2010-10-27T00:00:03 it seemed to do more harm than good 2010-10-27T00:00:17 When you find the magic formula, it will do a lot of good. 2010-10-27T00:00:41 *** yasith has joined #aichallenge 2010-10-27T00:01:49 yeah 2010-10-27T00:01:59 i want to get a good bot before i start trying to train parameters 2010-10-27T00:02:09 so, I decided to run my official version (#5) on tcp again with a couple of tweaks 2010-10-27T00:02:12 i still have a lot of things to iron out before i can add moves back in from the other player 2010-10-27T00:02:16 mcleopold 2010-10-27T00:02:23 which bot are you running 2010-10-27T00:02:27 my bot currently crashes on the main site, i need to address that too 2010-10-27T00:02:30 I think I lost 50 before I realized I had special code to make it lose. 2010-10-27T00:03:07 bhasker: closer to official than anything 2010-10-27T00:03:20 hmm k 2010-10-27T00:03:39 scotbot5? 2010-10-27T00:03:42 yeah 2010-10-27T00:03:51 did you see that epic win early in the day 2010-10-27T00:03:55 which one? 2010-10-27T00:04:01 the first one 2010-10-27T00:04:09 when you played my handicapped version 2010-10-27T00:04:25 ah 2010-10-27T00:04:30 i think i did 2010-10-27T00:04:30 and my bot looked like a sprinkler watering the lawn 2010-10-27T00:04:38 though http://72.44.46.68/canvas?game_id=723418 :) 2010-10-27T00:04:44 I really hate that map 2010-10-27T00:05:10 no, it was the map with 2 giant globs of planets separated by space 2010-10-27T00:05:11 I can't tell one map from another. 2010-10-27T00:05:41 i think i pasted that in irc chat today 2010-10-27T00:05:45 earler in the tday 2010-10-27T00:05:51 hmm 2010-10-27T00:06:04 right now my bot's are busy playing each other 2010-10-27T00:06:07 0.11 vs 0.12 2010-10-27T00:06:15 0.12 is a bit more aggresive 2010-10-27T00:06:25 0.11 seems to be doing better overall though 2010-10-27T00:06:39 http://72.44.46.68/canvas?game_id=715285 2010-10-27T00:06:58 ah yea saw that one 2010-10-27T00:07:03 it kept oscillating 2010-10-27T00:07:06 between two planets 2010-10-27T00:07:17 i think its an effect of my bots behaviour 2010-10-27T00:07:23 I though it was funny 2010-10-27T00:07:26 my bot tends to shuttle 2010-10-27T00:07:28 ships 2010-10-27T00:07:31 its actually a bug 2010-10-27T00:07:37 i tend to overestimate ships needed 2010-10-27T00:07:44 so i keep shuttling them back and forth 2010-10-27T00:07:53 thats what is probably causing your bot to behave like that 2010-10-27T00:08:07 if you see closely you will see ship counts keep going up and down between the two planets 2010-10-27T00:08:18 actually, I think yours was reacting to mine 2010-10-27T00:08:23 hmm 2010-10-27T00:08:29 i do know my bot over sends ships 2010-10-27T00:08:37 i basically try to account for rate of flow of ships 2010-10-27T00:08:42 but when there are spikes 2010-10-27T00:08:48 that calculation is way offo 2010-10-27T00:08:49 off* 2010-10-27T00:08:55 so i tend to send more than required 2010-10-27T00:08:58 has costed me games 2010-10-27T00:09:15 but i sadly haven't been able to come up with a good way to estimate that 2010-10-27T00:10:13 if i try to be exact i lose planets when enemy is closer than my main support planet 2010-10-27T00:10:35 if i overestimate i tend to sometimes keep shuttling ships back and forth and lose 2010-10-27T00:10:35 I'm usually the other way, not sending enough 2010-10-27T00:10:52 basically i am thinking maybe i will send 2x the ships 2010-10-27T00:12:04 Top 10 players: rsergio(3814), protocolocon(3762), dmj111(3759), Slin-.-(3685), felixcoto(3665), jambachili(3662), Hazard(3643), george(3603), sin_sun(3590), Simplex(3589) 2010-10-27T00:13:22 i have the attack part mostly figured out 2010-10-27T00:13:29 defense is where i am still having trouble 2010-10-27T00:13:31 :-\ 2010-10-27T00:14:33 17 wins in a row. 2010-10-27T00:14:39 nice 2010-10-27T00:15:02 It is nice to watch his bot in action :) 2010-10-27T00:15:13 whose bot? 2010-10-27T00:15:30 bocsimacko's 2010-10-27T00:15:40 ah 2010-10-27T00:15:41 yea 2010-10-27T00:15:42 his latest on he official is 17/17 2010-10-27T00:16:01 his bot is like a time travelling bot 2010-10-27T00:16:48 Yeah, there's some pretty impressive forecasting going on there. 2010-10-27T00:17:32 I like the cases where he uses planets that he doesn't own yet in his fleet calcs 2010-10-27T00:21:00 TheMacca: can you give an example? 2010-10-27T00:21:35 McLeopold: I'd have to go looking, it was on the TCP server, but basically 2010-10-27T00:21:42 there was a cluster of three planets 2010-10-27T00:21:45 far to one side 2010-10-27T00:21:52 he sent three fleets 2010-10-27T00:22:04 the largest fleet wasn't big enough to take the largest planet 2010-10-27T00:22:12 they all arrived roughly same time (ish) 2010-10-27T00:22:24 and he used the smaller two to augment the larger fleet 2010-10-27T00:22:28 and rolled over the cluster like a wave 2010-10-27T00:22:29 was cool 2010-10-27T00:22:34 will see if I can find, if you like 2010-10-27T00:22:42 I think accoun does that too. 2010-10-27T00:22:47 but it might have rolled off the list by now 2010-10-27T00:22:57 it was at that exact moment I knew I might be out of my depth :) 2010-10-27T00:23:05 Actually, AntBot does as well. 2010-10-27T00:23:41 I haven't seen Accoun do it 2010-10-27T00:23:42 i need to factor that in somehow 2010-10-27T00:23:46 accoun doesn't do it 2010-10-27T00:23:49 but haven't watched as many of his 2010-10-27T00:23:58 my bot has gone up a lot against accoun 2010-10-27T00:24:05 and his doesn't do that for sure 2010-10-27T00:24:44 bhasker: he absolutely does it for sure, I watched tonight. 2010-10-27T00:24:52 hmm let me watch 2010-10-27T00:24:56 if we are talking about the same thing... 2010-10-27T00:25:11 Hey Mc, http://72.44.46.68/canvas?game_id=705835 2010-10-27T00:25:22 there's an example in here, it's not quite as clean 2010-10-27T00:25:41 nw or se? 2010-10-27T00:25:53 i think nw 2010-10-27T00:25:56 he launches 15 2010-10-27T00:26:02 E edge 2010-10-27T00:26:03 adds 1 from the neutral he cpatures 2-3 turns later 2010-10-27T00:26:05 south ish 2010-10-27T00:26:11 yeah, account does that too. 2010-10-27T00:26:15 he sends a fleet of 7 to attack a 15 2010-10-27T00:26:20 My ant bot does as well. 2010-10-27T00:26:30 and picks up more from planets he doesn't own yet (at the time he sends the 7) 2010-10-27T00:26:50 McLeopold: do you have an example in one of your games? 2010-10-27T00:26:52 ah you are right mc 2010-10-27T00:26:54 he does it 2010-10-27T00:26:56 http://72.44.46.68/canvas?game_id=723450 2010-10-27T00:27:01 watch the top planet with 41 ships 2010-10-27T00:27:27 nice 2010-10-27T00:27:54 i shouldn't have lost that game though 2010-10-27T00:27:54 this version is aggresive 2010-10-27T00:28:00 ouch, and he left himself open 2010-10-27T00:28:16 bhasker: why didn't you take the center from him? 2010-10-27T00:28:29 not sure lol 2010-10-27T00:28:33 will have to look at the logs 2010-10-27T00:28:44 why my bot didnt attack is weird 2010-10-27T00:28:48 that was some prime cherry picking you missed 2010-10-27T00:29:11 actually let me run it with logging enabled 2010-10-27T00:29:11 and see 2010-10-27T00:29:38 oh i think i know why 2010-10-27T00:29:43 my bot is too conservative 2010-10-27T00:29:44 TheMacca, I'm looking for an example of AntBot doing it 2010-10-27T00:30:03 it calculates at start of turn how many ships it can spare without losing planet 2010-10-27T00:30:20 and as a result with 48 ships near it, it probably calculated it could only spare X ships 2010-10-27T00:30:28 where X<48+growth 2010-10-27T00:30:37 this is a bug i need to fix 2010-10-27T00:31:04 I'm doing an Human computer interface class and were redesigning youtube, if anybody can comment on my youtube video http://www.youtube.com/watch?v=7hWPTGEaZcE saying what they like or dont like about youtube interface and if their is anything to add or take away from it.!! thanks guys 2010-10-27T00:31:09 i never revisit or adjust my queue of available ships if the planet i am attacking could be affect my availability 2010-10-27T00:31:26 that is the reason i never attacked it 2010-10-27T00:31:34 if the balance of ships had been different i would have taken it 2010-10-27T00:31:40 :D http://72.44.46.68/canvas?game_id=715238 2010-10-27T00:32:00 oh wait, no, that didn't count 2010-10-27T00:32:40 interesting you are doing it too 2010-10-27T00:32:53 no, that 1 was extra at the beginning. 2010-10-27T00:33:11 no but you did launch 34 ships 2010-10-27T00:33:19 which got supplanted from your planet 3 turns later 2010-10-27T00:33:27 oh wait 2010-10-27T00:33:30 you launched it later 2010-10-27T00:33:44 hmm actually i should maybe try it out 2010-10-27T00:33:46 the sw corner 2010-10-27T00:33:50 time to tinker with my code 2010-10-27T00:34:06 you use it to attack the 60 in the center 2010-10-27T00:34:12 before you own the sw corner 2010-10-27T00:34:30 d3c1ph3r: Only one thing - on the side bar I wish they would give a hint on wether a video is enabled for my country or not. I hate to click through 10 music videos and every one turns out to be locked by BMG or Sony. 2010-10-27T00:34:47 interesting i wonder how i can rework my code to do that 2010-10-27T00:35:34 please post your comments on the youtube page so my professor can see it 2010-10-27T00:35:41 http://www.youtube.com/watch?v=7hWPTGEaZcE 2010-10-27T00:35:46 Thanks all 2010-10-27T00:36:21 *** irchs has joined #aichallenge 2010-10-27T00:36:30 ouch 2010-10-27T00:36:46 running two bots is a bad idea 2010-10-27T00:36:52 my laptop gets too busy 2010-10-27T00:36:56 just lost a bunch of games to timeouts 2010-10-27T00:37:03 d3c1ph3r: how is this related to PlanetWars? 2010-10-27T00:37:05 http://72.44.46.68/getplayer?player=malazanbot0.12 2010-10-27T00:37:06 :-\ 2010-10-27T00:39:09 I'd fire up AntBot so you could see the technique, but it's broken right now. :( 2010-10-27T00:40:23 i think i understand how to do it 2010-10-27T00:40:30 its just hard to work it into the way my code is written 2010-10-27T00:40:59 it's worth the effort 2010-10-27T00:41:05 yea i can see that 2010-10-27T00:41:48 its probably easier for me to work it into the setup for planets already under my control 2010-10-27T00:42:42 *** Accoun has quit IRC () 2010-10-27T00:42:43 bhasker: I've seen people send flleets sufficient for one neutral + another that has incoming enemy ships to the former. Then they launch ships from there to capture the second neutral which is then too close for the enemy to even react. 2010-10-27T00:43:26 that's fun too 2010-10-27T00:43:32 that is not hard to do, just need to see how many ships you need to capture bot planets 2010-10-27T00:43:35 basically 2010-10-27T00:43:37 It works as long as your opponent does only see incoming fleets and can't estimate what the enemy is about to do. 2010-10-27T00:43:39 ships to capture neutral A 2010-10-27T00:43:45 + ships to capture neutral B from A 2010-10-27T00:43:49 and send that many to A 2010-10-27T00:44:08 what's tricky is choosing the targets 2010-10-27T00:44:14 bhasker: but it takes a lot of record keeping. 2010-10-27T00:44:21 not it doesn't 2010-10-27T00:44:34 once enemy has launched his ships at that neutral 2010-10-27T00:44:39 you have perfect information 2010-10-27T00:44:45 to know how many to launch at A 2010-10-27T00:44:54 to capture A&B 2010-10-27T00:45:07 prediction is different from record keeping 2010-10-27T00:45:22 if you can program array, you can keep records 2010-10-27T00:45:28 ok, so I would write on special code path just for "snipe neutral from secondary planet after enemy has launched an attack". I dont like the idea ^^ 2010-10-27T00:45:35 no frontier 2010-10-27T00:45:45 what i do is 2010-10-27T00:45:53 i calculate how many ships i need to take the neutral 2010-10-27T00:46:08 as neutral + nearbyPlanets that are owned by enemy or will be in future 2010-10-27T00:46:10 and ships on them 2010-10-27T00:46:14 Frontier: how about this. You see 2 neutrals you would like to take, They are near each other. So send the amount needed to 1 planet only. 2010-10-27T00:46:43 You lose a bit of time, but if the worst happens, you'll be okay. 2010-10-27T00:47:46 McLeopold: I would still need to keep record of the fact that planet 2 doesn't need ships in the next turns as long as the ships to planet 1 are commited to it somehow. so the fleets would be split up somehow. 2010-10-27T00:47:51 Oh, wait, I think in your original description you should have said "later". 2010-10-27T00:47:59 *** xlk_ has joined #aichallenge 2010-10-27T00:48:13 Frontier: yep. you need that. 2010-10-27T00:48:32 You can do it! 2010-10-27T00:48:44 frontier in most cases you would not have enough ships on the next turn 2010-10-27T00:48:47 to double launch 2010-10-27T00:49:06 That puts me on the safe side I guess, haha 2010-10-27T00:49:06 or next few turns 2010-10-27T00:49:22 so i wouldn't worry too much about double launches 2010-10-27T00:49:37 at least early game it won't be an issue 2010-10-27T00:49:43 late game you can probably disable it 2010-10-27T00:49:49 since you may have enough ships to double launch 2010-10-27T00:49:57 http://72.44.46.68/canvas?game_id=724118 2010-10-27T00:50:05 bhasker: your bot is so boring 2010-10-27T00:50:10 lol 2010-10-27T00:50:11 yea 2010-10-27T00:50:13 its very conservative 2010-10-27T00:50:19 if it has an advantage 2010-10-27T00:50:21 it just sits 2010-10-27T00:50:28 till the advantage is large enough 2010-10-27T00:50:41 it won't acquire new neutrals 2010-10-27T00:50:52 and it will attack only when it can take the planet for sure 2010-10-27T00:51:05 a more interesting match for you 2010-10-27T00:51:05 http://72.44.46.68/canvas?game_id=724095 2010-10-27T00:51:08 i lost this one 2010-10-27T00:51:14 because i oversent to the back planet 2010-10-27T00:51:21 and then you launched at middle while my ships were in flight 2010-10-27T00:52:08 and routed them through the +5 2010-10-27T00:52:29 well it didnt route through them 2010-10-27T00:52:36 i have some logic which looks for closest enemy 2010-10-27T00:52:39 and sends towards it 2010-10-27T00:52:47 so it sent them there before you launched 2010-10-27T00:52:52 so ... 2010-10-27T00:52:58 you had a top right planet 2010-10-27T00:53:01 so it sent towards it 2010-10-27T00:53:04 instead of back to center 2010-10-27T00:53:39 let me find you an interesting one to watch though 2010-10-27T00:54:12 http://72.44.46.68/canvas?game_id=704540 2010-10-27T00:54:14 i lost this one 2010-10-27T00:54:18 but its a fun one to watch 2010-10-27T00:55:54 bhasker: funneling logic is very important, more important than bocsimacko's trick 2010-10-27T00:56:12 my funnelling logic works reasonably well 2010-10-27T00:56:19 the main problem though is overdefending 2010-10-27T00:56:32 when i defend accurately i lose when enemy has a planet closer 2010-10-27T00:56:35 when i over defend 2010-10-27T00:56:42 i get screwed like the game with you 2010-10-27T00:56:54 the real problem was i sent more ships to the back than required 2010-10-27T00:56:59 lol - I have to figure out how to stop underdefending 2010-10-27T00:57:05 next week when RamBot is done, well have to duel so you can see good funneling logic 2010-10-27T00:57:28 yea i have tried various combinations mc but none of them seem to work very reliabily 2010-10-27T00:57:35 the current one seems best 2010-10-27T00:57:40 seems to funnel reasonably correctly 2010-10-27T00:57:48 example game http://72.44.46.68/canvas?game_id=724198 2010-10-27T00:58:03 *** davidjliu has quit IRC (Quit: Page closed) 2010-10-27T00:58:41 http://72.44.46.68/canvas?game_id=724249 2010-10-27T00:58:50 my funnel logic is reasonably good 2010-10-27T00:59:02 if i could get my defense right i think the bot would do much better 2010-10-27T00:59:17 I wonder if "Monty" codes in Python 2010-10-27T00:59:25 my official just sends to the closest enemy as well. It does work "good", but it can be better. 