2010-11-01T00:01:04 demerzel_: so you are saying the branching factor is too high even for uct? 2010-11-01T00:01:37 how is the game not zero sum? 2010-11-01T00:01:47 i'm of the opinion that it very much is 2010-11-01T00:01:51 antimatroid: what happens when you attack a neutral planet? 2010-11-01T00:02:03 you lose ships or gain a planet 2010-11-01T00:02:21 how is that not zero sum? 2010-11-01T00:02:22 *** Tobu has joined #aichallenge 2010-11-01T00:02:29 attacking a planet is not the game :P 2010-11-01T00:02:39 ah 2010-11-01T00:02:50 attacking a planet it zero sum anyway 2010-11-01T00:02:58 in a way 2010-11-01T00:03:17 well, not in the long term. A difference of one ship changes who controls it, and then the growth makes it not zero sum 2010-11-01T00:03:43 how? 2010-11-01T00:04:00 your gains aren't always your opponents losses 2010-11-01T00:04:22 holding their ships/growth etc. constant and yeah it is 2010-11-01T00:04:35 I could be misunderstanding though 2010-11-01T00:04:40 so could i :P 2010-11-01T00:04:43 i'm just an undergrad 2010-11-01T00:04:50 me too 2010-11-01T00:05:05 i'm pretty sure it's reasonable to treat the game entirely as zero sum 2010-11-01T00:05:12 !eightball is attack a planet zero sum? 2010-11-01T00:05:28 .... 2010-11-01T00:05:29 eh? contestbot is not responding? 2010-11-01T00:05:40 @eightball is attack planet a zero sum game? 2010-11-01T00:05:40 antimatroid: It is possible. 2010-11-01T00:05:50 @eightball is attack planet a zero sum game? 2010-11-01T00:05:50 antimatroid: The answer is certainly yes. 2010-11-01T00:05:57 i guess that settles it then 2010-11-01T00:06:01 *** Frontier has quit IRC (Ping timeout: 276 seconds) 2010-11-01T00:06:12 hmmm... ! doesn't work anymore? 2010-11-01T00:06:14 !yay 2010-11-01T00:06:20 guess not 2010-11-01T00:06:22 yeah, i noticed that the other day when i tried to get rankings 2010-11-01T00:06:25 !rainbow rainbow 2010-11-01T00:06:30 @rainbow rainbow 2010-11-01T00:06:30 antimatroid: rainbow 2010-11-01T00:08:38 sigh: no i am saying it seems the main advantage of uct over mc is that it remembers 'good' moves and the likelihood of this being useful in a game where moves are unlikely to repeat means it's a small one at best 2010-11-01T00:09:21 demerzel_: what do you mean by "repeat" exactly? 2010-11-01T00:09:47 ? uct's advantage is that it uses mc to build a tree search 2010-11-01T00:13:03 Top 10 players: bocsimacko(4303), dmj111(4244), Hazard(4156), felixcoto(4129), joakim_mjardner(4125), FlameN(4124), Accoun(4122), luksian(4107), Slin-.-(4106), rsergio(4106) 2010-11-01T00:13:17 *** Gudradain has quit IRC (Quit: Page closed) 2010-11-01T00:17:47 zero sum if the planet is the enemies, not if it is a neutral 2010-11-01T00:17:57 McLeopold: thanks :) 2010-11-01T00:21:00 mcleopold i finally fixed most of my sniping related bugs ^ 2010-11-01T00:21:01 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-01T00:21:05 *** sigh_ has joined #aichallenge 2010-11-01T00:21:11 cool 2010-11-01T00:21:20 what about too much defense to the back? 2010-11-01T00:21:21 #45 on official now 2010-11-01T00:21:25 fixed that too 2010-11-01T00:21:43 http://72.44.46.68/canvas?game_id=800965 2010-11-01T00:21:47 sniping work 2010-11-01T00:21:48 what version on tcp? 2010-11-01T00:21:55 *** Rubicon-|-Cross has joined #aichallenge 2010-11-01T00:22:48 0.14 2010-11-01T00:23:04 though still bugs with my first move 2010-11-01T00:23:10 you must have had some security check on, because you didn't snipe right away 2010-11-01T00:23:13 my first move heuristic is just not good enough 2010-11-01T00:23:18 yes 2010-11-01T00:23:25 i do some calculations 2010-11-01T00:23:29 and don't snipe till the last minute 2010-11-01T00:23:37 when its not easy for him to counter 2010-11-01T00:23:39 haha i made ragebot: http://72.44.46.68/canvas?game_id=801072 2010-11-01T00:25:35 another game 2010-11-01T00:25:42 http://72.44.46.68/canvas?game_id=800575 no super defense to back 2010-11-01T00:26:35 i'm not convinced taking neutrals isn't zero sum 2010-11-01T00:29:17 antimatroid: it really depends on how you frame the question 2010-11-01T00:30:30 if you go down 30 ships, it's a plus 30 advantage to your opponent. 2010-11-01T00:30:45 if you attack him with 30 ships, you both go down 30 and it's even 2010-11-01T00:31:57 growth rates can complicate it, but I suppose you could say a position is like a good move each turn you hold it 2010-11-01T00:33:36 hah i like my snipe code check out this game and watch the top right 62 ship neutral 2010-11-01T00:33:38 http://72.44.46.68/canvas?game_id=801152 2010-11-01T00:35:15 I wouldn't be too proud of that 2010-11-01T00:35:32 It changed hands alot, and you sunk 10 into it with no affect. 2010-11-01T00:36:17 the 10 was a coordinated attack 2010-11-01T00:36:21 but since there was a 11 incoming 2010-11-01T00:36:27 it was better to let him burn it 2010-11-01T00:36:30 and then capture 2010-11-01T00:36:40 than capture by sending 22 ships first 2010-11-01T00:36:47 so it aborted the second the launch 2010-11-01T00:37:07 i meant it aborted the second launch* 2010-11-01T00:41:28 okay, that makes sense 2010-11-01T00:41:57 antimatroid: if you look at the wikipedia article on zero-sum, look at the extensions section 2010-11-01T00:42:55 you can reduce a non-zero-sum game to a zero-sum game by representing to global profit/loss 2010-11-01T00:43:20 so if you only care about having more ships at 200 turns, at that is your eval function, then it can be zero sum 2010-11-01T00:43:35 If you interesting in evaluating best planets to attack, then it is not 2010-11-01T00:44:45 I'm looking for critique on those ideas... 2010-11-01T00:46:37 hmm interesting 2010-11-01T00:46:47 i have that book in this room, i'll take a look 2010-11-01T00:47:54 McLeopold:i used zero-sum end game exclusively but you have to have some kind of midgame eval to make sure you will even get the full value 2010-11-01T00:49:03 haha i love wiki game theory articles 2010-11-01T00:49:04 For instance in the game of "chicken" if one player has the option of ripping the steering wheel from their car they should always take it because it leads to a "sub game" in which their rational opponent is precluded from doing the same thing (and killing them both). 2010-11-01T00:49:18 this book, for a first in it's field, has a butload of stuff 2010-11-01T00:50:49 in a pistol duel between 3, it is rational to purposely miss your target if you shoot first 2010-11-01T00:51:13 then the other 2 go for each other 2010-11-01T00:51:20 McLeopold: depends on the probablities, in that puzzzle 2010-11-01T00:51:31 *** sigh_ is now known as sigh 2010-11-01T00:51:49 sigh: like the probability that one has a repeating rifle? 2010-11-01T00:52:01 *** sigh is now known as Guest86556 2010-11-01T00:52:03 repeating rifle? 2010-11-01T00:52:16 um.. more than one bullet? 2010-11-01T00:52:24 *** Guest86556 is now known as sigh 2010-11-01T00:52:39 i can't find where they do that 2010-11-01T00:53:21 McLeopold: oh sorry, you are talking about where all 3 players have 100 % chance of hitting 2010-11-01T00:53:46 if you all have 100% chance of hitting and all know that each has that, then you know they purposely missed 2010-11-01T00:53:54 what you said made me immediatly think of a common puzzle where the three have different probabilities of making their shot 2010-11-01T00:56:32 I was just being silly. Yes, I was refering to that puzzle. 2010-11-01T00:57:07 :) 2010-11-01T00:57:53 there is not enough time :( 2010-11-01T00:57:59 someone needs to cure sleep already 2010-11-01T00:58:47 +1 2010-11-01T01:01:54 haha 2010-11-01T01:02:01 i spend hours tracking down a bug 2010-11-01T01:02:08 //GameState cState = gameState; 2010-11-01T01:02:08 GameState cState; 2010-11-01T01:03:44 ah, you are using the "calculate the best move then ignore it" approach 2010-11-01T01:05:13 no 2010-11-01T01:05:39 *** hornairs has joined #aichallenge 2010-11-01T01:05:47 i am ensuring the moves i make at the end very definitely valid in my starting state 2010-11-01T01:05:53 wait is this imperfect information because we're simultaneously moving or perfect because we know what has happened up to this point completely 2010-11-01T01:06:09 as an extensive form game it's imperfect information 2010-11-01T01:06:12 demerzel_: this game has "complete information" but not "perfect information" 2010-11-01T01:06:18 *** narnach has quit IRC (Ping timeout: 240 seconds) 2010-11-01T01:06:24 ah ty 2010-11-01T01:06:31 sigh: am i wrong there? 2010-11-01T01:06:32 getting lost in this game theory stuff again 2010-11-01T01:06:37 yeah, it gets confusing 2010-11-01T01:06:41 it's all so similar it's hard to keep straight 2010-11-01T01:07:03 hmm, just starts to make sense after a while 2010-11-01T01:07:21 i've spent a lot of time playing around with game theory though 2010-11-01T01:07:23 antimatroid: I don't know what subset imperfect information encompasses 2010-11-01T01:07:47 when information sets aren't singletons i think 2010-11-01T01:08:46 http://en.wikipedia.org/wiki/Information_set_(game_theory) i think that's right 2010-11-01T01:09:51 the way i consider information sets is one consists of a set of nodes for which a player could be at, and they wouldn't know which of the nodes in the information set they are at 2010-11-01T01:10:05 yeah, I think it is imperfect too 2010-11-01T01:10:14 but saying complete information is more precise 2010-11-01T01:10:27 but then that would mean all strategic games are imperfect information 2010-11-01T01:10:31 but i don't think that's right either 2010-11-01T01:10:32 http://en.wikipedia.org/wiki/Complete_information 2010-11-01T01:11:07 although this might be more relevent: http://en.wikipedia.org/wiki/Game_theory#Simultaneous_and_sequential 2010-11-01T01:11:27 *** narnach has joined #aichallenge 2010-11-01T01:11:57 "Thus, only sequential games can be games of perfect information, since in simultaneous games not every player knows the actions of the others." 2010-11-01T01:11:59 extensive form games with imperfect information are tricky to write functions to find equilibrium refinements for 2010-11-01T01:12:19 ah okay, cool 2010-11-01T01:12:44 *** hellman has joined #aichallenge 2010-11-01T01:13:03 Top 10 players: bocsimacko(4330), dmj111(4238), Hazard(4156), FlameN(4123), felixcoto(4121), joakim_mjardner(4120), luksian(4102), Slin-.-(4101), rsergio(4098), Accoun(4095) 2010-11-01T01:13:06 *** Azrathud has quit IRC (Ping timeout: 240 seconds) 2010-11-01T01:13:12 I only know about this because this topic came up once before 2010-11-01T01:13:42 i took a really crap course on game theory in my first year, most of it is self learnt 2010-11-01T01:14:17 I've never actually studied game theory properly 2010-11-01T01:14:39 whose bots are dimwit> 2010-11-01T01:14:53 phreeza 2010-11-01T01:15:01 the class didn't even use calculus, i would say technically neither have i :P 2010-11-01T01:15:12 hmmm... I think that's the UCT bot 2010-11-01T01:16:10 oh k 2010-11-01T01:16:10 *** sigh has quit IRC (Read error: Connection reset by peer) 2010-11-01T01:16:25 *** Azrathud has joined #aichallenge 2010-11-01T01:16:31 its not very smart i guess lives up to its name 2010-11-01T01:16:53 *** sigh has joined #aichallenge 2010-11-01T01:17:28 i don't think it matters how good the idea is that one throws at this, unless you also have a kickass move selection function, one isn't going to do very well 2010-11-01T01:17:59 *** delt0r___ has joined #aichallenge 2010-11-01T01:19:17 *** delt0r has quit IRC (Ping timeout: 245 seconds) 2010-11-01T01:20:42 *** wh1teside has quit IRC (Read error: Connection reset by peer) 2010-11-01T01:23:47 *** wh1teside has joined #aichallenge 2010-11-01T01:26:43 dammit i keep hitting run instead of build 2010-11-01T01:26:47 i feel rude if i just dc 2010-11-01T01:26:54 vc++? 2010-11-01T01:27:00 sorta 2010-11-01T01:27:02 i'm using vs ;) 2010-11-01T01:27:51 i almost put a vector in haha 2010-11-01T01:28:23 heh 2010-11-01T01:29:25 every time i try to read game theory stuff i just end up giving up 2010-11-01T01:29:33 too many options that make no sense to me 2010-11-01T01:29:57 i mean the theory makes sense but how to make the relative function is what is beyond me 2010-11-01T01:30:36 heh thats why i am sticking with a heuristic bot for now 2010-11-01T01:30:48 i cant wrap my head around how to implement a good minmax solution for something like this 2010-11-01T01:31:39 bhasker: you aren't the only one 2010-11-01T01:31:41 minmax is the only thing i can understand - except my min is pretty minimal 2010-11-01T01:32:04 minimax is easy to understand as an algorithm... hard to implement (effectivly) for this game 2010-11-01T01:32:26 i think my next trick would be to implement mega1's launching ships ahead of neutral acquistion 2010-11-01T01:32:37 need to start factoring in planets for attack before i acquire them 2010-11-01T01:32:48 bhasker: "launching ships ahead of neutral acquistion" ? 2010-11-01T01:32:49 sigh: it's the pruning of moves that's hard 2010-11-01T01:32:56 McLeopold: yes 2010-11-01T01:32:58 mega1's bot does that 2010-11-01T01:33:06 it launches partial fleets 2010-11-01T01:33:11 at start 2010-11-01T01:33:19 assuming that he will capture the neutral at X turns 2010-11-01T01:33:19 ah, I see 2010-11-01T01:33:25 and add ships from the neutral along the way 2010-11-01T01:33:39 i have the framework to do it but its a little tricky to get it right 2010-11-01T01:33:55 is it actually a good idea, or it is simply a side effect of his algorithm? 2010-11-01T01:34:07 well its a good idea at start at least 2010-11-01T01:34:12 not sure after mid game 2010-11-01T01:34:17 I can do it, and it's a side affect. 2010-11-01T01:34:18 at start you capture planets faster 2010-11-01T01:34:30 I don't think it would make too much difference. 2010-11-01T01:34:32 hmm how is it a side affect? 2010-11-01T01:34:34 i have what he has already 2010-11-01T01:34:43 what is hard is figuring out whether to keep it going as situations change 2010-11-01T01:34:48 well it does matter on games where start planets are far apart 2010-11-01T01:34:59 capturing a planet one turn ahead can mean win/loss 2010-11-01T01:35:18 ie you have a plan to launch 2/3 of the ships ahead of time 2010-11-01T01:35:24 but the last 1/3 is needed for defense 2010-11-01T01:35:38 but I don't see how sending a partial fleet to a neutral helps? 