2011-12-16T00:00:02 i think the trick is knowing when to move it off? 2011-12-16T00:00:03 and that's when you don't want to waste ants 2011-12-16T00:00:09 *** Paradoxial has quit IRC (Quit: :) 2011-12-16T00:00:20 capping them is easy 2011-12-16T00:00:37 to look at turn 75 the map looks very random 2011-12-16T00:00:46 so to pull a victory out of that soup is pretty impressive 2011-12-16T00:00:53 ant so far my system seems to work well in most cases, but I did have to rethink a bit about my strategic priorities of food vs. explore, vs. attack hills and enemy ants 2011-12-16T00:01:00 so rebalance it all 2011-12-16T00:01:09 s/so/to 2011-12-16T00:02:22 I don't know what happened, but my bot just played a whole bunch of games on the main server while we were chatting, and now I need to go digest how I jumped up to #24. 2011-12-16T00:02:56 :) 2011-12-16T00:03:48 BenJackson: I think top 20 is much more competitive than thought before 2011-12-16T00:03:54 beating xathis was +2 mu, wow 2011-12-16T00:04:40 *** codetiger has joined #aichallenge 2011-12-16T00:04:41 wow, 4:33, 34, 42, 44, 46 games 2011-12-16T00:04:53 ChrisH_: what did that cost you in donations? 2011-12-16T00:05:35 Flag beat xathis twice in a row 1 on 1 3 minutes apart 2011-12-16T00:05:48 oh won and stalemated 2011-12-16T00:06:14 <_flag> I should've won the stalemate, I had a clear advantage and was scared to push in :( 2011-12-16T00:06:55 yep 2011-12-16T00:07:51 _flag: pussy :P 2011-12-16T00:08:20 i'm up to 23rd on tcpants ehe 2011-12-16T00:08:24 *** cyphase has joined #aichallenge 2011-12-16T00:09:59 am on 22nd tcpants 2011-12-16T00:10:14 yesterday, I remember i was on 13th 2011-12-16T00:10:21 I was #1 until I timed out several games when I tried to kill the client :) 2011-12-16T00:10:25 codetiger: http://tcpants.com/player/SkyCaptain not sure if this run is a fluke or not 2011-12-16T00:10:30 refugees from fluxid are making competition stiffer 2011-12-16T00:10:47 *** ldlework has joined #aichallenge 2011-12-16T00:10:51 OMG 2011-12-16T00:10:57 I just learned about this challenge...! 2011-12-16T00:11:08 * matthewd is doing his part to make competition softer :) 2011-12-16T00:11:12 ldlework: unlucky :P 2011-12-16T00:11:13 <_flag> Oh, didn't realize tcpants was popular 2011-12-16T00:11:13 in just 4 games, then its awesome 2011-12-16T00:11:15 I have been writing this exact thing I can't believe there is a challenge 2011-12-16T00:11:21 <_flag> I guess I need to start playing there too 2011-12-16T00:11:35 _flag: it was definitely less popular than fluxid 2011-12-16T00:11:36 _flag: fluxid is so slow 2011-12-16T00:11:40 antimatroid, I have been writing a 2D cellular agent simulation system for a few weeks now 2011-12-16T00:11:41 until people realized fluxid's db was corrupt 2011-12-16T00:11:43 i just switched before cause of speed 2011-12-16T00:12:02 <_flag> antimatroid: I know, I just didn't know there were alternatives with top players on them 2011-12-16T00:12:12 antimatroid, I use genetic algorithms to breed creatures in the system. Ugh, I just can't believe it. 2011-12-16T00:12:15 <_flag> tcpants actually appears to have better games than fluxid 2011-12-16T00:12:23 i just got pwned 2011-12-16T00:12:59 * ldlework sulks. 2011-12-16T00:13:03 ok, a1k0n has a bug in his bot, watch what it does in the beginning of this game: http://aichallenge.org/visualizer.php?game=247705&user=10 2011-12-16T00:13:20 Is anyone using evolvers for this challenge? 2011-12-16T00:13:57 ChrisH_: yes, yes he does 2011-12-16T00:14:08 ldlework: not i 2011-12-16T00:15:08 ChrisH_: is that old? 2011-12-16T00:15:11 If anyone has python installed and wants to try my parallel project to this competition, https://github.com/dustinlacewell/eater-experiment 2011-12-16T00:15:15 I could swear we talked about it on irc 2011-12-16T00:15:19 it requires pyevolve 2011-12-16T00:15:39 BenJackson: no, that was my most recent game 2011-12-16T00:15:47 30 minutes ago 2011-12-16T00:15:53 ldlework: years ago (after getting mad at chapter 1 of _The Selfish Gene_) I wrote a generic algoritm that evolved ants that eat 2011-12-16T00:15:56 similar to this 2011-12-16T00:16:05 it actually evolved microcode to search and move and eat 2011-12-16T00:16:06 Nice. 2011-12-16T00:16:10 it was really interesting 2011-12-16T00:16:30 BenJackson, my agents use a look up table for what their neighbors are, to what action they should take 2011-12-16T00:16:37 except they have 10 of these stables 2011-12-16T00:16:54 I had something like 10 opcodes 2011-12-16T00:16:54 what they see maps to the action and the new state 2011-12-16T00:17:01 eat, go forward, turn left, turn right, skip, again, reset 2011-12-16T00:17:03 so they can have multi-step behaviors 2011-12-16T00:17:08 some of those pushed booleans on a stack 2011-12-16T00:17:10 some consumed them 2011-12-16T00:17:28 and it would just breed instruction strings until ants could find food on their own 2011-12-16T00:17:31 oh, "look" 2011-12-16T00:17:31 is food ahead of me 2011-12-16T00:17:48 BenJackson: In fact, it is the current game on the home page. ;) 2011-12-16T00:17:55 *** Areks has joined #aichallenge 2011-12-16T00:18:01 right the way I do it is simply a full map of every possible combination of what they can see to random actions 2011-12-16T00:18:16 oh, "look" there's a plant infront of me 2011-12-16T00:18:18 ChrisH_: your bot looks super dominant in that game and pretty much lost 2011-12-16T00:18:38 also I was pretty stoked when I noticed myself on the front page once 2011-12-16T00:18:41 BenJackson, I have two breeds in my system. PEaters that gain fitness from eating plants 2011-12-16T00:18:45 but I don't pay attention so it was a fluke I noticed 2011-12-16T00:18:52 and MEaters who gain fitness from eating the PEaters 2011-12-16T00:18:58 hah 2011-12-16T00:19:09 the PEaters evolve both plant seeking behaviors and MEater avoidance behaviors 2011-12-16T00:19:28 The progression of evolved behaviors in the system are drastically different depending on whether there are MEaters in the system 2011-12-16T00:19:28 BenJackson: yeah, it's the balance between early aggression and territory control 2011-12-16T00:20:10 http://tcpants.com/replay.1486 it ends the games pretty early aye? 2011-12-16T00:20:16 I really really wish I would have known about this :( 2011-12-16T00:20:21 <_flag> The turn limit on tcpants is only 750 though 2011-12-16T00:20:28 But here's the thing, teritory control will often lead to at least a middle of the pack finish, while early aggression will lead to an all or nothing bot 2011-12-16T00:20:55 i feel robbed of that game 2011-12-16T00:21:05 antimatroid: yeah at 750 2011-12-16T00:21:16 <_flag> Both my games so far didn't have enough turns 2011-12-16T00:21:19 _flag: but it got called off even before that, for not razing hills 2011-12-16T00:21:26 ChrisH_: yep 2011-12-16T00:21:30 look at the end of the game :| 2011-12-16T00:21:38 I've been migrating from an all-or-nothing to territory bot over time 2011-12-16T00:21:48 i broke through their barriers and just made it to almost claiming their hill, and no go apparently :( 2011-12-16T00:21:51 <_flag> antimatroid: Oh wow, I hate it when it does that 2011-12-16T00:22:17 and since a lot of the other top bots go for territory as well it is hard to get close enough to their hills to be successful against them with an aggressive attack 2011-12-16T00:22:37 antimatroid: OH that's not the turn limit, ahahahah 2011-12-16T00:22:48 janzert: http://tcpants.com/replay.1486 I don't think that game should have ended, hmph :( 2011-12-16T00:23:11 tcpants seems like it is using very aggressive cutoffs 2011-12-16T00:23:17 ChrisH_: even being "early aggressive" doesn't always help 2011-12-16T00:23:24 who runs tcpants? 2011-12-16T00:23:31 i wonder if the ants not razing is any more agressive 2011-12-16T00:23:37 I lost a 3-way battle when I actually got to the enemy first, drew off his defenders and let the 3rd guy in 2011-12-16T00:23:49 I stared at that game wondering wtf I could do differently 2011-12-16T00:23:49 yep 2011-12-16T00:24:16 but you can't win every single game unless the competition isn't any good 2011-12-16T00:24:23 so you just got to let some slide 2011-12-16T00:24:35 yeah, it's all about playing best in general 2011-12-16T00:24:38 antimatroid: well there's certainly no way you had >85% of ants for 150 turns 2011-12-16T00:24:41 and let the law of averages work in your favor 2011-12-16T00:25:12 okay cool 2011-12-16T00:25:53 looks like his cutoff might be >65% for 150 turns possibly 2011-12-16T00:26:08 janzert: what's up with my match schedule, I got a bunch of lame match quality games to start after I resubmitted, then all of a sudden I got six matches in 13 minutes and vaulted into the top 25 without even realizing. 