'm incredibly biased 2010-11-12T07:36:21 but changing the symmetries seems a bit harsh 2010-11-12T07:36:32 no it doesn't, people shouldn't make assumptions 2010-11-12T07:36:50 lol 2010-11-12T07:36:50 it's harsh to allow people who made assumptions to have had that extra time working under that constraint 2010-11-12T07:36:52 i think even people who didn't explicitly make that assumption tweaked their bots on these maps 2010-11-12T07:36:55 antimatroid: yes they should 2010-11-12T07:37:15 you can't let the official run the same type of maps for 80% of the contest time 2010-11-12T07:37:23 and then suddenly change everything 2010-11-12T07:37:32 the maps were provided only as an example 2010-11-12T07:37:44 it was always stated they are not to be used as an exhaustive testing set 2010-11-12T07:37:47 not even the generator 2010-11-12T07:37:49 but its all the contestants got.. 2010-11-12T07:37:54 make your own 2010-11-12T07:37:57 it's not that difficult 2010-11-12T07:38:03 no i mean... 2010-11-12T07:38:04 i've made a couple of different generators 2010-11-12T07:38:08 the space is so huge.... 2010-11-12T07:38:08 and plenty of small maps 2010-11-12T07:38:14 testing out lots of general cases 2010-11-12T07:38:29 the contestants need other opponents and they can't get hands on the binaries of other players so all they got is official 2010-11-12T07:38:32 i did that under the specified rules 2010-11-12T07:38:37 if the symmetries can be changed, whos to say the size of the planets isn't 2010-11-12T07:38:41 or the number 2010-11-12T07:38:48 now it's basically saying, people who made assumptions get to hold them and have had that extra time working with that 2010-11-12T07:39:09 phreeza: one option and generator switched between symmetries and number of planets etc. 2010-11-12T07:39:14 switches* 2010-11-12T07:39:19 i am actually in favour of that one 2010-11-12T07:40:01 also, with bilateral symmetry, you can have multiple neutrals starting at an equal distance from both players 2010-11-12T07:40:08 rather than just the central planet 2010-11-12T07:40:42 you can have that now too 2010-11-12T07:40:42 and the maps are way cooler 2010-11-12T07:40:46 no you can't 2010-11-12T07:40:48 because of the ceil 2010-11-12T07:41:00 no, you can do it without ceil 2010-11-12T07:41:11 but the games become unfair 2010-11-12T07:41:38 depending on distribution of planets on each half on each playeyers half of the map 2010-11-12T07:41:53 so split the map in 4 2010-11-12T07:42:24 to make it possible with radial symmetry, it needs to be bilaterally symmetric too 2010-11-12T07:43:11 my bots so far could probably deal with any change but this would be really good to know for the new one i'm working on... 2010-11-12T07:43:24 hence why i'm trying to finalise this 2010-11-12T07:43:28 *** cYmen has quit IRC (Ping timeout: 272 seconds) 2010-11-12T07:43:40 i've been waiting for it to change for ages, and it hasn't so i'm pushing for decisions to be made 2010-11-12T07:43:45 antimatroid: what if there are ppl with machinelearning algos that used data from the official server etc 2010-11-12T07:43:57 *** cYmen has joined #aichallenge 2010-11-12T07:44:01 Error323: maps are changing no matter what 2010-11-12T07:44:07 yes ofc 2010-11-12T07:44:10 it's an ai contest, not who can solve these 100 maps the best contest 2010-11-12T07:44:11 but not that radically 2010-11-12T07:44:31 the difference with mine really isn't that much imo 2010-11-12T07:44:31 just different point symmetry maps 2010-11-12T07:44:52 i don't expect any bots to really change ranking significantly 2010-11-12T07:45:08 mega1 sounds pretty concerned in the forum :D 2010-11-12T07:45:27 i'm pretty sure he is just opposed to changing it so late 2010-11-12T07:45:43 but the way i see it, it's always been expected to change, so it's not fucking people over at the last minute 2010-11-12T07:45:46 i'm surprised that you are not :P 2010-11-12T07:45:58 i've been expecting it to always change 2010-11-12T07:45:58 well probably because you are in the middle of a rewrite 2010-11-12T07:46:06 and the inventor of the new mapgen 2010-11-12T07:46:24 jeff gave me the idea for the maps months ago when he said he intended to write such a generator 2010-11-12T07:46:28 no one ever did so I did yesterday 2010-11-12T07:46:45 the generator is a result of my desire for new maps, not vice versa 2010-11-12T07:47:22 and the maps are cooler 2010-11-12T07:47:27 even mega1 admitted that 2010-11-12T07:48:58 well so do I 2010-11-12T07:49:09 and probably my bot will have a positive effect 2010-11-12T07:49:14 on the new set 2010-11-12T07:49:31 but for any player with fancy learning algos its a real nightmare 2010-11-12T07:49:33 imo the timeframe is just the killer argument here... a month ago i'm sure it would have been possible, but by now i think its just implicit to most people that it won't change, at least not so severely 2010-11-12T07:49:43 i think the ragebot rule might make sense still 2010-11-12T07:51:12 i'll grant people the not too close to integer distances too 2010-11-12T07:51:32 Error323: yes its very important to me, in my new bot i match up the mp with previously seen maps. the algorithm to do that has O(n^3), with the current symmetry i only need to match 11, now i would have to match all 23 2010-11-12T07:52:51 and i'm lucky, i haven't even started the learning yet, but still it meany i will only be abler to perform 1/8th of the matches i thought i would 2010-11-12T07:53:37 sorry for the typos, just started on one of these fancy microsoft natural keyboards 2010-11-12T07:56:07 it was never said people can assume these symmetries hold 2010-11-12T07:56:21 it's unfair for people who listended to that to then keep it as is 2010-11-12T07:56:41 that's the way i see it 2010-11-12T07:56:45 but yes, i'm biased :) 2010-11-12T07:57:01 well thats a valid argument to i guess 2010-11-12T07:57:54 but you have to agree that the fact that the official server hasn't used this map generator is a bit unfair 2010-11-12T07:58:06 we'll just have to have another season after this, with modified rules ;) 2010-11-12T07:58:23 how is that unfair? 2010-11-12T07:59:08 it has always been said that those maps should not be taken as representative of maps used in the final tournament 2010-11-12T07:59:51 that would defeat the entire benefit of having the server in the first place 2010-11-12T08:00:20 antimatroid: so whats to stop me from witing another map generator that just produces pahological cases where ragebot wins 90% of the time? :D 2010-11-12T08:00:30 pathological* 2010-11-12T08:00:36 yeah 2010-11-12T08:00:45 in lets say a day before the end of the contest 2010-11-12T08:01:16 \o/ 2010-11-12T08:01:23 phreeza: because that'd be stupid 2010-11-12T08:01:29 lol 2010-11-12T08:01:30 this map gen is entirely open for scrutiny 2010-11-12T08:01:31 :D 2010-11-12T08:01:35 it doesn't give anyone advantages 2010-11-12T08:01:52 the server is there as a guide 2010-11-12T08:01:58 it 2010-11-12T08:02:01 it does give some disadvantages 2010-11-12T08:02:08 s not even guaranteed that the same ranking would be used during the finals 2010-11-12T08:02:21 although it looks like it will be, and i agree with that 2010-11-12T08:02:49 it gives no one a disadvantage that didn't go making assumptions for themselves 2010-11-12T08:03:07 the alternative advantages people who made assumptions, whereas anyone who didn't should be on equal footing 2010-11-12T08:03:54 while i'm biased being in the group who didn't make assumptions, i think the fairer thing is to not disadvantage those who didn't make their own assumptions 2010-11-12T08:06:21 *** p4p4 has quit IRC (Ping timeout: 245 seconds) 2010-11-12T08:08:48 yea i guess the argument is not going anywhere... I see your point, to a degree... but I still think its a bad idea at this point. 2010-11-12T08:09:05 same here 2010-11-12T08:09:15 i don't see how it's bad, if people followed the guidelines, there bot would perform fine 2010-11-12T08:09:19 either way, it would be good to have an official decision asap 2010-11-12T08:09:27 if it doesn't, that's their fault for making assumptions 2010-11-12T08:09:34 their* 2010-11-12T08:09:49 antimatroid: its not just about assumptions 2010-11-12T08:09:57 learning algos take time 2010-11-12T08:10:01 it is, people were free to generate their own test maps 2010-11-12T08:10:14 yes but then they don't have the enemy bots 2010-11-12T08:10:21 and a lot don't know about tcp 2010-11-12T08:10:34 you think anyones done machine learning on the main site? 2010-11-12T08:10:36 since its nowhere on the site 2010-11-12T08:10:42 it's in the forums 2010-11-12T08:11:18 antimatroid: ppl will definitly have gathered statistics of it i'm sure 2010-11-12T08:12:16 and imho ppl using learning algos should not be discouraged because of this stuff since i think they deserve more to win then the heuristic versions 2010-11-12T08:12:26 but thats just my opinion +) 2010-11-12T08:12:34 and look on ai 2010-11-12T08:12:43 heh 2010-11-12T08:12:53 i think the bots should work on genreal maps 2010-11-12T08:12:55 that makes it better ai 2010-11-12T08:13:00 but they are learning from heuristic bots! :D 2010-11-12T08:13:09 whether it's using machine learning or heuristics 2010-11-12T08:13:23 i don't think it even changes game play very much 2010-11-12T08:13:40 *** delt0r___ has joined #aichallenge 2010-11-12T08:13:42 general maps is too general 2010-11-12T08:13:58 well i guess the fairest thing would be to add the poll on the main site and use the votecount 2010-11-12T08:14:07 for the decision 2010-11-12T08:14:12 maybe someone has something that has O(N_planets!) but tweaked it enough to work on 23 planets 2010-11-12T08:14:18 but then we bump it to 50? 