2010-11-01T01:35:41 the 2/3 are already in the air - so what do you gain/lose 2010-11-01T01:35:50 you send ahead of time 2010-11-01T01:35:54 *** fmeyer has quit IRC (Ping timeout: 240 seconds) 2010-11-01T01:36:01 knowing you will have ships at a closer planet to help out 2010-11-01T01:36:08 all arrive same turn 2010-11-01T01:36:34 lets say you will acquire 10 ships 2 turns from now 2010-11-01T01:36:36 and you need 12 to take the neutral 2010-11-01T01:36:39 you launch 2 now 2010-11-01T01:36:45 which travel 2 turns 2010-11-01T01:36:53 and then you add the 10 from the neutral you captured 2010-11-01T01:37:05 so you arrive at your destination with your full fleet 2 turns ahead 2010-11-01T01:37:09 oh sure, coordinating attacks from multiple sources, I do that already 2010-11-01T01:37:12 i did actually implement that - i just have to implement some kind of priority queue 2010-11-01T01:37:18 no its not coordinating 2010-11-01T01:37:21 its more 2010-11-01T01:37:23 it's not? 2010-11-01T01:37:28 the neutral is not converted yet 2010-11-01T01:37:31 its still neutral 2010-11-01T01:37:35 yea this involved queuing orders for a future turn 2010-11-01T01:37:37 when you launched the 2 2010-11-01T01:38:11 yeah, I delay orders so that they all arrive on the same turn... or at least they try to 2010-11-01T01:38:21 if you need to wait for the neutral, you might as well send enough for both 2010-11-01T01:38:26 one of my few wins 2010-11-01T01:38:30 well you don't hav eenough 2010-11-01T01:38:33 lets say it turn 1 2010-11-01T01:38:37 http://72.44.46.68/canvas?game_id=801072 you can see it 2010-11-01T01:38:38 you sent to capture 3 neutrals 2010-11-01T01:38:42 and there is a 4th 2010-11-01T01:38:54 which you could capture once you have some ships from the neutrals 2010-11-01T01:39:03 sigh: there's a difference between holding orders and actually launching this turn on faith that you will launch more 2 turns ahead 2010-11-01T01:39:04 you could launch your remaining at the 4th at turn 1 2010-11-01T01:39:14 and add in the ships from the neutrals as they convert 2010-11-01T01:39:25 well not at turn 1 2010-11-01T01:39:28 maybe at turn 3 2010-11-01T01:39:44 yea my bot does the holding part as well 2010-11-01T01:39:51 but it needs to have all the ships for the order 2010-11-01T01:40:02 in that game i launch at turn 29 2010-11-01T01:40:10 i don't factor into the calculations future ships 2010-11-01T01:40:21 ah, future arrivals, got you 2010-11-01T01:40:28 with the expectation that the neutral i acquire at turn 33 will help out 2010-11-01T01:40:43 yeah... mine only takes that into account to a very small extent 2010-11-01T01:40:53 it's difficult for a variety of reasons 2010-11-01T01:41:03 yup 2010-11-01T01:41:08 and even if you do implement it there's nothing to stop your plan from failing at the end 2010-11-01T01:41:11 yea well mega1's bot does it quite well 2010-11-01T01:41:13 true 2010-11-01T01:41:20 so its good at start on split maps 2010-11-01T01:41:26 where start positions are far apart 2010-11-01T01:41:34 can allow you to acquire a couple of neutrals faster 2010-11-01T01:41:46 yea my bot is seriously gimped atm i only allow it one target per turn but that's where it wins if it does win 2010-11-01T01:42:27 only one target per turn? o_O 2010-11-01T01:42:36 it's still very early alpha 2010-11-01T01:42:41 fair enough :) 2010-11-01T01:42:43 i mean it's only been working < 24 hours 2010-11-01T01:42:53 ah, i see 2010-11-01T01:42:55 and i just tracked down the last bug i know about like 4 hours ago lol 2010-11-01T01:43:44 this is the only game i'm semi-proud of so far: http://72.44.46.68/canvas?game_id=800923 2010-11-01T01:45:32 the rest are pretty horrible defeats because of my target limitation, bugs, and lack of defense lol 2010-11-01T01:45:54 that's not bad for such a new bot 2010-11-01T01:46:00 its interesting to watch the bot 2010-11-01T01:46:23 my bot is rather boring to watch 2010-11-01T01:46:25 :( 2010-11-01T01:46:27 it just makes me all warm and fuzzy to see 5 coordinated launches land 2010-11-01T01:46:41 it just piles up ships and steam rolls 2010-11-01T01:46:42 demerzel_: yeah, I know what you mean :) 2010-11-01T01:48:15 i think it's got potential but putting in more than 1 move is going to be ap ain 2010-11-01T01:48:19 or rather the branch part 2010-11-01T01:48:24 *** espes has joined #aichallenge 2010-11-01T01:48:33 branch part? 2010-11-01T01:48:38 because it will change every turn due to circumstances + figuring out which move to prune and if i've invested in them etc 2010-11-01T01:49:04 yea i'm going to set up a tree 2010-11-01T01:49:14 the moves it makes are completely isolated plans already 2010-11-01T01:49:31 gotta set up some kind of simworld to run them in and see results of movex -> movey etc 2010-11-01T01:49:37 i see... sounds similar to what I'm trying 2010-11-01T01:49:56 i had that in my old C# bot but it was isolated to 1 source -> 1 target 2010-11-01T01:50:38 sigh: what's your tcp name 2010-11-01T01:50:56 I haven't had a bot up on tcp for weeks 2010-11-01T01:51:19 ah well i know the feeling 2010-11-01T01:51:36 *** Palmik has joined #aichallenge 2010-11-01T01:52:17 I'm starting to get back into it, so hopefully after I make some improvements I'll start putting bots back on tcp 2010-11-01T01:53:22 problem is that all my ideas now need a lot of code re-writing and I keep putting it off 2010-11-01T01:56:10 *** TravisD has quit IRC (Quit: TravisD) 2010-11-01T01:56:37 *** virdo has quit IRC (Ping timeout: 255 seconds) 2010-11-01T02:02:36 demerzel_: brain finally registered, eto demerzel right? 2010-11-01T02:03:33 a.k.a. Oliver R. Daneel 2010-11-01T02:03:48 if my memory serves right 2010-11-01T02:05:26 *** zoo has quit IRC (Quit: Page closed) 2010-11-01T02:06:54 raedwulf: yes it's my sa forum name i chose like 7 years ago right after reading foundation so i use it sometimes and thought it was appropriate 2010-11-01T02:08:06 yes :P 2010-11-01T02:08:22 man i should have implemented a sorted by distance indices array a long time ago - gets rid of so many while loops it's not even funny 2010-11-01T02:08:26 i read foundation maybe before that ~ 2010-11-01T02:08:39 lol i did that immediately 2010-11-01T02:08:53 i actually prefer dune but it was fresh on the mind lol 2010-11-01T02:09:11 asimov is a little simplistic for me now but i was young and impressionable 2010-11-01T02:09:16 didn't manage to finish dune unfortunately, need to try it again 2010-11-01T02:09:20 it was too tense xD 2010-11-01T02:09:23 lol 2010-11-01T02:09:32 *** narnach has quit IRC (Remote host closed the connection) 2010-11-01T02:09:37 it chills way out and they basically have chapters of just 'what if' convos 2010-11-01T02:09:48 yea 2010-11-01T02:09:52 *** narnach has joined #aichallenge 2010-11-01T02:10:05 that's what happens when the main char is a basically invincible god lol 2010-11-01T02:10:31 *** McLeopold has quit IRC (Quit: Leaving) 2010-11-01T02:10:47 ^^ 2010-11-01T02:11:11 *** narnach has quit IRC (Client Quit) 2010-11-01T02:11:28 *** narnach has joined #aichallenge 2010-11-01T02:11:29 im writing a fantasy novel atm, but i've had so many other commitments, its been hard to find time for it 2010-11-01T02:11:54 is about 45 pages so far 2010-11-01T02:12:18 or 2010-11-01T02:12:25 well depends the size of page 2010-11-01T02:12:27 lol 2010-11-01T02:13:05 Top 10 players: bocsimacko(4322), dmj111(4233), Hazard(4147), shangas(4144), felixcoto(4107), FlameN(4107), luksian(4104), joakim_mjardner(4098), Accoun(4092), Slin-.-(4086) 2010-11-01T02:13:20 anyways i think i need to get my bot running today 2010-11-01T02:16:07 good luck! 2010-11-01T02:16:18 i got mine running but found about 8 hours worth of bugs lol 2010-11-01T02:16:23 was hoping to be a bit further 2010-11-01T02:16:31 i've been testing bits and pieces as i go along 2010-11-01T02:16:34 at least that's over with 2010-11-01T02:16:55 well me too but certain little things = big problems lol 2010-11-01T02:17:10 but the main search routine is completely untested and its parameters are random values which i probably need to tweak or "train" 2010-11-01T02:17:15 i do have a nice setup now where i can recreate the last tcp game i played over and over and figure out what i was thinking at least 2010-11-01T02:17:27 yea i have that too 2010-11-01T02:17:43 basically - i can take a text file with the moves 2010-11-01T02:17:48 and use that for stdin 2010-11-01T02:17:50 lol 2010-11-01T02:18:10 quite easy on linux just pipe it in using "<" 2010-11-01T02:18:13 *** Vasiliy_Semenov has quit IRC (Ping timeout: 255 seconds) 2010-11-01T02:18:31 i have a hookable 'nextline' proc 2010-11-01T02:18:41 either tcp or file or stdin can use it 2010-11-01T02:18:46 ah 2010-11-01T02:19:13 i have a bashscript called sniff.sh 2010-11-01T02:19:20 to record whatever is sent down stdin 2010-11-01T02:19:29 and saves it to text file 2010-11-01T02:20:00 hmmm probably don't need to do that - it would probably work with tee 2010-11-01T02:20:20 well it does use tee internally anyway 2010-11-01T02:31:02 *** sigh has quit IRC (Remote host closed the connection) 2010-11-01T03:12:41 you people talk too much 2010-11-01T03:13:05 wouldn't it be better for mega1 to not send those fleets first move? 2010-11-01T03:13:06 Top 10 players: bocsimacko(4331), dmj111(4235), Hazard(4169), shangas(4155), felixcoto(4114), FlameN(4111), luksian(4110), joakim_mjardner(4102), Accoun(4097), Slin-.-(4090) 2010-11-01T03:13:17 wait and see if it's still best to take those neutrals 2010-11-01T03:13:23 what game? 2010-11-01T03:13:37 you were talking about him sending out fleets first move to neutrals he intends to take later 2010-11-01T03:13:47 i do something like that, but i coordinate attacks 2010-11-01T03:14:13 so it wont send fleets from a planet until it's expected to reach the destination at the final time of the overall move 2010-11-01T03:14:24 i should change that for offense 2010-11-01T03:15:18 that also gives you an information advantage, in that the other bot can't entirely see what you're planning to do 2010-11-01T03:15:22 well in a pure cost-benefit analysis accounting for enemy possibility you would naturally favor closer 2010-11-01T03:15:42 but it wouldn't change the time in which you acquire the neutral 2010-11-01T03:15:46 on turn 1 that is 2010-11-01T03:15:52 yes that's true 2010-11-01T03:16:02 but why would you pick it as a target if you're factoring in enemy poss 2010-11-01T03:16:16 you might, but you don't know exactly what they're going to do 2010-11-01T03:16:25 those ships might be used somewhere better on the next turn 2010-11-01T03:16:36 if you don't already commit them you can make that move, if they're still optimal, send them then 2010-11-01T03:16:53 yes but if they weren't used somewhere you lost a turn's advantage 2010-11-01T03:17:12 woop my shitbot won against JamesMG(fun haha) 2010-11-01T03:17:13 no you didn't, cause you weren't planning to acquire the planet with that fleet 2010-11-01T03:17:26 only if the planet is in direct line extending one of the other neutrals you took 2010-11-01T03:17:31 it still has no concept of overcommitting 2010-11-01T03:18:54 mega1: if you can be bothered, ^^, yay or nay? 2010-11-01T03:19:17 antimatroid: what is the question exactly? 2010-11-01T03:19:34 people say you send fleets to neutrals on the first turn without intending to take it with that move 2010-11-01T03:19:42 i think it's better to wait and send those fleets the next turn 2010-11-01T03:19:45 i'm sure he intends to take it 2010-11-01T03:19:52 or, for them to arrive when you plan to take it 2010-11-01T03:19:53 he just might need to cancel that plan before completion 2010-11-01T03:20:06 I think my bot sends fleets at the latest possible time. 2010-11-01T03:20:23 i'm going strictly off the convo in here, nevermind ;) 2010-11-01T03:20:25 so if one is sent on the first turn then it's needed 2010-11-01T03:20:41 (or is a staging order) 2010-11-01T03:20:50 (which is unlikely on the first turn) 2010-11-01T03:20:52 i am not even going to touch staging for a while 2010-11-01T03:21:01 staging order? 2010-11-01T03:21:10 there is something wrong with my defense i can't quite figure out what it is but i can see it 2010-11-01T03:21:33 antimatroid: an order to send ships to the 'front line' 2010-11-01T03:21:55 hmmm, yeah i could see my bot doing that 2010-11-01T03:22:17 i tried doing a bot the other night that had each player pick moves and implement it against the state from the moves the other player just made 2010-11-01T03:22:32 but then if i had their bot "expand", my bot would try to offend against that 2010-11-01T03:22:42 hence my bot started trying to attack neutrals closest to them stupidly 2010-11-01T03:23:32 you should ask the opponent what he's going to play 2010-11-01T03:23:59 i just treat it like it's my bot and pick moves 2010-11-01T03:27:13 haha holy shit 2010-11-01T03:27:15 my bot turned into a sniping machine: http://72.44.46.68/canvas?game_id=803010 2010-11-01T03:28:36 their bot just sucked :P 2010-11-01T03:28:52 you probably shouldn't be waiting first move like that 2010-11-01T03:28:54 dont cheapen my victory i get very few atm lol 2010-11-01T03:29:03 :P sorry 2010-11-01T03:29:04 it's not intentional haha i just don't feel like debugging it atm 2010-11-01T03:29:18 my bot sucks at first move too 2010-11-01T03:29:34 what is your tcp name 2010-11-01T03:29:41 nothing atm 2010-11-01T03:29:45 oic 2010-11-01T03:29:56 usually antimatroid, youwillfail or something along those lines 2010-11-01T03:30:19 i had an exam this morning :P 2010-11-01T03:30:50 i have more exams saturday, monday and tuesday then i get to work on my bot till the end of the comp 2010-11-01T03:32:17 so naturally, i'm now going to start doing some work to my bot :) 2010-11-01T03:34:00 *** Arthur has joined #aichallenge 2010-11-01T03:36:08 *** Arthur has quit IRC (Remote host closed the connection) 2010-11-01T03:36:35 *** Arthur has joined #aichallenge 2010-11-01T03:39:34 *** Arthur has quit IRC (Client Quit) 2010-11-01T03:41:14 *** Aaaarg has joined #aichallenge 2010-11-01T03:44:08 woop my first 'real' victory short of really close map wins due to no defense consideration 2010-11-01T03:44:14 http://72.44.46.68/canvas?game_id=803194 - almost up to a real bot 2010-11-01T03:44:19 error323 haha 2010-11-01T03:58:25 *** demerzel_ has quit IRC (Quit: Page closed) 2010-11-01T04:05:01 *** hornairs has quit IRC (Ping timeout: 276 seconds) 2010-11-01T04:13:03 *** Jumpster has joined #aichallenge 2010-11-01T04:13:06 Top 10 players: bocsimacko(4333), dmj111(4237), shangas(4204), Hazard(4164), luksian(4126), felixcoto(4125), FlameN(4124), joakim_mjardner(4096), Slin-.