2011-12-16T00:26:10 yeah i just care about the official server settings :) 2011-12-16T00:26:35 janzert: yeah, what's up with that. I want that too 2011-12-16T00:26:48 the pairing is having a hard time keeping up with the game rate at times 2011-12-16T00:26:51 ChrisH_: but seriously I think what happens is your mu hits some threshold where you can be pulled into other people's games 2011-12-16T00:26:57 you don't want the first 8 lame matches I bet 2011-12-16T00:27:13 so ends up repeatedly giving the same person repeated seed 2011-12-16T00:27:17 BenJackson: I think that's part of it, but I've never seen this many in a row before. 2011-12-16T00:27:19 ChrisH_: if it got me into the top 1000... 2011-12-16T00:27:30 *** Jak_o_Shadows1 has quit IRC (Remote host closed the connection) 2011-12-16T00:29:05 janzert: any idea why I was getting so many matches against opponents with a mu 10 less than mine early in my submission's history? 2011-12-16T00:29:36 probably they had played significantly less games than the bots closer in mu 2011-12-16T00:29:44 in the past 24 hours that is 2011-12-16T00:29:53 that'd be my initial guess at least 2011-12-16T00:30:28 ah, we have too few mid-tier bots to support all the resubmits of the top tier 2011-12-16T00:30:30 makes sense, but it seemed to be a pattern, I would expect it to be more random: I would get some the other direction too, but they were always low 2011-12-16T00:30:45 My most recent match was against 2 deactivated bots. One ("virus") has had 6 matches since then and the other ("jocke") had one... and I'm still sitting there. 2011-12-16T00:30:54 ChrisH_: but! there are even fewer ones with higher mu 2011-12-16T00:30:55 the higher ranks have lots of games pretty evenly 2011-12-16T00:32:56 well, all is well that ends well, I have learned that my recent changes seem successful, but the matching system seems to be getting a bit unstable 2011-12-16T00:33:39 It will be pretty wild when we reset for finals. 2011-12-16T00:34:05 up to 13th 2011-12-16T00:35:03 for the finals I'm going to run a few rounds of random pairings to kickstart the pairing system 2011-12-16T00:35:29 that makes sense 2011-12-16T00:35:33 "virus" has now had 9 since his match with me. That's 3 more than when I whined up there at 21:30. 2011-12-16T00:36:33 janzert: that sounds good 2011-12-16T00:36:58 Scryer: wow, what the hell? 2011-12-16T00:37:20 Yes, exactly what I was wondering. 2011-12-16T00:37:35 Scryer: Technically, he hasn't had any matches... he's being used in them. 2011-12-16T00:37:48 ok, i've got tests to run and code to work on, IRC off. 2011-12-16T00:37:54 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-16T00:37:55 And two more since then. 2011-12-16T00:38:30 Yes, but he's been used in 11 matches. How does this makes sense again? 2011-12-16T00:39:43 i don't think it cares about how many games opponents have had when picking them 2011-12-16T00:39:48 in past 24 hours he's been in 16 total 2011-12-16T00:40:11 so he'll be past over pretty soon here until tomorrow again 2011-12-16T00:40:16 antimatroid: it does 2011-12-16T00:40:23 ah okay 2011-12-16T00:41:00 Hmm. 2011-12-16T00:42:23 he's favored by the pairing system because of a low sigma so he'll get picked whenever his game number in the past 24 hours drop senough to make him favorable again 2011-12-16T00:42:49 damn tcp is distracting :P 2011-12-16T00:42:59 favored for games occurring with a close mu of course 2011-12-16T00:43:15 OK. So working as intended? 2011-12-16T00:44:08 you'll notice he had no games between Dec 15th 6:08am and Dec 16th 4:49am 2011-12-16T00:44:27 Scryer: more an unintended side effect, but not harmful 2011-12-16T00:45:35 and none from 14th 5:13am to 15th 4:37am 2011-12-16T00:46:45 OK. I'll nap out and pay attention to my own games instead, then. :) 2011-12-16T00:49:53 i timed out on tcpants :( 2011-12-16T00:55:18 http://stackoverflow.com/questions/3601602/what-are-rvalues-lvalues-xvalues-glvalues-and-prvalues mind blown 2011-12-16T00:57:25 *** jstrong has quit IRC (Quit: Leaving.) 2011-12-16T00:58:58 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has quit IRC (Quit: Lost terminal) 2011-12-16T01:10:07 *** ldlework has quit IRC (Ping timeout: 240 seconds) 2011-12-16T01:19:05 *** Scryer has quit IRC (Quit: Page closed) 2011-12-16T01:19:17 *** pairofdice has joined #aichallenge 2011-12-16T01:22:07 *** ldlework has joined #aichallenge 2011-12-16T01:22:08 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-16T01:26:25 http://tcpants.com/replay.1538 stupid stupid bot 2011-12-16T01:28:45 lol, my bot does similar stuff too sometimes 2011-12-16T01:28:55 I have it fixed in a branch but it causes other bugs 2011-12-16T01:30:34 handling similar bug 2011-12-16T01:31:04 wondering if sacrificing means anything if ants are very near enemy hills 2011-12-16T01:31:44 depends on if other enemies are near 2011-12-16T01:31:56 if it's just you you could maximise enemy deaths and minimise friend deaths 2011-12-16T01:32:52 I mean, count enemies vs ally around k radius near enemy ant hill to decide sacrifices 2011-12-16T01:33:44 coz, razing is more valuable isn't it 2011-12-16T01:33:56 I see top bots sacrificing a lot 2011-12-16T01:36:09 FlagCapper resubmits again 2011-12-16T01:36:54 Google patents the self driving car 2011-12-16T01:37:53 who gets the blame of a crash if it's a self driving car? 2011-12-16T01:37:58 and how the hell would insurance work? 2011-12-16T01:38:40 Great questions 2011-12-16T01:38:55 *** replore has quit IRC (Remote host closed the connection) 2011-12-16T01:39:24 my thoughts are that the company who owns the software is at fault if it messes up? which makes insurance difficult 2011-12-16T01:39:27 *** ldlework has quit IRC (Ping timeout: 240 seconds) 2011-12-16T01:39:29 But atleast the goal is to have a lot less crashes 2011-12-16T01:39:32 presumably you want the driver to be paying the insurance 2011-12-16T01:39:37 but to cover the patent holder 2011-12-16T01:39:47 is it? or lazyness? :P 2011-12-16T01:40:21 if human lives were really worth any cost, nobody would drive anywhere except emergency/delivery vehicles 2011-12-16T01:40:22 Over a million fatalities a year is a good motivator 2011-12-16T01:41:03 Right... 2011-12-16T01:41:23 that's my response to people who say no expense is too much to save a human life 2011-12-16T01:42:21 antimatroid: No expense is too much... inconvenience, on the other hand.. :) 2011-12-16T01:42:48 inconvenience is an expense in this context :P 2011-12-16T01:43:50 What are you saying? Nobody should drive anywhere? 2011-12-16T01:44:08 Because that would be a bit naive 2011-12-16T01:44:54 People aren't going to stop living because they're afraid of death 2011-12-16T01:45:48 i'm not saying that 2011-12-16T01:45:55 *** ChrisH_ has joined #aichallenge 2011-12-16T01:45:56 i'm just saying society does place prices on peoples lives 2011-12-16T01:46:00 maybe not explicity, but it does 2011-12-16T01:46:19 Sure, you have to weigh riskis 2011-12-16T01:46:20 it's just my response to people who say something like above, i'm not proposing we stop driving 2011-12-16T01:47:07 Anyway, the Google car seems to be pretty safe 2011-12-16T01:47:16 I wonder how it handles winter 2011-12-16T01:47:33 @later tell janzert The combination of the scheduling baglog clumping games for players and the 24 hour time frame of the game limits seem to be resonating with each other. Take a look at http://aichallenge.org/profile.php?user=77. That bot gets a batch of 12 games every 12 hours and then nothing for another 12 hours. 2011-12-16T01:47:33 ChrisH_: Yes master! 