2010-11-12T08:14:26 *** hellman has joined #aichallenge 2010-11-12T08:14:37 Error323: agreed 2010-11-12T08:14:40 i'm sure it will break quite a lot of bots 2010-11-12T08:15:12 people made the bots under false assumptions then 2010-11-12T08:15:29 it's unfair on the people who made sure their bots would work more generally 2010-11-12T08:15:46 *** delt0r has quit IRC (Ping timeout: 272 seconds) 2010-11-12T08:15:48 maybe so but you can't say it has been stated very clearly through official channels 2010-11-12T08:15:52 i always get back to, it's more unfair to disadvantage those who listended rather than those who made assumptions 2010-11-12T08:16:30 i think ideally, such decisions should be in stone before the contest even starts, but they weren't 2010-11-12T08:16:37 i agree 2010-11-12T08:16:44 Top 10 players: bocsimacko(4323), asavis(4272), bix0r4ever(4263), shangas(4259), Raschi(4246), george(4220), GreenTea(4184), Doraj8(4122), _Astek_(4121), dmj111(4114) 2010-11-12T08:16:46 some people are disadvantaged whichever way 2010-11-12T08:16:51 so who is it fairer to disadvantage? 2010-11-12T08:17:02 thats why a vote would be the best thing to do imo 2010-11-12T08:17:08 i don't get what proportions are, which group most deserves to be at a disadvantage? 2010-11-12T08:17:20 care* what proportions are 2010-11-12T08:17:35 size of groups should be irrelevant imo 2010-11-12T08:18:34 in the end we're all doing this for fu, right? so why spoil it for many people who have put in the work by crippling their bots, even if it may be technically fair 2010-11-12T08:18:58 they'll feel screwed over, whatever the real intentions may have been 2010-11-12T08:19:08 for fun' 2010-11-12T08:19:15 not for fu haha 2010-11-12T08:19:16 i'm not the only person who has expected maps to change 2010-11-12T08:19:34 everyone has expected it 2010-11-12T08:19:38 *** boegel has joined #aichallenge 2010-11-12T08:19:45 its just the variety that differs 2010-11-12T08:19:49 thats the argument why proportions should count 2010-11-12T08:20:13 i'm sick of the majority getting their way, just because they're the majority 2010-11-12T08:20:19 it's a fucking stupid decision rule 2010-11-12T08:20:40 rules like that kept slavery alive 2010-11-12T08:20:50 extreme example :P 2010-11-12T08:21:31 * phreeza sees Godwin rubbing his hands 2010-11-12T08:22:28 is e.g. the second of cputime set in stone? 2010-11-12T08:23:23 yes 2010-11-12T08:25:10 directly on the specification page 2010-11-12T08:25:10 "Maps have no particular dimensions and are defined completely in terms of the planets and fleets in them." 2010-11-12T08:25:10 "Most maps on the server and in the starter packs were generated programmatically by a Python script. You may also create your own to use for testing." 2010-11-12T08:25:10 " the bots each have 1 second of wall-clock time to receive the game state, process it, and send their orders." 2010-11-12T08:25:59 it says nowhere people should assume the maps on the official site are representative of teh final maps 2010-11-12T08:26:04 even the generator used 2010-11-12T08:27:31 *** mceier has joined #aichallenge 2010-11-12T08:29:47 *** _Utkarsh <_Utkarsh!~admin@117.201.90.251> has quit IRC (Ping timeout: 265 seconds) 2010-11-12T08:32:59 *** perestrelka has quit IRC (Read error: Connection reset by peer) 2010-11-12T08:33:22 *** perestrelka has joined #aichallenge 2010-11-12T08:36:17 *** Utkarsh has joined #aichallenge 2010-11-12T08:44:40 I'm generating PNGs of all the benzedrine and anti's maps. Anyone else interested? It's just the planets and their indexes. 2010-11-12T08:47:28 i wouldn't have anything to do with them, but thanks :) 2010-11-12T08:47:52 *** p4p4 has joined #aichallenge 2010-11-12T08:48:45 i mainly made it to browse for cock 'n balls map pr0n in your set 2010-11-12T08:49:07 haha 2010-11-12T08:49:27 *** Utkarsh has quit IRC (Ping timeout: 250 seconds) 2010-11-12T08:49:44 what you don't realise is that i secretely made it so they all look like cock and balls 2010-11-12T08:49:55 it's the ultimate troll 2010-11-12T08:50:14 :D 2010-11-12T08:53:38 *** Naktibalda has left #aichallenge 2010-11-12T08:54:57 *** Utkarsh has joined #aichallenge 2010-11-12T09:03:13 *** bhasker has joined #aichallenge 2010-11-12T09:04:38 *** boegel has quit IRC (Ping timeout: 245 seconds) 2010-11-12T09:08:15 *** hellman has quit IRC (Remote host closed the connection) 2010-11-12T09:16:44 Top 10 players: bocsimacko(4333), Raschi(4261), asavis(4257), shangas(4248), george(4237), bix0r4ever(4227), GreenTea(4174), dmj111(4131), _Astek_(4131), davidjliu(4115) 2010-11-12T09:20:06 antimatroid: do you know where the point of origin lies for the maps? 2010-11-12T09:20:20 that is, 0.0,0.0 2010-11-12T09:20:48 Error323: its just somewhere in the corner i think, I transform my maps so center planet is at origin 2010-11-12T09:21:11 phreeza: which corner? :) 2010-11-12T09:21:32 *** Migi32 has joined #aichallenge 2010-11-12T09:21:34 you can render the maps any way you like 2010-11-12T09:21:35 :) 2010-11-12T09:21:50 i think the vewer makes it top left 2010-11-12T09:21:54 viewer 2010-11-12T09:27:08 *** bhasker has quit IRC (Read error: Connection reset by peer) 2010-11-12T09:27:34 *** bhasker has joined #aichallenge 2010-11-12T09:29:50 Error323: in my maps, all planets are centered around (0,0) then translated 14 in both the x and y direction 2010-11-12T09:30:10 max map radius is 12 2010-11-12T09:33:44 ok 2010-11-12T09:38:52 it's also picking polar coordinates randomly, then converting them to cartesian 2010-11-12T09:38:57 made it a million times easier 2010-11-12T09:40:42 *** Frontier has joined #aichallenge 2010-11-12T10:07:45 *** Azrathud has joined #aichallenge 2010-11-12T10:14:18 antimatroid, that would make the "planet density" higher at the center than at the edges. Did you put code to counter that? If not, I should be able to come up with the right math formula for uniform distribution in polar coordinates 2010-11-12T10:16:44 Top 10 players: bocsimacko(4394), asavis(4246), george(4214), shangas(4206), bix0r4ever(4179), _Astek_(4141), davidjliu(4131), dmj111(4124), GreenTea(4120), Error323(4117) 2010-11-12T10:17:26 wuuuuuuuut 2010-11-12T10:17:34 i'm in the top 10 again XD 2010-11-12T10:18:27 Migi32: i didn't 2010-11-12T10:18:43 i don't see why it would though 2010-11-12T10:19:12 okay i see what you mean 2010-11-12T10:19:17 but you are still likely to get planets far out 2010-11-12T10:19:55 yea, still likely, but it'll be more likely that planets will bunch up in the middle 2010-11-12T10:20:28 nice Error 2010-11-12T10:20:34 so what is new in your submission 2010-11-12T10:20:41 the formula to prevent this should be something along the lines of radius=max_radius * random(0.0 to 1.0)^2 2010-11-12T10:21:12 no wait sqrt() instead of ()^2 2010-11-12T10:22:15 yeah okay, i see what you're getting at 2010-11-12T10:22:20 i will do that at some point 2010-11-12T10:22:24 i'm working on my own bot atm 2010-11-12T10:22:58 sure, just wanted to mention it :) 2010-11-12T10:23:06 btw are you guys writing your bot in C++, i am thinking my python version will be too slow to do minmax kind of stuff 2010-11-12T10:23:18 just updating the state/resolving fleets for all planets is taking over 15 ms 2010-11-12T10:23:19 :-\ 2010-11-12T10:23:27 doing that a few hundred times will probably kill my bot 2010-11-12T10:23:43 i'm using c++ 2010-11-12T10:23:47 bhasker: ow i'll get kicked out soon enough, its 17.2 @ tcp 2010-11-12T10:23:53 although i also gave up on minimax 2010-11-12T10:23:59 pure heuristic now? 2010-11-12T10:24:08 i'm trying maximax :P 2010-11-12T10:24:15 i'm using a tree to generate my moves 2010-11-12T10:24:18 ah 17.2 is like an elo point above my bot but mine is at 58: -\ 2010-11-12T10:24:20 but it's not modelling the opponent 2010-11-12T10:24:28 that is mega1's approach. 2010-11-12T10:24:37 ah yea but to do a tree i still need to resolve battles 2010-11-12T10:24:38 that has alwaays been my heuristic appraoch 2010-11-12T10:24:39 bhasker: i just submitted, i'll drop down anyway 2010-11-12T10:24:41 I am too, but bhasker do you know c++ already? If not, it's probably too late to learn it and write a winning entry in it. 2010-11-12T10:25:00 got a new idea for the evaluation function. 2010-11-12T10:25:05 yea i know c++ 2010-11-12T10:25:09 code in it all the time @ work 2010-11-12T10:25:24 just that python is easy 2010-11-12T10:25:26 to tinker with 2010-11-12T10:25:27 ahh, the joys of holidays till end of contest 2010-11-12T10:25:48 python is easy, i'd never done anything with it before today and wrote out a map generator in it 2010-11-12T10:25:49 bhasker: i don't think you need to port it untill your minimax is fully working and you can't get it faster 2010-11-12T10:25:53 without looking a single thing up 2010-11-12T10:25:58 except how to comment 2010-11-12T10:25:58 mostly you want algorithmic optimizations 2010-11-12T10:26:09 they are >> then c++ vs pythooon 2010-11-12T10:26:18 hmm 2010-11-12T10:26:36 oh well let me try and implement it in python 2010-11-12T10:26:39 ah ok. Well, while it may be beneficial to switch, but computation speed doesn't seem to be the deciding factor in this contest. 