-(4096), Accoun(4092) 2010-11-01T04:18:29 *** boegel has joined #aichallenge 2010-11-01T04:20:01 *** Commandir has joined #aichallenge 2010-11-01T04:24:48 *** sigh has joined #aichallenge 2010-11-01T04:37:08 *** Itkovian has quit IRC (Quit: Itkovian) 2010-11-01T04:40:42 *** shades has quit IRC (Ping timeout: 240 seconds) 2010-11-01T05:04:42 *** shades has joined #aichallenge 2010-11-01T05:10:31 *** jaspervdj has joined #aichallenge 2010-11-01T05:12:22 *** Accoun has joined #aichallenge 2010-11-01T05:13:07 Top 10 players: bocsimacko(4362), dmj111(4275), shangas(4193), Hazard(4187), felixcoto(4150), FlameN(4143), luksian(4137), Accoun(4130), joakim_mjardner(4128), Slin-.-(4110) 2010-11-01T05:21:53 *** Accoun has quit IRC () 2010-11-01T05:25:28 *** Accoun has joined #aichallenge 2010-11-01T05:32:56 *** narnach has quit IRC (Quit: Leaving.) 2010-11-01T05:41:24 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-11-01T05:45:49 *** Appleman1234 has joined #aichallenge 2010-11-01T06:10:48 *** Jumpster has left #aichallenge 2010-11-01T06:13:06 Top 10 players: bocsimacko(4366), dmj111(4275), Hazard(4206), shangas(4197), felixcoto(4155), FlameN(4150), Accoun(4133), luksian(4131), joakim_mjardner(4131), Slin-.-(4107) 2010-11-01T06:17:14 *** boegel has joined #aichallenge 2010-11-01T06:35:13 *** Appleman1234 has quit IRC (Ping timeout: 276 seconds) 2010-11-01T06:36:31 *** Appleman1234 has joined #aichallenge 2010-11-01T07:09:08 *** narnach has joined #aichallenge 2010-11-01T07:12:13 *** Itkovian has joined #aichallenge 2010-11-01T07:13:07 Top 10 players: bocsimacko(4369), dmj111(4273), Hazard(4201), shangas(4181), FlameN(4148), felixcoto(4143), joakim_mjardner(4136), Accoun(4135), luksian(4119), Slin-.-(4106) 2010-11-01T07:14:48 is the current ranking page broken? 2010-11-01T07:15:07 *** Appleman1234 has quit IRC (Read error: Connection reset by peer) 2010-11-01T07:19:49 it's blocked by AdBlock in Firefox :) 2010-11-01T07:20:01 ah right 2010-11-01T07:20:09 thats what i added this morning 2010-11-01T07:20:25 because i was getting annoyed with popup ads on another site lol :P 2010-11-01T07:23:38 *** Appleman1234 has joined #aichallenge 2010-11-01T07:23:47 thanks Naktibalda, put it on my allow list ^^ 2010-11-01T07:25:39 *** sigh has quit IRC (Remote host closed the connection) 2010-11-01T07:36:23 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-11-01T07:43:11 *** Utkarsh has joined #aichallenge 2010-11-01T07:55:52 *** rangzen has joined #aichallenge 2010-11-01T08:01:22 *** Itkovian has quit IRC (Remote host closed the connection) 2010-11-01T08:01:30 *** Itkovian has joined #aichallenge 2010-11-01T08:13:06 Top 10 players: bocsimacko(4369), dmj111(4274), Hazard(4199), FlameN(4148), felixcoto(4148), joakim_mjardner(4144), Accoun(4142), shangas(4122), Slin-.-(4106), rsergio(4085) 2010-11-01T08:14:21 *** Itkovian has quit IRC (Quit: Itkovian) 2010-11-01T08:14:45 *** Appleman1234 has quit IRC (Ping timeout: 252 seconds) 2010-11-01T08:17:37 *** Appleman1234 has joined #aichallenge 2010-11-01T08:29:51 *** boegel has quit IRC (Quit: Leaving) 2010-11-01T08:43:30 *** Appleman1234 has quit IRC (Ping timeout: 250 seconds) 2010-11-01T08:49:31 My life is complete: I beat bocsimacko :) http://www.ai-contest.com/visualizer.php?game_id=6436339 2010-11-01T08:51:23 nicely done 2010-11-01T08:51:27 congrats:) 2010-11-01T08:52:22 *** Itkovian has joined #aichallenge 2010-11-01T08:57:15 *** RainCT has joined #aichallenge 2010-11-01T09:00:44 *** Migi32 has joined #aichallenge 2010-11-01T09:05:12 congratulation, you can now die with a smile :) 2010-11-01T09:05:23 lol 2010-11-01T09:06:28 very frustrating, our goal is to beat a top10 in just a game, we don't expect reach the top :) 2010-11-01T09:08:47 *** Ari2 has quit IRC (Quit: leaving) 2010-11-01T09:08:48 simple command queue : done, simple frontier supply : done. Java complet rewriting is going, will I come back in top500? :) 2010-11-01T09:11:56 Sadic bot keep victim alive :) http://72.44.46.68/canvas?game_id=806598 2010-11-01T09:13:07 Top 10 players: bocsimacko(4375), dmj111(4285), Hazard(4204), FlameN(4164), shangas(4155), joakim_mjardner(4150), felixcoto(4148), Accoun(4142), Slin-.-(4098), rsergio(4091) 2010-11-01T09:13:33 lol 2010-11-01T09:13:39 probably ignores zero growth planets 2010-11-01T09:13:48 mine does that too till the enemy takes it 2010-11-01T09:14:20 your bot just went up against mine 2010-11-01T09:14:28 rangzen-0.2.3 2010-11-01T09:15:59 http://72.44.46.68/canvas?game_id=806700 2010-11-01T09:16:13 you burnt over a 100 ships taking 3 neutrals right next to me and my bot sniped the 2010-11-01T09:16:14 them 2010-11-01T09:16:44 would you guys think it's bad if my bot surrendered when it sees itself hopelessly lost on TCP? 2010-11-01T09:17:10 it will say "invalid move" when it was a genuine win for you, but there will be more games. 2010-11-01T09:17:58 doesn't really matter on TCP i guesss 2010-11-01T09:18:53 *** delt0r has joined #aichallenge 2010-11-01T09:19:06 *** delt0r___ has quit IRC (Ping timeout: 240 seconds) 2010-11-01T09:20:07 let's do that then 2010-11-01T09:20:17 *** Itkovian has quit IRC (Quit: Itkovian) 2010-11-01T09:20:54 if I'm behind 10 growth and in ship numbers I don't even think there's a 0.01% chance of coming back 2010-11-01T09:21:28 hmm probably not, unless you manage to steal the planet in next couple of turns 2010-11-01T09:21:37 actually 10 growth can be made up 2010-11-01T09:21:40 if you steal a 5 2010-11-01T09:21:41 you are even 2010-11-01T09:22:00 i would say >10 is harder 2010-11-01T09:22:11 though as long as you have more ships you can always win 2010-11-01T09:23:14 hmm, I changed my mind. Surrendering will show up as "invalid move", so it will mask if there is a genuine invalid move sent... 2010-11-01T09:23:21 (even though that should never happen) 2010-11-01T09:23:32 if you really want to surrender just launch all your ships 2010-11-01T09:23:35 at some enemy targets 2010-11-01T09:23:48 most probably any good bot will just wipe you out 2010-11-01T09:24:14 ah yes, or send to every neutral exactly how many ships that neutral has 2010-11-01T09:25:35 bhasker: I don't take care of sniping right now, I don't even re-implement anti rage bot so … :) 2010-11-01T09:25:47 ah 2010-11-01T09:27:19 *** Frontier has joined #aichallenge 2010-11-01T09:27:40 still cherching a good score function, it's go too far away right now 2010-11-01T09:27:51 looking for 2010-11-01T09:28:01 French? :P 2010-11-01T09:28:08 :D 2010-11-01T09:28:09 :D 2010-11-01T09:28:30 Oh ça va … avec search c'est pas loin … 2010-11-01T09:29:26 uh.. 10,000,000 possibilities to attack the central 119 ships planet on map1 from 10 sourrounding planets seems a bit too much to evaluate in real-time. 2010-11-01T09:30:18 *** Azrathud has quit IRC (Ping timeout: 250 seconds) 2010-11-01T09:30:40 yeah, and people are planning minimax tree searches. For 2 turns deep that's 10000000^2 possibilities. For 5 it's unimaginable 2010-11-01T09:32:03 why would you need to evaluate every possibility 2010-11-01T09:32:14 If you could just go with approximations... but most of the time in planet wars you have to deal with exact numbers of ships. 2010-11-01T09:32:23 can't you like heuristically eliminate most of the launches 2010-11-01T09:32:39 that's harder than it sounds, I think 2010-11-01T09:32:57 bhasker: Because, it is cool. No honestly, because you might find a combination that leaves your attack source planets in a good constellation for the next set of attacks 2010-11-01T09:33:19 hmm 2010-11-01T09:33:54 In effect every single ship can make a difference many times later, I believe. Like a butterfly effect. 2010-11-01T09:34:34 That's quite academic though, as the enemy's decision bring much more unexpected behaviour into the graph. 2010-11-01T09:39:14 I have a feature request on bendedrine.cx server - can you keep games longer? All links to games beyond 12 hours or so on irc chat are invalid 2010-11-01T09:44:03 too many games happening 2010-11-01T09:44:13 also if its your game don't you just log the state? 2010-11-01T09:57:34 *** Azrathud has joined #aichallenge 2010-11-01T09:59:19 *** Migi32 has quit IRC (Quit: http://boinc.berkeley.edu) 2010-11-01T10:13:07 Top 10 players: bocsimacko(4373), dmj111(4284), Hazard(4203), shangas(4185), FlameN(4164), joakim_mjardner(4148), felixcoto(4145), Accoun(4139), Slin-.-(4094), rsergio(4082) 2010-11-01T10:15:25 *** JamesMG has joined #aichallenge 2010-11-01T10:31:18 *** superflit has quit IRC (Quit: superflit) 2010-11-01T10:34:26 *** bhasker has quit IRC (Read error: Connection reset by peer) 2010-11-01T10:35:04 *** lavalamp_ has quit IRC (Quit: Page closed) 2010-11-01T10:43:37 *** bhasker has joined #aichallenge 2010-11-01T10:48:42 *** fmeyer has joined #aichallenge 2010-11-01T10:54:13 bhasker: that's where i learnt to use game tree's with an ai bot 2010-11-01T10:54:14 http://www.sifflez.org/misc/tronbot/index.html 2010-11-01T10:54:19 *** krokkrok has joined #aichallenge 2010-11-01T10:59:46 *** krokkrok has quit IRC (Quit: Page closed) 2010-11-01T11:08:52 *** superflit has joined #aichallenge 2010-11-01T11:11:18 interesting thanks antimatroid 2010-11-01T11:13:08 Top 10 players: bocsimacko(4381), dmj111(4298), shangas(4223), Hazard(4195), FlameN(4166), Accoun(4155), joakim_mjardner(4154), felixcoto(4147), Slin-.-(4096), jambachili(4090) 2010-11-01T11:13:36 np, i'm pretty sure i wasn't the only one to work it out off that page last time ;) 2010-11-01T11:14:05 now if only i could work out why my bot doesn't like to attack very much anymore :| 2010-11-01T11:18:46 *** Azrathud has quit IRC (Ping timeout: 276 seconds) 2010-11-01T11:32:15 *** fmeyer has quit IRC (Ping timeout: 240 seconds) 2010-11-01T11:34:00 *** lavalamp_ has joined #aichallenge 2010-11-01T11:41:42 bhasker, how many games are happening and how big is the state (the games aren't usually mine) 2010-11-01T11:46:16 *** demerzel_ has joined #aichallenge 2010-11-01T11:47:01 one of my friends who was data mining them 2010-11-01T11:47:07 told me over a few thousand every day 2010-11-01T11:48:14 Naktibalda: if you're still here, thanks--was wondering why "eaderboard" was broken in ff but only at home 2010-11-01T11:48:18 datamining what? 2010-11-01T11:48:52 *** bhasker has quit IRC (Quit: bhasker) 2010-11-01T11:49:01 i have literally 8000 top 100 games and a script to search all top 100 vs top 100 games but they're worthless in a week lol 2010-11-01T11:49:04 *** Sarin has joined #aichallenge 2010-11-01T11:50:22 * janzert wonders why adblock is getting a false positive on our rankings page 2010-11-01T11:50:32 If there are 400 an hour and each one takes 500K to store (both very conservative estimates) that would be under 5GB per day, so a 1TB drive could store 200 days worth. 2010-11-01T11:51:28 janzert: i'm not sure but if i use firebug to wipe out the 'class="leaderboard"' it displays fine 2010-11-01T11:51:48 heh, wierd 2010-11-01T11:51:52 eburnette: there are 400 games per hour maybe but only so many that are within the top 100, it was 8000 last time 2010-11-01T11:52:15 janzert: could change the class to tblRanking or something else 2010-11-01T11:52:56 eburnette: still with 8000 games i have something like 5.2 million fleet states lol 2010-11-01T11:53:08 I'd just like links to games to last more than a few hours 2010-11-01T11:53:12 the only really useful data i got out of it was all the maps are now in my db 2010-11-01T11:53:44 even tho i have yet to use it except to provide mega1 with a list of all the starting planet distances by mapid 2010-11-01T11:53:49 looking back that may not have been smart 2010-11-01T11:54:37 demerzel: lol 2010-11-01T11:54:39 Fleet states are from, to, owner, length, remaining. 5 integers, maybe 15 bytes in ascii. 2010-11-01T11:55:34 under 100MB 2010-11-01T11:55:39 oh yea it's not huge 2010-11-01T11:56:01 i actually started out thinking i'd just do like i do with my fleets only once but it was too much of a pain 2010-11-01T11:56:04 it's under half a gig 2010-11-01T11:56:27 janzert: this is the thing that kills the "leaderboard" -- ~courier-journal.com,~sporza.be,~southparkstudios.com,~mpora.com,~metalsucks.net,~mygamercard.net,~trueachievements.com##.leaderboard 2010-11-01T11:56:41 idk why it triggers on there honestly 2010-11-01T11:58:26 i guess the address don't matter - anyway i'll post it up on the forums it's pretty easy to fix 2010-11-01T11:58:47 So whether it's 0.5GB or 5GB per day, isn't there room for a few more days of history to get more persistent game links? 2010-11-01T11:59:01 *** Rodya has left #aichallenge ("Adieu, dear Werther!") 2010-11-01T12:00:52 Like this one you posted 24 hours ago: http://72.44.46.68/canvas?game_id=793385 - can't see it now. 2010-11-01T12:00:53 that's not per day tho 2010-11-01T12:01:00 oh you mean tcp scrape 2010-11-01T12:01:11 i don't even try to do that his keep window is so short it's ridiculous 2010-11-01T12:01:15 er, bhasker posted that one 2010-11-01T12:01:32 i scraped the ai-contest games 2010-11-01T12:01:42 8000 is the number of top 100 vs top 100 over all time 2010-11-01T12:01:54 or was a week ago, probably lower due to all the resubmits 2010-11-01T12:02:30 http://www.ai-contest.com/forum/viewtopic.php?f=18&t=1049 2010-11-01T12:02:31 The ai-contest games are persistent. For example http://ai-contest.com/visualizer.php?game_id=6282482 is still there 2010-11-01T12:02:39 yea that's why i chose to scrape it 2010-11-01T12:04:11 guess i'll figure out why i do nothing on turn 1 2010-11-01T12:04:16 altho it has worked out a few times lol 2010-11-01T12:06:25 ai-contest seems to keep at least 1 million games, http://ai-contest.com/visualizer.php?game_id=5000000 and http://ai-contest.com/visualizer.php?game_id=6000000 still exist 2010-11-01T12:07:30 4500000 exists 2010-11-01T12:08:15 *** bhasker has joined #aichallenge 2010-11-01T12:09:04 my history object still has bugs, I like haven't made any progress :( 2010-11-01T12:09:58 what are you hoping to do with it 2010-11-01T12:10:26 that was a very easy bug - if only all of them were like that 2010-11-01T12:10:57 *** hornairs has joined #aichallenge 2010-11-01T12:11:43 ai-contest has all the games since the contest started, the format on how they are stored in the database changed early in the contest and I'm not sure games before that were converted to the new format but they are still there in the old format for sure 2010-11-01T12:12:41 mine is trying to send more ships from a planet than are stationed there during the evaluation function :( 2010-11-01T12:12:56 there are currently 2295362 games in the database in the current format 2010-11-01T12:13:08 Top 10 players: bocsimacko(4387), dmj111(4298), shangas(4224), Hazard(4193), FlameN(4164), Accoun(4154), joakim_mjardner(4152), felixcoto(4146), george(4111), Slin-.