2010-10-27T00:59:49 http://72.44.46.68/canvas?game_id=724285 2010-10-27T00:59:51 frontier 2010-10-27T00:59:55 watch my attack on the central planet 2010-10-27T01:00:03 i should have won that 2010-10-27T01:00:15 but i guess my stupid laptop keeps doing stupid stuff and time out my bot at times 2010-10-27T01:01:11 bhasker: Well, you sent EVERYTHING to the center planet. I can't tell if you had some ships already designated to capture the second planet ^^ 2010-10-27T01:01:27 the biggest launch is 134 ships 2010-10-27T01:01:28 at least you can claim you reacted at the right time 2010-10-27T01:01:28 or so 2010-10-27T01:01:45 *** narnach has joined #aichallenge 2010-10-27T01:02:53 That's true, Monty didn't see the fleet coming or just didn't have any spare ships to support the center at that time. 2010-10-27T01:03:33 bhasker: have you though of defending with just the right amount to see how that does? 2010-10-27T01:04:23 i tried that 2010-10-27T01:04:24 McLeopold: The right amount = the amount of incoming fleets on any specific turn? 2010-10-27T01:04:28 i lose when bots stream ships 2010-10-27T01:04:38 cause sometimes my planets are further away 2010-10-27T01:04:49 and i lose a critical planet 2010-10-27T01:04:53 and it goes downhill 2010-10-27T01:05:14 pretty much that frontier 2010-10-27T01:05:30 and to adjust for any growth if you are going to lose the planet 2010-10-27T01:05:40 i.e enemy was closer and launched massive fleet 2010-10-27T01:05:51 bhasker: It doesn't work there... you should have more ships on the front line anyway. 2010-10-27T01:06:05 well i don't redistribute ships 2010-10-27T01:06:15 so sometimes my edge planets don't have the optimum number 2010-10-27T01:06:17 bhasker: That is bad :p 2010-10-27T01:06:21 ^ 2010-10-27T01:06:33 haven't figured out a good heuristic for that :-\ 2010-10-27T01:06:55 Oh you can imagine the planets as some kind of web. 2010-10-27T01:07:09 well the problme is everytime you redistribute ships 2010-10-27T01:07:12 you lose travel time 2010-10-27T01:07:15 and those ships are lost 2010-10-27T01:07:17 You have to figure out how you can defend every node from as few nodes as possible 2010-10-27T01:07:21 flying between your planets 2010-10-27T01:07:31 yea so i do that via my staging code 2010-10-27T01:07:32 :( true again... 2010-10-27T01:07:42 since i stage to edge planets 2010-10-27T01:07:54 and most of my edge planets are close to the main central one 2010-10-27T01:08:04 so usually my central planet can send ships to support my edge 2010-10-27T01:08:20 but sometimes two nodes happen at extremeties of the map 2010-10-27T01:08:27 thats when my bot goes a bit nutty 2010-10-27T01:08:31 sending ships back and forst 2010-10-27T01:08:32 forth* 2010-10-27T01:08:39 from one edge to the other 2010-10-27T01:08:44 because it oversends ships 2010-10-27T01:08:58 Funny thing is: You shouldn't get into a situation where you have to stage to multiple planets from a tactical point of view. 2010-10-27T01:09:26 actually it makes complete sense to do so 2010-10-27T01:09:38 if enemy is closer then sending to central is pointless 2010-10-27T01:09:51 because if enemy attacks 2010-10-27T01:10:00 your central will never be able to reach the planet in time 2010-10-27T01:10:06 so its better to hold ships there 2010-10-27T01:10:08 and wait 2010-10-27T01:10:10 bhasker: if you already can funnel to planets closer to the enemy, then you can defend exactly to planets further from the enemy. 2010-10-27T01:10:23 oh the problem is never for planets further 2010-10-27T01:10:27 bocsimacko: 18/18 2010-10-27T01:10:30 the problem is for the edge planets 2010-10-27T01:10:36 it was in our game 2010-10-27T01:10:49 hmm maybe i should do that 2010-10-27T01:11:00 if its not an edge planet 2010-10-27T01:11:03 defend exactly 2010-10-27T01:11:07 http://72.44.46.68/canvas?game_id=724095 turn 49 2010-10-27T01:11:30 yea cause right now i don't differentiate 2010-10-27T01:11:36 i was talking when i do exact defense 2010-10-27T01:11:41 i never have a problem defending my back planets 2010-10-27T01:11:49 its always the edge planets that i have trouble with 2010-10-27T01:11:53 since i don't redistribute 2010-10-27T01:12:02 *** Accoun has joined #aichallenge 2010-10-27T01:12:05 Top 10 players: rsergio(3800), protocolocon(3774), dmj111(3757), felixcoto(3683), jambachili(3675), Hazard(3665), Slin-.-(3660), george(3621), sin_sun(3618), bocsimacko(3591) 2010-10-27T01:12:06 so sometimes edges can be low on ships than enemy 2010-10-27T01:12:12 and get attacked 2010-10-27T01:12:13 Once you have to defend 2 planets and the enemy only has to defend 1 planet you lose an the advantage. I would go into offense from there as soon as possible to avoid beeing caught in a situation where my enemy sends a huge fleet to one of those half secured planets. 2010-10-27T01:12:13 gtg 2010-10-27T01:12:21 *** delt0r___ has quit IRC (Read error: Operation timed out) 2010-10-27T01:12:22 *** Accoun has joined #aichallenge 2010-10-27T01:12:34 frontier enemy doesn't have to defend 1 2010-10-27T01:12:37 *** McLeopold has quit IRC (Quit: Leaving) 2010-10-27T01:12:44 he has to defend all of them 2010-10-27T01:13:17 bhasker: ok, i was thinking of your staging planets where you keep all your ships in mid game 2010-10-27T01:13:30 *** delt0r___ has joined #aichallenge 2010-10-27T01:16:22 *** virdo has quit IRC (Ping timeout: 252 seconds) 2010-10-27T01:20:55 Well first I have to copycat Accoun to get my growth go up as fast as possible. 2010-10-27T01:22:18 *** Olathe has quit IRC (Ping timeout: 240 seconds) 2010-10-27T01:25:55 *** Palmik has joined #aichallenge 2010-10-27T01:30:27 *** Accoun has quit IRC (Ping timeout: 240 seconds) 2010-10-27T01:33:13 *** jukkerknott has quit IRC (Quit: Page closed) 2010-10-27T01:35:07 *** seisatsu has quit IRC (Quit: Leaving) 2010-10-27T01:35:21 *** boegel has joined #aichallenge 2010-10-27T01:45:16 *** Olathe has joined #aichallenge 2010-10-27T01:51:40 Can it always be assumed that the game will be 200 moves? 2010-10-27T01:52:13 no 2010-10-27T01:52:20 .... 2010-10-27T01:52:22 hhhmmmm.... 2010-10-27T01:52:31 Is there anyway to find out the turn limit? 2010-10-27T01:52:35 though it is likely 2010-10-27T01:52:49 no, not yet 2010-10-27T01:52:59 Not yet? 2010-10-27T01:53:03 They are gonna change the spec? 2010-10-27T01:53:04 *** Appleman1234 has quit IRC (Ping timeout: 276 seconds) 2010-10-27T01:53:05 * irchs cries 2010-10-27T01:53:10 someone was working on a branch to send turn limits, but the thing kinda died 2010-10-27T01:53:27 ideally don't make your bot depend on turn limit 2010-10-27T01:53:28 I need to calculate end game conditions so I don't end up just sitting there being cagey :) 2010-10-27T01:53:32 and if you do, make sure it's easy to change 2010-10-27T01:54:58 http://ai-contest.com/profile.php?user_id=7843 2010-10-27T01:55:08 yay a C# programs breaks the top 100 2010-10-27T01:58:57 *** Appleman1234 has joined #aichallenge 2010-10-27T01:59:24 Would you agree that when you have 10 times the number of ships as the enemy you can attack freely? 2010-10-27T01:59:51 btw, I think I found the race condition that allowed a few people to end up with double entries. Basically if a second entry was made before the first was compiled they could both end up playing. 2010-10-27T02:00:10 So the 10 minute delay had the side effect of basically fixing it. 2010-10-27T02:00:33 yay 2010-10-27T02:00:34 janzert: Do you have the POWER to change org affiliation? 2010-10-27T02:00:49 irchs: post it in the thread 2010-10-27T02:00:57 I have :( 2010-10-27T02:00:57 *** boegel has quit IRC (Ping timeout: 245 seconds) 2010-10-27T02:00:58 Twice 2010-10-27T02:00:59 :( 2010-10-27T02:01:06 hmm, technically that would be yes but I'm not willing to do so sorry 2010-10-27T02:01:08 then wait.. 2010-10-27T02:01:16 fine 2010-10-27T02:01:39 soon though it is quite possible that I will be working on allowing you to edit it yourself 2010-10-27T02:01:58 I think I'm finally getting past the critical bugs 2010-10-27T02:02:00 there's a patch for this 2010-10-27T02:02:04 i think 2010-10-27T02:02:07 w/e 2010-10-27T02:02:14 amstan: just for password changes I believe 2010-10-27T02:02:26 new language support is kinda more important 2010-10-27T02:02:50 in some ways yes 2010-10-27T02:02:58 *** Olathe has quit IRC (Ping timeout: 276 seconds) 2010-10-27T02:03:01 the thread is ok by its own now, there's a script in my homedir which accepts those lines that people post 2010-10-27T02:03:25 mostly I'm a bit weary of language support at the moment after wrestling with C# for the last few days 2010-10-27T02:03:40 janzert: you could try the script then, lol 2010-10-27T02:04:00 once I recuperate I'm sure I'll probably give some more languages a go 2010-10-27T02:04:24 *** hornairs has quit IRC (Quit: hornairs) 2010-10-27T02:04:27 err, maybe I shouldn't mention 'go' :P 2010-10-27T02:04:35 lol 2010-10-27T02:05:22 janzert: you did a bad thing with the c# support: http://ai-contest.com/forum/viewtopic.php?f=18&t=1010 2010-10-27T02:05:45 hehe, I don't want to tell him that C# is working now 2010-10-27T02:05:54 yeah, i was about to 2010-10-27T02:05:57 but then i reconsidered 2010-10-27T02:06:02 I'm still waiting to see if anyone in the forums actually notices 2010-10-27T02:06:15 they have stopped complaining anyway :) 2010-10-27T02:06:19 kinda hard with that topic being in announcements 2010-10-27T02:06:31 *** Appleman1234 has quit IRC (Ping timeout: 255 seconds) 2010-10-27T02:08:35 so it took another 20 hours or so to actually get it working consistently after that announcement where I *finally* thought I had it 2010-10-27T02:08:53 lol 2010-10-27T02:10:15 I need to kick jeff again to create some new instances, been since friday or saturday that any of his were up 2010-10-27T02:10:46 and I think the code for them has somewhat stabilized minus language additions 2010-10-27T02:11:10 but this is pretty good though 2010-10-27T02:11:11 like.. 2010-10-27T02:11:22 if you plan to use such languages that have only half support 2010-10-27T02:11:29 and their compiler memory leaks.. 2010-10-27T02:11:39 it's only your fault 2010-10-27T02:11:51 :) yeah 2010-10-27T02:11:59 oh! let's use C#, it's cross platform and stuff, oh wait! 2010-10-27T02:12:06 Top 10 players: rsergio(3820), dmj111(3769), protocolocon(3750), Slin-.-(3697), felixcoto(3686), Hazard(3677), jambachili(3668), bix0r4ever(3646), bocsimacko(3638), george(3613) 2010-10-27T02:12:08 of course only my cost at the moment too :P 2010-10-27T02:13:30 *** Appleman1234 has joined #aichallenge 2010-10-27T02:13:40 same with java last contest 2010-10-27T02:13:49 it didn't go well with the sandbox and timed out.. 2010-10-27T02:13:56 yeah, I had that thought cross my mind more than once 2010-10-27T02:13:57 you're using java, it's a risk you have to take 2010-10-27T02:14:23 these kind of problems happen in real life too 2010-10-27T02:14:32 at least the timeouts aren't disqualifiers now 2010-10-27T02:14:52 it was fun when it was 2010-10-27T02:15:19 heh, yep. pushed me into using C++ for tron 2010-10-27T02:15:43 *** TravisD has joined #aichallenge 2010-10-27T02:16:10 I enjoyed making this bot: http://ai-contest.com/visualizer.php?game_id=6262315 (very cowardly! I think you can prove that it will never win a match, though) 2010-10-27T02:17:50 *** amstan_ has joined #aichallenge 2010-10-27T02:17:50 *** ChanServ sets mode: +o amstan_ 2010-10-27T02:18:08 *** onetwo3 has quit IRC (Ping timeout: 265 seconds) 2010-10-27T02:19:10 *** amstan_ has quit IRC (Remote host closed the connection) 2010-10-27T02:19:15 *** amstan_ has joined #aichallenge 2010-10-27T02:19:15 *** ChanServ sets mode: +o amstan_ 2010-10-27T02:19:53 TravisD: who are you in there? red? 2010-10-27T02:20:06 TravisD: sure it will just wait till you get down to the bots that always crash on first turn :P 2010-10-27T02:20:30 *** amstan has quit IRC (Ping timeout: 252 seconds) 2010-10-27T02:20:32 janzert: there's this epic visualizer upgrade i've seen 2010-10-27T02:20:38 amstan_: red 2010-10-27T02:21:01 amstan_: too bad you don't have that irc-nick to account bot service up and running :P 2010-10-27T02:21:11 *** amstan_ is now known as amstan 2010-10-27T02:21:12 amstan_: I think I saw that, had all sorts of information 2010-10-27T02:21:35 amstan: I thought it was too noisy for the main site though 2010-10-27T02:21:41 if it's the same one 2010-10-27T02:21:48 there's a few good features 2010-10-27T02:22:28 *** Titankiller has quit IRC (Ping timeout: 276 seconds) 2010-10-27T02:22:42 *** Appleman1234 has quit IRC (Ping timeout: 252 seconds) 2010-10-27T02:25:21 janzert: this one: http://spz.nofate.com/planetwars/visualizer/?game_id=6136676&server=Official%20Apparently 2010-10-27T02:25:21 amstan: do you have a link? 2010-10-27T02:25:21 oh. 2010-10-27T02:25:21 janzert: you can even click in the history bar thingy to go to that time 2010-10-27T02:25:36 yeah, that is nice 2010-10-27T02:25:38 janzert: it scrapes games from the main server to 2010-10-27T02:25:58 do away with all the extra planet numbers and it would probably work for the main site fine 2010-10-27T02:26:07 yeah 2010-10-27T02:26:49 3 of the top 4 usernames on tcp are going to expire in under an hour 2010-10-27T02:29:38 janzert: apparently this is just the latest pull from the repo 2010-10-27T02:30:01 oh really 2010-10-27T02:30:26 where does that repo live? 2010-10-27T02:30:26 http://ai-contest.com/forum/viewtopic.php?f=18&t=937&p=6136&hilit=http://spz.nofate.com/#p6136 2010-10-27T02:30:37 *** irchs has quit IRC (Quit: irchs) 2010-10-27T02:30:38 *** Appleman1234 has joined #aichallenge 2010-10-27T02:30:39 danielvf's visualizer repo 2010-10-27T02:32:11 *** sigh has quit IRC (Remote host closed the connection) 2010-10-27T02:32:13 must be sighs branch going by commit messages 2010-10-27T02:33:27 *** hellman has quit IRC (Ping timeout: 245 seconds) 2010-10-27T02:34:12 *** Palmik has quit IRC (Remote host closed the connection) 2010-10-27T02:35:15 *** Titankiller has joined #aichallenge 2010-10-27T02:35:28 *** Appleman1234 has quit IRC (Ping timeout: 276 seconds) 2010-10-27T02:36:56 *** wvdschel has joined #aichallenge 2010-10-27T02:36:56 *** wvdschel has joined #aichallenge 2010-10-27T02:37:32 *** wvdschel has left #aichallenge 2010-10-27T02:43:16 *** Utkarsh has quit IRC (Ping timeout: 252 seconds) 2010-10-27T02:44:37 up and down arrows dont work for me as long as the page has a scrollbar and what is the purpose of highlighting a single planet with a mouse hover? 2010-10-27T02:46:09 otherwise, I really like the growth rate display and that you can click on the timeline to see what was going on there! 2010-10-27T02:46:28 +the total production of course 2010-10-27T02:46:36 *** perror has joined #aichallenge 2010-10-27T02:47:52 *** antimatroid has joined #aichallenge 2010-10-27T02:48:19 *** antimatroid1 has quit IRC (Ping timeout: 276 seconds) 2010-10-27T02:49:51 *** Utkarsh has joined #aichallenge 2010-10-27T02:55:39 *** Palmik has joined #aichallenge 2010-10-27T02:56:34 *** Commandir has joined #aichallenge 2010-10-27T02:59:47 *** Appleman1234 has joined #aichallenge 2010-10-27T03:03:28 *** TravisD has quit IRC (Read error: Connection reset by peer) 2010-10-27T03:06:23 *** narnach has quit IRC (Quit: Leaving.) 