2010-11-12T10:26:46 but i think i need to optimize my battle resolution and state update 2010-11-12T10:26:48 its too slow 2010-11-12T10:26:53 15ms to update all planets is too slow 2010-11-12T10:27:13 is that for 200 turns or just the diameter of the map? 2010-11-12T10:27:14 well migi to do tree searches 2010-11-12T10:27:16 imo minimax is too hard in one second for this 2010-11-12T10:27:18 my heuristic version is blazing fast 2010-11-12T10:27:20 2x diameter 2010-11-12T10:27:21 even using c++ 2010-11-12T10:27:30 oops rather 2x maxdistance between planets 2010-11-12T10:27:33 then it's indeed quite slow. 2010-11-12T10:27:44 yup 2010-11-12T10:29:54 i think logging is also killing me 2010-11-12T10:30:04 just dumping state at every turn takes that from 15ms to 50ms 2010-11-12T10:30:05 ^ 2010-11-12T10:30:20 i think the mistake is i should just update all planets at the same time 2010-11-12T10:30:22 not one at a time 2010-11-12T10:30:28 just advance the world by one time step 2010-11-12T10:30:57 right now my engine is like for p in pw.Planets: updatePlanet(p) 2010-11-12T10:31:26 which basically gets a list of fleets orders them by time and resolves battles at all turns and just duplicates state for the rest 2010-11-12T10:32:12 no I think doing it planet per planet would be faster 2010-11-12T10:32:59 python does have a fast sorting algorithm right? 2010-11-12T10:33:24 *** superflit has quit IRC (Quit: superflit) 2010-11-12T10:35:18 wait I think I misunderstood you. Why does it duplicate states? 2010-11-12T10:44:23 *** lavalamp_ has joined #aichallenge 2010-11-12T10:46:19 i also haven't had anyone say they actually dislike the new maps 2010-11-12T10:46:23 just that they don't like change 2010-11-12T10:46:39 i really don;t think it even changes peoples stategies much 2010-11-12T10:46:52 i've had a number of people reluctantly admit they like them 2010-11-12T10:48:18 *** Azrathud has quit IRC (Read error: Connection reset by peer) 2010-11-12T10:48:56 *** siconize has joined #aichallenge 2010-11-12T10:49:34 Error323, does your alpha-beta search use iterative deepening? 2010-11-12T10:50:16 *** wh1teside has joined #aichallenge 2010-11-12T11:01:25 *** mega1 has joined #aichallenge 2010-11-12T11:02:15 @seen dhartmei 2010-11-12T11:02:15 lavalamp_: dhartmei was last seen in #aichallenge 6 weeks, 1 day, 23 hours, 17 minutes, and 37 seconds ago: project selection problem? 2010-11-12T11:02:36 that's long :( 2010-11-12T11:02:45 good thing he keeps his server running though :) 2010-11-12T11:03:51 yeah... 2010-11-12T11:05:31 Migi32: no 2010-11-12T11:07:01 I was just thinking that we should try and get some new maps on the server if there's much chance they'll be used 2010-11-12T11:07:43 Error323, ok. I was going to take a shot at tree searches myself, so I wondered. But I just realised alpha-beta isn't even possible with Nash equilibria as node evaluation 2010-11-12T11:09:34 it would be interesting to have the server generate a new map for every battle 2010-11-12T11:10:59 you could contact dhartmei at daniel -at- benzedrine -dot- cx 2010-11-12T11:11:36 that bounced my email today 2010-11-12T11:11:51 sent it to at openbsd.org instead 2010-11-12T11:12:00 well yea benzedrine.cx seems down 2010-11-12T11:14:09 *** superflit has joined #aichallenge 2010-11-12T11:15:22 *** fawek has quit IRC (Ping timeout: 265 seconds) 2010-11-12T11:16:22 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-11-12T11:16:38 lavalamp_: we decided it's better not to use lots of maps 2010-11-12T11:16:46 Top 10 players: bocsimacko(4424), asavis(4231), george(4204), shangas(4203), davidjliu(4151), bix0r4ever(4148), Error323(4133), _Astek_(4131), dmj111(4111), GreenTea(4108) 2010-11-12T11:16:47 that way you have lots of bots playing on the same map 2010-11-12T11:16:54 although you don't want bots repeating 2010-11-12T11:17:11 on the same map, so i think the allowable set is going to be extended as the contest runs as needed 2010-11-12T11:17:28 yeah, I know it's not real practical (or fair) to do a random one for each game but it would still be interesting 2010-11-12T11:17:49 i love the new maps, i hope the are allowed 2010-11-12T11:20:12 *** superflit has quit IRC (Ping timeout: 276 seconds) 2010-11-12T11:20:55 *** fawek has joined #aichallenge 2010-11-12T11:25:17 *** aerique has quit IRC (Quit: ...) 2010-11-12T11:25:31 *** perror has quit IRC (Quit: Bye all !) 2010-11-12T11:33:00 *** narnach has joined #aichallenge 2010-11-12T11:33:00 *** narnach is now known as narnach_mb 2010-11-12T11:33:25 *** Ice_Harley has quit IRC (Ping timeout: 265 seconds) 2010-11-12T11:36:38 *** Utkarsh has quit IRC (Ping timeout: 265 seconds) 2010-11-12T11:40:01 *** perror has joined #aichallenge 2010-11-12T11:41:44 *** Utkarsh has joined #aichallenge 2010-11-12T11:44:22 lol wut #6th 2010-11-12T11:45:01 *** vpetro has joined #aichallenge 2010-11-12T11:53:01 *** vpetro has quit IRC (Ping timeout: 245 seconds) 2010-11-12T12:02:16 *** paulmcq_ has joined #aichallenge 2010-11-12T12:04:52 Error323: nice 2010-11-12T12:05:09 heh 2010-11-12T12:05:13 well i'll drop any second now :P 2010-11-12T12:05:24 *** yasith has quit IRC (Ping timeout: 264 seconds) 2010-11-12T12:05:37 you already did and came back :P 2010-11-12T12:05:44 true 2010-11-12T12:06:06 perhaps its not that bad as on tcp 2010-11-12T12:06:10 since i was also running my old version there 2010-11-12T12:06:35 *** tapwater has joined #aichallenge 2010-11-12T12:10:05 *** spacebat has quit IRC (Ping timeout: 250 seconds) 2010-11-12T12:10:23 but i doubt it 2010-11-12T12:10:38 *** lovebot_ has quit IRC (Ping timeout: 252 seconds) 2010-11-12T12:11:47 *** lovebot has joined #aichallenge 2010-11-12T12:11:52 *** spacebat has joined #aichallenge 2010-11-12T12:16:45 Top 10 players: bocsimacko(4386), asavis(4211), shangas(4192), george(4158), _Astek_(4120), Accoun(4120), davidjliu(4116), GreenTea(4110), Error323(4092), dmj111(4084) 2010-11-12T12:17:17 \o/ 2010-11-12T12:20:30 *** yasith has joined #aichallenge 2010-11-12T12:36:39 *** shades has quit IRC (Ping timeout: 255 seconds) 2010-11-12T12:39:39 *** boegel has joined #aichallenge 2010-11-12T12:42:31 *** boegel has quit IRC (Client Quit) 2010-11-12T12:46:07 sigh, alright, antimatroid, you convinced me, I changed my vote on the forum thread... :/ 2010-11-12T12:46:30 :) 2010-11-12T12:46:55 *** fawek has quit IRC (Read error: Operation timed out) 2010-11-12T12:47:12 there isn't much difference in preferneces atm 2010-11-12T12:48:09 the linearly symmetric maps sure look weird if your window isn't square 2010-11-12T12:48:24 yeah, i have a 24 inch wide screen :P 2010-11-12T12:48:34 and window is far from square 2010-11-12T12:49:13 I think point symmetric is all people thought of because it's the only one that you can generalize to more players 2010-11-12T12:49:29 you can't do more than 2 players 2010-11-12T12:49:38 it complicates things beyond imagination 2010-11-12T12:49:57 yeah, I agree, but it seems like they were thinking of it at first 2010-11-12T12:50:05 never going to happen 2010-11-12T12:50:12 then people need to start worryig about coalitions 2010-11-12T12:50:33 yeah, it changes things completely 2010-11-12T12:50:56 the optimal move isn't to wait like some people thing 2010-11-12T12:51:08 because smart bots will collude to take you out then battle each other 2010-11-12T12:51:15 right 2010-11-12T12:51:35 and you can't just start attacking player 2 because then player 3 will take you both out 2010-11-12T12:51:41 yep 2010-11-12T12:51:48 and then when do you break your coalition? 2010-11-12T12:51:56 and if there's n players, it's just insane 2010-11-12T12:52:02 ...not to mention the fact that nD payoff tables get big fast 2010-11-12T12:52:11 yep 2010-11-12T12:52:26 i'm making a game theory library, i'm aware of how fast they blow out :| 2010-11-12T12:52:35 tensors are insane 2010-11-12T12:52:48 same as game trees 2010-11-12T12:52:58 or any tree for that matter :P 2010-11-12T12:53:01 we could even do non-symmetric maps if the server played 2 game sets 2010-11-12T12:53:07 lol yeah 2010-11-12T12:53:18 i think non-symmetric maps would be quite boring 2010-11-12T12:53:41 I suppose if the common case is that one player has a huge advantage, yeah 2010-11-12T12:54:07 but do we know that? no one as tried afaik 2010-11-12T12:54:14 *h 2010-11-12T12:54:18 *has 2010-11-12T12:54:34 tried non-symmetric maps? 2010-11-12T12:54:43 i have generated trivial ones testing out cases 2010-11-12T12:54:52 ah ok 2010-11-12T12:55:06 but that's more to check my move selection is working as intended 2010-11-12T12:55:11 maybe I should make a few and see what happens 2010-11-12T12:55:19 ...after I get my bugs squashed 2010-11-12T12:56:04 *** bhasker has quit IRC (Quit: bhasker) 2010-11-12T12:58:20 hmm, why do I keep playing low ranked opponents? 2010-11-12T12:59:28 tcp or official? 2010-11-12T13:00:04 official 2010-11-12T13:00:34 is it counting your old games against top bots now maybe? 2010-11-12T13:00:42 and then not giving you new pairings 2010-11-12T13:00:45 that's just a guess 2010-11-12T13:00:46 antimatroid: nope 2010-11-12T13:01:03 mega1: so far as I can see it's just an unlucky streak 2010-11-12T13:01:31 Raschi had a very lucky streak in pairings 2010-11-12T13:01:33 and won them to boot 2010-11-12T13:01:38 janzert: thanks. So I'm not back to be running on server #0 only? 2010-11-12T13:01:57 taking another look at the code to see if there is something that would skew pairings for the top player 2010-11-12T13:02:06 there doesn't even seem to be a server 0 anymore? 2010-11-12T13:02:11 123 games in the last minute 2010-11-12T13:02:15 mega1: nope, actually nobody is running on 0 anymore 2010-11-12T13:02:31 janzert: oh, cool. 2010-11-12T13:02:54 The distribution of ranks of opponents looked fine for my previous submission. 2010-11-12T13:03:03 maybe it's just laggggging 2010-11-12T13:03:30 btw, the new pairing code can be found in http://code.google.com/p/ai-contest/source/browse/trunk/planet_wars/backend/create_matchups.py 2010-11-12T13:03:32 using an oldish rank to compute the opponents 2010-11-12T13:04:02 the important function here would be choose_opponent 2010-11-12T13:04:20 but I'm not seeing anything off the top of my head that should skew the distribution 2010-11-12T13:06:14 gah, fuuu c++, let me define a priority queue of moves, i defined an inequality operator for you :( 2010-11-12T13:16:46 Top 10 players: bocsimacko(4380), Raschi(4363), asavis(4206), shangas(4181), george(4161), davidjliu(4135), bix0r4ever(4131), _Astek_(4118), Accoun(4100), GreenTea(4091) 2010-11-12T13:17:04 *** perror_ has joined #aichallenge 2010-11-12T13:19:12 *** CIA-36 has quit IRC (Ping timeout: 264 seconds) 2010-11-12T13:21:11 give me a tip of good comedy movie 2010-11-12T13:21:20 *** hellman has joined #aichallenge 2010-11-12T13:21:41 *** perror_ has quit IRC (Client Quit) 2010-11-12T13:21:47 *** perror has quit IRC (Quit: Bye all !) 