-(4096) 2010-11-01T12:13:32 And how big is the database in terms of disk space 2010-11-01T12:13:37 lavalamp_: that sucks, i've definitely got an overcompensation on defense and there's something wrong with it in general but i can't find a concrete problem 2010-11-01T12:15:01 it's just off - i had perfect defense on my C# bot lol - at least my c++ is picking up a few wins now makes me feel better 2010-11-01T12:15:07 the table containing the game playback information alone is just over 20GB 2010-11-01T12:15:29 you probably don't store fleet info directly ie you only have launch turn 2010-11-01T12:15:32 and it is being stored compressed 2010-11-01T12:15:35 i was going to do that but too much effort to make the functions 2010-11-01T12:18:05 total size of the contest database on disk is 26GB 2010-11-01T12:19:31 *** davidjliu has joined #aichallenge 2010-11-01T12:31:36 *** Rubicon-|-Cross has quit IRC (Quit: Leaving) 2010-11-01T12:39:38 damn i thought shitbot was gonna win against mega1 for a second but he fixed me up: http://spz.nofate.com/planetwars/visualizer/?game_id=808494&server=tcp 2010-11-01T12:39:54 you had the game won 2010-11-01T12:40:19 yea i have no 'stupid move' eval lol 2010-11-01T12:40:26 then played it too aggressive 2010-11-01T12:40:36 it has no consideration for leaving itself open yet ;) 2010-11-01T12:41:28 the only reins it has atm is 'don't take a neutral unless you can't take an enemy' and also 'only if you are going to lose without it' 2010-11-01T12:42:39 *** Blkt has joined #aichallenge 2010-11-01T12:43:10 good evening everyone! 2010-11-01T12:45:33 morning :) 2010-11-01T12:46:33 afternoon 2010-11-01T12:52:28 hello 2010-11-01T12:52:36 hi 2010-11-01T12:53:30 demerzel_: the 'only if you are going to lose without it' bit sounds hard 2010-11-01T12:54:04 it's not perfect 2010-11-01T12:54:11 it's temporary anyway until i get a tree going 2010-11-01T12:54:28 it was just spending all its fleets taking neutrals and losing horribly when it was ahead 2010-11-01T12:56:33 that's what mine does atm 2010-11-01T12:58:21 if you can't take an enemy juts compare turn 200 score totals and if you're going to lose then tell it to go ahead 2010-11-01T12:58:25 that's my temporary stopgap 2010-11-01T12:58:43 works pretty well 2010-11-01T12:58:55 i only have mine expand if i don't have final growth over like 1.7*theirs 2010-11-01T12:58:58 excluding first turn 2010-11-01T12:59:13 first turn your scores are even 2010-11-01T12:59:15 <= 2010-11-01T12:59:28 not when i'm updating info for each "move" i make :P 2010-11-01T12:59:45 yea same it's flawed for sure 2010-11-01T12:59:48 but better than nothing 2010-11-01T13:00:06 one option is to make that decision about expanding at the start of the move 2010-11-01T13:00:09 rather than as you go 2010-11-01T13:00:20 *** bhasker_ has joined #aichallenge 2010-11-01T13:01:51 i don't understand why, my bot refuses to send redistribution moves if i tell it to send all unused ships, but if i do 0.99* it's fine, but then it always sends 1 less than the amount of growth 2010-11-01T13:02:04 yea i would do that except i don't intend for the heuristics to stay and it doubles as a stupid move check until i get that set in because it can only pick one neutral per turn 2010-11-01T13:02:35 some heuristics will always have to remain 2010-11-01T13:02:37 *** bhasker has quit IRC (Ping timeout: 276 seconds) 2010-11-01T13:02:41 never! 2010-11-01T13:02:47 don't destroy my dream 2010-11-01T13:02:54 you're going to be exploring too many dead end avenues with this otherwise 2010-11-01T13:03:02 only the purest math bot 2010-11-01T13:03:06 it will be like the kid from powder 2010-11-01T13:03:11 explode into pure mathematical light 2010-11-01T13:03:29 haha, pure maths isn't going to help you that much for this i don't think 2010-11-01T13:03:38 *** Bobng has joined #aichallenge 2010-11-01T13:03:41 it's done me alright thus far 2010-11-01T13:03:48 none of my bots have had heuristics 2010-11-01T13:03:56 what are you doing? 2010-11-01T13:04:02 calcing pure score gain 2010-11-01T13:04:23 i hear pure maths and think of algebra/analysis etc. :P 2010-11-01T13:04:46 it has huge flaws like midgame and early game termination 2010-11-01T13:04:55 but that's what i intend to fix 2010-11-01T13:05:20 nah i don't really understand the 0.99999/mymom*thehouse eval functions honestly 2010-11-01T13:05:59 i am certain there is a basic arithmetic solution 2010-11-01T13:07:38 mine is edging towards more calculative approach 2010-11-01T13:08:06 my move selection algorithm is so inefficient it takes forever 2010-11-01T13:08:12 i can't even process like 500 nodes in a turn 2010-11-01T13:13:10 Top 10 players: bocsimacko(4368), dmj111(4277), shangas(4204), Hazard(4188), FlameN(4150), joakim_mjardner(4140), Accoun(4117), felixcoto(4106), george(4085), jambachili(4079) 2010-11-01T13:18:45 bhasker_: i had you :( 2010-11-01T13:19:09 *** rebelxt has joined #aichallenge 2010-11-01T13:20:14 *** rebelxt has left #aichallenge 2010-11-01T13:26:08 here's a test zero sum game bot 2010-11-01T13:26:09 http://72.44.46.68/getplayer?player=zBot 2010-11-01T13:26:19 seems to be doing well when it doesn't time out 2010-11-01T13:27:48 which it keeps bloody doing :| 2010-11-01T13:28:13 *** demerzel__ has joined #aichallenge 2010-11-01T13:28:20 *** roshan has joined #aichallenge 2010-11-01T13:29:43 *** roshan has left #aichallenge 2010-11-01T13:30:11 *** demerzel_ has quit IRC (Ping timeout: 265 seconds) 2010-11-01T13:34:57 this is an interesting game: http://72.44.46.68/canvas?game_id=809000 2010-11-01T13:36:06 *** Gudradain has joined #aichallenge 2010-11-01T13:37:44 ah zBot yea it owned my bot 2010-11-01T13:37:46 before it timed out 2010-11-01T13:38:08 it timed out against pretty much everyone 2010-11-01T13:38:13 i'm reworking some things 2010-11-01T13:38:40 how do you work zero-sum in this game 2010-11-01T13:38:46 i guess i sorta do it but it's not explicit 2010-11-01T13:39:06 my gamestate is evaluated for how much i care about it, i consider that - their payoff 2010-11-01T13:39:17 i use decision trees to generate pure strategies 2010-11-01T13:39:43 and your care in numerical format = ? 2010-11-01T13:39:52 ? 2010-11-01T13:40:08 you say you eval for how much you care about it 2010-11-01T13:40:17 what does that mean in numerical terms 2010-11-01T13:40:18 *** mceier has joined #aichallenge 2010-11-01T13:40:49 *** jaspervdj has quit IRC (Quit: NEVER GONNA GIVE YOU UP NEVER GONNA LET YOU DOWN) 2010-11-01T13:41:44 let current fleets run their course then, myShips - theirShips + remainingTime*(ourCumulativeGrowth - theirCumulativeGrowth) 2010-11-01T13:42:20 one could always just set it up as a non-zero sum game, but i don't really know what else to use for a payoff function yet anyway 2010-11-01T13:42:31 ah well that's pretty much what i do then 2010-11-01T13:42:37 a score comparison essentially 2010-11-01T13:44:08 i still have no idea how to do future states with both players making moves into the future 2010-11-01T13:45:22 once a planet is expected to switch ownership multiple times and players repeating attacks change who owns it at what time makes it very hard to work out how to resolve that 2010-11-01T13:49:24 antimatroid, yeah that gets kinda hairy... that's currently where my bugs are :( 2010-11-01T13:50:47 i'm curious as to whether anyone has that right 2010-11-01T13:50:51 probably only mega1 aha 2010-11-01T13:50:59 :P 2010-11-01T13:51:30 im going to have some fun debugging soon 2010-11-01T13:51:47 but i'm somewhat busy atm... have lots of other commitments 2010-11-01T13:52:05 in theory, my code should handle this 2010-11-01T13:52:19 how are you doing it? 2010-11-01T13:52:19 *** rebelxt has joined #aichallenge 2010-11-01T13:52:31 *** rebelxt has left #aichallenge 2010-11-01T13:52:35 that would be telling! 2010-11-01T13:52:38 constraints 2010-11-01T13:53:06 of which, i've not quite figured out yet 2010-11-01T13:53:32 I cancel conflicting orders 2010-11-01T13:53:40 but i think my method is way too conservative 2010-11-01T13:53:51 i think my bot might end up cowering at home 2010-11-01T13:53:53 scared to leave 2010-11-01T13:54:29 yeah, i considered trying to make it so that my bot NEVER lost a planet, but i think it'd get crippled too easily 2010-11-01T13:54:59 my bot thinks it's Marco friggin Polo 2010-11-01T13:55:06 if one of yuour planets is attacked you could consider attacking the sources of those attacks 2010-11-01T13:55:26 if you can gain 2 planets instead of saving 1 2010-11-01T13:55:41 i do defense above anything else atm 2010-11-01T13:55:54 eliminates a whole aspect of moves to consider in my trees 2010-11-01T13:55:57 same here, I'm starting to think it's crippling my offense in some cases 2010-11-01T13:55:59 davidd: yea, thats what i was figuring 2010-11-01T13:56:07 same with redistribution, i just tack that on at the end 2010-11-01T13:56:09 yup i made defense automatic and perfect 2010-11-01T13:56:18 you're right tho it does cause problems on offense 2010-11-01T13:56:31 yep, sending troops to the front is my last step too 2010-11-01T13:56:39 yep, another thing i do is never send out ships from a planet of mine that is to be taken or whatever 2010-11-01T13:56:44 which is very sub optimal in places 2010-11-01T13:56:48 yeah, defense & offense are the same value, what matters is how important the planets that you end up with are 2010-11-01T13:56:52 but if i remove that, it sends ships out just before being taken 2010-11-01T13:57:10 yeah, I let them last stand on defense, which is pretty dumb in many cases 2010-11-01T13:57:22 antimatroid: mine currently wastes ships like that by sending them to a neutral right before it loses a planet 2010-11-01T13:57:27 i use some suboptimal strategies for my testing 2010-11-01T13:57:35 tbf tho it's a rare case where the time you spend defending could be better used somewhere else 2010-11-01T13:57:38 my bot sucks atm anyway 2010-11-01T13:57:40 for 1 it's cheap and a guaranteed value 2010-11-01T13:57:43 i want back in the top 10 :( 2010-11-01T13:57:45 i have an optimal bot as opponent 2010-11-01T13:57:57 and i build my new features in the suboptimal bot 2010-11-01T13:58:05 see if it can overcome the difference 2010-11-01T13:58:34 than after developing a feature i add that to the optimal bot 2010-11-01T13:58:37 even simple sitations are so damn complicated, it's so hard to get everything right 2010-11-01T13:59:05 the distance between " do what i want" and writing the if then constraints to make it so 2010-11-01T14:07:30 that's why defense is automatic - the situation where it being optional is an advantage are so rare and everythign else is already compartmentalized 2010-11-01T14:10:11 demerzel__, I dunno. when I made attack an equal priority with defense in my bot it was worth 100-150 Elo on the TCP server 2010-11-01T14:10:33 well you have to have a damn good eval then 2010-11-01T14:10:59 no actually it sucks hardcore 2010-11-01T14:11:28 It does prioritize planets by growthrate and distance 2010-11-01T14:11:30 JamesMG: I thought you had me for a minute 2010-11-01T14:11:40 didn't notice we were playing 2010-11-01T14:11:41 http://72.44.46.68/canvas?game_id=809502 2010-11-01T14:11:44 this is a fun one though 2010-11-01T14:11:46 idk who knottabot is but thanks for testing out my defense fix thoroughly lol 2010-11-01T14:11:56 it's close again 2010-11-01T14:12:03 nah, i've got you 2010-11-01T14:12:18 nice, 6 is the best i've seen i think 2010-11-01T14:12:43 which of my bots are you playing? 2010-11-01T14:12:48 .deb 2010-11-01T14:12:50 dev* 2010-11-01T14:12:51 ah 2010-11-01T14:12:57 i'm zBot atm 2010-11-01T14:12:59 yeah it's not lookin good for me at the moment 2010-11-01T14:13:10 Top 10 players: bocsimacko(4366), dmj111(4273), shangas(4204), Hazard(4184), FlameN(4150), joakim_mjardner(4128), Accoun(4113), felixcoto(4101), george(4098), jambachili(4082) 2010-11-01T14:13:18 here we go again 2010-11-01T14:13:24 got me now 2010-11-01T14:14:00 my start was shocking 2010-11-01T14:14:07 bbl going home lecture over :P 2010-11-01T14:14:22 i have 3 exams left for my undergrad :P 2010-11-01T14:14:36 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-11-01T14:15:49 that was kinda a strange start there 2010-11-01T14:16:16 it's alright, i'm not expecting it to be great there yet 2010-11-01T14:16:22 probably woulda won again if it hadn't done that, yours seems better mid-conflict than mine 2010-11-01T14:16:34 that's all i want it to be good at now :) 2010-11-01T14:18:02 *** Tobu has quit IRC (Ping timeout: 250 seconds) 2010-11-01T14:24:30 *** Tobu has joined #aichallenge 2010-11-01T14:27:24 http://72.44.46.68/getplayer?player=zBot 2010-11-01T14:27:28 doing quite well now it's not timing out 2010-11-01T14:28:31 *** Itkovian has joined #aichallenge 2010-11-01T14:28:31 *** fawek has joined #aichallenge 2010-11-01T14:28:57 i say that and now it loses 2010-11-01T15:01:05 *** Gudradain has quit IRC (Quit: Page closed) 2010-11-01T15:03:23 *** Itkovian has quit IRC (Quit: Itkovian) 2010-11-01T15:11:01 Bummer, crashed out my last 3 games on the official server 2010-11-01T15:11:25 *** boegel has joined #aichallenge 2010-11-01T15:13:10 yay moved my gamestate successfully to a global instance pointer function 2010-11-01T15:13:11 Top 10 players: bocsimacko(4364), dmj111(4263), shangas(4232), Hazard(4189), FlameN(4141), joakim_mjardner(4117), george(4103), Accoun(4097), felixcoto(4087), jambachili(4079) 2010-11-01T15:13:21 my c++ knowledge is almost back to what it was lol 2010-11-01T15:13:34 *** Bobng has quit IRC (Ping timeout: 255 seconds) 2010-11-01T15:13:57 I think your bot just beat mine demerzel__ 2010-11-01T15:14:12 yup 2010-11-01T15:14:20 sweet 2010-11-01T15:14:41 i got no dist and no concept of leaving myself open and use a 1-ply pure score method 2010-11-01T15:14:46 digest from that what you can ;) 2010-11-01T15:15:56 I have a bot like that too, just not the one you just beat 2010-11-01T15:16:19 although it is less developed than yours for sure 2010-11-01T15:17:00 mine has only been working about a day and a half 2010-11-01T15:17:05 *** Bobng has joined #aichallenge 2010-11-01T15:17:11 and still has some bugs but i think i worked out defense finally 2010-11-01T15:18:31 it has almost zero of the actual logic i want ti to use but been working out all the state bugs 2010-11-01T15:20:04 I'm starting to think that if I port my bot to something not C#, and combine the lessons learned from all 3 approaches I've tried 2010-11-01T15:20:08 I'd have something decent 2010-11-01T15:20:14 i'm not sure why you let me keep 20 2010-11-01T15:20:21 anybody want to do me a favour? 