2010-10-27T03:07:02 *** narnach has joined #aichallenge 2010-10-27T03:10:59 *** hellman has joined #aichallenge 2010-10-27T03:12:07 Top 10 players: rsergio(3796), dmj111(3769), protocolocon(3737), bix0r4ever(3702), Hazard(3688), felixcoto(3677), Slin-.-(3668), jambachili(3661), bocsimacko(3639), sin_sun(3629) 2010-10-27T03:12:15 *** seisatsu has joined #aichallenge 2010-10-27T03:13:04 *** boegel has joined #aichallenge 2010-10-27T03:14:22 *** antimatroid1 has joined #aichallenge 2010-10-27T03:17:09 *** antimatroid has quit IRC (Ping timeout: 252 seconds) 2010-10-27T03:20:32 *** aerique has joined #aichallenge 2010-10-27T03:22:13 *** hellman has quit IRC (Ping timeout: 264 seconds) 2010-10-27T03:25:31 *** mega1 has quit IRC (Ping timeout: 276 seconds) 2010-10-27T03:27:55 *** mega1 has joined #aichallenge 2010-10-27T03:32:13 *** Accoun has joined #aichallenge 2010-10-27T03:41:16 mega1: you lost a game, shock horror 2010-10-27T03:41:20 everyone lost their bets 2010-10-27T03:42:53 it really stank on that map 2010-10-27T03:43:23 @repo 2010-10-27T03:43:24 lavalamp_: repo = http://code.google.com/p/ai-contest/ 2010-10-27T03:44:47 *** hellman has joined #aichallenge 2010-10-27T03:46:07 mega1: we're all disappointed in you 2010-10-27T03:46:10 :( 2010-10-27T03:46:22 xD 2010-10-27T03:46:44 * raedwulf goes back to writing his uber(good/bad) bot 2010-10-27T03:48:54 *** TheMacca has quit IRC (Quit: Page closed) 2010-10-27T03:51:01 * raedwulf looks at his time budget to play new games. 2010-10-27T03:51:09 * raedwulf has -100 hours to play games this week. 2010-10-27T03:51:13 *** Utkarsh has quit IRC (Ping timeout: 255 seconds) 2010-10-27T03:54:37 *** vizier has joined #aichallenge 2010-10-27T03:56:42 *** Utkarsh has joined #aichallenge 2010-10-27T04:05:26 *** iris1 has quit IRC (Ping timeout: 265 seconds) 2010-10-27T04:12:07 Top 10 players: rsergio(3808), dmj111(3770), protocolocon(3738), bix0r4ever(3732), bocsimacko(3730), Hazard(3688), felixcoto(3687), Slin-.-(3676), jambachili(3671), sin_sun(3653) 2010-10-27T04:19:36 *** jesionaj has joined #aichallenge 2010-10-27T04:22:03 woah new names 2010-10-27T04:33:38 *** Naktibalda has joined #aichallenge 2010-10-27T04:36:04 *** hellman_ has joined #aichallenge 2010-10-27T04:36:13 *** hellman has quit IRC (Ping timeout: 276 seconds) 2010-10-27T04:38:02 *** narnach1 has joined #aichallenge 2010-10-27T04:38:02 *** narnach1 is now known as narnach_mb 2010-10-27T04:42:39 *** vizier has quit IRC (Ping timeout: 265 seconds) 2010-10-27T04:54:06 *** RobotCaleb has quit IRC (Read error: Operation timed out) 2010-10-27T04:59:59 *** hellman_ has quit IRC (Ping timeout: 265 seconds) 2010-10-27T05:12:08 Top 10 players: rsergio(3835), dmj111(3792), bocsimacko(3768), protocolocon(3752), felixcoto(3704), Hazard(3696), Slin-.-(3687), jambachili(3678), sin_sun(3659), rs5147(3654) 2010-10-27T05:14:25 *** Bobng has joined #aichallenge 2010-10-27T05:16:16 *** hellman_ has joined #aichallenge 2010-10-27T05:16:24 *** Florian95 has joined #aichallenge 2010-10-27T05:17:09 *** Bobng has quit IRC (Client Quit) 2010-10-27T05:19:43 *** Itkovian has quit IRC (Ping timeout: 276 seconds) 2010-10-27T05:20:48 @top200 2010-10-27T05:20:49 Frontier: I'm sorry Dave, err Frontier; I cannot 'top200'. 2010-10-27T05:22:00 *** Utkarsh has quit IRC (Ping timeout: 252 seconds) 2010-10-27T05:22:59 *** jaspervdj has joined #aichallenge 2010-10-27T05:27:28 *** Utkarsh has joined #aichallenge 2010-10-27T05:37:05 *** SamothRefoe has joined #aichallenge 2010-10-27T05:37:21 Greetings! 2010-10-27T05:39:24 I'm logged in from work on my mobile phone and am thus likely to miss replies, please don't hod it against me! ;) 2010-10-27T05:41:23 I was wondering if anyone had any news regarding possible Scala support, and what more I could do to make it more likely to happen! 2010-10-27T05:45:36 i would like to play galcon as turn based 2010-10-27T05:45:50 human can 30 seconds time to think own move 2010-10-27T05:47:59 rabidus_: you can 2010-10-27T05:48:06 agaist your own bot too 2010-10-27T05:48:51 wut, how? 2010-10-27T05:49:02 http://www.ai-contest.com/forum/viewtopic.php?f=18&t=751 2010-10-27T05:49:31 its not polished... 2010-10-27T05:49:34 but it works 2010-10-27T05:49:42 mostly 2010-10-27T05:49:44 ;) 2010-10-27T05:49:48 hmm.. wow 2010-10-27T05:49:53 is there human vs human version? 2010-10-27T05:50:25 I didn't do one... no. However if you look at the code, such a thing could be done quite easily 2010-10-27T05:50:41 depending on how/if you want a network layer 2010-10-27T05:50:59 it would be fantastic if it is like tcp server 2010-10-27T05:51:24 it would be cool if you could challenge the bots on the official server to a human match 2010-10-27T05:51:32 yep 2010-10-27T05:51:33 :) 2010-10-27T05:51:34 although that would be awesome for debugging actually 2010-10-27T05:51:44 "play against THIS ONE!" 2010-10-27T05:52:02 I used my play human for debug.. that was the motive 2010-10-27T05:52:27 but it could easily be used to play on TCP... but without long timeouts... its very hard to win 2010-10-27T05:53:08 I think humans will suck at this game for interface reasons-- it's so hard to direct all your ships 2010-10-27T05:53:38 yea 2010-10-27T05:53:52 even bad bots kick my human arse... 2010-10-27T05:54:04 as bender would say... stupid humans 2010-10-27T05:54:06 ;) 2010-10-27T05:54:22 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2010-10-27T05:55:18 galcon.com/flash kicks my arse and .... it is not so good @ ai-contest.com 2010-10-27T05:55:45 well... there's a reason we work on artificial intelligence... 2010-10-27T05:56:35 "natural" intelligence is harder to find ;) 2010-10-27T05:56:57 *** iris1 has joined #aichallenge 2010-10-27T05:57:45 I did beat a few levels of the flash game, but I'm sure my bot would fare better, unpoished as it is! 2010-10-27T05:59:35 *unpolished 2010-10-27T06:01:08 *** Utkarsh has joined #aichallenge 2010-10-27T06:06:49 *** hellman_ has quit IRC (Ping timeout: 255 seconds) 2010-10-27T06:08:40 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2010-10-27T06:10:44 oh, delt0r___ , nice work on the "all in one" tool... simple to use and a nice TCP testing interface... 2010-10-27T06:12:08 Top 10 players: rsergio(3805), dmj111(3800), bocsimacko(3778), protocolocon(3746), felixcoto(3715), Hazard(3711), jambachili(3691), george(3684), Arekku(3683), bix0r4ever(3681) 2010-10-27T06:12:14 I was tired of beating on the example bots, t'was good to take a beating on TCP, shows me were I still need to work... 2010-10-27T06:13:43 *** Utkarsh has joined #aichallenge 2010-10-27T06:15:23 contestbot: later tell janzert I modified the server setup script to instal go. Also changed order of installed things, it tried to use curl before installing it. See here: https://bitbucket.org/lavalamp/pw-server-code/src/tip/game_server_build.sh 2010-10-27T06:15:23 lavalamp_: OK 2010-10-27T06:18:13 *** hellman has joined #aichallenge 2010-10-27T06:19:31 http://72.44.46.68/canvas?game_id=728399 2010-10-27T06:19:33 can steal win 2010-10-27T06:30:16 *** xlk_ has quit IRC (Quit: Page closed) 2010-10-27T06:34:15 *** Utkarsh has quit IRC (Ping timeout: 240 seconds) 2010-10-27T06:37:02 SamothRefoe: np 2010-10-27T06:41:57 *** smellyhippy2 has joined #aichallenge 2010-10-27T06:43:34 *** Titankiller has quit IRC (Read error: Connection reset by peer) 2010-10-27T06:43:59 *** Utkarsh has joined #aichallenge 2010-10-27T06:44:52 *** smellyhippy has quit IRC (Ping timeout: 265 seconds) 2010-10-27T06:49:58 *** Bobng has joined #aichallenge 2010-10-27T06:50:57 *** Olathe has joined #aichallenge 2010-10-27T07:02:25 *** hellman has quit IRC (Ping timeout: 240 seconds) 2010-10-27T07:03:04 *** Bobng has quit IRC (Quit: Leaving) 2010-10-27T07:05:44 *** seisatsu has quit IRC (Read error: Connection reset by peer) 2010-10-27T07:05:58 *** seisatsu has joined #aichallenge 2010-10-27T07:12:08 Top 10 players: bocsimacko(3856), dmj111(3808), rsergio(3802), protocolocon(3754), felixcoto(3722), Hazard(3719), jambachili(3703), george(3701), Arekku(3693), sin_sun(3684) 2010-10-27T07:24:59 *** Florian95 has quit IRC (Ping timeout: 265 seconds) 2010-10-27T07:27:00 *** irchs has joined #aichallenge 2010-10-27T07:29:38 *** Titankiller has joined #aichallenge 2010-10-27T07:30:19 *** hellman has joined #aichallenge 2010-10-27T07:50:12 *** lavalamp_ has quit IRC (Quit: Page closed) 2010-10-27T07:55:26 *** Florian95 has joined #aichallenge 2010-10-27T07:55:27 *** SamothRefoe has quit IRC (Read error: Connection reset by peer) 2010-10-27T07:57:57 *** SamothRefoe has joined #aichallenge 2010-10-27T07:58:49 *** Appleman1234 has quit IRC (Ping timeout: 264 seconds) 2010-10-27T07:59:51 *** Palmik has quit IRC (Remote host closed the connection) 2010-10-27T08:00:32 *** SamothRefoe has quit IRC (Read error: Connection reset by peer) 2010-10-27T08:05:42 *** irchs has quit IRC (Quit: irchs) 2010-10-27T08:07:12 *** hellman has quit IRC (Ping timeout: 245 seconds) 2010-10-27T08:12:09 Top 10 players: bocsimacko(3859), rsergio(3804), dmj111(3794), protocolocon(3753), Hazard(3736), felixcoto(3728), george(3715), jambachili(3705), bix0r4ever(3697), luksian(3692) 2010-10-27T08:13:18 *** nullkuhl has joined #aichallenge 2010-10-27T08:13:58 i am trying to do a java submission ( its my 3rd submission ) however am getting a compilation error, although it compiles and executes fine on my own machine, can any one help ? 2010-10-27T08:25:53 *** Palmik has joined #aichallenge 2010-10-27T08:28:19 fixed it, i had to fix some warnings regarding unsafe operations, although they aren't really compilation errors. 2010-10-27T08:28:25 *** nullkuhl has quit IRC (Remote host closed the connection) 2010-10-27T08:32:05 *** hellman has joined #aichallenge 2010-10-27T08:50:44 *** Titankiller has quit IRC (Ping timeout: 240 seconds) 2010-10-27T08:54:55 *** boegel has quit IRC (Read error: Operation timed out) 2010-10-27T08:56:39 morning 2010-10-27T08:58:55 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-10-27T08:59:01 *** narnach1 has joined #aichallenge 2010-10-27T08:59:01 *** narnach1 is now known as narnach_mb 2010-10-27T09:01:10 morning 2010-10-27T09:01:13 by a minute :P 2010-10-27T09:01:23 my bot is so retarded on the first move 2010-10-27T09:01:40 i either expand too much or too little, i can't get it to expand just when it can back it up properly 2010-10-27T09:04:28 follow the arrows: http://landoflisp.com/ 2010-10-27T09:09:14 *** smellyhippy2 is now known as smellyhippy 2010-10-27T09:12:10 Top 10 players: bocsimacko(3872), rsergio(3814), dmj111(3803), protocolocon(3763), felixcoto(3747), Hazard(3742), george(3726), luksian(3719), jambachili(3709), bix0r4ever(3707) 2010-10-27T09:14:45 *** delt0r has joined #aichallenge 2010-10-27T09:15:22 *** delt0r___ has quit IRC (Ping timeout: 255 seconds) 2010-10-27T09:22:47 *** Azrathud has quit IRC (Read error: Operation timed out) 2010-10-27T09:37:39 hmm 2010-10-27T09:38:02 how lame , i just added some new stuff to my bot and first match against ragebot 2010-10-27T09:38:03 bah 2010-10-27T09:38:32 haha 2010-10-27T09:38:51 my bot seems to be getting worse again :| 2010-10-27T09:38:56 and second game was a loss 2010-10-27T09:38:58 lol 2010-10-27T09:39:11 my bot is actually doing not too bad now 2010-10-27T09:39:29 was holding 400+ elo most of yesterday on tcp and can beat accoun's bot most of the time 2010-10-27T09:40:08 *** phreeza has joined #aichallenge 2010-10-27T09:40:40 trying to see if i can make it a bit more smarter when it comes to neutral acquisition 2010-10-27T09:41:18 i want first place :( 2010-10-27T09:41:26 it's so far out of my reach aha 2010-10-27T09:41:39 after watching bockismacko's bot 2010-10-27T09:41:46 i doubt i will get anywhere near that lol 2010-10-27T09:41:55 i will settle for top 20/30 2010-10-27T09:47:03 *** cfaftw has joined #aichallenge 2010-10-27T09:48:02 *** yasith has quit IRC (Ping timeout: 245 seconds) 2010-10-27T09:49:24 *** Bobng has joined #aichallenge 2010-10-27T09:50:32 *** hellman has quit IRC (Ping timeout: 265 seconds) 2010-10-27T09:54:07 *** Sarin has joined #aichallenge 2010-10-27T09:58:17 *** boegel has joined #aichallenge 2010-10-27T10:00:09 *** yasith has joined #aichallenge 2010-10-27T10:02:32 *** SamothRefoe has joined #aichallenge 2010-10-27T10:03:11 *** hellman has joined #aichallenge 2010-10-27T10:03:17 OMG 2010-10-27T10:03:26 did my bot kill the cloud servers?!? 2010-10-27T10:03:36 *** Azrathud has joined #aichallenge 2010-10-27T10:03:46 aha! it's your fault!!! 2010-10-27T10:03:55 * boegel hides in a corner 2010-10-27T10:04:58 haha 2010-10-27T10:04:58 *** iNtERrUpT has joined #aichallenge 2010-10-27T10:06:30 ? 2010-10-27T10:06:52 *** boegel has quit IRC (Remote host closed the connection) 2010-10-27T10:07:00 i don't get it 2010-10-27T10:07:12 run harder 2010-10-27T10:08:03 *** iNtERrUpT has quit IRC (Remote host closed the connection) 2010-10-27T10:12:11 Top 10 players: bocsimacko(3863), rsergio(3804), dmj111(3793), protocolocon(3753), felixcoto(3736), Hazard(3725), george(3710), luksian(3710), jambachili(3704), bix0r4ever(3696) 2010-10-27T10:12:26 *** Palmik has quit IRC (Read error: Operation timed out) 2010-10-27T10:12:55 *** jesionaj has quit IRC (Ping timeout: 240 seconds) 2010-10-27T10:16:10 i bow before boxsimacko's bot 2010-10-27T10:16:15 bocsimacko* 2010-10-27T10:16:19 *** Florian95 has quit IRC (Ping timeout: 276 seconds) 2010-10-27T10:16:40 *** JamesMG has joined #aichallenge 2010-10-27T10:17:10 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-10-27T10:17:41 *** Palmik has joined #aichallenge 2010-10-27T10:19:20 *** SamothRefoe has quit IRC (Read error: Connection reset by peer) 2010-10-27T10:24:03 *** SamothRefoe has joined #aichallenge 2010-10-27T10:25:42 *** boegel has joined #aichallenge 2010-10-27T10:29:14 cfaftw: everyone does 2010-10-27T10:29:27 but not willingly... his bot forces me to bow 2010-10-27T10:29:40 i'm compelled to obey 2010-10-27T10:30:03 and i love that it's written in Lisp 2010-10-27T10:30:27 figures the crazy Lisp AI guy would come in at the last minute and humble everyone :) 2010-10-27T10:30:48 there's a month elft still, not quite the last minute :) 2010-10-27T10:30:52 (last minute meaning one month left :) 2010-10-27T10:31:22 how's your bot coming, James? 2010-10-27T10:31:47 not well, it only works if I leave in a bug in my move generation code 2010-10-27T10:31:59 otherwise it generates too many possible moves and times out :) 2010-10-27T10:32:41 I might go rewrite my basic heuristics bot this weekend and step away from minimax 2010-10-27T10:33:20 ah 2010-10-27T10:33:27 i'm still struggling with neutral selection 2010-10-27T10:34:04 which... is pretty important -_- 2010-10-27T10:34:17 haha, yeah 2010-10-27T10:35:03 yeah i think everyone is 2010-10-27T10:35:09 except the above mentioned person :P 2010-10-27T10:35:34 i tried to come up with a simple rock paper scissors example yesterday and even that wasn't so straigth forward 2010-10-27T10:35:39 http://i.imgur.com/p2XGG.png 2010-10-27T10:35:42 that was what i knocked up 2010-10-27T10:36:06 did anyone go through lavalamp's scenario on there? 2010-10-27T10:36:15 antimatroid1: can you elaborate on this? 2010-10-27T10:36:36 I think he said 46 to the enemy planet, 27 to each neutral, but I didn't spend any time coming up with a counter to that 2010-10-27T10:36:36 antimatroid1: what's so rock paper scissors about this example? 