2011-12-16T01:48:17 ChisH_: see discussion with Scryer earlier ;) 2011-12-16T01:48:52 ok 2011-12-16T01:51:15 *** yoden-cloud has quit IRC (Quit: Page closed) 2011-12-16T01:51:27 *** Palmik has joined #aichallenge 2011-12-16T01:55:16 * antimatroid comes up with a conspiracy theory about this being the original purpose of google maps :P 2011-12-16T01:55:34 well, street stuff 2011-12-16T01:55:51 *** Kingpin13 has joined #aichallenge 2011-12-16T01:59:27 *** sigh has joined #aichallenge 2011-12-16T02:03:18 long game going on now 2011-12-16T02:04:23 broke into top 10 on tcpants 2011-12-16T02:13:57 mmmm, bbq time 2011-12-16T02:18:05 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-16T02:18:11 http://tcpants.com/replay.1572 :P 2011-12-16T02:18:13 now i can leave 2011-12-16T02:22:11 http://www.youtube.com/watch?v=4BkLKhIEPpc 2011-12-16T02:22:13 that explains my username 2011-12-16T02:50:54 OperationalError: database is locked 2011-12-16T02:50:57 BenJackson: 2011-12-16T02:50:57 ... 2011-12-16T02:52:08 restarted 2011-12-16T02:54:09 hey 2011-12-16T03:03:28 my first game after resubmit was against "6ixela" 2011-12-16T03:03:39 so I didn't realize I'd played again against "6legs" 2011-12-16T03:04:44 *** Thinq has quit IRC (Read error: Connection reset by peer) 2011-12-16T03:11:14 nice, when the tcp server restarted all 3 of my bots went into a busy loop and heated up my room 6 degrees 2011-12-16T03:11:16 and I woke up... 2011-12-16T03:13:09 I guess the 100% CPU fan whirring sound probably helped 2011-12-16T03:20:19 oh, X was also eating 100% of a CPU core 2011-12-16T03:20:46 computer in bedroom == bad idea 2011-12-16T03:22:11 *** thestinger has quit IRC (Quit: WeeChat 0.3.6) 2011-12-16T03:28:20 back when I had to do that I was poor and my harddrives were super noisy 2011-12-16T03:36:00 *** aerique has joined #aichallenge 2011-12-16T03:45:36 *** Israfel has quit IRC (Ping timeout: 240 seconds) 2011-12-16T03:50:53 *** sigh has quit IRC (Remote host closed the connection) 2011-12-16T03:51:20 *** Surya has joined #aichallenge 2011-12-16T03:57:35 *** Israfel has joined #aichallenge 2011-12-16T03:57:48 http://ants.fluxid.pl/replay.36543 2011-12-16T03:57:59 easier to tell it's a network problem when two bots by the same guy time out 2011-12-16T03:59:51 @later tell ChrisH_ get a load of this map: http://ants.fluxid.pl/replay.35878 2011-12-16T03:59:51 BenJackson: As you wish. 2011-12-16T04:03:19 *** mj41 has joined #aichallenge 2011-12-16T04:08:18 *** weezor has joined #aichallenge 2011-12-16T04:11:07 *** mj41 has quit IRC (Ping timeout: 252 seconds) 2011-12-16T04:11:59 thestinger, that reminded me of: http://www.bitcoinminingaccidents.com/?p=271 2011-12-16T04:12:00 *** Palmik has quit IRC (Read error: Connection reset by peer) 2011-12-16T04:12:06 ah, he's not here anymore 2011-12-16T04:12:16 *** Palmik has joined #aichallenge 2011-12-16T04:13:06 wow 2011-12-16T04:13:42 Can anyone look at http://paste.aichallenge.org/JgGuO/, and explain why there is a collision at row 28 col 20 at turn 4? I am clearly sending o 28 20 S 2011-12-16T04:13:43 you know what's more fun? watching a bunch of games your bot wins 2011-12-16T04:13:48 rather than poring over all the defeats 2011-12-16T04:16:08 weezor: you can't move onto food 2011-12-16T04:16:12 you gather food by standing next to it 2011-12-16T04:16:20 so one of your ants didn't move ("move blocked") 2011-12-16T04:17:08 Oh. I thought i could move onto food. Stupid me, thanks 2011-12-16T04:22:32 *** ikaros has joined #aichallenge 2011-12-16T04:24:15 *** replore_ has joined #aichallenge 2011-12-16T04:26:17 *** mj41 has joined #aichallenge 2011-12-16T04:27:41 BenJackson: because i stopped developing my bot, i'm watching mostly games in which i win 2011-12-16T04:27:43 *** Vaenom has joined #aichallenge 2011-12-16T04:27:50 there wasn't one lately tho ;) 2011-12-16T04:29:16 hey, is using visualizer.jar with tcp servers possible? 2011-12-16T04:30:28 it doesn't seem to simply like being piped a la "| java -jar visualizer.jar" 2011-12-16T04:32:27 *** mj41 has quit IRC (Ping timeout: 240 seconds) 2011-12-16T04:33:00 *** sigh has joined #aichallenge 2011-12-16T04:35:10 Fluxid? 2011-12-16T04:45:18 *** GeorgeSebastian has joined #aichallenge 2011-12-16T04:45:18 *** GeorgeSebastian has joined #aichallenge 2011-12-16T04:49:24 *** weezor has quit IRC (Ping timeout: 258 seconds) 2011-12-16T04:55:29 *** Kurnevsky has joined #aichallenge 2011-12-16T04:57:40 *** replore_ has quit IRC (Remote host closed the connection) 2011-12-16T05:08:40 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-16T05:13:21 *** g0llum has joined #aichallenge 2011-12-16T05:19:46 *** kire has joined #aichallenge 2011-12-16T05:25:22 *** nha has joined #aichallenge 2011-12-16T05:26:12 Vaenom: uh, what do you mean? what do you expect to get by piping (and what do you pipe to visualizer?)? 2011-12-16T05:26:56 Fluxid: TcpClient, on your server actually. (thanks) 2011-12-16T05:30:22 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-16T05:39:31 you can't pipe from tcpclient because afaik visualizer expects json piped in. and you can't "pipe on my server" because you don't have access ;) 2011-12-16T05:44:09 BenJackson: that's why you write a bot that never loses 2011-12-16T05:44:13 more fun 2011-12-16T05:53:12 *** AntDroid has quit IRC (Quit: Page closed) 2011-12-16T05:53:53 *** Areks has joined #aichallenge 2011-12-16T05:59:53 *** kaemo has joined #aichallenge 2011-12-16T06:08:52 *** iglo has joined #aichallenge 2011-12-16T06:16:13 Fluxid: Okay, thanks 2011-12-16T06:28:35 *** liberforce has joined #aichallenge 2011-12-16T06:29:03 *** codetiger has quit IRC (Remote host closed the connection) 2011-12-16T07:03:31 *** flowenol has joined #aichallenge 2011-12-16T07:36:35 *** TheLinker has joined #aichallenge 2011-12-16T07:54:52 *** ademar__ has joined #aichallenge 2011-12-16T07:55:08 *** ademar_ has quit IRC (Ping timeout: 252 seconds) 2011-12-16T07:56:50 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-12-16T08:03:36 *** iglo has quit IRC (Quit: .) 2011-12-16T08:05:03 *** zyberkiddy has joined #aichallenge 2011-12-16T08:06:20 *** u_ has quit IRC (Quit: u_) 2011-12-16T08:13:56 *** GeorgeSebastian is now known as eleanora 2011-12-16T08:14:10 *** eleanora has joined #aichallenge 2011-12-16T08:14:20 *** eleanora is now known as GeorgeSebastian 2011-12-16T08:14:27 *** GeorgeSebastian has joined #aichallenge 2011-12-16T08:17:35 *** zyberkiddy has quit IRC (Read error: Operation timed out) 2011-12-16T08:17:40 *** mcstar has quit IRC (Ping timeout: 240 seconds) 2011-12-16T08:18:09 *** SMJ has joined #aichallenge 2011-12-16T08:19:04 *** mcstar has joined #aichallenge 2011-12-16T08:19:12 HEHE 2011-12-16T08:19:19 my first useful haskell program just ate all my memory 2011-12-16T08:19:25 fortunately i could kill it 2011-12-16T08:19:57 rwest_: it was called rwest 2011-12-16T08:22:30 *** kaemo has quit IRC (Ping timeout: 255 seconds) 2011-12-16T08:25:57 *** sigh has quit IRC (Remote host closed the connection) 2011-12-16T08:28:36 *** u_ has joined #aichallenge 2011-12-16T08:31:57 *** u__ has joined #aichallenge 2011-12-16T08:34:38 *** u_ has quit IRC (Ping timeout: 240 seconds) 2011-12-16T08:34:38 *** u__ is now known as u_ 2011-12-16T08:38:34 @later tell rwest_ i have a haskell solution to your problem, but it doesnt scale well, honestly, its horribly slow 2011-12-16T08:38:34 mcstar: Yes master! 2011-12-16T08:45:33 *** Accoun has quit IRC (Read error: Connection reset by peer) 2011-12-16T08:46:24 *** Accoun has joined #aichallenge 2011-12-16T08:50:40 *** kevlar has quit IRC (Quit: This computer has gone to sleep) 2011-12-16T08:53:18 *** bob_ has joined #aichallenge 2011-12-16T08:54:43 *** k333es has joined #aichallenge 2011-12-16T08:55:40 *** bergmark has quit IRC (Ping timeout: 276 seconds) 2011-12-16T08:58:14 *** k333es has quit IRC (Client Quit) 2011-12-16T09:05:18 *** UncleVasya has joined #aichallenge 2011-12-16T09:11:05 *** besh has joined #aichallenge 2011-12-16T09:11:45 *** Antimony has joined #aichallenge 2011-12-16T09:13:45 mcstar: welcome back! We were missing you :) 2011-12-16T09:13:56 hi 2011-12-16T09:14:02 *** sigh has joined #aichallenge 2011-12-16T09:14:04 sweet 2011-12-16T09:14:16 you didnt miss much, believe me 2011-12-16T09:15:30 UncleVasya: i got into the first 100 2011-12-16T09:15:43 wont last long though 2011-12-16T09:16:10 Into the first 100 among the PyPy users? :P 2011-12-16T09:16:39 pypy? 2011-12-16T09:17:11 i never used pyp 2011-12-16T09:17:13 y 2011-12-16T09:17:23 http://aichallenge.org/language_profile.php?language=PyPy 2011-12-16T09:17:31 thats not me 2011-12-16T09:17:34 thats a polish guy 2011-12-16T09:17:40 im AgentSmith 2011-12-16T09:17:57 hahah 2011-12-16T09:18:01 my bad 2011-12-16T09:18:08 My bot has won 5 games in a row, and it's still not in the top 1000. What the heck? 2011-12-16T09:18:53 I was sure you are still alive vecause you upload bot frequently. I was wrong. 