2010-11-12T13:21:48 happy gilmore was the first thing to pop to mind :P 2010-11-12T13:23:33 sry, have seen it already :( 2010-11-12T13:25:22 *** choas has joined #aichallenge 2010-11-12T13:27:04 hmm, I did think of an unintended effect on the distribution of players at the top and bottom of the rankings. Basically the top and bottom will tend toward playing opponents twice the average distance than bots in the middle. 2010-11-12T13:27:27 mega1: raschi isn't far behind on elo 2010-11-12T13:27:57 this is because one side of the distribution is getting cut off since there aren't any players there 2010-11-12T13:28:11 janzert: that's a nice property 2010-11-12T13:28:52 antimatroid: he has only four games 2010-11-12T13:29:05 yes, but against reasoanble opponents too :P 2010-11-12T13:29:38 ooh, I see it's finally time to anchor the elo range 2010-11-12T13:29:47 bottom bot is now at 5 elo 2010-11-12T13:30:04 he, that took WAY longer than expected 2010-11-12T13:30:10 yeah 2010-11-12T13:30:27 what does anchoring means? 2010-11-12T13:30:38 bottom bots elo is fixed i think 2010-11-12T13:30:42 so it doesn't drop below 0 2010-11-12T13:30:46 yes 2010-11-12T13:31:11 right now the whole range just keeps the median(?) at 2000 2010-11-12T13:31:24 instead we can anchor the worst bot to always be 0 2010-11-12T13:31:29 yep 2010-11-12T13:31:51 or you could anchor the best bot at 4500 2010-11-12T13:32:00 that's boring 2010-11-12T13:32:18 we need irrational aesthetics thanks :) 2010-11-12T13:32:19 better yet, enter ragebot and anchor it at 600 2010-11-12T13:32:24 yeah, the absolute number really are meaningless 2010-11-12T13:34:01 if you anchor one or two constant bots, then the absolute numbers start to mean something 2010-11-12T13:36:45 minimally :) 2010-11-12T13:36:59 *** boegel has joined #aichallenge 2010-11-12T13:37:45 *** McLeopold has joined #aichallenge 2010-11-12T13:37:49 they would still depend on the # of bots entered, I think 2010-11-12T13:38:02 *** CIA-37 has joined #aichallenge 2010-11-12T13:38:57 maybe not if you had two anchor points and scaled appropriately, hm 2010-11-12T13:39:35 all elo scores would vary too wildly then, I think 2010-11-12T13:39:52 having multiple anchor players in a rating system can lead to real problems if you aren't very careful 2010-11-12T13:40:31 if ragebot gets a couple good maps and suddenly appears to do really good, and you used it as an anchor, everyone's ELO will drop really fast 2010-11-12T13:40:32 I think two would be OK since you just translate for one and scale for the other (and feed the rating algorithm the unmodified numbers) 2010-11-12T13:41:11 Hm you might be right about that since ragebot would be playing those near its level 2010-11-12T13:41:23 amstan: ping 2010-11-12T13:41:27 with enough games that effect should dissapear 2010-11-12T13:41:34 *disappear 2010-11-12T13:41:37 the problem is when there are intransitivities with cycles involving both anchors 2010-11-12T13:42:04 why would that cause a problem? 2010-11-12T13:42:29 I mean, a worse one than intransitivities already cause the system? 2010-11-12T13:43:12 haha yes if you anchor one bot at 600 (say, starter package) and another at 1000 (say, ragebot) and ragebot by bad luck drops in internal score to below the starter package, your ELO ratings would be inverted 2010-11-12T13:43:36 or is that not how anchors work? 2010-11-12T13:44:02 Lol ok I was tihnking more, say, rage bot and my old bot, which is enough better than ragebot that that would never happen 2010-11-12T13:44:15 migi32: yeah, that's the basic problem 2010-11-12T13:44:33 although the ratings don't actually invert but they will be very wierd 2010-11-12T13:45:08 suddenly first is rated .53 below ragebot, haha 2010-11-12T13:46:14 Just so you know, the elo problem is not solvable. Relative ratings are the best you can get. 2010-11-12T13:46:38 I'd say, just keep them like they are, fixed average and fixed standard deviation. The people with negative ELO won't notice anyway, I think. 2010-11-12T13:46:46 *** Utkarsh has quit IRC (Ping timeout: 245 seconds) 2010-11-12T13:48:29 the fact that top ELO keeps rising kinda reflects the fact that the contest is getting tougher and tougher 2010-11-12T13:48:46 McLeopold: yeah, and they're not even relative all the time due to intransitivities. But anchoring can be useful, IIRC the kgs (gokgs.com) rating system has a number of anchors to stabilize it 2010-11-12T13:49:11 I think the fact that the top Elo is rising represents the fact that more people have entered :P 2010-11-12T13:49:29 lavalamp_ is more correct 2010-11-12T13:49:38 *** CryBaby has joined #aichallenge 2010-11-12T13:51:01 yes, I know that. But as antimatroid said "we need irrational aesthetics thanks :)" 2010-11-12T13:51:21 #1 one is zero, everyone else is negative? 2010-11-12T13:51:33 where? 2010-11-12T13:51:53 I'm just saying to anchor the top bot at elo=0 2010-11-12T13:51:58 ohh 2010-11-12T13:52:16 I thought that was in reference to my KGS statement 2010-11-12T13:52:18 then everyone's goal is just to suck less 2010-11-12T13:52:49 Don't they list golf scores that way? how many stokes behind the leader you are? 2010-11-12T13:52:54 *strokes 2010-11-12T13:53:01 (I don't play golf) 2010-11-12T13:53:11 no 2010-11-12T13:53:17 and yet 2010-11-12T13:53:19 yes 2010-11-12T13:53:23 *** Utkarsh has joined #aichallenge 2010-11-12T13:53:27 it seems televised tournaments at least do 2010-11-12T13:53:37 baseball is that way, games behind 2010-11-12T13:54:01 how about this: keep elo as it is, but if elo<0, show elo as 0 instead 2010-11-12T13:54:21 I'm sure tv is the only place I've seen golf, and that accidentally :) 2010-11-12T13:54:35 *** superflit has joined #aichallenge 2010-11-12T13:54:40 why is negative elo bad? 2010-11-12T13:55:18 just for the feelings of those players and aesthetics of the list 2010-11-12T13:56:01 how about a positive number that says "you are this close to becoming #1" 2010-11-12T13:56:06 an advantage of anchoring the minelo is also that it easy at a glance of the top to tell the overall spread 2010-11-12T13:57:54 from which you can determine at least the theoretical number of levels of play, where each level is defined as winning against the next lowest a certain percentage of the time 2010-11-12T13:58:10 In go, 3/4ths of the ratings are negative 2010-11-12T13:58:53 oops, mixed up the order of commands sent to bayeselo 2010-11-12T13:59:00 *** Cold-Phoenix has joined #aichallenge 2010-11-12T13:59:09 now half our ranks are negative :P 2010-11-12T13:59:34 janzert: that may or may not work... I can come up with bots that suck on purpose 2010-11-12T13:59:54 the worst bots already lose every game immediately 2010-11-12T14:00:15 hard to get worse than that :) 2010-11-12T14:01:28 e.g., in go human players can suck at 30 or 40 kyu... random play is estimated to be about 150 kyu, and then you could write algorithms that tried to lose on purpose to do worse than that. If we already have bots that are as terrible as possible, that's distorting the "number of levels" calculation 2010-11-12T14:01:58 when you compare it to go or chess, anyway 2010-11-12T14:02:21 how is chess elo anchored? 2010-11-12T14:02:23 yeah, but I say that is because go and chess aren't calculating it correctly :P 2010-11-12T14:02:36 there has been recent talk about chess elo inflation 2010-11-12T14:02:43 janzert, that's a possible view :) 2010-11-12T14:03:01 mega1, isn't there always talk about that? 2010-11-12T14:03:04 mega1: I don't think it is just recent 2010-11-12T14:03:23 I don't believe chess is anchored 2010-11-12T14:03:29 unless they started it recently 2010-11-12T14:03:36 recent as in past couple of years 2010-11-12T14:03:49 lavalamp_: yeah, I don't think so 2010-11-12T14:04:08 mega1: It's been ongoing for as long as I can remember which is at least 12 years 2010-11-12T14:04:26 "what rating would XXX have today?" 2010-11-12T14:05:10 yeah, masters of old would always rule ;-) 2010-11-12T14:06:17 to me, only the top 10 and how many people are still better than me are important anyway :) 2010-11-12T14:09:16 Migi32: so, 11 total? :P 2010-11-12T14:09:37 yep :) 2010-11-12T14:16:47 Top 10 players: bocsimacko(2443), Raschi(2315), asavis(2194), bix0r4ever(2166), shangas(2145), george(2145), davidjliu(2120), Accoun(2108), _Astek_(2096), GreenTea(2070) 2010-11-12T14:18:00 wow 2010-11-12T14:18:07 elo just got cut in half? 2010-11-12T14:18:32 janzert: are you guys removing old bot entries entirely? 2010-11-12T14:18:36 basically the arbitrary offset is removed but not the 0 anchor 2010-11-12T14:19:04 *** hellman has quit IRC (Remote host closed the connection) 2010-11-12T14:19:04 http://ai-contest.com/forum/viewtopic.php?f=19&t=1090 2010-11-12T14:19:05 McLeopold: yeah, there is a job that runs and deletes submissions that are no longer active 2010-11-12T14:19:14 that's what is doing it 2010-11-12T14:19:29 Please join another map type discussion 2010-11-12T14:19:37 ^ above link 2010-11-12T14:20:50 I'm fond of my "jump map" idea. :) 2010-11-12T14:21:31 Also, everyone on IRC needs to vote for different map types... or else. 2010-11-12T14:22:28 I'm not really good a persuasion. 