2010-10-27T10:36:55 i created it with it being rock paper scissors if you consider each player just picking "wait", "expand to my neutral", "expand to their neutral" 2010-10-27T10:37:05 if you consider JUST those moves from each player, it's rock paper scissors 2010-10-27T10:37:15 i'm not sure if there are any strategies that dominate those though 2010-10-27T10:37:29 yeah i never got around to it, hold on i'll run th enumbers 2010-10-27T10:37:55 it would be easy to counter on turn2, but that would depend on you waiting on turn1 I think 2010-10-27T10:38:16 that counts as dominating it 2010-10-27T10:39:17 so you want to send 24 to their neutral turn 1 and 24 to your neutral turn 2 2010-10-27T10:39:45 leaving you with 54 + 10*5 = 104 when they get to you 2010-10-27T10:39:54 which is more than what they sent at you, so you'd win 2010-10-27T10:40:05 so it's not a "perfect" strategy 2010-10-27T10:41:48 i think closer starts (but not too close) are most interesting 2010-10-27T10:41:54 i also do quite badly on them so far 2010-10-27T10:41:59 I do struggle too with neutral selection and will lose most games with a dominant middle planet. 2010-10-27T10:42:04 certainly have a very "sudden death" feeling to them 2010-10-27T10:42:21 mega1: what are your thoughts on the above map? 2010-10-27T10:43:30 attack the closer neutral with 100 2010-10-27T10:43:50 if they attack that planet first move you lose 2010-10-27T10:44:05 i agree it's better on average with imperfect bots though :P 2010-10-27T10:44:29 then compute the mixed strategy 2010-10-27T10:44:35 as in nash equilibrium. 2010-10-27T10:44:43 as in good luck :P 2010-10-27T10:45:06 is that why you're doing so well :P 2010-10-27T10:45:15 big numerical msne approximations? :P 2010-10-27T10:45:27 no, nothing special I swear :-) 2010-10-27T10:45:51 I only have plans. 2010-10-27T10:45:59 don't say that :P 2010-10-27T10:46:03 you're already miles ahead 2010-10-27T10:46:05 that's why "Simplex" scares me so 2010-10-27T10:46:21 i don't think he's using simplex 2010-10-27T10:46:28 he was in last contest as same name 2010-10-27T10:46:35 really? 2010-10-27T10:46:38 and someone else said they tried and was far too slow for them 2010-10-27T10:46:48 was it jmcarthur? 2010-10-27T10:46:55 i tried doing a dynamic programming approach a while ago but that was far too slow 2010-10-27T10:46:57 not sure 2010-10-27T10:48:29 what would be cool about simplex is if you found a feasible solution that you wanted to use, you can iterate towards the optimal getting ever closer to it up to your time limit 2010-10-27T10:51:54 *** fairy44 has quit IRC (Quit: leaving) 2010-10-27T10:55:39 antimatroid1: in some linear subset of this problem 2010-10-27T10:55:42 that's the rub 2010-10-27T10:56:30 I haven't been able to think of a promising linear formulation 2010-10-27T10:56:40 be it the whole game or a subproblem. 2010-10-27T10:58:24 yeah there's way too much information 2010-10-27T11:03:20 *** fairy44 has joined #aichallenge 2010-10-27T11:04:03 *** fairy44 has joined #aichallenge 2010-10-27T11:06:14 *** aerique has quit IRC (Quit: ...) 2010-10-27T11:06:32 *** lavalamp_ has joined #aichallenge 2010-10-27T11:10:55 *** hellman has quit IRC (Ping timeout: 240 seconds) 2010-10-27T11:12:10 Top 10 players: bocsimacko(3868), rsergio(3808), dmj111(3798), protocolocon(3760), felixcoto(3738), Hazard(3729), bix0r4ever(3724), george(3716), jambachili(3712), luksian(3708) 2010-10-27T11:19:15 *** Azrathud has quit IRC (Ping timeout: 240 seconds) 2010-10-27T11:24:20 *** hellman has joined #aichallenge 2010-10-27T11:24:51 *** Sarin has quit IRC (Remote host closed the connection) 2010-10-27T11:30:01 *** fairy44 has quit IRC (Quit: leaving) 2010-10-27T11:30:24 *** fairy44 has joined #aichallenge 2010-10-27T11:31:10 hmmm, i'm throwing it back into a minimax tree 2010-10-27T11:31:16 i don't think it's ready for it yet though 2010-10-27T11:33:29 This is a new low for me, I think: http://72.44.46.68/canvas?game_id=732199 2010-10-27T11:34:04 considering my old code did close starts fairly well 2010-10-27T11:35:02 JamesMG, ouch! 2010-10-27T11:35:09 haha, awesome 2010-10-27T11:35:10 the "better" i make my bot the worse it gets a lot of the time 2010-10-27T11:35:14 it's very frustrating 2010-10-27T11:35:31 yeah, thinks that you know should work, but in practice they don't seem to 2010-10-27T11:35:33 things* 2010-10-27T11:35:48 I thought of a way to set up regression tests for for my eval function 2010-10-27T11:35:53 oh yeah? 2010-10-27T11:36:11 yeah, I'll feed it two positions and test that it considers the correct one superior 2010-10-27T11:36:46 I'd like to gather a mess of "turning point" game states for that at some point 2010-10-27T11:37:31 need a decent number of them due to my tendancy to modify algorithms to work perfectly on my example cases... only 2010-10-27T11:37:49 my bot is going to be made of win if I can get it to stop throwing its ships at a neutral right before it loses a planet 2010-10-27T11:39:07 i just set up slack ships i can send from each planet at each point in time 2010-10-27T11:39:52 antimatroid1, I do that too 2010-10-27T11:40:15 *** d3c1ph3r has quit IRC (Ping timeout: 265 seconds) 2010-10-27T11:40:22 problem is if the defense goal fails, it frees up all those ships that almost defended the planet to go do something incredibly stupid 2010-10-27T11:40:43 this is why my minimax bot is not running right now... 2010-10-27T11:41:04 i just through my stuff back into a minimax tree, but it wont do very well 2010-10-27T11:41:08 i did it in like 10 minutes 2010-10-27T11:41:24 (it was already set up previously, but i went back to heuristics cause it was so buggy) 2010-10-27T11:42:12 currently my "minimax" bot actually does something more like running a tournament between the various goals my generator comes up with 2010-10-27T11:42:43 and it's not even a minimax bot, I think it was janzert that said the term is "recursive payoff tables" 2010-10-27T11:43:18 payoff matrix is kinda what I'm leaning towards now 2010-10-27T11:43:49 yeah... and mine's not very recursive since it only gets one level deep... 2010-10-27T11:44:06 atm i'm making individual moves for each player and not incrementing states 2010-10-27T11:44:11 it doesn't work very well though 2010-10-27T11:44:42 when I tried to duplicate my decision process to predict the next enemy move 2010-10-27T11:44:52 it played itself pretty well, but nothing else acts like it 2010-10-27T11:45:02 it certainly wasn't me saying that simplex was too slow 2010-10-27T11:45:19 i was planning to use simplex for this contest too before i decided to drop out 2010-10-27T11:45:30 and in fact had already implemented it 2010-10-27T11:46:09 *** Florian95 has joined #aichallenge 2010-10-27T11:46:23 antimatroid1: your "iterate toward optimal" idea was pretty similar to what my plan was, actually 2010-10-27T11:46:26 why did you drop out? 2010-10-27T11:46:30 no time 2010-10-27T11:46:34 i have things i need to do 2010-10-27T11:46:45 fair enough 2010-10-27T11:46:47 still like to hang out in the channel, though? 2010-10-27T11:46:50 i have no job. i should be job hunting more aggressively. i also have been working on a web site that might make me some money, etc. 2010-10-27T11:46:52 i don't have time either, but i'm hooked 2010-10-27T11:46:53 yeah 2010-10-27T11:46:53 *** narnach1 has joined #aichallenge 2010-10-27T11:46:53 *** narnach1 is now known as narnach_mb 2010-10-27T11:47:33 *** dmj111 has quit IRC (Ping timeout: 252 seconds) 2010-10-27T11:47:42 cfaftw: i want to keep up with it. i just don't think by bot is a good priority for me right now 2010-10-27T11:47:51 *my 2010-10-27T11:48:01 if you won you might get a job because of it :P 2010-10-27T11:48:05 jmcarthur, i totally understand 2010-10-27T11:48:06 kill two birds with the one stone 2010-10-27T11:48:11 true, but there are a lot of contestants 2010-10-27T11:48:15 only one can win 2010-10-27T11:48:19 i have no time, but im working on my bot ^^ 2010-10-27T11:48:23 well, in theory anyway :) 2010-10-27T11:48:26 my friend should be about shortly though 2010-10-27T11:48:28 *** lavalamp_ has quit IRC (Ping timeout: 265 seconds) 2010-10-27T11:48:41 we were meant to do this as a team then he got busy with uni 2010-10-27T11:48:42 lol 2010-10-27T11:48:47 raedwulf, i forgot that you've been working on this by yourself as a "team" :P 2010-10-27T11:49:08 antimatroid1: it's not that i'm having trouble finding businesses that would want to hire me. it's that i'm having trouble finding businesses that i want to work for 2010-10-27T11:49:17 i have enough savings to be picky for a good while 2010-10-27T11:49:27 that's key 2010-10-27T11:49:56 yeah, i plan to be particularly picky 2010-10-27T11:50:00 my wife is going to grad school next fall... i just hope i can be picky when selecting a new employer once we move 2010-10-27T11:50:03 so picky that i'll probably just end up in academia 2010-10-27T11:50:06 so winning this contest might get me a few interviews, but not necessarily with companies i really want to deal with 2010-10-27T11:50:09 ^^ 2010-10-27T11:50:12 haha, me too antimatroid1 2010-10-27T11:50:26 i have a few years to decide such things yet though 2010-10-27T11:50:35 lately i've been mostly looking at research positions at unis 2010-10-27T11:50:44 that's my job :P 2010-10-27T11:50:49 yea thats always a lot nicer hehe :P 2010-10-27T11:50:51 but i very rarely work as i never find the time 2010-10-27T11:50:54 i feel like even being in the top 10 is resume-worthy... considering the scale of the contest 2010-10-27T11:50:55 my boss is nice like that 2010-10-27T11:50:57 my last job was commercial research, but still researchy 2010-10-27T11:51:01 lmao antimatroid1 2010-10-27T11:51:07 lol: http://ai-contest.com/forum/viewtopic.php?f=19&t=1014 2010-10-27T11:51:33 research is so much more fun than the usual deserialize-serialize-rejigger-send-receive-and-that's-all programming 2010-10-27T11:51:52 amstan, i am winner! 2010-10-27T11:52:12 the research i am paid to do isn't so fun (macroeconomics/programming lots of stats crap) 2010-10-27T11:52:23 oh alex champandard, i wrote an email to him a couple of years ago or so about his voxel terrain 2010-10-27T11:52:24 lmao 2010-10-27T11:52:46 voxel terrain? where? sounds like minecraft 2010-10-27T11:52:58 he wrote an article on gamedev 2010-10-27T11:52:59 lemme find it 2010-10-27T11:53:11 that's a terrible "advanced guide" 2010-10-27T11:53:29 I'll talk about what I did with the official AiGameDev.com bot to get into the Top 10%, 2010-10-27T11:53:34 i love voxel engines in theory 2010-10-27T11:53:35 10% is a lot of ppl 2010-10-27T11:54:05 *** Itkovian has joined #aichallenge 2010-10-27T11:54:07 he also went to my university 2010-10-27T11:54:45 well i can't find the article, but its called "terravox" 2010-10-27T11:54:48 written many many years ago 2010-10-27T11:55:38 sorry it wasn't gamedev, it was flipcode: http://www.flipcode.com/archives/Realtime_Voxel_Landscape_Engines-Part_4_Level_of_Detail_and_Optimisation.shtml 2010-10-27T11:55:49 well thats not the first page but anyways xD 2010-10-27T11:56:16 actually a couple... come to think of it, it might have been 3-4 years ago lol... 2010-10-27T11:56:51 ah flipcode 2010-10-27T11:56:54 i miss that site 2010-10-27T11:57:06 it's only archives now 2010-10-27T11:57:12 yea 2010-10-27T11:57:28 * raedwulf goes looks for a maths book 2010-10-27T11:58:10 a couple months ago i met the guy that wrote the Mostly Harmless Algorithms article series 2010-10-27T11:58:35 he wrote the graphics engine for that 1997 blade runner game. it was a software-only voxel engine. very impressive, IMO 2010-10-27T11:58:49 just to tie voxel engines and flipcode in another way :) 2010-10-27T11:59:16 :D 2010-10-27T12:01:13 this was stinking 1997 on a cpu http://www.youtube.com/watch?v=V9-hTaFqh_Q 2010-10-27T12:01:56 dang i should have linked to a cinematic 2010-10-27T12:01:57 oh well 2010-10-27T12:05:44 jpeck, If you can hold the ranking at the end of the contest -- yes! I have another version of the bot that's roughly 2x better and it should be ready for submitting in a few days hopefully ;-) Alex 2010-10-27T12:05:52 dated oct 12 2010-10-27T12:06:19 last submit oct 6 2010-10-27T12:06:42 he's going to have to give a lot of free memberships 2010-10-27T12:07:12 *** bhasker has quit IRC (Quit: bhasker) 2010-10-27T12:07:35 maybe he'll resubmit with a #1 placing bot soon 2010-10-27T12:08:06 he should stop making promises and do it 2010-10-27T12:08:59 I wonder if he was surprised by the level of competition 2010-10-27T12:11:41 the level of competition is stumping me :| 2010-10-27T12:12:11 Top 10 players: bocsimacko(3917), GreenTea(3914), rsergio(3828), dmj111(3814), luksian(3761), felixcoto(3750), protocolocon(3748), jambachili(3734), bix0r4ever(3726), george(3722) 2010-10-27T12:13:13 I'm hiding from the competition behind a flawed approach :) 2010-10-27T12:13:18 *** Bobng has quit IRC (Quit: Leaving) 2010-10-27T12:13:41 you're all getting owned by a bot named "george" 2010-10-27T12:14:12 how does it feel to loose to "george"? :D lol 2010-10-27T12:14:31 not as bad as "bob" 2010-10-27T12:14:48 can I change my name to John.Smith? 2010-10-27T12:15:04 Agent.Smith 2010-10-27T12:15:05 :P 2010-10-27T12:15:11 haha, that'd actually work 2010-10-27T12:15:13 *** Bobng has joined #aichallenge 2010-10-27T12:15:32 i speaketh of the devil and he cometh :o 2010-10-27T12:15:34 no name changes, i'm tired of those 2010-10-27T12:16:35 hmmm i should go home now 2010-10-27T12:17:02 you know you are lazy when you're in the library and can't be bothered to leave... 2010-10-27T12:17:47 what's a library? 2010-10-27T12:18:01 :p 2010-10-27T12:18:33 *** seisatsu has quit IRC (Quit: Leaving) 2010-10-27T12:18:38 lol 2010-10-27T12:18:41 bye! 2010-10-27T12:18:53 back later, once i cycle 5 miles ugh 2010-10-27T12:18:54 >.< 2010-10-27T12:23:23 look out, greenteas new entry is doing very well 2010-10-27T12:23:36 @rankings 2010-10-27T12:23:38 amstan: Top 10 players: bocsimacko(3910), GreenTea(3907), rsergio(3820), dmj111(3806), luksian(3754), felixcoto(3742), protocolocon(3740), jambachili(3726), bix0r4ever(3718), rs5147(3715) 2010-10-27T12:24:12 it hasn't lost yet 2010-10-27T12:24:41 nice! 2010-10-27T12:25:20 *** Bobng has quit IRC (Quit: Leaving) 2010-10-27T12:35:03 this is the best contest ever 2010-10-27T12:35:34 *** Bobng has joined #aichallenge 2010-10-27T12:36:01 cfaftw: yeah, tron was awesome too 2010-10-27T12:36:11 people tend to disagree: http://www.ai-contest.com/forums/viewtopic.php?f=18&t=996#p6616 2010-10-27T12:36:12 amstan: are you guys having another in february? 2010-10-27T12:36:19 and rant about c# at every opportunity 2010-10-27T12:36:24 antimatroid1: yeah, maybe 2010-10-27T12:37:34 oh man... i don't know if i'll have recovered by then 2010-10-27T12:38:47 *** bhasker has joined #aichallenge 2010-10-27T12:41:06 *** Frontier has left #aichallenge 2010-10-27T12:41:39 i do like that it is going for longer this time 2010-10-27T12:42:04 *** Frontier has joined #aichallenge 2010-10-27T12:58:36 me too, for the record 2010-10-27T12:58:43 although i think maybe splitting the difference would be better 2010-10-27T12:59:04 2 months. 1 is too short. 