2011-12-16T09:18:53 UncleVasya: i read a brief tutorial on ocaml 2011-12-16T09:19:03 why? 2011-12-16T09:19:23 why what? 2011-12-16T09:19:25 :D 2011-12-16T09:19:30 shouldnt i? 2011-12-16T09:19:46 i was looking around for functional languages 2011-12-16T09:19:47 what was your motivation? 2011-12-16T09:19:54 *** g0llum has joined #aichallenge 2011-12-16T09:19:55 and ocaml had a nice reputation speed wise 2011-12-16T09:19:55 just like me :) 2011-12-16T09:20:08 but it turned out im not gonna like it 2011-12-16T09:20:35 http://sds.podval.org/ocaml-sucks.html 2011-12-16T09:20:52 One guy from my country who writes in Pascal overpowered me :( 2011-12-16T09:21:01 i decided i should search for "why xyz language sucks" instead why does it rock 2011-12-16T09:21:50 There is an article 'Why OCaml language sucks'. 2011-12-16T09:22:43 do you have some secret weapon for the finals? 2011-12-16T09:22:47 do people randomise the order of directions in which they search? 2011-12-16T09:23:02 antimatroid: i go straight to the fridge 2011-12-16T09:24:01 so, xfire is probably the next challenge? 2011-12-16T09:25:55 Fluxid: http://pastebin.com/uasjYFK8 2011-12-16T09:26:02 *** Migi32 has joined #aichallenge 2011-12-16T09:26:07 have you considered that problem? and if so what have you done about it? 2011-12-16T09:26:56 hey everyone. I'm back. 2011-12-16T09:27:07 school's over now, exams aren't until another 3 weeks 2011-12-16T09:27:10 let's do this :P 2011-12-16T09:27:15 Migi32: :) 2011-12-16T09:27:24 another wannabe.... 2011-12-16T09:27:24 mcstar: No, I don't. I had some ideas for little improvements but I've got an important 2 weeks in study-place. What's the name of the week when you get your main marks for the whole semestr? 2011-12-16T09:27:58 UncleVasya: hell week 2011-12-16T09:28:04 rwest: hey 2011-12-16T09:28:09 have you seen my message? 2011-12-16T09:28:10 yeahhh 2011-12-16T09:28:47 UncleVasya: 2011-12-16T09:28:53 accidental enter 2011-12-16T09:28:56 mcstar: no, it was probably sent to rwest_ 2011-12-16T09:29:00 yep 2011-12-16T09:29:02 I am at work now heh 2011-12-16T09:29:17 i wrote a haskell program that solves your problem 2011-12-16T09:29:25 unfortunately it generates a lot of lists 2011-12-16T09:29:39 it solves the 48x10 in under a minute? 2011-12-16T09:29:46 and combinatorial explosion forbids it from giving a number for the problem size 2011-12-16T09:29:49 no 2011-12-16T09:29:56 but solves instantly the previous one 2011-12-16T09:30:05 my solution does too 2011-12-16T09:30:06 heh 2011-12-16T09:30:16 yeah, its not useful 2011-12-16T09:30:23 but I can't get 48x3 or higher 2011-12-16T09:30:36 i dont know how to solve it wo explicitely generating those lists 2011-12-16T09:31:29 *** bergmark has joined #aichallenge 2011-12-16T09:31:43 there has to be a way to make the matrix multiplication faster 2011-12-16T09:32:01 or some short cut I am missing 2011-12-16T09:32:17 especialy since it can be done in 0.1s in C 2011-12-16T09:33:42 i dont use matrices 2011-12-16T09:34:21 i compute the all the possible combinations of 3 and 4.5 for a given length 2011-12-16T09:34:23 -the 2011-12-16T09:34:42 after that i sum of those lists like this: 2011-12-16T09:35:00 [3,3,4.5,3] -> [3,6,10.5,13.5] 2011-12-16T09:35:05 drop the last element 2011-12-16T09:35:13 these are the basis lists 2011-12-16T09:35:46 you can put such lists on top of each other as long as their pair-wise itnersection is the empty set 2011-12-16T09:36:10 I do the same thing 2011-12-16T09:36:12 so i get a graph, that describes that i can go from ilst A to B 2011-12-16T09:36:19 yea 2011-12-16T09:36:23 my matrix is your graph 2011-12-16T09:36:23 and count... 2011-12-16T09:36:41 now the problem is with the first step 2011-12-16T09:36:44 it's an "incidence matrix" 2011-12-16T09:36:53 it takes too long to generate those lists 2011-12-16T09:37:01 yes, that is why I use the matrix 2011-12-16T09:37:02 lol 2011-12-16T09:37:10 I had a graph at first 2011-12-16T09:37:22 but needed something faster, looked around and found the matrix 2011-12-16T09:37:31 antimatroid: what problem? there is nothing to do? 2011-12-16T09:37:43 own hill, few own ants, water and no enemy? 2011-12-16T09:38:12 Fluxid: move an ant in line over so the stuck ant can continue moving outwards 2011-12-16T09:38:33 suppose the a's are going to go left again for their target 2011-12-16T09:38:39 and you have a ridiculously large hive 2011-12-16T09:38:47 ok, so the one outstanding ant want to go in line 2011-12-16T09:38:53 yes 2011-12-16T09:38:57 there could be more 2011-12-16T09:39:02 this is just to highlight it 2011-12-16T09:39:04 i didn't think of it, it depends on how high priority that ant has 2011-12-16T09:39:26 i sort move orders by priority/confidence/weight and to them in that order 2011-12-16T09:39:28 but the "optimal" thing to do would be to move the line over for one a 2011-12-16T09:39:43 but i don't really know how to tell my bot that 2011-12-16T09:39:47 so if that one ant really-really needs to go somewhere, it'll go first 2011-12-16T09:40:00 if it has lower weight, it will wait until it has a chance to move 2011-12-16T09:40:06 well, either ant can get the target first 2011-12-16T09:40:17 my problem is how do you make sure both ants can get out from the hill in minimal time 2011-12-16T09:40:28 i don't make sure 2011-12-16T09:40:39 i wonder if anyone does 2011-12-16T09:40:49 no 2011-12-16T09:40:55 thats micro optimization 2011-12-16T09:41:02 anyway ants are interchangeable 2011-12-16T09:41:07 yeah, my bot has much worse issues 2011-12-16T09:41:28 well i noticed my bot had pile up issues when i tried to make it streamline the ants out better 2011-12-16T09:41:37 i'm going to try and resolve what they do at the other end i think 2011-12-16T09:41:49 i think one ant would go to right, and the waiting one take it's place 2011-12-16T09:42:13 just because even if all ants have the same confidence in reaching a goal, there is slight randomness involved 2011-12-16T09:42:30 so at some time waiting ant gets higher weight and gets in line 2011-12-16T09:43:20 oooor, because ants try to go "around" themselves 2011-12-16T09:44:06 waiting ant adds weight to heightmap, so ants going from north may go more to east eventually 2011-12-16T09:44:20 line would spread to sides 2011-12-16T09:44:22 ;P 2011-12-16T09:51:47 seeing as though it always takes 24-36hours for ranking and skill of a new bot to converge, should I only submit an update at a minimum of 36hours ahead of the Sunday deadline? 2011-12-16T09:52:40 bob_: pretty sure ranking is reset before final contest is run 2011-12-16T09:55:51 OK - thanks. maybe I'll carry on after tonight then :) 2011-12-16T09:57:47 ug, my skill is going up so slowly because every game has 2 or 3 opponents 2011-12-16T09:58:35 O_o 2011-12-16T10:00:55 My bot has won it's first 6 games but it still only rank 825 2011-12-16T10:02:52 http://aichallenge.org/visualizer.php?game=251446 whoa, the hell 2011-12-16T10:02:56 that graph 2011-12-16T10:04:37 a bit suicidal 2011-12-16T10:05:51 http://aichallenge.org/visualizer.php?game=250469&user=13978 oh, burger shows nice behaviour 2011-12-16T10:05:58 * Fluxid is game-walking 2011-12-16T10:06:38 why? 2011-12-16T10:06:44 he is expanding too slowly 2011-12-16T10:07:06 more like, interesting 2011-12-16T10:07:06 and that kind of "repulsive" thing appeared very early in the contest 2011-12-16T10:07:10 *** Areks|2 has joined #aichallenge 2011-12-16T10:07:25 well, i didn't manage to do such nice effect 2011-12-16T10:07:50 i think its only useful for territory control 2011-12-16T10:08:14 i had similar results with a simple potential summation 2011-12-16T10:08:32 *** stocha has joined #aichallenge 2011-12-16T10:09:14 * Fluxid googling for potential summation 2011-12-16T10:09:19 uh, neurobiology? 2011-12-16T10:09:59 no 2011-12-16T10:10:05 physics.. 