2010-11-01T15:20:32 yea a fresh recode was a good thing in my case 2010-11-01T15:20:35 link me to that cool visualizer with the planet IDs again demerzel__? 2010-11-01T15:20:43 and what do you need amstan? 2010-11-01T15:20:46 http://spz.nofate.com/planetwars/visualizer/?game_id=810108&server=tcp 2010-11-01T15:20:51 i need all of these concatenated: http://ai-contest.com/forum/viewtopic.php?f=4&t=881 2010-11-01T15:20:53 one per line 2010-11-01T15:21:26 lol what 2010-11-01T15:21:27 though 2010-11-01T15:21:32 i guess i could do them myself.. 2010-11-01T15:21:36 that's like 30 seconds 2010-11-01T15:22:02 JamesMG: you should have stomped me out of 20 - 6 was farther than 19 2010-11-01T15:22:09 maybe I'll try golang to support lavalamp_ 2010-11-01T15:22:33 *** Naktibalda has joined #aichallenge 2010-11-01T15:23:03 oh I see it now, yeah 2010-11-01T15:23:20 JamesMG, lol :) 2010-11-01T15:23:24 why didn't you if you don't mind me asking 2010-11-01T15:23:31 you went for 0 instead 2010-11-01T15:23:33 because I just coded a bug in 2010-11-01T15:24:00 oh nice 2010-11-01T15:24:15 notice it sent way more ships to 21 than it needed to also? 2010-11-01T15:24:49 ah no i know why 2010-11-01T15:24:54 *** cfaftw has joined #aichallenge 2010-11-01T15:25:06 you launched before i actually took it - do you consider non-neutral future enemies? 2010-11-01T15:25:12 by the time i took it you couldn't have taken it 2010-11-01T15:25:32 i guess that's a defense failure haha 2010-11-01T15:25:53 couldn't defend it but didn't consider it as a target until i took it 2010-11-01T15:26:08 I was tweaking my defense code just now to try and fix another issue, but my change was obviously ill-considered ;) 2010-11-01T15:26:57 i am pretty proud of my defense in that game - not bad considering it was very bugged 5 games ago 2010-11-01T15:27:21 wow, the more I look at it, the more I realize how much I just broke 2010-11-01T15:28:02 on turn 19 for some reason it's treating planet 0 as if it's mine 2010-11-01T15:28:29 oh weird 2010-11-01T15:28:37 i was wondering why you went there 2010-11-01T15:28:59 it's treating it like a frontline planet, so it's shuffling ships there... 2010-11-01T15:30:19 My first bot had a bug where if it didn't know what to do it routed everything to planet 0 2010-11-01T15:30:27 haha 2010-11-01T15:30:43 fortunately the only time that happened was when the enemy no longer had any planets 2010-11-01T15:30:50 ah well that's ok then 2010-11-01T15:30:56 i actually coded that into my dist code 2010-11-01T15:30:57 that's better than crashing when there are no enemy planets 2010-11-01T15:31:05 yup that's why i did it 2010-11-01T15:31:14 yes but it was funny when it was winning to see it send everything to the middle ^^ 2010-11-01T15:31:25 mine just stops issuing orders when it sees the enemy won't survive the fleets in the air 2010-11-01T15:33:12 welp i guess i gotta code the hard stuff now 2010-11-01T15:33:13 lol 2010-11-01T15:33:21 *** Itkovian has joined #aichallenge 2010-11-01T15:33:28 I really want to start over again 2010-11-01T15:33:32 do it 2010-11-01T15:33:50 i did it like 3x in C# although the 3rd time was more a reorg 2010-11-01T15:34:51 *** Itkovian has quit IRC (Client Quit) 2010-11-01T15:36:48 *** Blkt has quit IRC (Read error: Connection reset by peer) 2010-11-01T15:37:31 if you recode the whole thing once or twice it's way easier to abstract parts of it 2010-11-01T15:37:48 i've got state, supply, defense all as their own abstracted instances 2010-11-01T15:38:40 haha oops i guess i'm not done debugging 2010-11-01T15:38:57 I finally have a working order queue and state with future state and ship counts at turns as far as there is relevant data to 2010-11-01T15:39:13 I need to tie those together though 2010-11-01T15:39:30 my current dev code is mostly a vehicle for the order queue to be tested on 2010-11-01T15:41:01 I don't know if I can re-learn c++ quick enough for the comp though 2010-11-01T15:41:10 *** Olathe has quit IRC (Ping timeout: 265 seconds) 2010-11-01T15:41:22 *** Olathe has joined #aichallenge 2010-11-01T15:43:11 go is easy to learn but I don't think it's go-from-nothing-to-writing-a-top-10-bot-in-3-weeks easy... ;) 2010-11-01T15:43:24 :P 2010-11-01T15:43:42 I'm pretty sure I won't be writing a top 10 bot in any language 2010-11-01T15:43:43 i think it depends how good you are at working with new languages 2010-11-01T15:43:57 yea what JamesMG says is more likely 2010-11-01T15:43:59 :P 2010-11-01T15:44:16 maybe english if they let me pause the game every turn 2010-11-01T15:44:26 wish i knew of the tournament earlier 2010-11-01T15:44:43 preparation is overrated 2010-11-01T15:44:50 as is planning 2010-11-01T15:45:08 i started my bot mid-last month 2010-11-01T15:45:11 just gotta codecodecode 2010-11-01T15:45:38 demerzel__: thats what i I tell my boss 2010-11-01T15:45:45 which would have been nicer if i started in september, before my term started xD 2010-11-01T15:46:22 honestly I get more work done if I don't plan 2010-11-01T15:46:32 it just takes more code 2010-11-01T15:48:02 i plan in my head 2010-11-01T15:48:04 as i write 2010-11-01T15:48:11 *** hellman has quit IRC (Remote host closed the connection) 2010-11-01T15:48:14 my code is my plan 2010-11-01T15:49:39 imo, plans are only really really necessary when you have a team of people who work on completely different wavelengths and you need to cook something up that works together 2010-11-01T15:50:01 i.e. all those funky modelling techniques 2010-11-01T15:52:23 i think it's more assuming you know the best way to implement it now 2010-11-01T15:52:37 which assumes you have perfect understanding of the language and the problem 2010-11-01T15:53:44 *** davidjliu has quit IRC (Ping timeout: 265 seconds) 2010-11-01T15:59:46 man reading some of these ai progs in c from the 90s is crazy 2010-11-01T15:59:57 :P 2010-11-01T16:00:05 on my system, a pentium 2, i can pull 175k nodes per sec using this optimization 2010-11-01T16:00:11 what happened to devs lol 2010-11-01T16:00:38 i didn't quite understand that 2010-11-01T16:01:01 they have super heavy optimizations and are pulling more nodes per second on their pentium 2s than modern ones 2010-11-01T16:01:10 ah ^^ yes 2010-11-01T16:01:12 got lazy in optimization 2010-11-01T16:01:31 that is true 2010-11-01T16:01:46 well they have to invest on reasons to sell new computers 2010-11-01T16:01:56 a computer 15 years ago could run a word processor 2010-11-01T16:02:00 *** RainCT has quit IRC (Ping timeout: 255 seconds) 2010-11-01T16:02:07 but probably not microsoft word 2010 2010-11-01T16:02:08 lol 2010-11-01T16:02:20 well that's usage of the platform that's fine 2010-11-01T16:02:36 i'm talking the devs not givign a crap about the fact that they don't HAVE to use a wrapper for everything 2010-11-01T16:02:43 in fact sometimes it's even easier not to 2010-11-01T16:02:46 ah yes 2010-11-01T16:02:49 KISS you mean 2010-11-01T16:03:06 maybe it's just because development is more mainstream now 2010-11-01T16:03:16 well, tbh i prefer the KISS concept 2010-11-01T16:03:30 the only time i abstract is for good reason, i.e. if i want something to plug into the same interface 2010-11-01T16:04:05 oh i'm talking about the builtin wrappers 2010-11-01T16:04:10 ie arrays are bad use vectors 2010-11-01T16:04:12 builtin? 2010-11-01T16:04:16 i abstract all the time ;) 2010-11-01T16:04:54 ah vectors in C++ are special case, C++ easily removes most of the cruft 2010-11-01T16:05:03 so they aren't much different to arrays in terms of speed 2010-11-01T16:05:10 no but that's an exception to the rule 2010-11-01T16:05:25 and the fact that devs push their usage to the point of declaring arrays obsolete is telling 2010-11-01T16:05:55 im confused which end of the stick we're dicussing about 2010-11-01T16:05:57 but lol :D 2010-11-01T16:06:01 I like the ELO system, makes my bot look awesome when it's only games are losing to mega1's bot 2010-11-01T16:06:02 hahah don't worry about i 2010-11-01T16:06:07 nice 2010-11-01T16:06:13 haha 2010-11-01T16:06:19 540 ELO for losing 3 games, nice 2010-11-01T16:06:32 haha 2010-11-01T16:06:53 i'm debating whether to start on the tree system or set up the 'how much i should hold back to not give away my planet' state 2010-11-01T16:08:45 ok now to watch my bot crash and burn 2010-11-01T16:09:14 w00t 2010-11-01T16:09:16 hey raed since youre a c opt freak what's the disadvantage of holdign a static instance like this (if any)? http://pastebin.com/d1ivwYgQ 2010-11-01T16:09:23 Channel Zero op Canvas 2010-11-01T16:09:47 nvm, wrong channel 2010-11-01T16:10:01 hmm wait a second it doesn't compile 2010-11-01T16:10:03 :D 2010-11-01T16:10:52 demerzel__: i don't see it often 2010-11-01T16:11:11 but if i remember correctly, its essentially like a global variable in terms of performance 2010-11-01T16:11:14 what do people do? singletons? that seems prone to problems 2010-11-01T16:11:29 i tried to just set up a static instance but it didnt' work out lol 2010-11-01T16:11:31 so you lose the perf of locally defined variables that are on the stack 2010-11-01T16:11:35 yeah, demerzel__, that's a wordy way of making a global 2010-11-01T16:11:59 *** bhasker has joined #aichallenge 2010-11-01T16:12:01 i would declare it gameState State; 2010-11-01T16:12:09 i couldn't init it tho 2010-11-01T16:12:12 it was having a lot of issue 2010-11-01T16:12:17 ok 2010-11-01T16:12:34 my C++ isn't as good as my C 2010-11-01T16:12:37 but it was a class with static vars before 2010-11-01T16:12:44 so might just be the way i was using it 2010-11-01T16:13:07 i'm not confident about whats going on behind the scenes 2010-11-01T16:13:10 static vars are annoying because you have to initialize them correctly 2010-11-01T16:13:11 Top 10 players: bocsimacko(4401), dmj111(4268), shangas(4233), Hazard(4200), FlameN(4149), joakim_mjardner(4126), george(4091), Accoun(4091), felixcoto(4081), Slin-.-(4081) 2010-11-01T16:13:54 yea static methods were working great 2010-11-01T16:14:12 i just realized i'd either have to copy the whole thing and make a new class which was stupid or turn it into an instance 2010-11-01T16:14:21 *** bhasker_ has quit IRC (Ping timeout: 252 seconds) 2010-11-01T16:15:06 *** Tobu has quit IRC (Ping timeout: 240 seconds) 2010-11-01T16:24:00 *** cfaftw has quit IRC (Quit: Leaving) 2010-11-01T16:26:00 *** Accoun has quit IRC () 2010-11-01T16:36:23 *** boegel has quit IRC (Quit: Leaving) 2010-11-01T16:45:53 *** Accoun has joined #aichallenge 2010-11-01T16:54:28 awesome, code no longer crashes :D 2010-11-01T16:54:45 so all the source files have to be in the same folder when we submit? 2010-11-01T16:54:49 doesn't do anything useful but hehe 2010-11-01T16:54:56 think so antimatroid 2010-11-01T16:55:00 thats what i did 2010-11-01T16:58:47 nice 2010-11-01T16:59:12 would you say a class variable instantiated in main() is a global? does it ever go out of scope? 2010-11-01T16:59:42 ie i create gameState in main and then call it 2010-11-01T16:59:43 demerzel__ it's not a global but it won't go out of scope. It is on the stack. 2010-11-01T16:59:58 so that's a better impl than just declaring it global then? 2010-11-01T17:00:04 Well, unless you new it. 2010-11-01T17:00:09 No, make it a global. 2010-11-01T17:00:13 i have never used new in c++ i'm not even sure why i would 2010-11-01T17:00:19 lol 2010-11-01T17:00:19 *** Palmik has quit IRC (Remote host closed the connection) 2010-11-01T17:00:44 lavalamp_: why global over main? if the class is essentially the basis of the entire prog 2010-11-01T17:00:53 wouldn't it being on teh stack be a good thing? 