2010-11-12T14:23:44 janzert: does it concern you that I went from 78ish to 168ish in one day? 2010-11-12T14:23:57 I like the different ideas but whatever you decide on please make it official quite some time before the deadline (which means, like, right now) 2010-11-12T14:24:45 Migi32: I don't want you to miss the point. Talk about "how" it would affect bots. Not if we should or not. 2010-11-12T14:25:00 McLeopold: yes, but I also understand the problem and am working to correct it :) 2010-11-12T14:25:14 ok, well one thing is that bots using minimax will probably start faltering if you add maps with 50+ ships 2010-11-12T14:25:26 Does it mean I had an inflated rank? From being a really old bot? 2010-11-12T14:25:26 50+ planets* 2010-11-12T14:25:26 at an admittedly slow and often diverted pace 2010-11-12T14:25:54 Yeah, or adding more planets. 2010-11-12T14:26:04 All bots would be slower with more planets. 2010-11-12T14:26:10 McLeopold: no, it means the current method of using only the past 3 days of games for ranking is too volatile 2010-11-12T14:26:27 3 days? that's it? 2010-11-12T14:26:45 yep, and it now takes 4 minutes just to run the query for that 2010-11-12T14:27:36 the table has very very many games in it 2010-11-12T14:27:43 I wouldn't mind waited 30 minutes between rank updates. 2010-11-12T14:27:46 what I would like though is planets started with 1000 ships instead of 100 and then growth would be random between 0 and 50, not only multiples of 10, to give a hard time to algorithms that scale with absolute numbers (like 0-1 knapsack) 2010-11-12T14:28:11 Migi32: I agree 2010-11-12T14:28:22 oef that would hurt my system :P 2010-11-12T14:28:23 That was my growth map suggestion. 2010-11-12T14:29:00 ah yes, I only read up to "all growths where multiples of 10" 2010-11-12T14:29:20 That was just the agrument to say that it doesn't need to be limited. 2010-11-12T14:29:42 Or at least wouldn't change much. 2010-11-12T14:29:58 how about floating point numbers ;) 2010-11-12T14:30:19 sent 26.66666666 ships directly to the opponent's planet 2010-11-12T14:30:47 2.6x10^7 ships to planet 1 2010-11-12T14:30:55 and I will send 26.66666667 ;) 2010-11-12T14:30:55 engage! 2010-11-12T14:31:30 i will send i^9 2010-11-12T14:32:03 3 dimensional maps would be fun to watch. I don't think it would change the game much. 2010-11-12T14:32:52 i^9 is just i 2010-11-12T14:32:54 :) 2010-11-12T14:33:16 :D 2010-11-12T14:33:30 janzert: I think I've gotten 700 games with my official bot. 2010-11-12T14:35:52 *** irchs has joined #aichallenge 2010-11-12T14:36:38 ouch, another thing that would break my bot, >+5 planets 2010-11-12T14:37:05 I feel another map generator run coming on when I get home... 2010-11-12T14:37:58 huh, that could be really interesting... 2010-11-12T14:38:43 would it be possible to make enough small fleets to time out your opponent? 2010-11-12T14:42:00 there will still be the same limit on number of fleets thanks to coalescing orders 2010-11-12T14:43:32 the leaderboard hasn't been updated in 25 minutes 2010-11-12T14:43:35 janzert: whats the elo update frequency now? 2010-11-12T14:43:38 lol mega1 +) 2010-11-12T14:44:27 lavalamp_: it don't think so 2010-11-12T14:44:45 Error323: never right now, I'm working on it manually at the moment :/ 2010-11-12T14:44:51 I don't create fleet objects, I just add them together based on destination and turn 2010-11-12T14:44:56 I guess it depends on the number of planets there are, not teh number of ships you have 2010-11-12T14:45:36 *** boegel has quit IRC (Quit: *poof!*) 2010-11-12T14:45:39 janzert: working on it manually? 2010-11-12T14:45:43 the problem is each test iteration takes 5 minutes and my attention wanders in that time :) 2010-11-12T14:45:46 *** McLeopold has quit IRC (Read error: Connection reset by peer) 2010-11-12T14:45:58 like computing the scores by hand ^_^ 2010-11-12T14:46:14 *** McLeopold has joined #aichallenge 2010-11-12T14:46:18 hehe, no but working on the system that does it 2010-11-12T14:46:55 on the live server interesting :P 2010-11-12T14:47:20 janzert, if I updated the scripts to 1. make an additional table (leaderboard_calc) 2. replace the data in that table instead of adding to it 3. do the queries out of that table: would that help you at all? 2010-11-12T14:47:58 hmm? what's in leaderboard_calc? 2010-11-12T14:48:17 Only the most recent leaderboard, instead of every single one 2010-11-12T14:48:33 My memory may be fuzzy, it's been a few days since I looked at it 2010-11-12T14:48:38 nah, there isn't any bottleneck there 2010-11-12T14:48:54 the bottleneck is getting the game result data 2010-11-12T14:48:56 where is the bottleneck? 2010-11-12T14:48:58 ok I must have misunderstood which table 2010-11-12T14:49:07 I'll probably split up the games table soon 2010-11-12T14:49:17 that's probably what you want to do 2010-11-12T14:49:19 ok then I'll stop bothering you 2010-11-12T14:49:21 basically archive games that are between old submissions 2010-11-12T14:49:40 that works too 2010-11-12T14:52:05 I love the truely awesome amount of gpm 2010-11-12T14:53:56 mega1: ya there? 2010-11-12T14:54:05 Error323: yes 2010-11-12T14:54:22 did you run the bot i send yet? :D 2010-11-12T14:55:26 I'm running work related stuff now, then when windows can be shut down in virtualbox I'll set up a linux vm 2010-11-12T14:55:38 @ratings 2010-11-12T14:55:39 janzert: User error, it's not my fault. 2010-11-12T14:55:43 @rankings 2010-11-12T14:55:44 janzert: Top 10 players: Raschi(4347), bocsimacko(4345), asavis(4197), davidjliu(4147), shangas(4139), george(4139), Accoun(4111), _Astek_(4103), GreenTea(4080), dmj111(4074) 2010-11-12T14:56:07 ah ok 2010-11-12T15:01:28 @rankings 20 2010-11-12T15:01:30 McLeopold: Top 20 players: Raschi(4347), bocsimacko(4345), asavis(4197), davidjliu(4147), shangas(4139), george(4139), Accoun(4111), _Astek_(4103), GreenTea(4080), dmj111(4074), hutabhu(4053), jambachili(4027), Doraj8(4021), felixcoto(3997), Error323(3996), Slin-.-(3996), oldman(3996), flowerbot(3971), _iouri_(3970), FlameN(3969) 2010-11-12T15:01:56 \o/ 2010-11-12T15:02:01 * lavalamp_ really tempted to ask for enough rankingns to see myself 2010-11-12T15:02:09 Nice, I didn't realize you were ranked so high. 2010-11-12T15:02:29 It make me feel better that I can beat you on tcp. :P 2010-11-12T15:02:44 yeah Error323 is ahead of the crowd :/ 2010-11-12T15:03:06 if i asked for enough ranking to see myself contestbot would spam the channel for several minutes 2010-11-12T15:03:10 Error323: which tcp version is official? 2010-11-12T15:03:19 maybe it's his fault the TCP ratings are so brutal >:| 2010-11-12T15:03:20 McLeopold: 17.2 2010-11-12T15:03:23 Zannick: it would refuse 2010-11-12T15:03:28 @rankings 1000 2010-11-12T15:03:28 McLeopold: Error: Cannot do that many players. 2010-11-12T15:03:29 or that :P 2010-11-12T15:03:49 what's the max? 2010-11-12T15:03:55 @rankings 50 2010-11-12T15:03:55 Error323: Error: Cannot do that many players. 2010-11-12T15:04:02 @rankings 30 2010-11-12T15:04:03 Error323: Top 30 players: Raschi(4347), bocsimacko(4345), asavis(4197), davidjliu(4147), shangas(4139), george(4139), Accoun(4111), _Astek_(4103), GreenTea(4080), dmj111(4074), hutabhu(4053), jambachili(4027), Doraj8(4021), felixcoto(3997), Error323(3996), Slin-.-(3996), oldman(3996), flowerbot(3971), _iouri_(3970), FlameN(3969), joakim_mjardner(3952), dabino(3952), Polsky(3951), jimrogerz(3950), (1 more message) 2010-11-12T15:04:05 it's 25 i think 2010-11-12T15:04:08 or 30 2010-11-12T15:04:15 @rankings 35 2010-11-12T15:04:17 Error323: Top 35 players: Raschi(4347), bocsimacko(4345), asavis(4197), davidjliu(4147), shangas(4139), george(4139), Accoun(4111), _Astek_(4103), GreenTea(4080), dmj111(4074), hutabhu(4053), jambachili(4027), Doraj8(4021), felixcoto(3997), Error323(3996), Slin-.-(3996), oldman(3996), flowerbot(3971), _iouri_(3970), FlameN(3969), joakim_mjardner(3952), dabino(3952), Polsky(3951), jimrogerz(3950), (1 more message) 2010-11-12T15:04:55 but he can't fit more than 28 in one line 2010-11-12T15:05:05 er, 24 2010-11-12T15:05:45 hey, my bot climbed about 20 ranks since yesterday 2010-11-12T15:05:57 hooray top 650 2010-11-12T15:06:51 My bot will meet you half way, I'm going the other direction... :/ 2010-11-12T15:07:02 *** RainCT has joined #aichallenge 2010-11-12T15:12:50 i love when my bot decides on this strategy http://ai-contest.com/visualizer.php?game_id=7059885 2010-11-12T15:14:08 well, maybe not all optimal decisions there... :/ 2010-11-12T15:14:25 there was a link to that official server stats page earlier, can't find it. 2010-11-12T15:14:43 http://ai-contest.com/game_stats.php 2010-11-12T15:14:58 that's the one, thanks! 2010-11-12T15:15:07 @learn gpm as http://ai-contest.com/game_stats.php 2010-11-12T15:15:07 lavalamp_: Aye, aye, sir 2010-11-12T15:15:21 @stats 2010-11-12T15:15:21 janzert: I have 1 registered users with 0 registered hostmasks; 1 owner and 0 admins. 2010-11-12T15:15:32 I figured stats was taken ^^ 2010-11-12T15:16:26 @game_stats 2010-11-12T15:16:27 janzert: I do not know about 'game_stats', but I do know about these similar topics: 'gamestats' 2010-11-12T15:16:27 ? 