2011-12-16T10:10:09 for all nbors of my ants: 2011-12-16T10:10:14 for all myants: 2011-12-16T10:10:16 *** besh has quit IRC (Quit: Page closed) 2011-12-16T10:10:19 *** Areks has quit IRC (Ping timeout: 268 seconds) 2011-12-16T10:10:25 add+=1/(1+r^2) 2011-12-16T10:10:34 and choose the min nbor 2011-12-16T10:10:45 it resulted in a nice triangular lattice 2011-12-16T10:10:48 *** olexs has joined #aichallenge 2011-12-16T10:10:53 you can tune the constants 2011-12-16T10:11:28 damn, i tried something similar... but yeah, experimenting without knowing stuff leads nowhere after all 2011-12-16T10:11:34 jabbbotts visualizer is a godsend 2011-12-16T10:11:54 yeah, it helped me track a segfault, ;D 2011-12-16T10:12:06 im using it to draw strategy 2011-12-16T10:12:11 *** Areks|2 has quit IRC (Ping timeout: 268 seconds) 2011-12-16T10:12:25 it seems i messed up my response function 2011-12-16T10:12:32 that calculates necessary retaliation 2011-12-16T10:14:17 segfault? 2011-12-16T10:14:27 dont you use python? 2011-12-16T10:14:49 ah ok 2011-12-16T10:15:00 I'm really excited for the incoming final phase. I hope it will be a success (lot of games, stable final ranking, good visibility and so on) 2011-12-16T10:15:35 what is mtf? 2011-12-16T10:15:39 mcstar: yeah, mostly 2011-12-16T10:15:51 Fluxid: i just forgot about your c extensions 2011-12-16T10:15:54 ;) 2011-12-16T10:15:54 It's my software company. 2011-12-16T10:16:34 You looked for my ranking right away ;) 2011-12-16T10:16:48 *** Antimony has quit IRC (Ping timeout: 248 seconds) 2011-12-16T10:17:00 My goal is to be top 100. I'm almost sure I won't make it. But I still got 2 days :} 2011-12-16T10:17:08 *** keith_ has joined #aichallenge 2011-12-16T10:17:16 http://paste.aichallenge.org/gBbfF/ 2011-12-16T10:17:21 can someone explain that to me? 2011-12-16T10:17:24 2 days still ? 2011-12-16T10:17:28 when does it start ? 2011-12-16T10:17:36 Surya: i have the same goal 2011-12-16T10:17:37 On monday night 2011-12-16T10:17:41 exact utc time ? 2011-12-16T10:17:49 ah stocha 2011-12-16T10:17:58 -5 I think 2011-12-16T10:17:59 its been so long 2011-12-16T10:18:00 so people have the week end to build a bot if they want to 2011-12-16T10:18:01 its EST 2011-12-16T10:18:07 at midnight sunday 2011-12-16T10:18:19 23:59 rather 2011-12-16T10:18:29 I wonder if i should try to hack a bot. 2011-12-16T10:18:30 so 18:59 UTC? 2011-12-16T10:18:36 maybe i'll ask permission to my wife :p 2011-12-16T10:18:47 anybody take a look at that paste? 2011-12-16T10:18:56 rwest: i see nothign wrong with it 2011-12-16T10:19:08 the turn time limit is still 500 ms ? 2011-12-16T10:19:14 stocha: yes definitely ask away and pls play 2011-12-16T10:19:20 wow stocha! 2011-12-16T10:19:22 mcstar: both same distance, yet the ones on top don't fight 2011-12-16T10:19:30 *** keith_ has quit IRC (Client Quit) 2011-12-16T10:19:31 mcstar: there is few chances i'll reach the top 100 2011-12-16T10:19:33 at turn 1 2011-12-16T10:20:01 i don't even know if i can have a bot with fight code and the new protocol handling by then. 2011-12-16T10:20:04 rwest: i dont get it 2011-12-16T10:20:07 Nothing odd there, rwest 2011-12-16T10:20:10 A snake-bot creater, a replieCrush owner, sto_greed maker and something more I guess :) 2011-12-16T10:20:28 Mcstar, what's your nickname IG? 2011-12-16T10:20:34 UncleVasya: i like it the new game, looking at those bots :) 2011-12-16T10:20:37 pairofdice: both sets of ants are 3 spaces apart, vertical ones fight, horizontal do not? 2011-12-16T10:20:49 battle doesn't happen 3 spaces apart 2011-12-16T10:20:53 it became strategic in the end. i wish i had a free month to play with the new rules :p 2011-12-16T10:20:54 rwest: they fight when they get in ar2 2011-12-16T10:20:58 they arent in ar2 2011-12-16T10:21:06 they will be after they step 2011-12-16T10:21:08 bottom group moves closer together, use the ui buttons to see it 2011-12-16T10:21:13 It's ironic that the battle system was the most resilient parameter in the end :p 2011-12-16T10:21:13 those that will be that is 2011-12-16T10:21:33 rwest, it's this close to the finals and yoiu notice this behaviour now? 2011-12-16T10:21:38 pairofdice: combat animations shows up first, I see 2011-12-16T10:21:41 UncleVasya: how have your bot been doing ? 2011-12-16T10:21:46 mcstar: what about yours ? 2011-12-16T10:21:52 pairofdice: I didn't even start my bot til dec 2011-12-16T10:21:53 what? 2011-12-16T10:21:57 ah 2011-12-16T10:21:59 im AgentSmith 2011-12-16T10:22:06 mcstar: how have your bot performed ? 2011-12-16T10:22:10 pairofdice: now it's combat time hah 2011-12-16T10:22:12 stocha: Surya 2011-12-16T10:22:22 take a look if you want to 2011-12-16T10:22:35 i can't right now. i'm at work ... 2011-12-16T10:23:08 im 98th right now 2011-12-16T10:23:21 *** sigh has quit IRC (Remote host closed the connection) 2011-12-16T10:23:28 has everyone put their bot online yet ? 2011-12-16T10:23:32 no 2011-12-16T10:23:39 antimatroid is waiting for the end 2011-12-16T10:23:44 what is the game rate at the moment ? 2011-12-16T10:23:57 that's how i roll :P 2011-12-16T10:24:29 if i can find 20 hours or so this week end i'll try to hack up something to submit :p 2011-12-16T10:24:35 16:17:28 < stocha> when does it start ? 2011-12-16T10:24:45 IT'S IN TOPIC 2011-12-16T10:24:49 stocha: It was somewhere between 22 and 28 when I submitted new version few days ago. The New version is way stronger but now got only 51st place (althought 33 games already played ). 2011-12-16T10:24:52 in any event, i'll be happy to look at all the post contest chatting :) 2011-12-16T10:25:01 Also on the frontpage 2011-12-16T10:25:08 UncleVasya: nice. 2011-12-16T10:25:17 I want a Java bot to dissect 2011-12-16T10:25:46 pairofdice: ideas will be more interesting than code. Unless a lot of optimizing was needed. 2011-12-16T10:25:52 I think it is because I optimized new version too hard to fight well old one (I can test my bot only locally). 2011-12-16T10:26:03 I'm only interested in the code 2011-12-16T10:26:13 *** stocha has quit IRC (Quit: Page closed) 2011-12-16T10:26:23 stocha bye 2011-12-16T10:26:32 he always leaves in a hurry 2011-12-16T10:26:32 starting with what skill bots start to d something meaningful? 2011-12-16T10:26:42 i think his wife caught him chatting 2011-12-16T10:26:48 *** Accoun has quit IRC () 2011-12-16T10:26:48 haha 2011-12-16T10:27:17 http://aichallenge.org/visualizer.php?game=250203&user=6081 perfect! 2011-12-16T10:27:28 BEST GAME EVER 2011-12-16T10:27:39 hoaxer 2011-12-16T10:27:44 dont watch the link!! 2011-12-16T10:27:52 lol 2011-12-16T10:27:56 watch->follow 2011-12-16T10:29:39 *** Accoun has joined #aichallenge 2011-12-16T10:30:08 Accoun: whats your bot? 2011-12-16T10:30:15 Accoun: Hi. Do you have a bot? 2011-12-16T10:30:17 are you participating? 2011-12-16T10:30:18 Fluxid: starter bots much? 2011-12-16T10:30:19 lol 2011-12-16T10:30:40 UncleVasya: reading my mind, eh? 2011-12-16T10:31:01 no, it was you 2011-12-16T10:32:13 pairofdice: yeah, but it's great, non-starter bot didn't have to do ANYTHING 2011-12-16T10:32:24 and won with huge bonus 2011-12-16T10:32:25 i love it 2011-12-16T10:32:53 Those north-seeking bots are something 2011-12-16T10:33:58 Didn't even have any of the modifications of the tutorial 2011-12-16T10:35:29 around skill 41 bots start to do something more than just go north 2011-12-16T10:37:03 :) 2011-12-16T10:37:06 *** amstan has joined #aichallenge 2011-12-16T10:37:06 *** ChanServ sets mode: +o amstan 2011-12-16T10:37:48 http://aichallenge.org/visualizer.php?game=248635&user=4140 another interesting game, this time bot "doing something more" timeouts, and thus he's 9th and 8 starterbots tied as 1st 2011-12-16T10:38:40 haha 2011-12-16T10:39:26 Man, I never realized how expensive exponential functions are 2011-12-16T10:39:30 How do you find those games? You just browse 4k ranked bots? :o 2011-12-16T10:39:47 How long before closing time? 2011-12-16T10:39:50 loadtime 5000 2011-12-16T10:39:50 turntime 5000 2011-12-16T10:39:55 You timed out. 2011-12-16T10:39:57 cmon now 2011-12-16T10:40:08 @roulette 2011-12-16T10:40:08 mcstar: *click* 2011-12-16T10:40:16 @roulette 2011-12-16T10:40:16 pairofdice: *click* 2011-12-16T10:40:23 @roulette 2011-12-16T10:40:23 *BANG* Hey, who put a blank in here?! 2011-12-16T10:40:24 * contestbot reloads and spins the chambers. 