3 seems a bit long, to me 2010-10-27T12:59:25 i dunno, this game can certainly take the 3 months 2010-10-27T12:59:29 considering how much of my productivity in other areas it tends to steal 2010-10-27T12:59:33 all the bots have a lot of improvements to be made yet 2010-10-27T12:59:35 this game could take 3 years 2010-10-27T12:59:40 :D 2010-10-27T12:59:47 but i don't want it to 2010-10-27T13:00:56 i only mean the actual contest deadline. i still think it would be nice to continue accepting entries even after the contest deadline just for fun and casual competition 2010-10-27T13:01:13 yeah 2010-10-27T13:01:21 and people should be able to challenge bots to a dual 2010-10-27T13:01:30 where the human gets as long as they want to make each turn 2010-10-27T13:01:40 could even let them reverse and remake their moves maybe 2010-10-27T13:01:41 oh you mean human vs bot? 2010-10-27T13:01:44 yeah 2010-10-27T13:01:58 eek, reversal means bots either can't be stateful or have to know how to rewind 2010-10-27T13:02:18 my bot resets information up each turn 2010-10-27T13:02:29 you could start my bot at any point in any game and it'd be good to go 2010-10-27T13:02:30 yeah mine wasn't going to do that at all 2010-10-27T13:02:41 mine wasn't, but then it's so messy i decided it was easier 2010-10-27T13:02:48 given i had to process the string anyway 2010-10-27T13:02:50 it was critical to retain information from turn to turn because it was going to be doing a full game analysis the whole time 2010-10-27T13:03:07 ha, there is way too much information for that 2010-10-27T13:03:12 well yeah 2010-10-27T13:03:16 we had a self updating tree for tron 2010-10-27T13:03:23 by "full game" i mean "to completion" not "all possibilities" 2010-10-27T13:03:25 but you could start generating the whole tree for that 2010-10-27T13:03:46 each turn it'd just throw out what wasn't used then keep generating from where it was at 2010-10-27T13:04:05 it was going to be an iterative algorithm that refines the "game plan" over time 2010-10-27T13:04:50 it differed from tree search in that it also refined far future moves almost independently of near future moves 2010-10-27T13:05:11 yeah fair enough 2010-10-27T13:05:23 choice of near future moves would affect the desirability of far future moves, but the search was otherwise independent 2010-10-27T13:05:25 i've considered trying to predict the other bots characteristics, but i can't see it working very well 2010-10-27T13:05:32 yeah i wasn't trying that at all 2010-10-27T13:05:44 i was just searching for an equilibrium, and then it doesn't really matter what the other bot does 2010-10-27T13:05:56 at the moment i'm making future moves at each turn, and making one move per node, so it will in a way look into the future 2010-10-27T13:06:01 i could just update the search with the actual game state as it progresses and it could adapt 2010-10-27T13:06:32 but to throw it into minimax i'm having to only execute moves to be sent at time 0 2010-10-27T13:07:06 i haven't dealt with the mess that is doing minimax with future moves 2010-10-27T13:07:12 i'm certain that tree search is pretty much completely impractical for this 2010-10-27T13:07:31 even with the most awesome pruning or opponent prediction algorithms 2010-10-27T13:07:32 i don't think so, i have built the heuristics in pretty well 2010-10-27T13:07:49 * jmcarthur shrugs 2010-10-27T13:07:59 if you say so. i abandoned tree search pretty early 2010-10-27T13:08:00 i'm basically just picking my moves, throwing each in as a child, then repeating the process 2010-10-27T13:08:17 so it will sort of chain moves together in the tree and see if that works 2010-10-27T13:10:07 basically my idea was to use particle swarm optimization to determine, for each turn of the game, what proportion of my currently available ships do i want to send to each planet. it would be a separate swarm for each player. it would simulate an entire game for each pair of particles, one from each player, making a game matrix. then it solves the game matrix and advances the particles in each 2010-10-27T13:10:09 swarm according to the weights in the result 2010-10-27T13:10:33 then repeat 2010-10-27T13:10:56 so the swarms would tend to converge on an equilibrium 2010-10-27T13:11:41 i wasn't sure how fast i could get my simulation to run. that's why i was trying to so hard to write really efficient code 2010-10-27T13:11:46 it's a lot of simulations! 2010-10-27T13:12:10 it ended up being pretty fast, although i never actually benchmarked it 2010-10-27T13:12:11 Top 10 players: GreenTea(3925), bocsimacko(3879), rsergio(3848), dmj111(3812), felixcoto(3764), protocolocon(3763), luksian(3750), rs5147(3738), Hazard(3734), jambachili(3727) 2010-10-27T13:12:40 then i quit because i was spending too much time on it 2010-10-27T13:13:05 yeah fair enough 2010-10-27T13:13:39 there were lots of little details i had to get right too. the idea itself could be fragile without exactly the right heuristics, etc. 2010-10-27T13:13:47 i've considered trying to do it as a flow problem, but couldn't work out how 2010-10-27T13:13:55 same with a kind of supply/demand type thing 2010-10-27T13:14:18 the most important heuristic being: given that i have this many total ships available and these proportions to send to each planet, from which planets do i send these ships? 2010-10-27T13:15:13 "send to" also could mean "keep on" as well. like if i had 100 ships on planet A and i wanted to send 100 ships to planet A that could be a no-op 2010-10-27T13:16:28 i was also thinking of making it such that the swarm would converge more quickly on turns that are coming up soon than on turns that are still a good distance away and other little tweaks like that 2010-10-27T13:16:47 anyway, i think i have now explained most of how my bot was intended to work here 2010-10-27T13:17:53 the particle update algorithm was also slightly tweaked to favor this mixed strategy approach too. normally partical swarm assumes there is one best position, but that doesn't hold in this case 2010-10-27T13:18:15 so i had a more probabilistic update algorithm 2010-10-27T13:20:15 even if the game turns out to always have pure best strategies i think using mixed strategies in the core of my algorithm was a good idea just because it was an approximation in the first place 2010-10-27T13:20:42 any mixed strategy that comes up as a result of a poor set of games to choose from would be quickly corrected in a future iteration 2010-10-27T13:21:01 but the mixed strategy would still ensure that it doesn't try to converge on something stupid 2010-10-27T13:33:23 antimatroid1: that was my initial approach 2010-10-27T13:33:27 somewhat 2010-10-27T13:33:28 *** lavalamp_ has joined #aichallenge 2010-10-27T13:33:45 i got so far, then realised i was complicating it too much 2010-10-27T13:34:29 meh, i can't think atm, whats the initial branching factor if you start off with 100 ships and have 23 planets 2010-10-27T13:34:44 i.e. how many different ways can you allocate 100 into 23 2010-10-27T13:35:57 122 choose 22 2010-10-27T13:36:44 yeah, i'm not making moves like that 2010-10-27T13:37:04 each node is making like 5 moves max, and each move has a single destination planet/time 2010-10-27T13:37:17 Zannick: ah... how do you work that out? :P 2010-10-27T13:37:56 it's something i was taught as "pirates and gold" 2010-10-27T13:38:10 :P 2010-10-27T13:38:12 *** bhasker_ has joined #aichallenge 2010-10-27T13:38:20 i could go into a lot of depth about it 2010-10-27T13:38:25 oh ok :P 2010-10-27T13:38:37 its 9.41*10^23 2010-10-27T13:38:42 interesting :D 2010-10-27T13:38:43 but basically, if you have k pirates, and n gold, the number of ways to parcel out the gold such that each pirate gets >= 0 pieces, 2010-10-27T13:38:56 it's (n + k - 1) choose (k - 1) 2010-10-27T13:39:03 ah ok 2010-10-27T13:39:11 nice thanks 2010-10-27T13:39:14 sure 2010-10-27T13:39:36 9.41*10^23 is gonna take a while hmm lol :P 2010-10-27T13:39:48 *** bhasker__ has joined #aichallenge 2010-10-27T13:39:48 unless i limit it 2010-10-27T13:40:00 anyways, was just a passing thought 2010-10-27T13:40:52 *** bhasker has quit IRC (Ping timeout: 255 seconds) 2010-10-27T13:42:15 *** bhasker_ has quit IRC (Ping timeout: 240 seconds) 2010-10-27T13:42:15 yes. yes it will. 2010-10-27T13:43:36 *** irchs has joined #aichallenge 2010-10-27T13:43:37 *** Naktibalda has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]) 2010-10-27T13:43:44 I'll come get you before the sun burns out. 2010-10-27T13:45:02 ahhh, i just had a cool idea 2010-10-27T13:45:16 *** irchs has quit IRC (Client Quit) 2010-10-27T13:46:26 *** lavalamp__ has joined #aichallenge 2010-10-27T13:47:06 *** RainCT has joined #aichallenge 2010-10-27T13:47:20 *** irchs has joined #aichallenge 2010-10-27T13:48:20 *** lavalamp_ has quit IRC (Ping timeout: 265 seconds) 2010-10-27T13:48:45 *** lavalamp__ is now known as lavalamp_ 2010-10-27T13:51:16 *** Florian has joined #aichallenge 2010-10-27T13:54:22 *** Florian95 has quit IRC (Ping timeout: 250 seconds) 2010-10-27T13:54:34 *** narnach has quit IRC (Remote host closed the connection) 2010-10-27T13:55:21 *** irchs has quit IRC (Quit: irchs) 2010-10-27T13:55:35 *** lavalamp_ has quit IRC (Ping timeout: 265 seconds) 2010-10-27T13:58:23 antimatroid1: stop trying to burn out the sun 2010-10-27T13:59:01 it's 5am :P 2010-10-27T13:59:07 the sun is almost back up again 2010-10-27T13:59:22 benefits of exam period, doesn't matter when i sleep 2010-10-27T13:59:52 lol 2010-10-27T13:59:56 *** McLeopold has joined #aichallenge 2010-10-27T14:00:04 i tend to drop off about 9pm and wake around 3-4am 2010-10-27T14:00:15 if i drop off earlier i tend to get up at 2am 2010-10-27T14:00:17 :P 2010-10-27T14:01:04 anyway, i'm rewriting some of my game theory library to use :P 2010-10-27T14:01:06 see how that goes 2010-10-27T14:01:09 particularly because i sometimes want to talk to yankees and canadians which are not usually on during normal sleeping hours 2010-10-27T14:01:15 :D 2010-10-27T14:01:46 and i do stuff at 2am constructively 2010-10-27T14:02:28 yeah, it's strange how productive i can be during the middle of the night 2010-10-27T14:02:39 although not so much for this, as it's usually a result of having nothing to do 2010-10-27T14:02:42 irc doesn't help that 2010-10-27T14:02:47 xD 2010-10-27T14:02:53 yea i know 2010-10-27T14:02:59 << has same problem 2010-10-27T14:03:14 though i spent a good part of 4 hours doing uni work 2010-10-27T14:03:18 without a deadline! 2010-10-27T14:03:20 :D 2010-10-27T14:03:25 earlier 2010-10-27T14:03:45 and weirdly it felt a lot less stressful than doing my non-uni related stuff 2010-10-27T14:03:46 lol 2010-10-27T14:04:04 maybe because it was easy 2010-10-27T14:06:21 i just have exams 2010-10-27T14:06:28 last lot for undergrad 2010-10-27T14:08:13 nice 2010-10-27T14:08:23 40 exams in 4 years :| 2010-10-27T14:08:36 im in my 3rd academic year 2010-10-27T14:08:39 ain't it fun? 2010-10-27T14:08:40 i have 1 more after this 2010-10-27T14:09:01 i'm doing honours in maths next year, masters in maths the year after and not sure after that 2010-10-27T14:09:32 i'll get a masters at the end of my next year ^^ 2010-10-27T14:09:56 its a 4 years+1 MEng course 2010-10-27T14:12:13 Top 10 players: bocsimacko(3875), rsergio(3854), GreenTea(3815), dmj111(3804), felixcoto(3757), protocolocon(3751), rs5147(3747), Hazard(3730), jambachili(3728), luksian(3705) 2010-10-27T14:12:39 oh boy, i even have half of simplex set up for msne in 2 player zero sum games 2010-10-27T14:12:45 i have minimax set up for psne 2010-10-27T14:13:17 i'm intending to do my honours thesis next year on computing msne for non-cooperative n-player normal form games 2010-10-27T14:13:29 hello grobner bases 2010-10-27T14:16:32 *** eburnette has joined #aichallenge 2010-10-27T14:17:48 *** perror has quit IRC (Quit: Bye all !) 2010-10-27T14:18:02 *** bhasker has joined #aichallenge 2010-10-27T14:20:34 *** bhasker__ has quit IRC (Ping timeout: 276 seconds) 2010-10-27T14:24:05 *** bhasker_ has joined #aichallenge 2010-10-27T14:26:00 *** bhasker has quit IRC (Ping timeout: 250 seconds) 2010-10-27T14:32:56 *** janzert has quit IRC (Ping timeout: 250 seconds) 2010-10-27T14:33:43 *** example has joined #aichallenge 2010-10-27T14:34:43 does anybody here know how the initial ranking is calculated? (after uploading a new source) 2010-10-27T14:35:13 wondered why i was rank 190 after miserably losing my first game :D 2010-10-27T14:35:30 you probably got paired against a higher ranked bot 2010-10-27T14:35:45 your ranking is based off both the result of the game and the ranking of the player you played 2010-10-27T14:35:56 i'm not sure exactly how 2010-10-27T14:36:06 it's not like that rank is unrealistic.... but my first upload started as rank 1500 or something 2010-10-27T14:36:31 and all games stayed in that range... now all my games are against rank 100-220 2010-10-27T14:37:05 it's the same for the top 100 bots... they pretty much only fight each other 2010-10-27T14:37:27 yeah, pairings are biasing like that this time 2010-10-27T14:37:35 not sure what they're planning for the final tournament 2010-10-27T14:37:49 *** Titankiller has joined #aichallenge 2010-10-27T14:39:04 *** lavalamp_ has joined #aichallenge 2010-10-27T14:40:37 Welp, time to revert to yesterday morning again 2010-10-27T14:41:56 for some reason i thought you were talking about something else... 2010-10-27T14:42:12 real life revert to yesterday 2010-10-27T14:42:22 I'd be ok with that, but Monday morning would be better 2010-10-27T14:44:24 lol 2010-10-27T14:45:25 lost a table that hadn't been backed up since Oct 13th on Monday at work 2010-10-27T14:45:31 don't say that 2010-10-27T14:45:33 real professional like 2010-10-27T14:45:34 i had an exam tuesday 2010-10-27T14:45:42 oh ok, you can stay in the future then 2010-10-27T14:46:14 no, don't worry, you'll still have only taken it once ;) 2010-10-27T14:47:15 hmmm, i wonder if one could set it up as a cooperative game 2010-10-27T14:47:25 where each planet is a separate player 2010-10-27T14:47:31 then form coalitions etc. 2010-10-27T14:47:53 writing bots for that would be insane 2010-10-27T14:48:04 yep 2010-10-27T14:48:07 private void BackstabFriendlies(){} 2010-10-27T14:49:00 no, no, i mean same two player game 2010-10-27T14:49:05 no, that should be written in c++ 2010-10-27T14:49:12 so you can have "friend" classes :D 2010-10-27T14:49:16 haha 2010-10-27T14:49:16 your bot represents it's planets as separate players 2010-10-27T14:50:01 i've heard suggested that each planet is its own bot 2010-10-27T14:50:19 i don't like that though 2010-10-27T14:50:23 i think coordination is important 2010-10-27T14:50:29 my old code almost works like that, except it coordinates for multiplanet attacks 2010-10-27T14:50:32 that's more or less the same thing 2010-10-27T14:50:38 which really hamstrung it 2010-10-27T14:50:42 you have them coordinate 2010-10-27T14:50:51 you could even have another bot be a "general" type 2010-10-27T14:51:06 a bot to plan grand strategy? 2010-10-27T14:51:29 My official bot is essentially every planet for itself. 2010-10-27T14:51:32 "Left flank, weaken your front planet to draw his ships in... ok, right flank strike!" 2010-10-27T14:53:01 that would require enemies intelligent enough to fall for you trap - >90% aren't... 2010-10-27T14:53:30 *** mceier has joined #aichallenge 2010-10-27T14:53:53 http://72.44.46.68/canvas?game_id=734399 I really like the finish on this game. 2010-10-27T14:54:03 It just looks cool. 2010-10-27T14:54:05 sigh.. all good usernames on youtube are taken 2010-10-27T14:54:50 I like the borderline pincer attack you put on there McLeopold 2010-10-27T14:54:53 McLeopold: get a fireworks mode 2010-10-27T14:54:57 yeah i wanted to do that, once you can win, move to the centre planet then fire off 2010-10-27T14:55:03 McLeopold: at the end, when you're about to win, yeah 2010-10-27T14:55:03 in a cirlce 2010-10-27T14:55:14 pew pew pew pew 2010-10-27T14:55:25 finish him 2010-10-27T14:55:48 you could even do patterns with 1 sized fleets 2010-10-27T14:55:54 during the game 2010-10-27T14:56:02 voidptr already did the fireworks thing 2010-10-27T14:56:04 it would surely confuse the enemy 2010-10-27T14:56:04 voidptr already did fireworks 2010-10-27T14:56:11 antimatroid1: link? 2010-10-27T14:56:15 i saw a couple of people with fireworks 2010-10-27T14:56:16 too old 2010-10-27T14:56:22 a star would be cool 2010-10-27T14:56:24 a while ago 2010-10-27T14:56:37 but from each planet separately 2010-10-27T14:56:55 i think the centre winning sequence would be cool 2010-10-27T14:57:01 i want links! 