2010-11-12T15:16:34 @gamestats 2010-11-12T15:16:35 janzert: gamestats = http://ai-contest.com/game_stats.php 2010-11-12T15:16:40 I thought it was in there 2010-11-12T15:16:48 Top 10 players: Raschi(4347), bocsimacko(4345), asavis(4197), davidjliu(4147), shangas(4139), george(4139), Accoun(4111), _Astek_(4103), GreenTea(4080), dmj111(4074) 2010-11-12T15:18:04 ah ok 2010-11-12T15:18:19 gpm is probably a good alias though anyway 2010-11-12T15:19:00 yeah, for gamestats I would expect... I dunno, some win/loss ratios or something 2010-11-12T15:19:38 yeah, that page should probably really be something like contest stats 2010-11-12T15:20:06 contest stats should include languages available and languages used 2010-11-12T15:20:45 that could be added :) 2010-11-12T15:21:00 it's just not something I'm generally worried about all the time 2010-11-12T15:21:24 i've been considering adding that for the next one, after this one 2010-11-12T15:21:45 I've thought about adding to the start packages page 2010-11-12T15:22:38 *** Accoun has quit IRC () 2010-11-12T15:24:04 *** ColdPhoenix has joined #aichallenge 2010-11-12T15:24:40 hey, i still think this stats page is lacking the obvious thing, the total number of bots/players involved 2010-11-12T15:25:09 active bots is easy to check on the leaderboard 2010-11-12T15:25:14 k. 2010-11-12T15:25:18 ...not really 2010-11-12T15:25:28 but yeah, it could be integrated 2010-11-12T15:25:34 I've got some really weird problems going on. Don't seem to be getting errors or anythin' but games are "dying" on me 2010-11-12T15:25:42 er, now it gives me the actual last page 2010-11-12T15:25:43 like this, http://72.44.46.68/canvas?game_id=989527 2010-11-12T15:25:45 what happened :( 2010-11-12T15:26:05 *** Cold-Phoenix has quit IRC (Ping timeout: 240 seconds) 2010-11-12T15:26:30 http://72.44.46.68/getplayer?player=d0ugal 2010-11-12T15:26:34 you are sending invalid orders 2010-11-12T15:26:51 ohh 2010-11-12T15:27:04 That would make sense then 2010-11-12T15:27:28 Now to find out where my numbers went wrong :/ 2010-11-12T15:28:22 lol, i'm on the last day of my prepaid/allinonewithusbstick dataplan, and my provider definitely cuts me down, so timeouts/disconnected 2010-11-12T15:29:06 *** ColdPhoenix has quit IRC () 2010-11-12T15:31:30 thank god, irc isn't sending more than 512 at once ! 2010-11-12T15:34:32 *** Accoun has joined #aichallenge 2010-11-12T15:35:30 just for fun, how many bytes you ppl are spending on tcp on an average day ? 2010-11-12T15:40:05 * p4p4 died of a tuesday 2010-11-12T15:45:52 *** irchs has quit IRC (Quit: irchs) 2010-11-12T15:54:00 *** irchs has joined #aichallenge 2010-11-12T15:54:20 i have no clue 2010-11-12T15:54:29 i have a single bot running 24/7 2010-11-12T15:54:36 ho much would that be 2010-11-12T15:54:43 it uses the full second though 2010-11-12T15:55:34 *** McLeopold has quit IRC (Ping timeout: 272 seconds) 2010-11-12T15:55:55 I've got 3 bots running 24/7 2010-11-12T15:56:04 they each play about 1100 games a day 2010-11-12T15:56:36 *** Snowplan has joined #aichallenge 2010-11-12T15:57:30 so a planet can have exactly zero ships 2010-11-12T15:57:54 yes 2010-11-12T15:58:36 4 others have joined me, that makes me happy: http://ai-contest.com/language_profile.php?lang=Go 2010-11-12T15:59:02 @repo 2010-11-12T15:59:03 Frontier: repo = http://code.google.com/p/ai-contest/ 2010-11-12T16:02:41 *** CryBaby has left #aichallenge 2010-11-12T16:05:52 Snowplan: It should only happen after any kind of attack that neutralizes the growth on that planet or when both players attack a neutral with equal forces. 2010-11-12T16:06:36 I noticed my first game on the server the distance was just right to cause it 2010-11-12T16:07:04 it's all cool now 2010-11-12T16:11:44 *** irchs has quit IRC (Quit: irchs) 2010-11-12T16:14:00 * janzert wishes someone would use Groovy 2010-11-12T16:14:18 so the effort put into adding it doesn't feel wasted :/ 2010-11-12T16:15:13 *** delt0r has joined #aichallenge 2010-11-12T16:16:12 *** delt0r___ has quit IRC (Ping timeout: 264 seconds) 2010-11-12T16:16:49 Top 10 players: bocsimacko(4371), Raschi(4287), asavis(4254), bix0r4ever(4205), shangas(4191), george(4189), davidjliu(4180), _Astek_(4147), dmj111(4115), GreenTea(4108) 2010-11-12T16:17:29 will use groovy 2010-11-12T16:17:57 heh, thanks 2010-11-12T16:18:10 whats groovy 2010-11-12T16:18:23 *** Frontier has quit IRC (Ping timeout: 245 seconds) 2010-11-12T16:18:26 another programming language 2010-11-12T16:22:33 *** Frontier has joined #aichallenge 2010-11-12T16:22:43 janzert: If you added Pascal I knew someone who would use it. 2010-11-12T16:37:39 why does the server close the connection 2010-11-12T16:41:43 *** Frontier has quit IRC (Ping timeout: 245 seconds) 2010-11-12T16:49:02 *** lavalamp_ has quit IRC (Ping timeout: 265 seconds) 2010-11-12T16:50:04 hey, janzert, you've done a great thing, setting up all those languages ! 2010-11-12T16:51:07 thanks, I only added a couple of them though. DanielVF did most of them. 2010-11-12T16:51:17 *** McLeopold has joined #aichallenge 2010-11-12T16:52:21 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-11-12T16:54:15 *** fawek has joined #aichallenge 2010-11-12T16:54:16 *** p4p4 has quit IRC (Ping timeout: 245 seconds) 2010-11-12T16:59:57 McLeopold: so do you like the new maps? 2010-11-12T17:00:11 Haven't really looked yet. 2010-11-12T17:00:45 i don't think the concept of growth maps, it's basically turning it into who can solve a decision problem most optimally, not who can play a a game the best 2010-11-12T17:01:00 same with my problem with mario ai, it's not really a game, but a decision problem 2010-11-12T17:01:30 think/like* :P 2010-11-12T17:01:43 that's part of the game. How would you take that out? 2010-11-12T17:02:13 my final bot is being written as something that just tries to maximise anyway 2010-11-12T17:02:18 now we have 2 9n places http://ai-contest.com/rankings.php 2010-11-12T17:03:04 but on those maps, your moves don't really affect the other player until the game is pretty much decided 2010-11-12T17:03:24 and 2 31s, and two 41s, and three 55s... 2010-11-12T17:03:24 hmm... 2010-11-12T17:03:59 there has always been draws 2010-11-12T17:04:02 antimatroid: so you would prefer maps where you can affect your opponent sooner than later? 2010-11-12T17:06:12 i want maps where the outcome isn't strictly determined by ones ability to expand before even reaching the enemy 2010-11-12T17:06:30 otherwise i feel the problem is no longer much of a game but more of a decision problem 2010-11-12T17:06:46 i wouldn't be opposed to some maps like that i guess, but it doesn't seem very gamy 2010-11-12T17:07:35 antimatroid: I agree, my idea of growth maps wasn't disconnected clusters, but rather planets that are spread out and just take longer to aquire. 2010-11-12T17:07:56 are there new maps on the server? 2010-11-12T17:08:03 I really dislike the maps which are to giant blobs on either side of a map. 2010-11-12T17:09:17 ahhh okay, amstan suggested two big glog maps yesterday and i wasn't a fan 2010-11-12T17:09:21 antimatroid: I'm looking at the maps with some test games right now... 2010-11-12T17:09:22 that's what i was referring to 2010-11-12T17:09:36 300-400 are coolest imo 2010-11-12T17:10:11 you will get both kinds of symmetry in there, and non radially symmetric maps can have multiple neutrals starting at an equal distance to each player 2010-11-12T17:10:22 and varying number of planets :) 2010-11-12T17:10:52 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-11-12T17:10:58 ooh, these maps are revealing weaknesses in my bot! 2010-11-12T17:11:33 besides the ones mega1 already revealed... 2010-11-12T17:13:07 i don't think it's fair to say some people assumed they would only have maps like the examples, therefore everyone else should be restricted to that 2010-11-12T17:13:18 even though many of us haven't been working under any such restriction the whole time 2010-11-12T17:13:28 *** superflit has quit IRC (Ping timeout: 272 seconds) 2010-11-12T17:13:51 If we could push the deadline, that would be awesome. Everyone get's an extra week for dealing with the new maps. 2010-11-12T17:13:59 i also think the closer the players starting angles, the better one needs to be at expanding without being sniped 2010-11-12T17:14:34 these maps will definatly change the game. 2010-11-12T17:14:42 but it's still a very valid form of map, i don't see how it matters what kind of maps are used in the final contest, provided they're symmetric 2010-11-12T17:14:57 Often, the best expansion route is the opposite of you opponent, so you don't really meet up right away. 2010-11-12T17:15:04 people were told the example maps aren't exhaustive testing tools 2010-11-12T17:15:09 Now it's the same direction, like big center maps. 2010-11-12T17:15:31 I think these will reduce the average number of turns per game. 2010-11-12T17:15:33 is it not unfair to people who didn't make assumptions to now restrict the domain of possible maps? 2010-11-12T17:16:14 antimatroid: well, everyone has to make assumptions somewhere, and processing time has been an issue 2010-11-12T17:16:33 so, it is slightly unfair if someone has to do a re-write to support this. 2010-11-12T17:16:41 the new maps shouldn't really change processing time 2010-11-12T17:16:49 Top 10 players: bocsimacko(4363), asavis(4198), shangas(4171), george(4158), davidjliu(4157), bix0r4ever(4154), _Astek_(4121), dmj111(4082), Accoun(4078), GreenTea(4077) 2010-11-12T17:17:00 we're not asking to extend max dist to 1000 and have 100 planets 2010-11-12T17:17:02 If the planet count is the same or less, probably not too much. 2010-11-12T17:17:36 i think planets <= 30 is good 2010-11-12T17:17:53 If everything is not "official" yet anyway, just do a couple of days on the official servers with these maps. 2010-11-12T17:18:07 it's tricky to remove maps once they're added 2010-11-12T17:18:27 we really need dhartmei to stick some on tcp, but i don't know how to get ahold of him 2010-11-12T17:18:32 i tried an email address, but it bounced 2010-11-12T17:18:51 probably went in his tarpit :P 2010-11-12T17:19:58 j3camero: cough ahem splatter, you about? 2010-11-12T17:20:42 I wonder if waterloo could spare an ec2 instance for a tcp server? 2010-11-12T17:22:00 *** Azrathud has joined #aichallenge 2010-11-12T17:22:03 i think if my map gen is used i would prefer to increase the min starting distance between players from 2 2010-11-12T17:22:21 it's very boring when they're right next to each other on one edge of the map 2010-11-12T17:27:38 not that boring, but maybe at least 4 2010-11-12T17:28:28 *** Rubicon-|-Cross has joined #aichallenge 2010-11-12T17:32:19 just make sure that doing nothing or attacking the enemy aren't the only viable moves. 