2011-12-16T10:40:28 @roulette 2011-12-16T10:40:28 rwest: *click* 2011-12-16T10:41:06 anyway, Surya, yeah, i'm browsing through random bots starting from lat place 2011-12-16T10:41:12 every few pages 2011-12-16T10:41:15 Samvara: where are you using exponential functions if i may ask? 2011-12-16T10:41:41 Fluxid : You should make a list of best games available somewhere :] 2011-12-16T10:41:53 http://aichallenge.org/visualizer.php?game=249098&user=2820 again, two bots fail against starterbots :( 2011-12-16T10:42:11 Surya: it'll be hard to list them all 2011-12-16T10:42:45 Fluxid: poor guys, must be disappointing to watch those games 2011-12-16T10:43:09 at least they tried 2011-12-16T10:43:12 *** Migi32 has quit IRC (Quit: bye) 2011-12-16T10:43:37 cruel bastards :P 2011-12-16T10:44:22 damn, I am gonna be higher than top 200 soon :( 2011-12-16T10:44:30 or lower 2011-12-16T10:44:35 depending how you wanna look at it heh 2011-12-16T10:45:00 my guess is searching for meaningful bots under skill of 50 is meaningless 2011-12-16T10:45:17 :> 2011-12-16T10:45:22 so my bot is timing out on any tcp game hmm 2011-12-16T10:46:01 It's a shame we can't just throw away starter pack bots. I would love to know how many people actually tried to implement something. 2011-12-16T10:46:13 much less 2011-12-16T10:46:27 i suspect 2-3k 2011-12-16T10:46:46 i guess we will know when finals start 2011-12-16T10:46:54 oh haha, sending invalid text = timeout 2011-12-16T10:46:56 whoops 2011-12-16T10:46:59 i hope they wont upload starters there 2011-12-16T10:47:06 there are bots that do something like trying to gather food but are mixed with starters 2011-12-16T10:47:36 I like the bots that get a second ant 2011-12-16T10:47:39 mcstar: uh, upload where? 2011-12-16T10:47:40 then crash them together 2011-12-16T10:47:58 aren't submitted "just going" to finals? 2011-12-16T10:48:01 Fluxid: all the bots that are now uploaded will play int he finals? 2011-12-16T10:48:09 *** bqf has quit IRC (Read error: Operation timed out) 2011-12-16T10:48:19 i think so, i remember someone saying that 2011-12-16T10:48:30 all bots restart from zero 2011-12-16T10:48:31 thats a waste of some resources 2011-12-16T10:48:40 because how do you determine which bot go to finals? 2011-12-16T10:48:52 top 300? 2011-12-16T10:48:55 no 2011-12-16T10:49:02 everybody's who resubmit 2011-12-16T10:49:07 or activate or whatever 2011-12-16T10:49:10 then someone might miss it 2011-12-16T10:49:13 yep 2011-12-16T10:49:15 :D 2011-12-16T10:49:21 it's too late to inform "hey, reupload NOW QUICKLY GOGOGOGO" 2011-12-16T10:49:22 hopefully xathis? 2011-12-16T10:49:26 no 2011-12-16T10:49:33 if xathis is good let hime win 2011-12-16T10:49:42 The lowest mu bots will be cut incrementally 2011-12-16T10:49:45 i mean dont "let" 2011-12-16T10:49:52 (yeah, just let xathis win and be over with aichallenge! ;D) 2011-12-16T10:50:05 *** Antimony has joined #aichallenge 2011-12-16T10:50:13 pairofdice: sounds good 2011-12-16T10:50:52 *** aerique has quit IRC (Quit: leaving) 2011-12-16T10:52:02 http://aichallenge.org/visualizer.php?game=247119&user=1574 yocha looks like starter bot, but... it's 10th version of bot... it even had "Test Error: compiled, but failed test cases"... so someone tried to do something but actually did nothing? 2011-12-16T10:52:03 *** bqf has joined #aichallenge 2011-12-16T10:53:40 http://aichallenge.org/profile.php?user=13505 2011-12-16T10:53:44 coworkers bot in tcl 2011-12-16T10:53:59 yes, tcl 2011-12-16T10:55:00 >Won at 1 2011-12-16T10:57:26 i'd like to see histogram presenting how much bots are there with given skill 2011-12-16T10:58:33 ok, i'm past half of the bots. still mostly starters 2011-12-16T11:00:04 Just got my new version in... really though, when do the door close? 2011-12-16T11:00:22 @topic 2011-12-16T11:00:22 Fluxid: AI Challenge (sponsored by Google): http://aichallenge.org || Channel Logs: http://contestbot.aichallenge.org || Code Repo: http://github.com/aichallenge/aichallenge || About the game (Ants): http://aichallenge.org/problem_description.php || Submission deadline: December 18 23:59 EST 2011-12-16T11:00:40 pairofdice, "The lowest mu bots will be cut incrementally" Do you have more precise infos? I wish my 250~ ranked bot will play some games before getting kicked! 2011-12-16T11:01:02 No, I don't 2011-12-16T11:01:15 It's what Janzert said 2011-12-16T11:01:19 Surya: on the first game you fail you're kicked! 2011-12-16T11:01:20 ;) 2011-12-16T11:01:43 Fluxid: I hope it won't be against xathis and cie 2011-12-16T11:01:48 :] 2011-12-16T11:01:48 ;) 2011-12-16T11:02:05 I hope they will release their code, especially this Xathis 2011-12-16T11:02:17 this is boring, i could binary search instead of going from last page 2011-12-16T11:04:21 first game: stuck between xathis and pguillory 2011-12-16T11:04:22 gg 2011-12-16T11:05:10 crossfired and gangbanged? 2011-12-16T11:05:50 sounds about right 2011-12-16T11:06:25 I hate when you get stuck between strong bots that fight over your hill kill you and take off in different directions 2011-12-16T11:06:46 you are just a fly to them 2011-12-16T11:07:05 pretty much 2011-12-16T11:07:24 swat! 2011-12-16T11:07:40 halleluja, my new bot has improved on my older one 2011-12-16T11:08:04 my new bot beats my old one, but can't beat 3 of my old one 2011-12-16T11:08:12 it beats 5 of them 2011-12-16T11:08:21 something about how my old bot just suicides into the hill 2011-12-16T11:08:32 so when 3 bots suicide on you 2011-12-16T11:08:34 - lose 2011-12-16T11:08:48 oh 2011-12-16T11:08:48 *** kire has quit IRC (Remote host closed the connection) 2011-12-16T11:08:51 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-12-16T11:09:02 i dont differentiate between enemies anymore 2011-12-16T11:09:14 (om a side-note) 2011-12-16T11:09:16 i only do when resolving combat 2011-12-16T11:09:20 i dont 2011-12-16T11:09:28 how do you know which ants live/die? 2011-12-16T11:09:30 previously i did, with the enumeration 2011-12-16T11:09:40 antimatroid: my battle is too simple to describe 2011-12-16T11:09:45 you would laugh at my face 2011-12-16T11:09:51 mcstar: that sounds like an oxymoron 2011-12-16T11:09:52 lol 2011-12-16T11:09:59 too simple to describe 2011-12-16T11:10:01 ok, bots at page 24, skill 50, place around 2357 usually do something more 2011-12-16T11:10:25 rwest: i dont want to expose it, and if i tell anything, i basically explained it 2011-12-16T11:10:32 Fluxid: that's not too bad 2011-12-16T11:10:48 you have to remember a lot of people register, submit a bot then we never see them again 2011-12-16T11:10:59 i know 2011-12-16T11:12:30 Fluxid: You can now write a book: 'The scientific expedition to the 1500+ and describing of the behaviour of primitive tribes' 2011-12-16T11:12:57 ;) 2011-12-16T11:13:06 lol 2011-12-16T11:13:19 olex.s said he is doing this for his thessi 2011-12-16T11:13:23 thesis* 2011-12-16T11:13:25 Main goal of the expedition - to figure out the line between humans and animals :) 2011-12-16T11:13:33 any words from him? 2011-12-16T11:14:16 He comes here. Not too often, but you'll see him in few days if you stay here. 2011-12-16T11:14:55 *** avdg has joined #aichallenge 2011-12-16T11:15:51 *** KestasL has joined #aichallenge 2011-12-16T11:16:30 *** KestasL has left #aichallenge 2011-12-16T11:18:31 http://swizec.com/blog/why-programmers-work-at-night/swizec/3198 2011-12-16T11:19:11 *** treeform has joined #aichallenge 2011-12-16T11:19:29 http://www.reddit.com/r/AskReddit/comments/nf8w7/you_are_killed_and_death_lets_you_choose_a_game/c38lz98 winner 2011-12-16T11:19:53 Fluxid: for me it's because you guys are mostly awake then :P 2011-12-16T11:19:56 and all the other reasons 2011-12-16T11:19:58 For there is already talks about picking next game what do you guys think about something like Worms? 2011-12-16T11:20:18 UncleVasya: would be sweet, but dunno how the mechanics would work 2011-12-16T11:20:19 Fluxid: that shit like freezes my chrome tab haha 2011-12-16T11:20:25 and i'm pretty keen to stick to simultaneous turns 2011-12-16T11:20:26 there it goes 2011-12-16T11:20:36 If you don't know what the Wormsis you can check thir oss clone: Warmux. 2011-12-16T11:20:39 but that's merely my opinion 2011-12-16T11:21:09 we could base it off the open source worms for linux? 2011-12-16T11:21:59 rwest: works great under Opera here 2011-12-16T11:22:42 something real time with over a continuous space woudl be sweet 2011-12-16T11:23:12 Fluxid: I had to close and reopen the tab, dunno what happened 2011-12-16T11:23:31 >Chrome 2011-12-16T11:23:37 that happened :) 2011-12-16T11:29:41 so at this point the contest organizers need to decide in which direction to move forward: more formal, studied games with possible optimal solutions, or more life-like games where only heuristics is applicable 2011-12-16T11:29:55 because i think Ants is inbetween this two worlds 2011-12-16T11:29:57 They don't want optimal solutions 2011-12-16T11:30:11 these* 2011-12-16T11:31:42 strict requirements are it being simple enough for beginners 2011-12-16T11:31:57 my personal next requirements are that it has enough depth to be competitive at to thetop 2011-12-16T11:32:21 yeah, we had enough simple games :) 2011-12-16T11:34:44 Well you could take a vote on what features you want it to have. Observability, Stochasticity, discrete/continuous, number of agents ... 