2010-10-27T14:57:05 i'm not doing well enough to spend time on such things yet 2010-10-27T14:57:24 i have an idea for a good flashy thing to show people 2010-10-27T14:57:32 get a huge map, 100 planets maybe 2010-10-27T14:57:38 let them play for 5000 rounds 2010-10-27T14:57:39 my bot would die 2010-10-27T14:57:52 any bot that grep 200 returns nothing 2010-10-27T14:58:27 amstan did you see this map? 2010-10-27T14:58:34 http://i.imgur.com/p2XGG.png 2010-10-27T14:58:36 no, what map? 2010-10-27T14:58:40 try working out a best strategy for that 2010-10-27T14:59:26 i think it's isomorphic to rock paper scissors 2010-10-27T14:59:28 but i could be wrong 2010-10-27T14:59:29 antimatroid1: send 100 ships, first turn to planet a 2010-10-27T14:59:34 i mean to middle planet 2010-10-27T14:59:43 if the opponent sends all their ships their first move then you lose 2010-10-27T14:59:43 that's easy, 100 to the closest and hold for a draw 2010-10-27T14:59:56 if they try to attack you, you're closest so you can get there first 2010-10-27T15:00:03 yeah, but it costs you 9 2010-10-27T15:00:12 if you both move there first turn and send all growth you lose 2010-10-27T15:00:16 if they wait and you do that, you win 2010-10-27T15:00:24 opp sends 100 to your neutral first turn. 2010-10-27T15:00:25 etc. etc. 2010-10-27T15:00:39 so, follow up with a 5 and you hold 2010-10-27T15:00:47 not if they follow up too 2010-10-27T15:01:02 wait, no McLeopold's right. 2010-10-27T15:01:21 no he's not 2010-10-27T15:01:23 you land with 91 2010-10-27T15:01:36 then you have 96 when they land 2010-10-27T15:01:36 Oh, yes I am. Growth happens before the battle. 2010-10-27T15:01:46 opp is 1 turn out with 100, you are one turn out with 5 and growth of 5 2010-10-27T15:01:49 If it was an 11, then it would be tricky. 2010-10-27T15:02:07 no, you don't get growth when you claim a planet 2010-10-27T15:02:14 no, this is the next turn 2010-10-27T15:02:19 you get it the next turn, before the incoming 2010-10-27T15:02:26 you claim it at turn 4 with 91 ships there, then you arrive at turn 5 when they then have 96 planets 2010-10-27T15:02:32 oh wait 2010-10-27T15:02:35 yeah, you're right :P 2010-10-27T15:02:46 add 1-4 to middle planets 2010-10-27T15:03:02 at 10, it's still even. 2010-10-27T15:03:08 at 11 it's tricky 2010-10-27T15:03:17 yeah shit :P 2010-10-27T15:03:19 it's late 2010-10-27T15:03:43 okay, put it at 11, then what do you do? :P 2010-10-27T15:03:50 either way.. moving all your ships in first turn would be best bet 2010-10-27T15:03:59 since you'll still have growth on the planet you left 2010-10-27T15:04:08 not if it's 11 2010-10-27T15:04:14 what? 2010-10-27T15:04:16 then they can snipe you by moving there first turn 2010-10-27T15:04:31 how though? 2010-10-27T15:04:41 they can attack your homebase 2010-10-27T15:04:47 but they'll loose a few ships 2010-10-27T15:04:52 then you can send all your ships back 2010-10-27T15:04:54 no, if you both go for the neutral closest to you 2010-10-27T15:04:55 and conquer him 2010-10-27T15:04:56 they win 2010-10-27T15:04:56 if it's 11, then you can abandon to the furthur planet and you gain 5 ships overall 2010-10-27T15:05:13 change distances? :P 2010-10-27T15:05:33 my example is crumbling aha 2010-10-27T15:06:13 if they attack the home base, they lose time. You would take the center and hold with 25-35 extra ships. 2010-10-27T15:08:11 Holy crap! That's a great example. I just thought of something! 2010-10-27T15:08:22 *** McLeopold has quit IRC (Quit: Page closed) 2010-10-27T15:08:26 hahaha 2010-10-27T15:08:32 and he's off 2010-10-27T15:08:32 lol 2010-10-27T15:08:39 *** irchs has joined #aichallenge 2010-10-27T15:08:47 lol, i have 9 symmetry functions in my old matrix class 2010-10-27T15:08:49 that's insane 2010-10-27T15:09:04 I can't help but imagine McLeopold as Doc from back to the future 2010-10-27T15:09:11 after that little exit right there 2010-10-27T15:09:13 Lol. 2010-10-27T15:09:15 i got a bit carried away playing away with the fano plane last year, it's group table meets all symmetries i'm pretty sure 2010-10-27T15:09:26 last semester** 2010-10-27T15:09:52 1.21 gigabytes 2010-10-27T15:11:58 1.21 jigabytes 2010-10-27T15:12:13 Top 10 players: bocsimacko(3889), rsergio(3872), GreenTea(3842), dmj111(3822), felixcoto(3784), rs5147(3777), protocolocon(3771), jambachili(3746), Hazard(3746), Arekku(3728) 2010-10-27T15:12:33 *** irchs has quit IRC (Client Quit) 2010-10-27T15:14:40 *** marcelo has joined #aichallenge 2010-10-27T15:17:23 *** boegel has quit IRC (Quit: Leaving) 2010-10-27T15:21:26 Accoun, what happened? http://72.44.46.68/canvas?game_id=733362 2010-10-27T15:22:59 opening gamble didn't pay off 2010-10-27T15:23:25 someone tell voidptr to post in my thread 2010-10-27T15:23:25 hm, I beat him a few times actually-- weird 2010-10-27T15:25:26 interesting match 2010-10-27T15:25:27 http://72.44.46.68/canvas?game_id=734914 2010-10-27T15:25:36 i have no clue what my bot was doing though 2010-10-27T15:26:26 that was an interesting comeback 2010-10-27T15:28:55 that is a very interesting match 2010-10-27T15:35:46 *** SamothRefoe has quit IRC (Ping timeout: 250 seconds) 2010-10-27T15:36:13 My latest move generator now rates higher than my best old bot... that's a good sign 2010-10-27T15:37:16 It's #85 on the tcp server -- what do you think it would be on the main server? 2010-10-27T15:38:59 my code is around 170 on TCP and 600s on official 2010-10-27T15:39:27 147/695 now actually 2010-10-27T15:39:40 but that includes a string of special C# losses 2010-10-27T15:39:46 haha 2010-10-27T15:40:03 was in mid 300s before that 2010-10-27T15:40:38 hm so I'm probably somewhere between 150 and 300 2010-10-27T15:40:44 that'd be my guess, yeah 2010-10-27T15:40:58 hopefully janzert will accept my script and we can find out :) 2010-10-27T15:41:27 *** Accoun has quit IRC () 2010-10-27T15:41:44 seems your TCP ELO should still be climbing also 2010-10-27T15:41:51 unless your wins against Accoun are unusual 2010-10-27T15:42:28 I think they are, or he broke his code or something... and I still lose to weaker bots too 2010-10-27T15:42:51 it won't climb any more today, my script apparantly played all the games I told it to 2010-10-27T15:44:29 now if I can just get my eval function working... 2010-10-27T15:47:09 I think I recoded my old bot's highlight characteristics in about 20 lines of code 2010-10-27T15:47:14 like just now. 2010-10-27T15:48:17 it must be time for some sleep 2010-10-27T15:48:19 *** Utkarsh has quit IRC (Ping timeout: 276 seconds) 2010-10-27T15:52:53 Argh.. .I had this one but let it slip http://ai-contest.com/visualizer.php?game_id=6282482 2010-10-27T15:53:23 *** Accoun has joined #aichallenge 2010-10-27T15:56:19 *** Appleman1234 has joined #aichallenge 2010-10-27T15:56:48 *** Utkarsh has joined #aichallenge 2010-10-27T15:57:35 great..... i spend hours on finding a good heuristic to generate moves - and now my bot calculates for about 2 seconds. timeout each game :-( 2010-10-27T15:57:36 this one too up until move 29 I think 2010-10-27T15:57:50 http://72.44.46.68/canvas?game_id=735234 2010-10-27T15:58:45 guess i'll go to bed... no need to redo my bot right now.... 2010-10-27T15:59:03 *** example has quit IRC (Quit: Page closed) 2010-10-27T16:08:12 *** antimatroid has joined #aichallenge 2010-10-27T16:10:28 *** antimatroid1 has quit IRC (Ping timeout: 240 seconds) 2010-10-27T16:12:12 Top 10 players: bocsimacko(3886), rsergio(3881), GreenTea(3846), dmj111(3829), rs5147(3792), felixcoto(3785), protocolocon(3776), jambachili(3738), Arekku(3738), Hazard(3731) 2010-10-27T16:15:36 *** cfaftw has quit IRC (Quit: home) 2010-10-27T16:15:41 *** marcelo has quit IRC (Ping timeout: 265 seconds) 2010-10-27T16:15:52 *** hellman has quit IRC (Remote host closed the connection) 2010-10-27T16:26:41 *** hightime has joined #aichallenge 2010-10-27T16:28:38 *** hightime has quit IRC (Client Quit) 2010-10-27T16:29:32 I suppose this should be reassuring: http://72.44.46.68/canvas?game_id=735654 2010-10-27T16:33:05 *** marcelo has joined #aichallenge 2010-10-27T16:36:46 *** McLeopold has joined #aichallenge 2010-10-27T16:37:50 JamesMG, nice show at the end there... is that your minimax code or something else? 2010-10-27T16:39:21 rewritten basic bot 2010-10-27T16:39:45 giving minimax a rest until I come up with some brilliant idea to make it work 2010-10-27T16:40:20 cool 2010-10-27T16:40:30 heuristics? 2010-10-27T16:40:33 yeah 2010-10-27T16:41:47 it's much simpler now, just adding stuff back in to bring it up to speed 2010-10-27T16:41:57 *** Cold-Phoenix has joined #aichallenge 2010-10-27T16:42:16 @seen j3camero 2010-10-27T16:42:16 Mathnerd314: j3camero was last seen in #aichallenge 3 weeks, 1 day, 18 hours, 22 minutes, and 31 seconds ago: amstan: yes, but no names, haha. 2010-10-27T16:42:36 where is he? he needs to make decisions 2010-10-27T16:43:10 I'll send an email in ALL CAPS to get his attention 2010-10-27T16:44:22 Mathnerd314: do it! :) 2010-10-27T16:44:56 I still don't believe he exists 2010-10-27T16:45:24 why am I getting opponents of rank 3500 and 200ish? 2010-10-27T16:46:04 they were getting uppity? 2010-10-27T16:47:26 got to keep the riff raff in their place 2010-10-27T16:57:02 mega1: why wouldn't you? 2010-10-27T16:58:19 I thought there is a very strong preference towards opponents of similar scores. 2010-10-27T16:59:37 was it the other bots first match? 2010-10-27T17:00:07 you're looking at their rank after they play you 2010-10-27T17:00:16 their rank may have been different before 2010-10-27T17:00:36 mega1: what's your elo? 2010-10-27T17:00:48 3884 2010-10-27T17:01:06 *** shivateam has joined #aichallenge 2010-10-27T17:01:23 mega1: oh congrats 2010-10-27T17:01:27 I mean this guy didn't exactly have a hot streak: http://ai-contest.com/profile.php?user_id=3895 2010-10-27T17:01:43 i don't think there's a prefference for similar bots, there should be 2010-10-27T17:01:44 he must be close to dead last 2010-10-27T17:01:57 clearly there is some 2010-10-27T17:02:06 mega1: notice how he submitted last in sept 14 2010-10-27T17:02:19 and he never starts a game :-) 2010-10-27T17:02:19 max elo was about 400 at that time 2010-10-27T17:02:42 talk about shooting fleas with cannons 2010-10-27T17:02:43 *** shivateam has quit IRC (Client Quit) 2010-10-27T17:02:46 ulope: dude.. fix your bot, lol 2010-10-27T17:04:47 *** bhasker has joined #aichallenge 2010-10-27T17:05:04 *** sigh has joined #aichallenge 2010-10-27T17:06:22 amstan: so I have turn limit, binaries, and prizes on there. Anything else people were asking? 2010-10-27T17:06:33 *** dmj111 has joined #aichallenge 2010-10-27T17:06:34 Mathnerd314: what? 2010-10-27T17:06:51 Mathnerd314: what are you talking about? 2010-10-27T17:06:56 what prizes? and binaries? 2010-10-27T17:07:07 the angry email... I want to get all the questions people have been asking 2010-10-27T17:07:20 jmcarthur has wanted binary submissions for a while 2010-10-27T17:07:27 there's already email conversations related to those 2010-10-27T17:07:30 just bump them 2010-10-27T17:07:52 but then he won't notice 2010-10-27T17:08:22 i'm not even participating now 2010-10-27T17:08:55 *** bhasker_ has quit IRC (Ping timeout: 276 seconds) 2010-10-27T17:09:25 jmcarthur: what? given up in disgust? 2010-10-27T17:09:33 nah, i just need more time 2010-10-27T17:09:37 free time 2010-10-27T17:09:39 to make money :P 2010-10-27T17:12:13 Top 10 players: GreenTea(3898), bocsimacko(3879), rsergio(3877), dmj111(3834), felixcoto(3789), rs5147(3786), protocolocon(3767), Arekku(3736), jambachili(3733), sin_sun(3725) 2010-10-27T17:12:15 amstan: the prizes are from the FAQ... it says "there might be prizes" 2010-10-27T17:12:37 the prizes are cake 2010-10-27T17:13:05 the cake is a lie 2010-10-27T17:13:33 and therefore... 2010-10-27T17:14:07 *** delt0r___ has joined #aichallenge 2010-10-27T17:14:17 no prizes. but I want it official 2010-10-27T17:14:34 there were no prizes last time other than fame 2010-10-27T17:14:38 *** delt0r has quit IRC (Read error: Operation timed out) 2010-10-27T17:14:42 *** Palmik has quit IRC (Remote host closed the connection) 2010-10-27T17:14:47 and a1k0n is pretty famous now 2010-10-27T17:14:51 so there you go 2010-10-27T17:15:07 Mathnerd314: I'm interested in the map generator for the finals. 2010-10-27T17:15:22 mega1: what about it? 2010-10-27T17:15:31 indeed... constraints on the map. I know that there are <500 planets 2010-10-27T17:15:33 how are finals going to be done? will it be a tournament or based off the rankings on the server? 2010-10-27T17:15:36 Let's declare that the current one will be used or something. 2010-10-27T17:15:53 We need something to optimize for. 2010-10-27T17:15:58 lavalamp_: let the servers run for a while without new submissions, then same as before 2010-10-27T17:16:05 wait, my program depends on there being no more than 23 planets... 2010-10-27T17:16:11 And changing the map generator favors heavily certain styles. 2010-10-27T17:16:20 I wasn't here for the last one, what was it? 2010-10-27T17:16:43 mega1: I think the point is that the bot can handle anything; specifically, it can do maps made after submissions are over 2010-10-27T17:16:59 but... added to THE LIST 2010-10-27T17:17:19 Mathnerd314: that's why I said "in the map _generator_" and not the maps themselves 2010-10-27T17:17:41 I really hope janzert is ok with my additions to his build script... *bites nails* 2010-10-27T17:19:05 if the generator is decent it should make maps of all kinds... 2010-10-27T17:19:29 the map generator is probably fine 2010-10-27T17:19:40 yay, angry email sent. 2010-10-27T17:20:02 maybe you can do some statistics or something and choose an eval function that's best for the map 2010-10-27T17:20:03 unless you mean changing it by changing the number of planets, growth sizes, or max dimensions 2010-10-27T17:20:32 if number of planets changes (well, grows) it will totally screw my bot 2010-10-27T17:21:05 Mathnerd314: oh, forgot the computational limits 2010-10-27T17:21:08 memory/cpu 2010-10-27T17:21:24 who's getting the angry email? 2010-10-27T17:21:46 lavalamp_: j3camero 2010-10-27T17:21:47 I really hope the web server won't be used for the finals 2010-10-27T17:22:19 ah ok... 2010-10-27T17:22:45 *** bhasker_ has joined #aichallenge 2010-10-27T17:23:13 is there a link to the last one? I want to know how the finals are done 2010-10-27T17:24:35 *** bhasker has quit IRC (Ping timeout: 240 seconds) 2010-10-27T17:25:50 check the old forums; I think it was some rankings followed by a round-robin on the top n 2010-10-27T17:30:45 judging from the results on the tcp server 2010-10-27T17:30:58 ... with 500 games you get +-30 elo 2010-10-27T17:31:57 *** marcelo has quit IRC (Quit: Leaving) 2010-10-27T17:32:24 last time around it was a round robin for the top 100 2010-10-27T17:32:25 eek if planet count will change for the final I need to know now... :/ 2010-10-27T17:33:06 but playing only about 100 games is +-70 elo 2010-10-27T17:33:34 what kind of confidence interval is that? 2010-10-27T17:33:47 I don't think you can get it much lower than +/- 30... 