2010-11-12T17:32:35 already done 2010-11-12T17:32:52 ok then no map is boring, just slow to start 2010-11-12T17:35:51 *** Mathnerd314 has joined #aichallenge 2010-11-12T17:37:42 *** RainCT has quit IRC (Remote host closed the connection) 2010-11-12T17:46:26 *** superflit has joined #aichallenge 2010-11-12T17:50:00 *** choas has quit IRC (Quit: leaving) 2010-11-12T17:57:31 *** jaspervdj has quit IRC (Quit: NEVER GONNA GIVE YOU UP NEVER GONNA LET YOU DOWN) 2010-11-12T18:16:49 Top 10 players: bocsimacko(4392), asavis(4179), shangas(4179), Raschi(4156), george(4141), davidjliu(4137), Accoun(4124), bix0r4ever(4123), GreenTea(4086), dmj111(4075) 2010-11-12T18:24:03 *** Apophis has joined #aichallenge 2010-11-12T18:27:29 anyone have a planet switching heuristic? 2010-11-12T18:31:47 *** Sylph has quit IRC (Read error: Connection reset by peer) 2010-11-12T18:32:45 *** Bobng has quit IRC (Ping timeout: 255 seconds) 2010-11-12T18:50:52 yes 2010-11-12T18:56:26 Mathnerd314: so, how does it work? 2010-11-12T18:56:39 do you try and switch planets with the same growth rate? 2010-11-12T18:56:53 it just works into the strategy 2010-11-12T18:57:19 how is that? 2010-11-12T18:57:40 explaining it would give away my entire strategy 2010-11-12T18:58:55 *** superflit has quit IRC (Quit: superflit) 2010-11-12T18:59:05 which is probably enough information to figure it out 2010-11-12T18:59:35 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-11-12T19:04:03 *** Sylph has joined #aichallenge 2010-11-12T19:04:06 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-11-12T19:08:26 *** Apophis has quit IRC (Read error: Connection reset by peer) 2010-11-12T19:15:00 *** siconize has quit IRC (Ping timeout: 265 seconds) 2010-11-12T19:16:49 Top 10 players: bocsimacko(4433), asavis(4204), shangas(4189), george(4150), davidjliu(4148), Accoun(4144), bix0r4ever(4095), GreenTea(4091), dmj111(4087), jambachili(4063) 2010-11-12T19:17:05 *** McLeopold has quit IRC (Quit: Leaving) 2010-11-12T19:20:08 *** mceier has quit IRC (Quit: leaving) 2010-11-12T19:24:41 *** fawek has quit IRC (Ping timeout: 240 seconds) 2010-11-12T19:31:28 *** wdouglas` has quit IRC (Ping timeout: 265 seconds) 2010-11-12T19:33:00 *** iris1 has joined #aichallenge 2010-11-12T19:48:08 *** Mathnerd314 has quit IRC (Remote host closed the connection) 2010-11-12T19:48:17 *** asdffasf has quit IRC (Quit: Page closed) 2010-11-12T20:16:50 Top 10 players: bocsimacko(4462), asavis(4195), shangas(4184), Raschi(4163), george(4157), davidjliu(4150), Accoun(4114), bix0r4ever(4113), GreenTea(4108), dmj111(4089) 2010-11-12T20:47:08 *** amriedle has joined #aichallenge 2010-11-12T20:51:46 *** Migi32 has quit IRC (Quit: www.ai-contest.com) 2010-11-12T20:53:50 *** _xvinyl <_xvinyl!~xvinyl@ANantes-256-1-33-33.w86-199.abo.wanadoo.fr> has joined #aichallenge 2010-11-12T20:57:33 *** Rubicon-|-Cross has quit IRC (Ping timeout: 245 seconds) 2010-11-12T21:14:24 *** bhasker has joined #aichallenge 2010-11-12T21:16:50 *** eburnette has joined #aichallenge 2010-11-12T21:17:00 amstan: Error: There was a problem accessing the interface to ai-contest.com 2010-11-12T21:17:15 I failed the captcha test getting into irc, lol 2010-11-12T21:17:34 @rankings 2010-11-12T21:17:35 amstan: Top 10 players: bocsimacko(4455), Raschi(4210), asavis(4191), shangas(4177), george(4148), davidjliu(4126), GreenTea(4086), Accoun(4085), bix0r4ever(4074), dmj111(4068) 2010-11-12T21:17:38 there 2010-11-12T21:22:10 @rankings 50 2010-11-12T21:22:10 eburnette: Error: Cannot do that many players. 2010-11-12T21:22:14 @rankings 40 2010-11-12T21:22:15 *** yasith has quit IRC (Ping timeout: 255 seconds) 2010-11-12T21:22:16 eburnette: Top 40 players: bocsimacko(4455), Raschi(4210), asavis(4191), shangas(4177), george(4148), davidjliu(4126), GreenTea(4086), Accoun(4085), bix0r4ever(4074), dmj111(4068), hutabhu(4034), jambachili(4028), oldman(4014), felixcoto(4014), Slin-.-(4013), _Astek_(4008), Doraj8(4005), flowerbot(4002), luksian(4002), Xrillo(3966), SUA.khb(3963), murrayr(3948), Polsky(3947), joakim_mjardner(3943), (1 more message) 2010-11-12T21:32:41 *** bhasker has quit IRC (Read error: No route to host) 2010-11-12T21:33:27 *** bhasker has joined #aichallenge 2010-11-12T21:35:58 *** bhasker has quit IRC (Client Quit) 2010-11-12T21:44:01 zzzzzzzzzzzzzz 2010-11-12T21:44:21 Did anything get decided on the generator? 2010-11-12T21:46:03 eburnette: still waiting i think 2010-11-12T21:47:03 looks like poll preference is slightly in favour of using them atm, and I know jeff and amstan liked them, so i would say there's a reasonable chance of it happening though 2010-11-12T21:48:09 i like how people are complaining and saying these things should be announced 1 or even 2 months before the competition ends.. 2010-11-12T21:48:11 so.. 2010-11-12T21:48:26 last competion was 3 weeks long 2010-11-12T21:49:06 what was the link to sample maps generated by the latest generator? 2010-11-12T21:53:23 i like how people suggest it's just trying to mess the leaderboard up 2010-11-12T21:53:31 i've been pushing for these changes for ages 2010-11-12T21:53:47 no one did it, i finally get on holidays and do it myself, and everyone bitches about somebody doing something 2010-11-12T21:54:17 they have had nothing in stone yet, that's what i'm trying to get, with the generality that had sort of always been promised 2010-11-12T21:55:50 when the status quo lasts for significant lengths of time people start assuming it's really stone 2010-11-12T21:57:04 those people would be wrong 2010-11-12T21:57:30 i still claim it's unfair not to implement more general maps 2010-11-12T21:57:53 people shouldn't get to enfore their own advantage by having made assumptions 2010-11-12T22:02:25 *** iFire has quit IRC (Remote host closed the connection) 2010-11-12T22:02:45 *** Snowplan has left #aichallenge 2010-11-12T22:02:57 I want to try the new maps, where are some samples? 2010-11-12T22:07:48 http://ai-contest.com/forum/download/file.php?id=177 there's 300 maps 2010-11-12T22:08:29 1-100 are with the original generator, 101-200 are with line symmetric maps, 201-300 are combinations of both, and line symmetric maps can have multiple "central neutrals" 2010-11-12T22:11:20 oh god, i just managed to cause like the most horribly triple post ever 2010-11-12T22:11:41 *** Snowplan has joined #aichallenge 2010-11-12T22:14:24 *** Snowplan has left #aichallenge 2010-11-12T22:16:50 Top 10 players: bocsimacko(4449), dmj111(4277), Raschi(4199), asavis(4189), shangas(4189), george(4156), SUA.khb(4107), davidjliu(4095), GreenTea(4085), bix0r4ever(4071) 2010-11-12T22:21:36 *** Barry-Nichols has joined #aichallenge 2010-11-12T22:25:53 everyone keeps saying this is changing the maps for the contest too 2010-11-12T22:26:13 but i was always under the impression that we never knew what maps would be used in the final contest, wasn't everyone else in here? 2010-11-12T22:30:07 ergh, someone submitted an ocaml bot that compiles on the main server but fails on the workers for some reason. 2010-11-12T22:33:14 ugh how did this code ever work 2010-11-12T22:35:14 *** dmj111` has joined #aichallenge 2010-11-12T22:35:39 *** dmj111 has quit IRC (Remote host closed the connection) 2010-11-12T22:35:57 *** shades has joined #aichallenge 2010-11-12T22:36:07 dmj111': have you taken a look at the new maps at all? 2010-11-12T22:38:49 *** Barry-Nichols has left #aichallenge ("Leaving") 2010-11-12T22:41:58 Is there a problem with the compile success / failure emails? I am not getting them any more and I am not sure if it's my email or the server not sending them out. 2010-11-12T22:49:33 *** phreeza has joined #aichallenge 2010-11-12T23:08:04 *** lavalamp_ has joined #aichallenge 2010-11-12T23:10:37 I generated some maps with differing growth and ship starting counts 2010-11-12T23:12:05 *** iFire has joined #aichallenge 2010-11-12T23:16:52 Top 10 players: bocsimacko(4445), dmj111(4323), Raschi(4274), oldman(4273), shangas(4153), bix0r4ever(4149), asavis(4131), george(4124), jambachili(4087), SUA.khb(4072) 2010-11-12T23:18:47 lavalamp_: do you think it's better if people don't know where the "centre" of the map was for generation 2010-11-12T23:19:12 for testing, maybe... 2010-11-12T23:19:25 the point maps are symmetric around planet 0 2010-11-12T23:19:43 yes, and my generator then translates them so all coordinates are positive 2010-11-12T23:19:48 I don't know if anyone relies on that or no... 2010-11-12T23:19:53 i'm just wondering if that translate amount should be random 2010-11-12T23:20:07 i was told coordinates need to be strictly positive 2010-11-12T23:20:13 ah, isn't it already if it depends on the planet positions? 2010-11-12T23:20:18 the current generator simply uses the min value of each coord 2010-11-12T23:20:29 effectively, that means some planets have coords of 0 2010-11-12T23:20:36 try it in the viewers to see if it works there 2010-11-12T23:20:44 they don't have to be strictly positive, but nonnegative is good 2010-11-12T23:20:56 antimatroid's maps all seem to work in jbotmanager 2010-11-12T23:21:11 but with a line symmetric map, i dont' think you can deduce the centre without knowledge of the amount transposed 2010-11-12T23:21:18 because there isn't necessarily a centre neutral 2010-11-12T23:21:34 didn't you make a center neutral in your generator? 