2011-12-16T11:35:25 how about mortal combat fighters? :D 2011-12-16T11:35:52 continue; 2011-12-16T11:35:53 :D 2011-12-16T11:39:46 * antimatroid is against stochastic game mechanics 2011-12-16T11:39:55 i don't mind like how food spawning is picked 2011-12-16T11:40:07 but i despise the random ant spawn rule 2011-12-16T11:40:15 and would hate random battle resolution etc. 2011-12-16T11:40:49 the battle resolution is remarkable in this game 2011-12-16T11:41:00 remarkably awesome? 2011-12-16T11:41:12 it is very good 2011-12-16T11:41:33 randomness is good 2011-12-16T11:41:33 i dont know how did mcleo come up with it, or whoever 2011-12-16T11:41:41 jmcarthur i think 2011-12-16T11:41:58 i dont want randomness in the battle res 2011-12-16T11:42:19 maybe in a more real-world like space a little bit of it is acceptable 2011-12-16T11:42:31 but on a low-res space like ants, it would suck 2011-12-16T11:45:21 what about connect k in an mxn grid? 2011-12-16T11:45:30 is that solved? 2011-12-16T11:46:14 dunno 2011-12-16T11:46:20 but i dont want that for the next comp. 2011-12-16T11:46:25 me neither 2011-12-16T11:46:31 its too...dry 2011-12-16T11:46:33 or something 2011-12-16T11:46:36 Not sexy at all 2011-12-16T11:46:42 not at all :D 2011-12-16T11:47:05 mario kart would be awesome, but i don't see how 2011-12-16T11:48:04 *** xathis has joined #aichallenge 2011-12-16T11:48:39 xathis: feeling confident? :) 2011-12-16T11:49:10 Trackmania.. mmmm 2011-12-16T11:49:28 antimatroid: i am! But i know pguillory will reupload and _flag's bot is climbing up the rankings atm 2011-12-16T11:49:52 i'm starting to hope for top 10 2011-12-16T11:49:58 but i doubt i'll be able to touch you guys 2011-12-16T11:51:15 antimatroid: what's your name on the officiel server? 2011-12-16T11:51:22 antimatroid 2011-12-16T11:51:25 haven't uploaded yet 2011-12-16T11:51:29 ah okay 2011-12-16T11:51:44 i'm just going to let it loose in the finals 2011-12-16T11:51:49 are you on fluxid tcp? 2011-12-16T11:52:02 ...Betting starts now... 2011-12-16T11:52:02 i was, then switched to tcpants cause fluxid seemed to implode 2011-12-16T11:52:24 http://tcpants.com/player/SkyCaptain 2011-12-16T11:52:32 i'm trying to make it better still 2011-12-16T11:52:42 it's fairly stupid how it sends its ants out 2011-12-16T11:54:24 xathis: does java has anything to do with your success, or is it just the language you know the best? 2011-12-16T11:54:35 Xathis: Do you plan to release your code after the end of the challenge? I would be really interested in seeing how complex it is. 2011-12-16T11:54:51 mcstar: it's the language i know best 2011-12-16T11:55:32 Surya: yes, i will write about my bot and release the code 2011-12-16T11:55:39 Awesome 2011-12-16T11:55:49 oh 2011-12-16T11:55:52 That's really nice of you, Hope you'll win ;) 2011-12-16T11:55:57 so after all it isnt self-aware?? 2011-12-16T11:56:16 oh man.. xathis is way ahead 2011-12-16T11:56:45 i assume flag is resubmitting 2011-12-16T11:56:53 antimatroid: looks pretty good, but i noticed the sending-out-in-one-direction-only as well 2011-12-16T11:57:25 xathis: sending-out-in-one-direction? 2011-12-16T11:57:39 a lot of people are resubmitting a lot 2011-12-16T11:57:43 hence i'm in 3rd 2011-12-16T11:57:48 antimatroid: your bot sends the ants away from the hill strangely 2011-12-16T11:57:52 yeah 2011-12-16T11:57:56 i'm trying to fix that atm 2011-12-16T11:58:06 it's because it is trying to pathfind not over ants 2011-12-16T11:58:19 i'm trying to get the desired affect without doing that 2011-12-16T11:58:42 and in the process better streamlining the ants out from the hills 2011-12-16T11:59:02 then i'ma work on my combat a bit more, it might be a little too offensive atm 2011-12-16T11:59:08 then i'ma submit :) 2011-12-16T11:59:15 wait what? you're actually showing your bot? 2011-12-16T11:59:32 a1k0n: http://tcpants.com/player/SkyCaptain 2011-12-16T11:59:36 that was it playing earlier 2011-12-16T12:00:16 poor guy: http://forums.aichallenge.org/viewtopic.php?f=24&t=2135 2011-12-16T12:01:29 i'm not good at leaving back up in case enemies break through my defensive walls 2011-12-16T12:01:48 and i'm not defensive with them, they get treated like any other battle partition 2011-12-16T12:03:02 *** besh has joined #aichallenge 2011-12-16T12:04:42 antimatroid: i hope you get that done, it's not much time anymore 2011-12-16T12:05:05 i just hope it actually runs when i submit it to the server :P 2011-12-16T12:05:35 that would suck/be funny 2011-12-16T12:06:32 my bot really needs a high turn limit, i just won a game in which i had 3 pts while delineate had 11 pts at turn 350. 2011-12-16T12:07:03 yeah my bot doesn't seem amazing when there's lots of hills up for grabs at the start 2011-12-16T12:07:07 it's more in it for the long haul 2011-12-16T12:12:04 *** foRei has joined #aichallenge 2011-12-16T12:13:07 *** JorgeB has joined #aichallenge 2011-12-16T12:13:14 Approx. how many games do we get in the finals ? 2011-12-16T12:13:38 no idea 2011-12-16T12:13:38 enough 2011-12-16T12:13:53 We all want to have infos about the final :) 2011-12-16T12:14:47 I just hope my bot gains sentience by then 2011-12-16T12:15:04 *** codetiger has joined #aichallenge 2011-12-16T12:16:53 i resubmitted the same bot twice and seem to have 3 points difference after about 40 games 2011-12-16T12:17:22 BenJackson: online? 2011-12-16T12:17:31 submission isnt time-invariant 2011-12-16T12:17:54 *** TL__ has quit IRC (Ping timeout: 258 seconds) 2011-12-16T12:18:39 well it is who you are paired with that matters more. 2011-12-16T12:20:08 *** ikaros has joined #aichallenge 2011-12-16T12:21:18 *** treeform has quit IRC (Remote host closed the connection) 2011-12-16T12:22:08 *** amstan has quit IRC (Ping timeout: 240 seconds) 2011-12-16T12:22:21 *** Surya has quit IRC (Quit: Page closed) 2011-12-16T12:22:33 besh: the entire skill scale is drifting over time too 2011-12-16T12:23:03 *** praveen_ has joined #aichallenge 2011-12-16T12:23:13 i guess that's what mcstar was trying to say 2011-12-16T12:23:19 Absolute skill doens't matter very much in TrueSkill 2011-12-16T12:23:29 The number that is 2011-12-16T12:23:32 right 2011-12-16T12:24:22 *** codetiger has quit IRC (Read error: Connection reset by peer) 2011-12-16T12:24:35 yes bots strength change, new bots get submitted but with the same conditon you can get very different results if you dont take enough samples. 2011-12-16T12:24:45 3 points is pretty different though 2011-12-16T12:25:13 in the finals there wont be re-submissions 2011-12-16T12:25:17 so its moot 2011-12-16T12:25:29 well yeah, but anyway it still takes something like 100 games for it to really stabilize, i've noticed 2011-12-16T12:25:44 since trueskill just updates once after each game 2011-12-16T12:25:56 even if trueskill isnt perfect, you have to attack a number to a bot somehow 2011-12-16T12:26:05 as opposed to a whole-history reranking kind of thing like bayeselo 2011-12-16T12:26:06 what matters is that at least the top be correct 2011-12-16T12:26:17 attach* 2011-12-16T12:26:39 *** bmh has joined #aichallenge 2011-12-16T12:29:53 It is difficult to get correct rankings with just 100 games. Xathis seems like he is going to be at least 5 points better so we won't have problems there. 