2010-10-27T17:34:14 a round robin of 100 is 4950 games 2010-10-27T17:34:50 each player has 99 games, though 2010-10-27T17:34:56 so that's not quite the same 2010-10-27T17:35:04 that's going to be pretty random 2010-10-27T17:35:17 considering just how close scores are 2010-10-27T17:35:22 i was advocating round robin matches, where a match is multiple games 2010-10-27T17:35:37 eg 100 maps 2010-10-27T17:35:51 that's incredibly extreme, though 2010-10-27T17:35:55 yeah you have to settle this with a tournament, you can't choose a winner based on rating 2010-10-27T17:35:56 with the same amount of games in total? 2010-10-27T17:36:25 well, it's kinda weird when you throw elo in there 2010-10-27T17:36:47 you could do a tournament (single/double-elim) like that 2010-10-27T17:36:57 but with elo it'd just be better to keep the games separate 2010-10-27T17:37:08 and then everyone has 9900 games 2010-10-27T17:37:20 or 99n games for given n 2010-10-27T17:37:33 *** Error323 has quit IRC (Read error: Connection reset by peer) 2010-10-27T17:37:58 maybe pick 10 maps by hand, don't tell people which ones beforehand 2010-10-27T17:38:06 bah 2010-10-27T17:38:26 that only increases noise in the rating 2010-10-27T17:38:35 the fewer maps the higher the noise 2010-10-27T17:39:06 i'm not sure about that 2010-10-27T17:39:49 *** bhasker has joined #aichallenge 2010-10-27T17:40:20 I am quite sure that the chance of misrepresenting the total distribution of maps is greater with fewer maps. 2010-10-27T17:40:39 And we are to optimize for a certain distributions of maps. 2010-10-27T17:41:11 that's why I'd like the map generator set in stone. 2010-10-27T17:42:05 well, the optimize for certain distributions isn't a requirement, it's a good idea for the tournament, though 2010-10-27T17:42:13 where tournament = contest 2010-10-27T17:42:38 the real goal would be to make a bot that can handle any map, but that's more difficult 2010-10-27T17:42:44 *** bhasker_ has quit IRC (Ping timeout: 250 seconds) 2010-10-27T17:43:33 All bots have their weaknesses. Looking at their history I can handpick certain maps now to favor any of them. 2010-10-27T17:43:34 amstan: formatted irc logs like http://mg.pov.lt/irclog2html/ generates would be nice 2010-10-27T17:48:41 I have a vector storing objects of class A which have a member int b, how come int x = theVector[i].b; changes what's stored in b? 2010-10-27T17:49:02 *** Appleman1234 has quit IRC (Excess Flood) 2010-10-27T17:49:15 what? 2010-10-27T17:49:20 the assignment changes the value? 2010-10-27T17:49:35 I've managed to determine accessing the element of a class in a vector changes it for some reason 2010-10-27T17:49:50 or modifying x changes it? 2010-10-27T17:50:01 *** Appleman1234 has joined #aichallenge 2010-10-27T17:50:18 Zannick, if x is never used in my code, just initialized to that, it changes how my code runs 2010-10-27T17:50:42 The exact line in question is: int x = incomingOrders[i].amount; 2010-10-27T17:50:58 i happen not to believe you about that :P 2010-10-27T17:51:02 I never use x anywhere, yet when I comment out that line it works differently 2010-10-27T17:51:42 does it work differently if you put int x = 0; instead? 2010-10-27T17:51:48 I think I'm missing something about vectors or doing something really weird elsewhere, but I discovered this when I commented out some new debugging lines I added and they changed when my bot lost 2010-10-27T17:52:21 *** bhasker has quit IRC (Ping timeout: 265 seconds) 2010-10-27T17:53:12 you're probably doing something really weird 2010-10-27T17:53:41 I'm going to try and print the value 3 times and see if it changes every time 2010-10-27T17:54:12 are you sure you aren't using C++? 2010-10-27T17:54:33 *** javagamer is now known as javagamer_cpp 2010-10-27T17:54:37 that is c++ :P 2010-10-27T17:54:50 oh lol that's not confusing at all 2010-10-27T17:55:10 heh, I had my name like this before, just reset when I reconnected 2010-10-27T17:55:26 no one uses vector in java everyone uses lists 2010-10-27T17:55:30 :P 2010-10-27T17:55:36 in that case it could be any number of things, like you're overrunning your stack some where and an extra variable changes things 2010-10-27T17:55:54 ah, I don't use java so wouldn't know 2010-10-27T17:56:17 or neglecting to initialize a variable 2010-10-27T17:56:54 that's why i suggested the int x = 0; thing 2010-10-27T17:57:07 yeah it's the first thing I'd try 2010-10-27T17:58:38 *** Frontier has left #aichallenge 2010-10-27T17:58:51 *** bhasker has joined #aichallenge 2010-10-27T17:59:35 *** Titankiller has quit IRC (Ping timeout: 240 seconds) 2010-10-27T17:59:38 It seems to be specific to assignment, when I try and print the value it doesn't change it, at least not the same way as when I try and assign x to it. 2010-10-27T17:59:39 ...and that sort of thing is why I'm writing in go whenever I possibly can 2010-10-27T17:59:43 *** jesionaj has joined #aichallenge 2010-10-27T17:59:50 my other guess would be you're hitting somewhere out of the array bounds 2010-10-27T18:00:44 Zannick, just adding in int x=0; causes the same change, so this would be stack related then? 2010-10-27T18:00:52 oh yeah. 2010-10-27T18:00:53 yes definitely 2010-10-27T18:00:58 you blew your stack 2010-10-27T18:01:07 is this array on the stack? 2010-10-27T18:02:19 Uhhh... maybe, my formal CS education is really lacking (still in HS) 2010-10-27T18:02:31 *** nann has joined #aichallenge 2010-10-27T18:02:50 did you new it? did you declare it in a function? 2010-10-27T18:03:37 oh, it's a vector, he probably has it in the heap (used new) 2010-10-27T18:03:39 the vector is a member of my planet class 2010-10-27T18:04:01 actually, C++... you really need to know the difference between stack and heap... go google that. :) 2010-10-27T18:04:32 what about your planets list? vector/array? 2010-10-27T18:05:15 *** Apophis has quit IRC (Ping timeout: 240 seconds) 2010-10-27T18:06:56 My planets are stored in a hash table and several vectors have references to them 2010-10-27T18:07:15 references or copies? 2010-10-27T18:07:23 or pointers? 2010-10-27T18:07:51 hm, I don't think you need a hash table for 23 planets :) 2010-10-27T18:07:57 Aren't pointers references? I have several vectors of pointers to the planets 2010-10-27T18:08:10 okay 2010-10-27T18:08:19 and no, they're not the same :) 2010-10-27T18:08:22 confusingly, no not in the technical sense 2010-10-27T18:08:41 int *p = &i // p is a pointer 2010-10-27T18:08:47 but pointers are fine. 2010-10-27T18:08:51 int& r = i; //r is a reference 2010-10-27T18:10:23 unfortunately, it isn't going to be clear where your problem is. it could be that the declaration is doing something funky to your planets, but it's just as possible that something else is happening with your stack 2010-10-27T18:10:46 agree-- this bug will be hard to find 2010-10-27T18:11:12 :/ 2010-10-27T18:11:23 welcome to C++ 2010-10-27T18:11:37 common reasons for blowing your stack include: declaring large arrays in functions, recursing too deeply, some combination of both 2010-10-27T18:12:01 Could it be related to code like this: http://pastebin.com/DRrAEVxJ 2010-10-27T18:12:12 I spent my morning fixing a heap corruption... 2010-10-27T18:12:13 Top 10 players: rsergio(3894), GreenTea(3894), bocsimacko(3875), dmj111(3832), felixcoto(3786), rs5147(3781), protocolocon(3762), Arekku(3753), jambachili(3733), shangas(3717) 2010-10-27T18:12:30 I did stuff like that a few times and I felt really worried it would break something 2010-10-27T18:13:05 hm 2010-10-27T18:13:23 planet->incomingOrders.begin() + i 2010-10-27T18:13:26 seems suspect to me 2010-10-27T18:13:28 The idea there is to remove all orders from the vector that meet a certain qualification 2010-10-27T18:13:49 that doesn't quite work like you'd expect, i think, though i'm not sure it's the same problem 2010-10-27T18:13:55 Mathnerd314: you missed a perfect opportunity for 'WHAT is your favorite colour?' 2010-10-27T18:13:58 Zannick, that seems to be how it's done, http://www.cplusplus.com/reference/stl/vector/erase/ 2010-10-27T18:14:16 a1k0n: yeah, was wondering if anyone would catch that :p 2010-10-27T18:14:18 also lol i'm famous 2010-10-27T18:14:25 it looks weird, yeah, but I think it will work 2010-10-27T18:14:34 a1k0n: famous? where and how? 2010-10-27T18:14:47 16:14 < JamesMG> and a1k0n is pretty famous now 2010-10-27T18:14:54 maybe not efficiently, but I think it will work 2010-10-27T18:15:03 * Zannick shrug 2010-10-27T18:15:26 lavalamp_, I'd love to hear if you know of a more efficient way to accomplish the same thing 2010-10-27T18:15:57 I'd considered sorting first, but I figured it would result in a net loss of time, though the code would certainly look nicer 2010-10-27T18:16:05 if it's a short list, it doesn't matter 2010-10-27T18:16:25 Yeah, I assumed it should stay rather short 2010-10-27T18:16:26 but i don't think that is the problem 2010-10-27T18:16:41 then don't worry about it 2010-10-27T18:16:54 "it's easier to optimize correct code than to correct optimized code" 2010-10-27T18:17:11 I'm pretty sure it has to do with the two vectors of orders b/c afaik it was working before that 2010-10-27T18:17:23 I'll try commenting them out and seeing if it starts working then 2010-10-27T18:17:24 this is not your corruption source, agreed 2010-10-27T18:17:50 with stack corruption, any modification will tear something new 2010-10-27T18:18:32 yeah that may not be a good test. 2010-10-27T18:19:03 you could try moving the int x=0; line around in your code, try to narrow down where you can place it and still cause the issue 2010-10-27T18:19:28 are you using recursion anywhere? 2010-10-27T18:20:23 Zannick, not purposefully, but I might be accidentally, I'll check for that 2010-10-27T18:20:57 well, i mean functions that call themselves (or other functions that call them back...) 2010-10-27T18:21:29 every function call adds to the stack, so it's a very easy way to blow it 2010-10-27T18:21:38 (i used to do that often in java) 2010-10-27T18:21:41 Zannick, I have lots of functions calling lots of other functions, I didn't plan any recursion yet, but I might accidentally have something calling something that recalls the original thing or something 2010-10-27T18:21:47 *** RainCT has quit IRC (Remote host closed the connection) 2010-10-27T18:22:03 yeah, be on the lookout for that 2010-10-27T18:22:46 I originally found this when I noticed one thing would time out when it should go really quickly, so something like that could very well be happening 2010-10-27T18:23:04 my guess is it's an out of bound array access to something on the stack... 2010-10-27T18:24:55 *** JamesMG has quit IRC (Quit: Leaving.) 2010-10-27T18:25:09 you might consider using indexes to an array of planets instead of direct pointers... easier to catch if you do something wrong. 2010-10-27T18:25:19 lavalamp_, could that be caused with vectors as well b/c I'm using very few arrays 2010-10-27T18:25:44 std::vector should catch that but who knows what exactly you're doing... 2010-10-27T18:26:09 anyway I gotta go... good luck 2010-10-27T18:26:56 thanks for your help 2010-10-27T18:27:26 np 2010-10-27T18:27:35 *** lavalamp_ has quit IRC (Quit: so long, suckers) 2010-10-27T18:45:03 *** narnach has joined #aichallenge 2010-10-27T18:45:20 *** narnach_mb has quit IRC (Quit: Leaving.) 2010-10-27T18:46:48 *** narnach has quit IRC (Client Quit) 2010-10-27T18:48:55 *** Bobng has quit IRC (Ping timeout: 240 seconds) 2010-10-27T18:50:37 *** Azrathud has joined #aichallenge 2010-10-27T18:51:58 should i celebreate 2010-10-27T18:52:01 celebrate* 2010-10-27T18:52:01 http://72.44.46.68/canvas?game_id=737411 2010-10-27T18:52:29 victory by inaction 2010-10-27T18:54:05 a regular cold war there 2010-10-27T18:54:26 *** Appleman1234 has quit IRC (Remote host closed the connection) 2010-10-27T18:54:37 heh 2010-10-27T18:54:40 my bot does that a lot 2010-10-27T18:54:48 a lot of my games are 200 turn victory:-\ 2010-10-27T18:54:52 i get an advantage and just sit 2010-10-27T19:04:32 *** justin_pdx has joined #aichallenge 2010-10-27T19:09:41 *** Titankiller has joined #aichallenge 2010-10-27T19:12:13 Top 10 players: rsergio(3906), GreenTea(3892), bocsimacko(3872), dmj111(3847), rs5147(3790), felixcoto(3777), luksian(3761), protocolocon(3752), Arekku(3735), shangas(3729) 2010-10-27T19:12:33 *** Naktibalda has joined #aichallenge 2010-10-27T19:12:54 j3camero: ping. 2010-10-27T19:14:08 you are very unlikely to catch him here 2010-10-27T19:16:46 *** mceier has quit IRC (Quit: leaving) 2010-10-27T19:16:53 he just replied to my email... he should be around on IRC too 2010-10-27T19:17:21 and... out of the blue, he changes turn limit to 400 :p 2010-10-27T19:17:39 unfortunately, though, he isn't. 2010-10-27T19:17:54 though we would like him to be if he's going to change things like that 2010-10-27T19:19:09 yeah what the hell, 400 turns? i hope i just talked him out of that 2010-10-27T19:19:43 data: http://a1k0n.net/temp/galcon_histogram.png 2010-10-27T19:19:58 (i actually have since accumulated a lot more games and so i guess i could re-run that) 2010-10-27T19:22:03 what was the turn limit for Tron? IIRC the maps were 20x20, so... 400/2 = 200 2010-10-27T19:22:27 the initial maps were 15x15 2010-10-27T19:23:01 and they all had border walls 2010-10-27T19:23:06 a1k0n: how did you make that histogram? scraping the site again? 2010-10-27T19:23:22 is the vertical axis number of games? 2010-10-27T19:23:25 Mathnerd314: keep in mind that we have 7 times more people and crappier servers 2010-10-27T19:23:33 well, it's scraped from benzedrine.cx so it could arguably be biased 2010-10-27T19:23:34 Zannick: number of games ending 2010-10-27T19:23:44 yeah, what i thought 2010-10-27T19:23:56 yeah the green line is on a separate scale 2010-10-27T19:24:05 a1k0n: this is decent, benzedrine players would be more concerned with it 2010-10-27T19:24:12 right, that's what i figured 2010-10-27T19:24:18 oh 2010-10-27T19:24:26 so yeah, 90% games end before 200 turns 2010-10-27T19:24:48 and of the remainder, probably a good 5-8% of them would just go on forever 2010-10-27T19:24:56 *** seisatsu has joined #aichallenge 2010-10-27T19:25:01 maybe change it to 250 2010-10-27T19:25:09 there are games where they clearly would have ended on ilke turn 250 2010-10-27T19:25:13 but in that case the winner is still the winner 2010-10-27T19:25:29 good point 2010-10-27T19:25:30 i don't think the benefits of 400 turns outweight the costs 2010-10-27T19:25:37 they don't 2010-10-27T19:25:38 i don't think it's even possible to turn the game around after, say, turn 100 2010-10-27T19:25:46 wat? j3camero claims "the firewalls" are preventing him from IRC. 2010-10-27T19:25:47 he should just make 200 static and official 2010-10-27T19:25:51 wtf? 2010-10-27T19:25:58 doesn't he work in finance now? 2010-10-27T19:26:11 it's 7pm though 2010-10-27T19:26:20 he can't connect to his server, maybe? 