2010-11-01T17:00:58 for one, stack space is not unlimited 2010-11-01T17:01:10 true but i think i'm eating up something like 65k atm 2010-11-01T17:01:32 almost all pod arrays 2010-11-01T17:01:36 second, if you ever overrun it you don't risk corrupting everything else on the stack (so your bug might be easier to find) 2010-11-01T17:01:47 hmm 2010-11-01T17:02:04 i'm condensing my code now and just trying to figure out best practice 2010-11-01T17:02:12 i've got a couple of very similar structs etc 2010-11-01T17:02:14 hm well I guess an underrun would be more dangerous that way, but still 2010-11-01T17:02:33 how would i overrun a class on the stack by calling its functions tho 2010-11-01T17:02:55 anyway, if you're passing pointers to it all over the place it would be more efficient to have a global 2010-11-01T17:03:00 i mean any more than if it was global 2010-11-01T17:03:18 yea i'm debating between having a stub object that is handled by static methods w/ pointers or instances of the class 2010-11-01T17:03:27 it doesn't raise the risk of a bug, it just changes the consequence of having one. 2010-11-01T17:04:05 my gut says that having a simple struct to pass to global static methods is the best way 2010-11-01T17:04:13 my personal opinion is that designs with stub objects usually have a better alternative 2010-11-01T17:04:15 but i'm wondering if i'm right since i also like c strings lol 2010-11-01T17:04:47 You're basically describing how to do things in C 2010-11-01T17:04:59 yea i'm a c programmer masquerading as a c++ dev 2010-11-01T17:05:06 *** RainCT has joined #aichallenge 2010-11-01T17:05:14 only reason i like c++ is the org and the casts - plus the streams are nice for quick & dirty 2010-11-01T17:05:39 so what is the "c++ way" then 2010-11-01T17:05:46 C++ way is classes :) 2010-11-01T17:05:49 take gamestate as an example 2010-11-01T17:05:57 i use classes but i mostly use them as containers for static methods 2010-11-01T17:06:05 XP 2010-11-01T17:06:18 or structs with arrays 2010-11-01T17:06:30 *** nullkuhl_ is now known as nullkuhl 2010-11-01T17:06:31 well I guess if you're just doing it for the organization 2010-11-01T17:06:32 *** RainCT has quit IRC (Remote host closed the connection) 2010-11-01T17:06:51 but if that's the case namespaces might be more appropriate 2010-11-01T17:06:53 well what would you do if you wanted to recreate the gameState class for tree searches 2010-11-01T17:07:06 *** RainCT has joined #aichallenge 2010-11-01T17:07:31 lavalamp_:true about namespaces 2010-11-01T17:08:08 i would argue that i am mostly oop though lol 2010-11-01T17:09:09 demerzel__, that could be a long and interesting conversation which I don't have time for atm :/ 2010-11-01T17:09:28 awww 2010-11-01T17:09:40 the how to make the gameState qusetion, that is ^^ 2010-11-01T17:09:47 it's ok i'd probably just ignore you and do it my way anyway but i was interested in your viewpoint 2010-11-01T17:10:19 i am a c dev in a .NET world lol 2010-11-01T17:10:36 which is also my job incidentally lol 2010-11-01T17:10:47 short version: if you're doing OOP in C++ without using classes something is very wrong lol 2010-11-01T17:11:54 classes ^ as they're intended 2010-11-01T17:13:11 Top 10 players: bocsimacko(4401), dmj111(4272), shangas(4199), Hazard(4187), FlameN(4156), joakim_mjardner(4124), george(4089), Accoun(4087), felixcoto(4084), Slin-.-(4068) 2010-11-01T17:13:51 And the issue of how to deal with hypothetical game states is an interesting design problem, I think... 2010-11-01T17:14:15 I just store them :( 2010-11-01T17:14:44 well in C# i used a static class with invididual planets 2010-11-01T17:15:05 then while checking out hypos i would just copy the planet classes and hack them up without touching the original 2010-11-01T17:15:41 you could do the same in C++ I think 2010-11-01T17:15:59 well it was optimized to do it that way tho because i was only looking at 1 source -> 1 dest 2010-11-01T17:16:02 and only 1 ply 2010-11-01T17:16:18 when you get into 2+ ply and multi source then you have to look at all impacts 2010-11-01T17:16:37 *** nigg_ has joined #aichallenge 2010-11-01T17:16:44 so i am kinda torn between making a data object or copying the whole gamestate which has all these functions attached 2010-11-01T17:16:48 ah yeah 2010-11-01T17:17:14 I copy the entire map, including fleets, unfortunately 2010-11-01T17:17:15 given my design I would copy my entire history object if it didn't already handle hypothetical states 2010-11-01T17:17:40 and then step the map object until I'm where I want to be 2010-11-01T17:17:43 *** TravisD has joined #aichallenge 2010-11-01T17:17:45 ok im happy my code doesn't actually crash... actually amazed 2010-11-01T17:17:46 *** Bobng has quit IRC (Ping timeout: 255 seconds) 2010-11-01T17:17:51 demerzel__ C++ does not copy functions, they exist in one place in memory 2010-11-01T17:17:56 i only needed to fix about 5 obvious seg faults 2010-11-01T17:18:16 lavalamp_:thanks that helps actually - i know in C# each instance creates more 2010-11-01T17:18:22 i.e. 30 seconds to find discover and work out why it crashes 2010-11-01T17:18:41 *** delt0r___ has joined #aichallenge 2010-11-01T17:18:42 no real mysterious bugs 2010-11-01T17:18:54 yeah every instance of a class shares the same functions 2010-11-01T17:19:05 ok need to sleep now 2010-11-01T17:19:05 ugh refactoring c++ is such a pain i think i'm gonna just create a new project and c/p it in 2010-11-01T17:19:19 lol 2010-11-01T17:19:39 cya raed 2010-11-01T17:20:21 *** delt0r has quit IRC (Ping timeout: 265 seconds) 2010-11-01T17:25:52 yeeeeeehaw my new bot is kicking ass 2010-11-01T17:27:16 15.2? 2010-11-01T17:27:21 yes 2010-11-01T17:27:30 let me watch 2010-11-01T17:28:02 nice you are beating piter 2010-11-01T17:28:09 and shampoo 2010-11-01T17:28:15 mega1 is 6 of the top 10 right now 2010-11-01T17:28:17 on tcp 2010-11-01T17:28:21 sorry about that 2010-11-01T17:28:23 not you though ;) 2010-11-01T17:28:31 well not the last 2 games 2010-11-01T17:28:50 I do have a serious issue that sometimes makes me lose even from noobs 2010-11-01T17:28:51 :D 2010-11-01T17:28:55 not complaining mega1, just observing :) 2010-11-01T17:28:58 welcome to the club error 2010-11-01T17:29:10 i keep attacking a planet which i can't cap 2010-11-01T17:29:19 and that makes me fail in the long run 2010-11-01T17:29:19 i think my bot beat mega1's some bot today a few times:d 2010-11-01T17:29:31 I'm trying to test, but it's almost impossible. 2010-11-01T17:29:35 http://72.44.46.68/canvas?game_id=811004 2010-11-01T17:29:38 how's that? 2010-11-01T17:29:51 you should have won that one 2010-11-01T17:29:56 not sure why your bot never attacked 2010-11-01T17:29:59 or acquire more neutrals 2010-11-01T17:30:14 nice bhasker 2010-11-01T17:30:32 I have reason to beleive that b0.41 is better than b0.36, although it seems about 120 points weaker. 2010-11-01T17:30:52 thanks Error...lucky wins! 2010-11-01T17:31:34 timeouts and especially randomness of openings, map selection has a great effect on the outcome 2010-11-01T17:31:41 though i am beating shampooo more regularly now 2010-11-01T17:31:46 i am happy about that 2010-11-01T17:31:50 indd 2010-11-01T17:31:56 testingfrompiter still kicks my butt:-\ 2010-11-01T17:32:05 well his elo is quite good 2010-11-01T17:32:58 btw bhasker since your proposal on a differen value metric for neutrals it has become better :) 2010-11-01T17:33:07 *** Sarin has quit IRC (Remote host closed the connection) 2010-11-01T17:33:08 s/on/of 2010-11-01T17:33:14 i believe it was you 2010-11-01T17:33:16 which metric is that 2010-11-01T17:33:24 i have been trying multiple versions 2010-11-01T17:33:31 and i keep tweaking my metric:( 2010-11-01T17:33:39 *** fmeyer has joined #aichallenge 2010-11-01T17:34:12 std::numeric_limits::max() / ((numships/growthrate) + log(2^distance)) iirc 2010-11-01T17:34:30 which others have you got :P 2010-11-01T17:34:51 ah k 2010-11-01T17:34:59 heh i use a breakeven point metric as well 2010-11-01T17:35:02 simply 2010-11-01T17:35:03 i thought it might be a good idea to also factor in the neighbouring planets 2010-11-01T17:35:16 oh i do that as a second step 2010-11-01T17:35:26 the first is just a way to order neutrals in terms of desirability 2010-11-01T17:35:37 ugh lost again to you >_> 2010-11-01T17:35:47 and then when i actually decide to process the neutral i check nearby Planets etc 2010-11-01T17:35:53 and decide whether to attack or not 2010-11-01T17:35:56 and how much to attack with 2010-11-01T17:36:14 the value function is just to make sure I try for higher value neutrals before trying for crappy ones 2010-11-01T17:36:23 i see 2010-11-01T17:36:28 you won 2010-11-01T17:36:32 the latest one 2010-11-01T17:36:44 i don't factor in the map layout at all atm 2010-11-01T17:36:57 and then i exclude bad neutrals if there are neutrals that i can acquire in a turn or two 2010-11-01T17:36:58 etc 2010-11-01T17:37:29 because otherwise you would end up burning 45 ships on a 1 growth when after two turns you could have probably got a 5 growth with 55 ships 2010-11-01T17:37:33 at least my bot used to do that 2010-11-01T17:37:36 i pop the best neutral of the priority queue 2010-11-01T17:37:42 and wait until i can get it 2010-11-01T17:37:45 ah 2010-11-01T17:37:47 i don't do that 2010-11-01T17:37:52 my bot is aggresive 2010-11-01T17:38:13 *** sigh has joined #aichallenge 2010-11-01T17:38:25 ah yes mine aswell 2010-11-01T17:38:31 i attack before "saving up" 2010-11-01T17:38:48 its just that my routing algorithm will save up 2010-11-01T17:39:08 bhasker: where are you at official server? 2010-11-01T17:40:12 #45 last i checked 2010-11-01T17:40:13 i think 2010-11-01T17:40:20 46 2010-11-01T17:40:23 nice 2010-11-01T17:40:30 http://ai-contest.com/profile.php?user_id=5822 2010-11-01T17:40:30 *** Itkovian has joined #aichallenge 2010-11-01T17:40:34 i hope to be in the top 100 now too 2010-11-01T17:40:34 thanks 2010-11-01T17:41:25 i'm just below you now at tcp XD 2010-11-01T17:41:44 *** Azrathud has joined #aichallenge 2010-11-01T17:42:06 my routing algorithm has improved quite a lot 2010-11-01T17:42:55 :) 2010-11-01T17:43:07 :D 2010-11-01T17:45:09 *** bhasker_ has joined #aichallenge 2010-11-01T17:45:46 http://72.44.46.68/canvas?game_id=811566 bhasker why no attack? 2010-11-01T17:48:24 *** bhasker has quit IRC (Ping timeout: 252 seconds) 2010-11-01T17:48:34 *** McLeopold has joined #aichallenge 2010-11-01T17:50:10 *** bhasker has joined #aichallenge 2010-11-01T17:53:06 *** bhasker_ has quit IRC (Ping timeout: 240 seconds) 2010-11-01T17:53:07 *** delt0r___ has quit IRC (Quit: Leaving) 2010-11-01T17:57:13 my bot is conservative 2010-11-01T17:57:22 it won't launch if it can't do it without losing the planet 2010-11-01T18:01:32 bhasker: I've got the same problems 2010-11-01T18:03:00 yup 2010-11-01T18:03:05 i tried to make it a bit more aggressive 2010-11-01T18:03:10 but its risky 2010-11-01T18:03:14 since then you are trading planets 2010-11-01T18:03:21 and it gets complicated 2010-11-01T18:03:34 mmm 2010-11-01T18:03:41 i'm just gonna attack when i can 2010-11-01T18:03:49 perhaps i should build in some conservativity 2010-11-01T18:06:31 *** roshan has joined #aichallenge 2010-11-01T18:09:50 bhasker: put your bot back in please? ^_^ 2010-11-01T18:09:55 in sometime 2010-11-01T18:11:40 *** JamesMG has quit IRC (Quit: Leaving.) 2010-11-01T18:13:12 Top 10 players: bocsimacko(4409), dmj111(4288), shangas(4205), Hazard(4187), FlameN(4167), joakim_mjardner(4124), Accoun(4101), felixcoto(4085), george(4084), protocolocon(4081) 2010-11-01T18:13:33 whoah 2010-11-01T18:13:37 my bot is on a rampage 2010-11-01T18:13:40 holy crap ^_^ 2010-11-01T18:14:00 310 elo 2010-11-01T18:14:07 \o/ |o| /o/ 2010-11-01T18:18:30 *** lavalamp_ has quit IRC (Quit: Page closed) 2010-11-01T18:20:01 *** Migi32 has joined #aichallenge 2010-11-01T18:26:01 great job dmj111 2010-11-01T18:26:48 *** Mathnerd314 has joined #aichallenge 2010-11-01T18:27:19 hah you beat me 323 2010-11-01T18:37:28 *** nigg_ has quit IRC (Quit: Leaving) 2010-11-01T18:37:54 sorry to bring this offtopic, but does anyone here play StarCraft? 2010-11-01T18:39:27 bhasker: :D 2010-11-01T18:39:29 *** RainCT has quit IRC (Quit: good night :)) 2010-11-01T18:39:58 *** Appleman1234 has joined #aichallenge 2010-11-01T18:40:03 ah not anymore bhasker 2010-11-01T18:40:10 when you are not there i perform better :P 2010-11-01T18:42:05 bhasker: how significant is two hours of play on tcp 2010-11-01T18:42:13 i hope to be in the top 100 now 2010-11-01T18:42:20 i think i will... 2010-11-01T18:42:59 Error323, what do you mean? You're #25 on TCP 2010-11-01T18:44:14 well I don't know what that means 2010-11-01T18:45:46 Ah, like that. Well I don't know what it means either, but yea I guess it'd be in top 100. 2010-11-01T18:47:16 2 hrs is good enough 2010-11-01T18:47:23 a 300+ elo on tcp 2010-11-01T18:47:27 would normally end up in top 100 2010-11-01T18:48:01 though testingFromPiter is really screwing up my elo 2010-11-01T18:50:42 *** fmeyer has quit IRC (Ping timeout: 240 seconds) 2010-11-01T18:53:07 cool 2010-11-01T18:54:54 *** Itkovian has quit IRC (Quit: Itkovian) 2010-11-01T18:55:19 its soooooooo difficult to accuratly predict with how much to attack 2010-11-01T18:55:20 >_< 2010-11-01T18:57:41 Error323: my official bot is running on tcp and is about rank 58 on both. 2010-11-01T18:58:44 really 2010-11-01T18:58:45 wut 2010-11-01T18:58:59 i'm @ 23 on tcp now 2010-11-01T18:59:02 o_O 2010-11-01T18:59:06 removed two lines 2010-11-01T18:59:10 lol 2010-11-01T18:59:12 :D 2010-11-01T19:00:02 yea, my bot also does better sometimes if I remove this line: 2010-11-01T19:00:13 defendIncomingAttacks(); 2010-11-01T19:00:23 Error323: we don't seem to have any matches 2010-11-01T19:00:38 not anymore 2010-11-01T19:00:41 we had some in the past 2010-11-01T19:00:55 oh wait: http://72.44.46.68/canvas?game_id=811510 2010-11-01T19:00:56 or not atm i should say 2010-11-01T19:01:54 Just wait until I finally fix my current one. :) 2010-11-01T19:02:09 good god i am basically c/p winsock2 and it's somehow breaking line by line 2010-11-01T19:02:15 *** davidjliu has joined #aichallenge 2010-11-01T19:02:16 *** TravisD has left #aichallenge 2010-11-01T19:02:32 *** TravisD has joined #aichallenge 2010-11-01T19:02:44 for example i have pragma "winsock32.lib" in my first proj, 2nd refused to work until i corrected it to ws_32 2010-11-01T19:02:50 i have no idea how the first one is even working 2010-11-01T19:03:07 pure willpower i guess but damn this is stupid time 2010-11-01T19:04:05 You WIN against bocsimacko-b0.42 2010-11-01T19:04:06 whoot :P 2010-11-01T19:04:11 nice :P 2010-11-01T19:05:54 now trying out (growthrate^2 / numships) / distance aswell with E323-15.2-1-gec91a12 2010-11-01T19:06:54 no configurable constants? 2010-11-01T19:07:01 mmm? 2010-11-01T19:07:26 why the 2 and not, say, 1.9? 2010-11-01T19:07:43 oh 2010-11-01T19:07:48 just a guess or mathematically deduced, 2010-11-01T19:07:50 ? 2010-11-01T19:08:13 stetson harrison method 2010-11-01T19:08:53 what is ^1.9 some base normalization? 