2010-11-12T23:21:39 or is this another one? 2010-11-12T23:21:46 it can do multiple centre neutrals for line symmetric maps 2010-11-12T23:21:52 and they don't need to be in the dead centre 2010-11-12T23:21:55 honestly I don't think it's an important detail 2010-11-12T23:21:58 ah, you added more 2010-11-12T23:22:05 you can do more for point symmetric, but they need to be doubled on the other side of the centre line 2010-11-12T23:22:11 i...don't want to worry about adding those 2010-11-12T23:22:14 people should do the math and figure out what planets are equidistant already 2010-11-12T23:22:40 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-11-12T23:23:31 Zannick and eburnette: what do you guys think of the maps? 2010-11-12T23:23:43 i haven't really looked 2010-11-12T23:24:12 i like the line symmetric for maps with a center neutral, from the pictures you posted a while back 2010-11-12T23:24:29 i'll screen shot one with more centre neutrals for you 2010-11-12T23:24:48 i'll look as soon as I unbreak my bot 2010-11-12T23:26:40 *** jmpespxoreax is now known as jmacmaho 2010-11-12T23:26:45 http://imgur.com/vp1fA.png there's a line synmetric map with multiple "centre neutrals" 2010-11-12T23:26:49 I posted a bunch of maps in one of the threads 2010-11-12T23:26:56 that's one of them 2010-11-12T23:27:08 yeah those are interesting 2010-11-12T23:27:11 *** jmacmaho is now known as jmpespxoreax 2010-11-12T23:32:03 *** jmpespxoreax has left #aichallenge 2010-11-12T23:33:20 hm 2010-11-12T23:33:28 is that 17 planet number 0? 2010-11-12T23:35:07 *** jmpespxoreax has joined #aichallenge 2010-11-12T23:35:09 player 1's planet is planet 0, player 2's is planet 1 2010-11-12T23:35:17 then the centre planets come next, then the pairs 2010-11-12T23:35:52 shouldn't center come first, by convention if not rule? 2010-11-12T23:35:53 *** jmpespxoreax has left #aichallenge 2010-11-12T23:36:18 yeah, i kinda dislike that 2010-11-12T23:36:18 i have to generate the players locations first before doing the centers like that 2010-11-12T23:36:20 My bot doesn't care 2010-11-12T23:36:28 it's just easier, it shouldn't break any bot that followed the rules 2010-11-12T23:36:44 it can be changed if people really care 2010-11-12T23:36:55 but there isn't necessarily a planet at (0,0) on that map 2010-11-12T23:37:18 planets at (0,0) don't matter, i think, so much as 0=neutral, 1=p1, 2=p2 2010-11-12T23:37:25 why? 2010-11-12T23:37:53 well just for fun then put home planets at random locations in the list 2010-11-12T23:37:57 the latter is an easier optimization to make to figure out the center planet and starters 2010-11-12T23:37:57 people seem to be under the impression that the example map generator was meant to be taken as the official one 2010-11-12T23:38:14 it was the official one 2010-11-12T23:38:20 it was never an example 2010-11-12T23:38:25 it was always an example 2010-11-12T23:38:32 jeff cameron has said numerous times it will not be kept 2010-11-12T23:38:37 let's see how many bots we can break - change everything possible not specified by the rules 2010-11-12T23:38:49 well, copybot and my current submission will break 2010-11-12T23:38:52 these were the rules 2010-11-12T23:39:05 everyone has been told the whole time, don't make assumptions 2010-11-12T23:39:07 because i optimized for the other maps 2010-11-12T23:39:10 maybe if enough bots break i'll have a better chance :) 2010-11-12T23:39:14 it's unfair to peple who didn't to then restrict the domain of maps 2010-11-12T23:39:26 i'm more upset because i told other people those things :/ 2010-11-12T23:39:47 (I'm being sarcastic btw) 2010-11-12T23:40:01 i could probably make a one-line fix, though 2010-11-12T23:40:08 and disable my special first-turn logic 2010-11-12T23:40:12 Zannick: how would your bot break? 2010-11-12T23:40:17 *** jmpespxoreax has joined #aichallenge 2010-11-12T23:40:27 and is the first turn logic not based off an assumption you made? 2010-11-12T23:40:42 before this morning, my bot assumed 5 was teh max growth and 23 was the number of planets 2010-11-12T23:40:52 fortunately it wasn't too hard to fix 2010-11-12T23:41:07 my bot will just read what it's given 2010-11-12T23:41:12 on the first turn it says, who owns planet 1? if 1, then it's mine, if 2, then i own 2. no owns 0, and that's equidistant so ignore it for now 2010-11-12T23:41:19 i think it'd probably die if planet size went far above 50 2010-11-12T23:41:20 make a map that has all 0-growth planets, that would be fun 2010-11-12T23:41:23 but probably not time out 2010-11-12T23:41:28 and copybot will break entirely 2010-11-12T23:41:32 but that's less important 2010-11-12T23:41:34 that would actually be quite fun 2010-11-12T23:42:22 copybot could be fixed easily enough 2010-11-12T23:42:26 that should make people get lots of divide by zero exceptions, and eliminate a few more losers 2010-11-12T23:42:29 it just needs to go through and pair planets at the start 2010-11-12T23:42:43 i'm not fixing it :P i don't have time 2010-11-12T23:42:52 but i think anyone that assumed map planets would take any particularly order made as stupid an assumption as those who expected only one kind of symmetry 2010-11-12T23:43:11 :( 2010-11-12T23:43:12 tron maps had multiple symmetries 2010-11-12T23:43:26 okay, i made stupid assumptions 2010-11-12T23:43:29 :P, sorry, but i don't want the domain of maps restricted 2010-11-12T23:43:32 * Zannick -> 2010-11-12T23:43:35 here's an idea - add a couple of extra planet owners that just sit there and see if anybody will conquer them 2010-11-12T23:43:49 extra planet owners? 2010-11-12T23:43:54 the game can't go to more than 2 players 2010-11-12T23:44:01 that is an entirely new game 2010-11-12T23:44:06 players 3, 4, 5, rules allow it 2010-11-12T23:44:12 not the contest rules 2010-11-12T23:44:28 plus they would be ejected on turn one for not making moves 2010-11-12T23:44:32 as soon as n>2 you start having to form coalitions 2010-11-12T23:44:37 it's a completely different game 2010-11-12T23:44:47 it's like saying, right, your bots are now fighting in a mario kart battle 2010-11-12T23:45:07 no i mean these aren't AIs or people, it's like some dead civilization with stationary ships 2010-11-12T23:45:19 eburnette: "neutral" 2010-11-12T23:45:20 so it's just another neutral? 2010-11-12T23:45:33 or growing neutrals? 2010-11-12T23:45:37 the engine won't readily support that 2010-11-12T23:46:02 sort of like neutrals except owner is not == 0 2010-11-12T23:46:27 i do not believe the engine will support that * 2010-11-12T23:46:39 sure, lettem grow too, that would be interesting.. timing exactly when to capture them 2010-11-12T23:47:03 i like the 0 growth map idea 2010-11-12T23:47:08 i have a script that lets you run 3 players on a map 2010-11-12T23:47:13 early when they're cheap? or later if you have to expand 2010-11-12T23:47:20 so long as you specify player 3 start planet 2010-11-12T23:47:28 actually, 0 growth map is an easy tie 2010-11-12T23:47:38 always jump around with your full 100 ships 2010-11-12T23:47:48 no, just never move 2010-11-12T23:47:54 yeah or that 2010-11-12T23:47:59 :P 2010-11-12T23:48:33 although, if you trusted yourself, you could try to tempt them into trying to take you 2010-11-12T23:48:35 hmm yeah guess you're right 2010-11-12T23:48:35 and take the win 2010-11-12T23:48:42 and possibly do bette than all draws 2010-11-12T23:49:15 too bad we can't have negative growth, black holes 2010-11-12T23:49:17 although if their bot strictly does nothing, you lose 2010-11-12T23:49:24 no one would move there 2010-11-12T23:49:26 pretty simple 2010-11-12T23:49:42 no sucking action from planets :P 2010-11-12T23:51:11 actually wouldn't it be better to attack the other home planet? you might get a 1 move growth advantage 2010-11-12T23:51:30 cause they can't react until next move 2010-11-12T23:52:10 i'm confused. a 0 growth map has all 0 growth planets, no? 2010-11-12T23:52:25 so the players won't ever gain ships? 2010-11-12T23:52:26 home planet has growth 2010-11-12T23:52:49 oh 2010-11-12T23:52:55 then either stay or attack 2010-11-12T23:52:58 oh, i thought that was 0 too 2010-11-12T23:53:04 yeah, ^^ that 2010-11-12T23:53:09 either way, stay or attack 2010-11-12T23:53:11 maybe you could make home planets and 1 other have growth to prevent a tie 2010-11-12T23:53:24 draw unless someone goes for a neutral and loses ships 2010-11-12T23:53:26 my bot shouldn't be having epic battles with ragebot >:( 2010-11-12T23:54:59 haha 2010-11-12T23:55:15 how does one have an epic battle with ragebot? 2010-11-12T23:55:27 I broke defense completely but it can still attack fine 2010-11-12T23:55:47 how old is that revision? 2010-11-12T23:55:58 i started a fresh one for the final submission 2010-11-12T23:56:02 of my bot? about 10 minutes 2010-11-12T23:56:12 so i can try and but out all the bugs and short cuts i previousloy made 2010-11-12T23:56:21 no, from a fresh starter pack 2010-11-12T23:56:23 I'll put it up on tcp when I get defense fixed 2010-11-12T23:56:34 i'm days away from even testing this new bot 2010-11-12T23:56:37 I missed something... revision of what? 2010-11-12T23:56:48 like, start from a bare starter package 2010-11-12T23:57:06 anyway, g2g... later 2010-11-12T23:57:14 alright I'm trying your weirdmaps2 2010-11-12T23:57:25 they don't disappoint :P 2010-11-12T23:57:31 my bot is not crashing 2010-11-12T23:57:53 no bot should crash unless they made errors 2010-11-12T23:58:23 but it's not winning much for some reason 2010-11-12T23:58:26 although it took me a bit of work to be able to feel my bot a map starting with fleets 2010-11-12T23:58:41 is your defense still broken? 2010-11-12T23:58:51 i think that becomes more important as the starting angles become closer 2010-11-12T23:59:00 and you need to be better at expanding without becoming vulnerable 2010-11-12T23:59:11 that's my initial thoughts on them anyway 2010-11-12T23:59:21 i haven't got a reasonable bot to test out on them yet 2010-11-12T23:59:22 we don't need no steenkeng defense