2010-10-27T19:26:53 * Mathnerd314 finds a wall and bashes his head against it 2010-10-27T19:27:00 Zannick: he uses screen+ssh 2010-10-27T19:27:10 as do i 2010-10-27T19:27:58 w/e, work is work 2010-10-27T19:28:01 is his screen session on a machine that has since been firewalled to prevent his access? o_O 2010-10-27T19:28:04 a1k0n: you want to reply or should I? 2010-10-27T19:28:08 i am at work :P 2010-10-27T19:28:08 you shouldn't be sshing to your server and irssing from there 2010-10-27T19:28:15 nvm, lol 2010-10-27T19:28:21 * amstan did the same thing this summer 2010-10-27T19:28:29 Mathnerd314: done 2010-10-27T19:28:38 though maybe my company is cool about that 2010-10-27T19:28:47 @seen j3camero 2010-10-27T19:28:47 amstan: j3camero was last seen in #aichallenge 3 weeks, 1 day, 21 hours, 9 minutes, and 2 seconds ago: amstan: yes, but no names, haha. 2010-10-27T19:29:29 haha indeed... I guess he had the last laugh :p 2010-10-27T19:29:51 lol 2010-10-27T19:30:08 i would bet a very large percentage of googlers ssh into some personal colocated machine regularly 2010-10-27T19:30:30 he's not a googler anymore though 2010-10-27T19:30:34 oh what? 2010-10-27T19:30:37 er 2010-10-27T19:30:41 jeff isn't. zannick is. 2010-10-27T19:30:43 i am, though :P 2010-10-27T19:30:49 Zannick: oh yay! 2010-10-27T19:31:01 Zannick: had an interview with them, i hope i work for them next term 2010-10-27T19:31:04 google recruiter called me back yesterday heh 2010-10-27T19:31:16 i was considering connecting to an mpd instance at home, but i figure it'll be easier to just drop all the music on this box 2010-10-27T19:31:27 my workstation, i mean 2010-10-27T19:31:47 amstan: cool, good luck 2010-10-27T19:31:52 i'll...still be here 2010-10-27T19:32:08 here where? 2010-10-27T19:32:13 here, google. 2010-10-27T19:32:20 * Zannick full time 2010-10-27T19:32:28 what branch? 2010-10-27T19:32:31 which building btw? 2010-10-27T19:32:45 i'm in mountain view 2010-10-27T19:32:46 they're in like 10 cities 2010-10-27T19:32:50 cool 2010-10-27T19:32:56 yeah i know.. :) 2010-10-27T19:33:08 10 sounds low :) 2010-10-27T19:33:14 which building zannick? 2010-10-27T19:33:31 why need me to be so specific? 2010-10-27T19:33:38 just curious 2010-10-27T19:33:49 *** irchs has joined #aichallenge 2010-10-27T19:33:51 w/e 2010-10-27T19:34:27 *** lavalamp_ has joined #aichallenge 2010-10-27T19:34:44 lavalamp_: i've talked to the legendary jeff! 2010-10-27T19:34:51 lavalamp_: oh man, he's real! 2010-10-27T19:34:52 what! no way 2010-10-27T19:35:03 pic or it didn't happen ;) 2010-10-27T19:35:11 *snrk* 2010-10-27T19:35:31 lavalamp_: does email count? 2010-10-27T19:35:34 my analysis just doubled the ranking system's effiency! 2010-10-27T19:36:03 email counts if he's responding to mine about getting go installed on the server from weeks ago ^^ 2010-10-27T19:36:24 *** Commandir has left #aichallenge 2010-10-27T19:38:13 *** justin_pdx has quit IRC (Quit: justin_pdx) 2010-10-27T19:38:54 lavalamp_: i have proof: http://home.hypertriangle.com/~alex/legendaryjeff.png 2010-10-27T19:39:09 *** nullkuhl has joined #aichallenge 2010-10-27T19:39:26 amstan, lol 2010-10-27T19:39:40 lolol 2010-10-27T19:39:51 hack dem Planeten! 2010-10-27T19:40:30 now I wonder where janzert went... haven't seen him since I made the install script for him... 2010-10-27T19:41:39 lavalamp_: he ran away when he saw it 2010-10-27T19:42:08 he said something among the lines "nooooooo!" 2010-10-27T19:42:12 ;) 2010-10-27T19:43:22 real story: "Sorry, my power is out at the moment and not likely to come back this evening and the phone battery is about out of juice. ;) So I'm just going to reply to this for now." 2010-10-27T19:43:40 *** irchs has quit IRC (Quit: irchs) 2010-10-27T19:44:20 so he's stuck, powerless, in a stinking hovel? 2010-10-27T19:46:12 bummer, lol.... bbl 2010-10-27T19:46:19 janzert just replied 2010-10-27T19:46:24 oh hi 2010-10-27T19:47:02 is this the mailing list or the forum? 2010-10-27T19:47:10 ai-contest mailing list 2010-10-27T19:47:16 Zannick: mailing list 2010-10-27T19:47:18 hm, i should get on that 2010-10-27T19:47:36 i should get jeff to make me admin for the mailing list and google code 2010-10-27T19:47:45 so i can add people 2010-10-27T19:48:24 *** yasith has quit IRC (Ping timeout: 250 seconds) 2010-10-27T19:49:02 amstan: doit! 2010-10-27T19:49:38 amstan: also, did you catch the line about formatting irc logs with http://mg.pov.lt/irclog2html/ ? 2010-10-27T19:49:50 i was leaving for pizza when i saw it 2010-10-27T19:50:24 *** McLeopold has quit IRC (Quit: Leaving) 2010-10-27T19:50:49 ok. it should be pretty easy to set up 2010-10-27T19:50:51 Mathnerd314: i don't see it though, what does it add to all of this? 2010-10-27T19:51:23 and it's not that easy to setup, the bot actually dumps the log in real time 2010-10-27T19:51:35 unless i have a cronjob to convert all of them every 5 minutes 2010-10-27T19:51:54 right.. you just run the script. (and daily works fine IMO) 2010-10-27T19:52:18 but then you can't say cheese at the bottom: http://contestbot.hypertriangle.com/channellogs/%23aichallenge/%23aichallenge.2010-10-27.log 2010-10-27T19:53:12 i find myself using the log of the day more than i use the previous ones 2010-10-27T20:00:20 if people realtime, they should use an irc client... 2010-10-27T20:00:25 *** yasith has joined #aichallenge 2010-10-27T20:00:33 *people want 2010-10-27T20:01:35 * Zannick realtime 2010-10-27T20:01:38 i'm just saying, this is a lot of work and sacrifices for just a few colours 2010-10-27T20:05:23 *** seisatsu has quit IRC (Quit: Leaving) 2010-10-27T20:07:03 it's not... it does line wrapping 2010-10-27T20:09:21 *** Accoun has quit IRC (Ping timeout: 272 seconds) 2010-10-27T20:09:40 *** Accoun has joined #aichallenge 2010-10-27T20:12:15 Top 10 players: GreenTea(3946), bocsimacko(3911), rsergio(3906), dmj111(3840), rs5147(3793), felixcoto(3766), luksian(3764), protocolocon(3753), Hazard(3739), Arekku(3732) 2010-10-27T20:13:02 *** Mathnerd314 has quit IRC (Ping timeout: 245 seconds) 2010-10-27T20:13:12 oh boy.. he logged in again! 2010-10-27T20:17:05 *** jaspervdj has quit IRC (Quit: NEVER GONNA GIVE YOU UP NEVER GONNA LET YOU DOWN) 2010-10-27T20:35:53 ok... time to get my goal evaluating code in order... 2010-10-27T20:39:34 *** Cold-Phoenix has quit IRC () 2010-10-27T20:52:27 *** phreeza has quit IRC (Quit: Computer has gone to sleep) 2010-10-27T20:59:15 *** bhasker has quit IRC (Quit: bhasker) 2010-10-27T21:01:38 *** davidjliu has joined #aichallenge 2010-10-27T21:12:14 Top 10 players: GreenTea(3965), bocsimacko(3923), rsergio(3920), dmj111(3856), rs5147(3811), felixcoto(3786), protocolocon(3768), luksian(3755), Hazard(3753), Arekku(3747) 2010-10-27T21:27:58 *** McLeopold has joined #aichallenge 2010-10-27T21:30:46 *** jesionaj has quit IRC (Read error: Connection reset by peer) 2010-10-27T21:30:55 *** jesionaj has joined #aichallenge 2010-10-27T21:52:29 *** hornairs has joined #aichallenge 2010-10-27T21:54:05 *** cfaftw has joined #aichallenge 2010-10-27T22:00:05 *** Naktibalda has quit IRC (Remote host closed the connection) 2010-10-27T22:06:18 *** rebelxt_ has joined #aichallenge 2010-10-27T22:09:48 *** Appleman1234 has joined #aichallenge 2010-10-27T22:12:15 Top 10 players: GreenTea(3967), bocsimacko(3939), rsergio(3907), dmj111(3865), rs5147(3836), felixcoto(3790), Hazard(3765), luksian(3765), protocolocon(3763), Arekku(3755) 2010-10-27T22:13:49 *** rebelxt_ has quit IRC (Quit: Page closed) 2010-10-27T22:14:52 *** cfaftw has quit IRC (Quit: Leaving) 2010-10-27T22:18:13 *** jarfhy has quit IRC (Read error: Connection reset by peer) 2010-10-27T22:44:20 *** Appleman1234 has quit IRC (Quit: Leaving) 2010-10-27T22:48:32 hmm, i'm not sure representing it as a normal form zero sum game is any different to usign the tree 2010-10-27T22:48:38 unless chasing after mixed strategies to sue 2010-10-27T22:51:16 how's the treebot coming along? 2010-10-27T22:51:21 yeah, it is not real clear. 2010-10-27T22:51:53 tree bot sucks 2010-10-27T22:51:57 i'm rethinking aha 2010-10-27T22:53:18 i'm pretty sure minimax on the tree will get you all psne anyway wont it? 2010-10-27T22:58:28 *** eburnette_ has joined #aichallenge 2010-10-27T23:00:16 psne? 2010-10-27T23:02:06 pure strategy nash equilibrium 2010-10-27T23:04:26 If one exists, then the minimax tree should find it. The simultaneous moves makes things confusing though. 2010-10-27T23:05:34 Maybe a payoff matrix would make sense, just for the simul moves. But, searching multiple ply with full payoff matrices seems harder. 2010-10-27T23:06:05 I tried figuring out a way to to alpha-beta pruning of mixed strategies for tron. Every now and then I got my hopes up. And then dashed them to pieces. 2010-10-27T23:07:27 actually, you can go through and eliminate pure dominated strategies before finding mixed strategy nash equilibria 2010-10-27T23:07:29 that might help 2010-10-27T23:08:58 I don't think this game has NE because there's no best unchanging strategy 2010-10-27T23:10:00 eburnette_, individual moves do though, probably most of the time 2010-10-27T23:10:37 Argh: INFO You LOSE against bocsimacko-b0.30 2010-10-27T23:11:53 eburnette: there exists a mixed strategy nash equilibrium :P 2010-10-27T23:12:16 Top 10 players: bocsimacko(3956), rsergio(3937), GreenTea(3927), dmj111(3869), rs5147(3852), felixcoto(3808), protocolocon(3788), luksian(3777), Arekku(3773), Hazard(3771) 2010-10-27T23:13:43 antimatroid: oh you actually found a case? 2010-10-27T23:14:12 jmcarthur: it's a famous theorem :P 2010-10-27T23:14:41 Ok guys where did I go wrong: http://72.44.46.68/canvas?game_id=740899 2010-10-27T23:14:48 antimatroid: oh, you are assuming that mix strategies are a superset of pure strategies? 2010-10-27T23:14:52 (mixed 2010-10-27T23:14:55 every finite normal form game has a mixed strategy nash equilibria 2010-10-27T23:14:55 **mixed 2010-10-27T23:14:57 gah 2010-10-27T23:15:03 yeah, well they are :P 2010-10-27T23:15:24 normally i say mixed strategy to mean "not a pure strategy" 2010-10-27T23:15:31 but i suppose that is incorrect 2010-10-27T23:15:31 finding it is not so easy, that's my intended honours thesis topic next year, hello grobner bases 2010-10-27T23:15:44 although gambit uses newtons method for getting ne 2010-10-27T23:16:06 pure strategies are mixed strategies in a way 2010-10-27T23:16:20 yeah, a degenerate form of mixed strategy 2010-10-27T23:16:28 still a mixed strategy :P 2010-10-27T23:16:34 no disagreement 2010-10-27T23:19:00 i have no idea how to do this game well though 2010-10-27T23:19:35 *** espes has quit IRC (Ping timeout: 240 seconds) 2010-10-27T23:25:53 *** Apophis has joined #aichallenge 2010-10-27T23:26:49 eburnette_: still looking for advice? 2010-10-27T23:26:50 *** jesionaj` has joined #aichallenge 2010-10-27T23:27:06 yeah, seems like I almost had him for a while 2010-10-27T23:27:56 at turn 37, you were even 2010-10-27T23:28:45 *** jesionaj has quit IRC (Ping timeout: 265 seconds) 2010-10-27T23:28:46 on turn 39, you sent 70 up north. By the time they hit, you had lost 4 planets. 2010-10-27T23:29:01 *** nann has quit IRC (Ping timeout: 255 seconds) 2010-10-27T23:29:20 on turn 40, you sent 59 ships backwards to take a neutral 2010-10-27T23:29:32 twice 2010-10-27T23:31:02 on turn 36, you send 49 to your main planet, which opened a small hole, but would have been recoverable 2010-10-27T23:32:07 so, in a matter of 2 turns (39, 40) you attempt to take 2 back line neutral and an enemy planet far away. 2010-10-27T23:32:16 I sent the 70 because the planet looked vulnerable 2010-10-27T23:32:44 Yeah, maybe, but you can't leave yourself open when it's that close. 2010-10-27T23:33:01 Maybe just taking 1 planet at a time would've worked. 2010-10-27T23:34:09 If that 70 had done up to the 8+4 planet that just sent the 89, that might have been better 2010-10-27T23:34:13 It was closer 2010-10-27T23:36:17 Or if I hadn't sent either of the 59's 2010-10-27T23:37:04 Those were really bad. You might have still lost the center because of that 70. 2010-10-27T23:37:17 Ooh I think I know why I did the 59's, I can fix that 2010-10-27T23:37:44 You also need a simple check to make sure you don't capture a planet twice. 2010-10-27T23:37:57 Actually, my official bot does that. :( 2010-10-27T23:39:27 http://72.44.46.68/canvas?game_id=741025 2010-10-27T23:40:46 turn 49, you give up the center again by sending 142 across the map 2010-10-27T23:41:02 lucky for you, my bot was dumber. :P 2010-10-27T23:41:02 I don't think you're doing dup moves because I keep taking back 2010-10-27T23:42:01 So, I re-uploaded my bot, and it's... rank 3674. oh yeah 2010-10-27T23:42:11 and climbing? 2010-10-27T23:42:17 heh, well it played one game 2010-10-27T23:42:30 a clutch first-turn timeout victory 2010-10-27T23:42:42 I'm thinking a quick reupload is in order in another 10 minutes... 2010-10-27T23:42:45 that probably dropped you rank alot... 2010-10-27T23:42:58 That just means you start over 2010-10-27T23:43:09 heh, yeah 2010-10-27T23:43:12 I hate the official server, sooooo sloooow 2010-10-27T23:44:19 my bot crashes on the main server at the moment 2010-10-27T23:44:23 i need to fix that at some point 2010-10-27T23:44:30 the main server has been horrible this time around 2010-10-27T23:44:32 davidjliu: play on tcp! 2010-10-27T23:44:49 heh, sounds like a plan 2010-10-27T23:45:04 I'd be willing to fire up AntBot if you do. 2010-10-27T23:45:07 be sure you get delt0r's viewer 2010-10-27T23:45:19 I should check that out. 2010-10-27T23:45:35 i'm trying a new approach 2010-10-27T23:45:52 so it'll be a while, i'll throw the horrible thing i made last night up though 2010-10-27T23:45:55 it's goin up.. .probably just for 10-15 minutes-ish 2010-10-27T23:46:23 hopefully my changes didn't make it suck 2010-10-27T23:46:31 hmm... let's see of AntBot can play a game without a timeout... 2010-10-27T23:46:39 there you go :P 2010-10-27T23:46:40 first matching 2010-10-27T23:46:43 heh 2010-10-27T23:47:01 i'll probably lose 2010-10-27T23:47:20 I dunno, I haven't actually run this bot before... 2010-10-27T23:47:41 that bot has some changes i made rigth before bed last night to try out 2010-10-27T23:47:43 who does medrimonia? 2010-10-27T23:47:46 i have no idea if they hurt or not 2010-10-27T23:47:52 medrimonia, probably 2010-10-27T23:47:54 they are medrimonia? :P 2010-10-27T23:47:54 sweet! 2010-10-27T23:48:03 lol 2010-10-27T23:48:03 davidjlui is my opponent 2010-10-27T23:48:08 oh no, disconnection victory 2010-10-27T23:48:18 did i timeout? 2010-10-27T23:48:28 on map 7, my favorite 2010-10-27T23:48:30 for McLeopold 2010-10-27T23:48:41 i like 7 too :P 2010-10-27T23:48:42 so weird 2010-10-27T23:48:46 that's my number one test map 2010-10-27T23:48:58 shoot, it is broken still 2010-10-27T23:48:59 and 44 and 79 2010-10-27T23:49:38 oh crap playing cccf 2010-10-27T23:50:10 lol, I've played bocsimacko ... 3 times 2010-10-27T23:50:30 blech, tied for a while and lost it again 2010-10-27T23:51:05 yay, I won 2010-10-27T23:51:06 heh 2010-10-27T23:51:06 :P 2010-10-27T23:52:16 shoudl have kept expanding you idiot 2010-10-27T23:52:47 perpetual attack strategy 2010-10-27T23:52:59 yeah, i ruined it last night with redistribution 2010-10-27T23:53:12 AntBot uses 91% of my cpu 2010-10-27T23:53:14 it's sending stupid amounts at the enemy with no hope to claim 2010-10-27T23:53:45 and I still timeout 2010-10-27T23:53:46 Bone headed move at 62: http://72.44.46.68/canvas?game_id=741333 2010-10-27T23:55:18 eburnette_: you could have held that planet he was streaming to 2010-10-27T23:55:39 aw, our bots don't appear to be encountering 2010-10-27T23:56:03 ha, i beat ccf 2010-10-27T23:56:06 yeah it's like "Hey, come take my planet please" 2010-10-27T23:56:10 woah, nice 2010-10-27T23:56:11 grats 2010-10-27T23:56:34 http://72.44.46.68/canvas?game_id=741369 2010-10-27T23:57:34 wasn't a very impressive win 2010-10-27T23:57:42 and it's losing to people it didn't used to 2010-10-27T23:58:10 yeah... excitingly, my bot just lost to dhartmei