2010-11-01T19:08:59 seen it before 2010-11-01T19:09:56 no, it was also a random number :P 2010-11-01T19:10:32 Error323: i am going to fix the Drake equation in somehow 2010-11-01T19:10:42 lmao 2010-11-01T19:10:45 heh that would be awesome :P 2010-11-01T19:10:47 xD 2010-11-01T19:11:21 that's a really odd formula 2010-11-01T19:12:19 the units of that formula are ships/turns^3 btw 2010-11-01T19:12:21 Error323: can you please explain how (growthrate^2 / numships) / distance is any more useful than (growthrate * (turnsremaining - distance)) 2010-11-01T19:12:37 i really don't understand why all these log(n) and exponents 2010-11-01T19:12:45 is there some kind of attempt at normalization or? 2010-11-01T19:12:55 well the growthrate is more important in the long run 2010-11-01T19:13:09 ahh so you are purposefully weighting it 2010-11-01T19:13:12 Top 10 players: bocsimacko(4367), dmj111(4286), Hazard(4168), FlameN(4162), shangas(4159), joakim_mjardner(4123), george(4089), jambachili(4084), Accoun(4081), felixcoto(4079) 2010-11-01T19:13:14 indeed 2010-11-01T19:13:20 but it's a tiny tiny number 2010-11-01T19:13:29 or what is numships ? the number on the planet? 2010-11-01T19:13:40 of the neutral planet, yes 2010-11-01T19:13:41 i have long forgotten the starter pack classes sorry 2010-11-01T19:14:03 ah so 5 / 5 = 1 / 3 away is .3333 2010-11-01T19:14:14 with that scoring function, it would score a planet with 2 ships twice as high as one with 1 ship. Strange. 2010-11-01T19:14:14 err 5/3 lol 2010-11-01T19:14:23 yea i don't get the numships thing 2010-11-01T19:14:25 But I guess it's enough to beat bocsimacko 2010-11-01T19:14:49 well just once :P 2010-11-01T19:14:52 once 2010-11-01T19:14:58 out of 99 prolly XD 2010-11-01T19:15:00 dunno 2010-11-01T19:15:18 I couldn't beat him given a million tries 2010-11-01T19:15:25 yet :p 2010-11-01T19:15:29 I've always geared my formulas to look for the lowest number. 2010-11-01T19:15:31 my shitbot actually had him cornered once 2010-11-01T19:15:41 of course it jacked it up but i was like holy shit am i going to win 2010-11-01T19:16:01 ah demerzel__ you also use the JBotManager? 2010-11-01T19:16:08 no 2010-11-01T19:16:08 http://72.44.46.68/canvas?game_id=808494 2010-11-01T19:16:18 i write tcp code directly into the bot 2010-11-01T19:16:40 I can't figure out a way to let JBotManager continuously play so I don't have to press "start" every time I want to fight against a new opponent on TCP 2010-11-01T19:16:48 oh, nice 2010-11-01T19:16:54 also a visualiser? 2010-11-01T19:17:03 nah i just use the js one 2010-11-01T19:17:27 i have rarely had a bot so stable i want to put it on loop though 2010-11-01T19:17:58 Migi32: there is a number you can set for match count 2010-11-01T19:18:02 haha at the end there it all turns red, you can almost hear *poof* 2010-11-01T19:18:11 woooooooooow 2010-11-01T19:18:20 yea haha that match was funny 2010-11-01T19:18:25 reps 2010-11-01T19:18:33 i kept getting more incredulous and it was looking so strangely good 2010-11-01T19:18:45 then it screwed up ofc 2010-11-01T19:18:51 but holy crap 100 turns lol 2010-11-01T19:19:30 *** mceier has quit IRC (Quit: leaving) 2010-11-01T19:20:28 lol 2010-11-01T19:20:45 *** roshan has left #aichallenge 2010-11-01T19:20:49 mcleopold http://ai-contest.com/visualizer.php?game_id=6463251 2010-11-01T19:21:40 who is bocsimacko 2010-11-01T19:21:43 ? 2010-11-01T19:21:49 my *poof* is a lot less effective than bocsimacko's 2010-11-01T19:21:55 hehe 2010-11-01T19:22:00 superflit: mega1 2010-11-01T19:22:28 damn.. the bot seems good, nobody beat him 2010-11-01T19:22:57 he's the one currently ruling the league with an iron hoof 2010-11-01T19:23:18 you don't have dist leopold? 2010-11-01T19:23:25 i have gone back on forth on its necessity 2010-11-01T19:23:49 anyone is using a local serve to train? 2010-11-01T19:24:37 McLeopold, what happened to your bot? You used to be on the spot bocsimacko is now. 2010-11-01T19:24:37 ? dist? 2010-11-01T19:24:57 I haven't run my latest version on tcp in 2 weeks. 2010-11-01T19:24:58 distribution to frontline 2010-11-01T19:25:20 I'm re-writing the entire framework. 2010-11-01T19:25:28 nice 2010-11-01T19:25:35 everything is broken 2010-11-01T19:25:41 my bot on the site is literally over a month old and in a completely different language 2010-11-01T19:25:43 wow. Still in python? 2010-11-01T19:25:51 I'm only running my official version to make me feel like I'm still in it. 2010-11-01T19:26:08 *** Commandir has left #aichallenge 2010-11-01T19:26:12 Yeah, all python 2010-11-01T19:26:23 I want to do javascript, but I don't think I have time. 2010-11-01T19:26:28 o_O 2010-11-01T19:26:31 srsly? 2010-11-01T19:26:41 does that server even support that? 2010-11-01T19:26:47 Yeah, it's not a fast language, but it can be pretty cool. 2010-11-01T19:26:52 NodeJS, yes it does. 2010-11-01T19:27:02 IMO, javascript < java < everything else :P 2010-11-01T19:27:15 It's closer to lisp, which, btw, is winning. 2010-11-01T19:27:30 mega1 is lisp? 2010-11-01T19:27:43 superflit: yes 2010-11-01T19:27:45 http://www.amazon.com/JavaScript-Good-Parts-Douglas-Crockford/dp/0596517742 2010-11-01T19:27:48 it's also the only lisp in the top 100 2010-11-01T19:28:07 the language fits the problem.. 2010-11-01T19:28:11 top 200, for that matter 2010-11-01T19:28:16 I did not expect so goodly. 2010-11-01T19:28:22 yah, next lisp is 220 2010-11-01T19:28:25 is that how you pipe stdin/out in js? node.js? 2010-11-01T19:28:44 yeah 2010-11-01T19:28:55 i was wondering 2010-11-01T19:29:02 there are other javascript interpreters out there 2010-11-01T19:29:08 was going to do a js bot just to be esoteric originally but then got stuck at that little detail lol 2010-11-01T19:29:29 like ringojs 2010-11-01T19:29:52 i like js but the typing might get a little crazy 2010-11-01T19:29:57 IMO, javascript will be the language of the future. 2010-11-01T19:30:07 i do love js for the same reason i like c++ - you can program it in 5 different ways and they all work 2010-11-01T19:30:19 well, a lot of wrk is going into it, so maybe 2010-11-01T19:30:23 Take a look at libraries like jQuery or prototype. They are works of art. 2010-11-01T19:30:34 i don't really like jQuery but probably a bit biased 2010-11-01T19:30:45 tired of answering hey how do i - whoa whoa i only know jquery 2010-11-01T19:30:59 haha 2010-11-01T19:31:05 maybe you don't like jquery programmers? 2010-11-01T19:31:34 when i have to explain the dom and how to call it in javascript to a professional dev it's a little bizarre 2010-11-01T19:31:45 but i just put a $ and it appears?? 2010-11-01T19:31:56 :) 2010-11-01T19:32:41 usually $ = document.getElementById, but there are other shortcuts, and jquery does overload it 2010-11-01T19:32:48 yea i know 2010-11-01T19:32:57 i'm not overly fond of their dom-ignoring injection library 2010-11-01T19:33:09 I didn't know JavaScript was so complex 2010-11-01T19:33:10 okay, forget jquery, look at the prototype lib 2010-11-01T19:33:30 javascript is a crazy language 2010-11-01T19:33:38 you can make it do shit when you press a button in like 3 lines 2010-11-01T19:33:43 or you can make constructors, prototypes, oop 2010-11-01T19:33:50 Migi32: if you use it correctly, you need to know very little syntax, and you can start to mold it as you need. 2010-11-01T19:34:22 demerzel__: I haven't potty trained mine yet... 2010-11-01T19:34:45 mostly what annoys me about javascript is the really weak typing 2010-11-01T19:34:53 other than that, I like it 2010-11-01T19:35:01 McLeopold, I'm just reading up on JS a bit, but... Classes are functions? Can't wrap my head around that... 2010-11-01T19:35:09 well, that's kinda how languages are, strong typed or weak typed. 2010-11-01T19:35:26 I like weak-typed languages the best, which is why I like python. 2010-11-01T19:35:47 McLeopold: python is stronger than javascript... I don't mind python 2010-11-01T19:36:03 Migi32: that's an ugly part of javascript you just pretend doesn't exist. :P 2010-11-01T19:36:03 python doesn't automatically convert a string to an int or vice-versa 2010-11-01T19:36:37 I think it does... 2010-11-01T19:36:43 also, javascript is weak-typed but you do have to type the word "var" in front of a variable. While you're at it why not just type the damn "int"? 2010-11-01T19:36:50 "bla" + 0? 2010-11-01T19:37:11 McLeopold: TypeError: cannot concatenate 'str' and 'int' objects 2010-11-01T19:37:34 Migi32: another weak point of javascript. The var ensures you don't bleed into the global scope. 2010-11-01T19:37:35 I'm more concerned about: "1" + 2 2010-11-01T19:37:46 It could have been the other way around, but it isn't. 2010-11-01T19:38:01 which is a really big gotcha when doing stuff with javascript 2010-11-01T19:38:25 in JS, "a"+1+2 != "a"+(1+2) 2010-11-01T19:38:40 which can cause unexpected things 2010-11-01T19:38:44 *** TravisD has quit IRC (Quit: Leaving.) 2010-11-01T19:38:54 though in general I hate languages using + for string concat 2010-11-01T19:38:55 well, that's also the case for java 2010-11-01T19:39:10 *** TravisD has joined #aichallenge 2010-11-01T19:39:38 as in I hate hte fact they do that, but not hte language as a whole 2010-11-01T19:39:46 s/hte/the/g 2010-11-01T19:40:01 sigh, hurrah for c++ where you have to set up a std::stringstream for that! Or use a templated function. 2010-11-01T19:40:19 sigh: haven't seen a /g on irc before. awesome. 2010-11-01T19:40:27 :P 2010-11-01T19:40:30 very consistant 2010-11-01T19:41:21 i have used s/../../ in IM's at a Windows shop. they think I am weird. 2010-11-01T19:41:54 dmj111, well, it took me a long time before I knew what they meant. Can't exactly google that. 2010-11-01T19:42:30 http://www.google.com.au/search?q=s%2F..%2F..%2Fg 2010-11-01T19:42:33 works for mew 2010-11-01T19:42:37 *me 2010-11-01T19:42:58 oh no 2010-11-01T19:43:27 sorry, was looking at the wrong page, that doesn't do anything useful :D 2010-11-01T19:43:30 "Welcome to Syngenta Flowers" - second result 2010-11-01T19:44:14 odd... my rank has gone from 45 to 23 over the last few days, these rankings are really unstable 2010-11-01T19:44:25 * sigh changes topic, quickly 2010-11-01T19:45:00 :P 2010-11-01T19:45:18 yeah sigh, I've thought the same thing. Only dmj111 is a constant :) 2010-11-01T19:45:46 yeah, dmj111 just says in the top5 no matter what anyone else does 2010-11-01T19:46:51 *** rangzen has quit IRC (Quit: Quitte) 2010-11-01T19:47:14 on the official? 2010-11-01T19:47:22 I went from 99 to 58 in like a day 2010-11-01T19:47:22 yeah 2010-11-01T19:47:33 yup 2010-11-01T19:48:20 makes me wonder how consistent the final results will be 2010-11-01T19:49:29 "The final rankings will be determined using a computerized tournament designed by the contest organizer." 2010-11-01T19:49:51 i went from 42 to 200 something in a day 2010-11-01T19:50:27 Zannick: how old was the submission? 2010-11-01T19:50:34 I hope it's not tournament as in winner advances to the next round, loser leaves the contest 2010-11-01T19:50:51 *** Mathnerd314 has left #aichallenge 2010-11-01T19:50:53 well, i was only at 42 by awesome luck, being 1-1 a day after submission 2010-11-01T19:50:56 i submitted on saturday 2010-11-01T19:51:06 Migi32: my understanding was that it is the same format as now, except starting from a clean slate 2010-11-01T19:51:14 it is not 1-6 2010-11-01T19:51:38 sigh, how was it on the Tron contest? 2010-11-01T19:51:57 Zannick: yeah, that makes sense... however, my bot was over 3 weeks old and McLeopolds is months old 2010-11-01T19:52:11 your bots are probably good, though 2010-11-01T19:52:14 Migi32: I wasn't around for the tron contest, but I think it was the same 2010-11-01T19:52:19 mine didn't deserve top 50 2010-11-01T19:52:35 i was amused 2010-11-01T19:53:00 :) 2010-11-01T19:54:57 *** demerzel_ has joined #aichallenge 2010-11-01T19:55:24 *** demerzel__ has quit IRC (Ping timeout: 265 seconds) 2010-11-01T19:56:28 *** rebelxt has joined #aichallenge 2010-11-01T19:56:33 *** rebelxt has left #aichallenge 2010-11-01T19:56:41 re: js -- if you're serious about your oop just set up an object with an int get/set and use it as a type -- i'm sure at least one of the frameworks does that 2010-11-01T19:57:53 does albertz's cpp viewer do tcp or not i'm trying to find the code for it but not seeing it 2010-11-01T20:00:40 grr antimatroid's bot is very very slow!! 2010-11-01T20:00:52 takes the full 1 second per turn 2010-11-01T20:00:53 or more 2010-11-01T20:01:04 flavor used to be my bane - i swear it let him go like 3s per turn 2010-11-01T20:02:26 tcp has a 5 second timeout i think 2010-11-01T20:02:27 or 10 2010-11-01T20:02:28 not sure 2010-11-01T20:04:17 E323 2010-11-01T20:04:45 10 second first move, 5 second rest IIRC 2010-11-01T20:05:11 *** TravisD has joined #aichallenge 2010-11-01T20:06:20 E323? 2010-11-01T20:08:54 woah, it's sigh 2010-11-01T20:09:09 bhasker: sup 2010-11-01T20:09:10 hello? 2010-11-01T20:09:15 yes? 2010-11-01T20:09:55 I'm just wondering why davidjliu finds my presence so surprising :P 2010-11-01T20:10:04 I haven't seen you in, like, weeks 2010-11-01T20:10:05 maybe 2010-11-01T20:10:26 yeah, I basically took a break from the contest for 2 weeks 2010-11-01T20:10:46 i should be working more on my thesis >_< 2010-11-01T20:11:00 :) 2010-11-01T20:11:15 *** McLeopold has quit IRC (Quit: Leaving) 2010-11-01T20:12:54 k going to bed. See y'all tomorrow 2010-11-01T20:12:56 *** Migi32 has quit IRC (Quit: http://boinc.berkeley.edu) 2010-11-01T20:13:13 Top 10 players: bocsimacko(4352), dmj111(4275), Hazard(4161), FlameN(4154), shangas(4153), joakim_mjardner(4116), jambachili(4086), george(4081), felixcoto(4077), Accoun(4062) 2010-11-01T20:28:37 bhasker: whoaah 2010-11-01T20:28:41 new valuefunction 2010-11-01T20:29:09 growth^2 / (numships*distance + epsilon) 2010-11-01T20:29:18 epicnezz 2010-11-01T20:29:20 :P 2010-11-01T20:30:21 heh 2010-11-01T20:32:32 *** AlliedEnvy has quit IRC (Remote host closed the connection) 2010-11-01T20:34:40 *** exton has joined #aichallenge 2010-11-01T20:34:51 *** exton has left #aichallenge 2010-11-01T20:34:53 *** exton has joined #aichallenge 2010-11-01T20:35:35 hello. 2010-11-01T20:35:35 meta1's bot makes my bot look so bad:( 2010-11-01T20:36:13 *** exton has left #aichallenge 2010-11-01T20:37:35 meta1 ? 2010-11-01T20:37:54 mega1, I assume 2010-11-01T20:38:25 *** fawek has quit IRC (Ping timeout: 276 seconds) 2010-11-01T20:40:57 *** phreeza_ has quit IRC (Quit: kthxbai) 2010-11-01T20:41:50 mega1 sorry 2010-11-01T20:41:57 47 turns /cry 2010-11-01T20:42:39 but then i come right back and beat him at 200 turns with 7 ships! 2010-11-01T20:42:39 http://72.44.46.68/canvas?game_id=813345 2010-11-01T20:42:40 lol 2010-11-01T20:42:44 most interesting game ever 2010-11-01T20:44:25 lol 2010-11-01T20:44:39 wonder why bocsimacko launches that huge fleet at the end 2010-11-01T20:45:07 bhasker: why aren't you rerouting at the bottom 2010-11-01T20:45:08 ? 2010-11-01T20:55:22 ok i have a stupid c++ question - if i pass a class in as a reference ie a(const classname& param) - is it possible to call classname functions? i can access member vars fine but not sure how the hell to call functions 2010-11-01T20:55:52 sure 2010-11-01T20:56:00 demerzel_: the functions need to be const as well, I believe 2010-11-01T20:56:12 ahhh that might be it! 2010-11-01T20:56:24 tbh i don't even know what declaring a function const does 2010-11-01T20:56:32 static? 2010-11-01T20:56:39 its a contract saying that you won't modify the state of the object 2010-11-01T20:56:40 noo 2010-11-01T20:56:51 demerzel_: http://www.possibility.com/Cpp/const.html 2010-11-01T20:57:21 constness in c++ can be really confusing if you don't know how it works exactly 2010-11-01T20:57:21 ty - i get stuck on stupid semantics while refactoring lol 2010-11-01T20:57:35 liike hmm maybe i could just pass a ref since i wno't modify this obj 2010-11-01T20:57:45 but then an hour later my eyes are bleeding and i have no idea what i'm doign any more 2010-11-01T20:58:39 whenever you have a class method which doesn't change an object's state you should make it const. That way your const references aren't useless :P 2010-11-01T20:59:41 lol 2010-11-01T20:59:55 change state = alter member vars? 2010-11-01T21:00:00 yes 2010-11-01T21:00:01 if so that just sounds like static method to me 2010-11-01T21:00:13 no thats sth completely different 2010-11-01T21:00:15 or at least the way i use them 2010-11-01T21:00:15 demerzel_: It might read member variables 2010-11-01T21:00:37 demerzel_: Like, "getter" functions are good examples of functions which should be const 2010-11-01T21:00:59 god i cannot digest that page 2010-11-01T21:01:17 perhaps its time to take a break ;) 2010-11-01T21:01:33 i get why people push const but it's so bizarre how many things it can mean 2010-11-01T21:01:39 back to a pointer for me 2010-11-01T21:01:51 that's c++ for you 2010-11-01T21:02:05 janzert: You there? 2010-11-01T21:02:07 About the rating instability: I am pretty sure it is because games from old submissions are not used in the rating process. When GreenTea resubmitted a few days ago, I went back up because I no longer had a bunch of losses to him. When 50+ games disappear from the history, things can change a lot. 2010-11-01T21:02:08 the const issues come up with pointers as well... 2010-11-01T21:02:15 around anyway 2010-11-01T21:02:25 oh well i just make the contract with myself not to modify it not the compiler haha 2010-11-01T21:02:51 janzert: Did you have any luck getting Groovy running? :) 2010-11-01T21:03:06 yep see annoncement in forums 2010-11-01T21:03:19 announcement even 2010-11-01T21:03:23 demerzel_: I mean, there's no need to go back to pointers if all you want to do is remove the const 2010-11-01T21:03:28 janzert: Thanks :) I appreciate the effort on your part 2010-11-01T21:03:36 just waiting to see a few successfully running bots before closing the issue 2010-11-01T21:03:46 although my test with the starter package worked fine 2010-11-01T21:03:58 sigh: yea that's true i guess i'll just be happy that i'm using references instead somewhat at least 2010-11-01T21:04:33 references lead to less shoot-self-in-foot moments, I think 2010-11-01T21:04:48 constness is an orthogonal issue 2010-11-01T21:05:12 Hahaha, janzert: Thanks for the mention in the post ;) I'm not sure whether I should be flattered or feel sheepish :P 2010-11-01T21:06:53 well ty for the help i guess i'll conquer const another day i need to just finish reorging and got stuck on that crap for like an hour 2010-11-01T21:07:05 :) 2010-11-01T21:07:25 about the elo on tcp, is there any significance to an elo < 0 vs > 0 ? 2010-11-01T21:07:30 :) 2010-11-01T21:07:31 or meaning 2010-11-01T21:07:34 0 = median iirc 2010-11-01T21:08:02 main site has an an offset so that elo isn't negative 2010-11-01T21:08:06 i think it is 2000 2010-11-01T21:08:13 yup 2010-11-01T21:08:16 everybody's a winner! 2010-11-01T21:08:21 :P 2010-11-01T21:09:46 bhasker: around? 2010-11-01T21:10:08 Thanks again janzert. You guys are doing a good job of running the competition :) 2010-11-01T21:10:28 Thanks, doing what I can anyway 2010-11-01T21:10:50 *** superflit has quit IRC (Quit: superflit) 2010-11-01T21:12:23 *** superflit has joined #aichallenge 2010-11-01T21:13:15 Top 10 players: bocsimacko(4351), dmj111(4243), Hazard(4159), FlameN(4145), shangas(4122), george(4081), joakim_mjardner(4075), jambachili(4065), Accoun(4049), felixcoto(4049) 2010-11-01T21:15:19 *** AlliedEnvy has joined #aichallenge 2010-11-01T21:25:36 *** virdo has joined #aichallenge 2010-11-01T21:27:28 *** superflit has quit IRC (Quit: superflit) 2010-11-01T21:27:28 *** AlliedEnvy has quit IRC (Remote host closed the connection) 2010-11-01T21:30:01 *** AlliedEnvy has joined #aichallenge 2010-11-01T21:32:32 *** bhasker has quit IRC (Quit: bhasker) 2010-11-01T21:51:15 *** lavalamp_ has joined #aichallenge 2010-11-01T22:13:15 Top 10 players: bocsimacko(4353), dmj111(4244), Hazard(4161), FlameN(4146), shangas(4125), george(4083), jambachili(4075), joakim_mjardner(4062), felixcoto(4060), Accoun(4054) 2010-11-01T22:16:32 *** AlliedEnvy has quit IRC (Remote host closed the connection) 2010-11-01T22:17:19 *** AlliedEnvy has joined #aichallenge 2010-11-01T22:19:47 *** yasith has joined #aichallenge 2010-11-01T22:22:11 *** McLeopold has joined #aichallenge 2010-11-01T22:23:25 *** hornairs has joined #aichallenge 2010-11-01T22:44:31 *** TravisD has quit IRC (Read error: Connection reset by peer) 2010-11-01T22:49:19 *** Naktibalda has quit IRC (Ping timeout: 272 seconds) 2010-11-01T22:55:38 *** nullkuhl has quit IRC (Read error: Connection reset by peer) 2010-11-01T22:56:08 *** nullkuhl has joined #aichallenge 2010-11-01T23:13:15 Top 10 players: bocsimacko(4371), dmj111(4263), Hazard(4188), bix0r4ever(4156), shangas(4152), FlameN(4143), george(4093), joakim_mjardner(4074), felixcoto(4071), Accoun(4067) 2010-11-01T23:14:35 who is shampoo on the main site? 2010-11-01T23:16:26 *** Cyndre has quit IRC (Read error: Connection reset by peer) 2010-11-01T23:16:41 *** Cyndre has joined #aichallenge 2010-11-01T23:22:34 *** bhasker has joined #aichallenge 2010-11-01T23:23:38 afternoon 2010-11-01T23:23:51 evening 2010-11-01T23:23:55 its 8:30pm here 2010-11-01T23:23:56 ! 2010-11-01T23:24:21 2 25pm here :) 2010-11-01T23:24:34 my bot didn't do veyr well on tcp 2010-11-01T23:24:37 refinement time i guess 2010-11-01T23:24:38 its noon somewhere. 2010-11-01T23:24:41 it beat mine:p 2010-11-01T23:24:57 i want it to beat everyone :) 2010-11-01T23:25:01 ah high aims 2010-11-01T23:25:07 yep 2010-11-01T23:25:19 mine on official is upto 40 2010-11-01T23:25:46 *** Rubicon-|-Cross has joined #aichallenge 2010-11-01T23:26:05 mine wasn't running on official so i haven't had it going, hopefully it shouldn't crash anymore when i pout it up next 2010-11-01T23:26:13 i was sitting around there last time i had it up 2010-11-01T23:26:21 i wonder if python if fast enough to implement minmax 2010-11-01T23:26:26 or should i look at c++ 2010-11-01T23:27:03 i've no idea of the speed of different languages, i would say it should be fine 2010-11-01T23:27:11 i tried minimax over the weekend, using my current heuristics to guide the move selection. It didn't look promising. 2010-11-01T23:27:22 i don't get very far at all cause i waste all my time calculating reasonable moves 2010-11-01T23:27:23 hmm 2010-11-01T23:27:36 dmj111: neither, i'm currently doing a zero sum game approach 2010-11-01T23:27:46 and use decision trees for each player to generate pure strategies 2010-11-01T23:27:51 i think it sort of seems to work wel 2010-11-01T23:28:13 atm, i'm only picking my pure strategy that maximises the min row too, i plan to change that 2010-11-01T23:30:23 it will be interesting to see what kind of strategies you are selecting. I considered having a (N+1) x (N+1) payoff matrix , where each player all-out attacks a single planet, one at a time, and then solving that. But, that doesn't seem right. 2010-11-01T23:32:38 nope, i'm using a decision tree, where at each node i'll pick what looks like the best 3-4 moves, then i pick the best x terminal nodes, and the string of moves to get there are my pure strategies, if that makes sense? 2010-11-01T23:33:22 once i start doing so many things i can't allocate much time to each though 2010-11-01T23:33:23 antimatroid: what constitutes a move? 2010-11-01T23:33:35 an entire set of orders? 2010-11-01T23:33:43 single destination, multiple sources (can even even multiple fleets from the same source, just at different times) 2010-11-01T23:33:59 yup, i see 2010-11-01T23:34:15 *** adu has joined #aichallenge 2010-11-01T23:34:18 bhasker: python is the slowest benching lang that i know of atm 2010-11-01T23:34:22 bhasker: vs you know c++ 2010-11-01T23:34:28 hi 2010-11-01T23:34:33 that way the decision tree looks at moves it plans to make in the future from it's past stringed moves too 2010-11-01T23:34:43 adu: hi :) 2010-11-01T23:34:51 what's the latest challenge? 2010-11-01T23:34:52 hm k 2010-11-01T23:34:55 planet wars 2010-11-01T23:35:09 it's rather tricky 2010-11-01T23:35:30 bhasker: but java isn't bad either honestly - i'd rather do c++ since it's compiled instead of relying on a vm on a decrepit server ;) 2010-11-01T23:35:39 i detest java 2010-11-01T23:35:42 antimatroid: i think your explanation is fine. I am just trying to fit it into my preconceived ideas of how to do things. 2010-11-01T23:35:43 it's not bad 2010-11-01T23:35:53 and from what i am hearin about minmax not sure there is much benefit in implementing one 2010-11-01T23:36:02 and my crappy math background i would probably screw it up 2010-11-01T23:36:11 i have no math background - it's not too bad 2010-11-01T23:36:13 i think it could work if set up right, that's really hard though 2010-11-01T23:36:24 i am almost positive shampoo is minma 2010-11-01T23:36:33 his playstyle is the ideal of my prev bots 2010-11-01T23:36:35 bhasker: the nice thing about minimax is there is not much math. 2010-11-01T23:36:45 minimax? 2010-11-01T23:37:02 hmm 2010-11-01T23:37:04 minimize loss, maximize gain 2010-11-01T23:37:17 pure arithmetic calc 2010-11-01T23:37:19 yea but i really don't want to code in c++ :-\ i do it at wokr 2010-11-01T23:37:25 and i hate compiling every time i make a change 2010-11-01T23:37:36 mine takes a good 10 seconds to compile now :( 2010-11-01T23:37:37 python is sooo flexibile 2010-11-01T23:37:40 haha well i use vs even though i don't use VC 2010-11-01T23:37:56 i can't wrap my head around the fact that your internal dilemna is python and C++ tho 2010-11-01T23:37:58 and testing takes forever 2010-11-01T23:38:07 <---lazy programmer 2010-11-01T23:38:12 implementing it in python will be much faster 2010-11-01T23:38:15 <--- not a proper programmer :) 2010-11-01T23:38:17 but the speed would suck 2010-11-01T23:38:22 antimatroid: true! i am refactoring all my code for the 10th time 2010-11-01T23:38:28 i actually have fun doing stupidly tedious things still 2010-11-01T23:38:33 haha me too 2010-11-01T23:38:42 bhasker: it depends on how deep you want to go 2010-11-01T23:38:43 antimatroid: sorry, i meant this: http://en.wikipedia.org/wiki/Minimax#Minimax_algorithm_with_alternate_moves , even though the moves are simultaneous. 2010-11-01T23:38:43 and i still can't get my stupid planet selection criteria to work correctly 2010-11-01T23:38:50 example 2010-11-01T23:38:51 http://72.44.46.68/canvas?game_id=815097 2010-11-01T23:39:01 stupid bot never went for the planet with 8 ships 2010-11-01T23:39:06 and goes for the 83 ship one instead 2010-11-01T23:39:24 wait you have that bot already? in what lang 2010-11-01T23:39:31 dmj111: yeah, i have minimax implemented for a game tree (not with alpha beta pruning, yet) but it doesn't seem to work well either 2010-11-01T23:39:45 * adu doesn't like java either 2010-11-01T23:39:47 i have minimax written for simultaneous games too, but i'm not sure about just playing pure strategy nash equilibria 2010-11-01T23:39:53 dmj was that a question for me? 2010-11-01T23:39:59 me has an irrational hate of java, but i refuse to touch it 2010-11-01T23:40:03 i think minimax has to be used with simultaneous prediction 2010-11-01T23:40:12 which i have yet to implement 2010-11-01T23:40:24 simultaneous prediction? 2010-11-01T23:40:38 yes you will need to model choices from the other bot, any ideas with game theory will need that 2010-11-01T23:41:01 my gamestate is set up that i can just pass in a player parameter and it'll take that as the person it's making decisions for 2010-11-01T23:41:08 yes but using minimax for the prediction is the problem haha 2010-11-01T23:41:30 i am going more for maximize gain minimize potential gain - minimize loss is just autopilot defenses 2010-11-01T23:41:36 i have functions to pick out individual moves 2010-11-01T23:41:50 they're really inefficient atm though 2010-11-01T23:42:05 *** fmeyer has joined #aichallenge 2010-11-01T23:42:09 antimatroid yea i did that for a while but now i can just switch which bool it's looking at ;) 2010-11-01T23:42:38 i need to change the way i do closestfriends/enemies 2010-11-01T23:42:48 it needs to be indexible by player number 2010-11-01T23:43:53 dmj111: for the record, i'm under the thought that mega1 is trying the simultaneous game rather than extensive 2010-11-01T23:44:53 i also have a play game function that i run befor evaluation nodes in my payoff matrix, where each player makes defense and offense moves each turn before incrementing the state 2010-11-01T23:44:57 It will be interesting to see what he was doing when this is over. 2010-11-01T23:45:03 it will 2010-11-01T23:45:07 and what he will do next :P 2010-11-01T23:45:16 he still doesn't have any competition 2010-11-01T23:45:21 maybe go on vacation. 2010-11-01T23:45:24 i hope. 2010-11-01T23:45:29 haha 2010-11-01T23:45:31 and submit a buggy bot. 2010-11-01T23:45:34 haha 2010-11-01T23:45:51 that crash the other day would have had him disqualified if it was during the contest last time 2010-11-01T23:45:57 that happened to me last time :( 2010-11-01T23:54:00 *** AlliedEnvy has quit IRC (Remote host closed the connection)