2011-12-16T12:32:01 Hm... can I implement combat in a day? Let's see! 2011-12-16T12:33:21 *** mviel_ has quit IRC (Remote host closed the connection) 2011-12-16T12:44:05 *** bmh has quit IRC (Quit: Page closed) 2011-12-16T12:44:16 bmh good luck with that 2011-12-16T12:45:53 (i did) 2011-12-16T12:45:57 heh 2011-12-16T12:46:09 it's almost not fair how good my bot is relative to how much time spent 2011-12-16T12:46:20 i don't even get it 2011-12-16T12:46:21 so I made a change to my bot that seemed iffy 2011-12-16T12:46:25 but I couldn't pin down why it was wrong 2011-12-16T12:46:34 now I keep wandering by hills I coudl take 2011-12-16T12:46:35 oops 2011-12-16T12:50:39 http://aichallenge.org/visualizer.php?game=252102&user=3473 2011-12-16T12:50:44 wow, that victory was xathis-esque 2011-12-16T12:52:48 *** Murashka has joined #aichallenge 2011-12-16T12:55:16 *** jstrong has joined #aichallenge 2011-12-16T12:57:46 Murashka: Hi. 2011-12-16T12:58:32 I was hoping tofinish second in Ukraine's ranking. But your bot is better :( 2011-12-16T12:58:56 UncleVasya: I was joking about trying to win *any* category but I'd have to change programming languages :) 2011-12-16T12:59:16 Murashka: And it is very annoying that some bot written in Pascal is better than mine :) 2011-12-16T13:00:28 BenJackson: Heheh, in language category I am first with huge advantage. 2011-12-16T13:00:49 *** janzert has quit IRC (Read error: Connection reset by peer) 2011-12-16T13:01:11 wow, I played at 4:40, 49, 5:07, 07 (!), 08, 09, 09 (!), 11, 18, 21 2011-12-16T13:01:22 at that poitn I think trueskill is getting a bit random 2011-12-16T13:01:30 since my skill depends on the order the games finish 2011-12-16T13:01:34 But solifugid (starterpack creator and my *teacher* in OCaml) updated his bot and rising up. 2011-12-16T13:01:36 i'm winning in the google category. i can't believe joakim's bot fell that far down 2011-12-16T13:02:01 *** Israfel has quit IRC (Read error: Operation timed out) 2011-12-16T13:02:04 we have kind of a pathetic showing :( 2011-12-16T13:02:14 I really didn't expect to be up to #63 overnight 2011-12-16T13:02:18 but matchmaking bugs help! 2011-12-16T13:02:30 heh, did you get race-conditioned into several games at once? 2011-12-16T13:02:35 i love it when that happens 2011-12-16T13:02:49 yeah, did you see my list above? 2011-12-16T13:02:55 oh yeah, heh 2011-12-16T13:02:59 a1k0n: You have a taste to famous internet-search-companies. I remember you were working in Yahoo during Tron. 2011-12-16T13:03:10 sometimes i'll even play with the exact same people at the same time on two different servers 2011-12-16T13:03:13 I played Speedy_Consoles 4 times 2011-12-16T13:03:20 UncleVasya: yep. 2011-12-16T13:03:36 i got all sorts of interviews after tron. 2011-12-16T13:03:50 *** janzert has joined #aichallenge 2011-12-16T13:04:24 janzert: matchmaking error in my favor, collect 10 games :) 2011-12-16T13:04:28 thanks! 2011-12-16T13:04:38 *** olexs has quit IRC (Ping timeout: 252 seconds) 2011-12-16T13:04:54 *** olexs has joined #aichallenge 2011-12-16T13:04:56 *** liberforce has left #aichallenge 2011-12-16T13:05:20 *** dom7b5 has joined #aichallenge 2011-12-16T13:05:26 hey a1k0n , good luck! good luck to everyone, I'm not discriminating 2011-12-16T13:05:36 you too! good to see you again dmj111 2011-12-16T13:05:53 i'm rooting for flagcapper, personally 2011-12-16T13:06:02 just for the upset? 2011-12-16T13:06:06 though.. xathis you aren't 16 years old are you? 2011-12-16T13:06:12 I can't wait for the next challenge where I will get more than 3 weeks to work on it 2011-12-16T13:06:13 thx. my only hope is that everyone else puts in bugs the next couple of days.... 2011-12-16T13:06:24 dmj111: oh I totally did 2011-12-16T13:06:28 yeah, its great that flag is 16 2011-12-16T13:06:35 in my current climb I saw at least two times I walked by an unattended hill 2011-12-16T13:06:38 BenJackson: yes! one day. 50 - 150 to go 2011-12-16T13:06:45 a1k0n: no, i'm 4 years older than _flag 2011-12-16T13:06:48 oops... one done, 50-150 to go 2011-12-16T13:06:49 old man 2011-12-16T13:06:59 one down 2011-12-16T13:07:01 gosh 2011-12-16T13:07:02 seems google employees werent interested in this contest except you 2011-12-16T13:07:12 xathis: still pretty young then :) 2011-12-16T13:07:22 xathis: yes, nice work! 2011-12-16T13:07:34 a1k0n, dmj111: thanks 2011-12-16T13:08:08 UncleVasya :) 2011-12-16T13:09:18 we have google offices in hamburg, hint hint 2011-12-16T13:09:42 a1k0n: hey, he's the one winning 2011-12-16T13:09:56 I think google only gets to recruit contestants that finish below the top google finisher 2011-12-16T13:10:02 oh ok 2011-12-16T13:10:03 xathis can recruit you 2011-12-16T13:10:44 Murashka: btw, my friend has very big imagination and decided your nick is a ukranian trolling for russians: my Rashka. I said no way he's right. :) 2011-12-16T13:10:47 since there are many professionals here, what do you think the lowes ranking is that you would put in your resume? 2011-12-16T13:10:56 thats an important question for me 2011-12-16T13:11:12 mcstar: hard to answer. only contestants really know what the ranks mean 2011-12-16T13:11:22 mcstar: the ranking doesn't matter as much as being able to put together a good description of your bot 2011-12-16T13:11:37 what do you mean by that? 2011-12-16T13:11:40 yeah, I'd say it's good fodder as an answer to lots of interview questions 2011-12-16T13:11:45 tell me about a difficult problem you debugged 2011-12-16T13:11:48 plus there's like 7000 entries so any working bot looks good ranking-wise 2011-12-16T13:11:55 tell me about a program that had a performance problem that you resolved 2011-12-16T13:11:57 a1k0n: so can ypu please upload some of your earlier version so we can find a job in Google 2011-12-16T13:12:25 i'll refer all of you, and then if any of you get hired i get a referral bonus 2011-12-16T13:12:42 I'd say participating (beyond uploading a starter kit) is an attribute I would find encouraging in a candidate 2011-12-16T13:12:48 (if bmh doesn't do it first) 2011-12-16T13:12:56 UncleVasya no trolling, just funny name 2011-12-16T13:13:01 if you are in the top 20 I'll throw out my other questions and we'll talk bot strategy :) 2011-12-16T13:13:17 that probably wont happen 2011-12-16T13:13:36 there are only 20 people who will get into top 20 2011-12-16T13:13:37 the few times I have been an interviewer, enthusiasm for learning new things was high on my list of qualities of the inteviewee. 2011-12-16T13:13:41 And after some amound of hiring company will fire you because there is so many another talented guys here! :p 2011-12-16T13:13:47 *** GeorgeSebastian has quit IRC (Ping timeout: 252 seconds) 2011-12-16T13:15:02 yeah if you can code a BFS on the map and you participate in programming contests, companies will be fighting to hire you 2011-12-16T13:15:11 especially in silicon valley 2011-12-16T13:15:25 I've certainly interviewed guys who I doubt would be able to get into the top 1000 2011-12-16T13:16:23 i hope you didnt hire them 2011-12-16T13:16:26 no offense 2011-12-16T13:16:34 I hope I didn't too! 2011-12-16T13:17:04 my friend took the c/c++ tests at ericsson yesterday 2011-12-16T13:17:12 i taught him 2011-12-16T13:17:15 *** Israfel has joined #aichallenge 2011-12-16T13:17:22 they probably wont hire him :D 2011-12-16T13:17:53 UncleVasya also... sasha.sochka looks like better then yours last version too. 2011-12-16T13:18:15 ukraine is very much present in the comp. 2011-12-16T13:18:24 im very disappointed that there are <20 hungarians 2011-12-16T13:18:42 yeah, too bad no mega1 2011-12-16T13:19:18 somehow the news of his winning didnt reach the university hostels 2011-12-16T13:19:21 it's probably because popular social network habrahabr talk about it very much 2011-12-16T13:19:33 Murashka: Yeah... but he writes in C++ so this is acceptable. No shame in losing to C++ bot. 2011-12-16T13:20:12 it's a social network? looks more like.. slashdot 2011-12-16T13:20:18 UncleVasya: obviously you had some good ideas, your ranking is really nice wrt your language of choice 2011-12-16T13:20:24 *** kire has joined #aichallenge 2011-12-16T13:20:25 UncleVasya no shame to lose to Pascal 2011-12-16T13:20:27 or is it more like reddit 2011-12-16T13:21:05 *** treeform has joined #aichallenge 2011-12-16T13:21:23 UncleVasya actualy for this contest possibilities of C++ and Pascal is the same 2011-12-16T13:21:28 I still have lots to improve and have many ideas too but am too afraid to make any change now - only two days left 2011-12-16T13:21:36 Murashka: Will you upload one more version? 2011-12-16T13:22:11 any chance of extending it by one more week? 2011-12-16T13:22:40 NO 2011-12-16T13:23:04 UncleVasya don't know. I have couple of ideas....but I always got them...to few time to implement something important 2011-12-16T13:23:55 *** Dlayne has joined #aichallenge 2011-12-16T13:24:07 yep increase for half year please I have a huge todo list :) 2011-12-16T13:24:15 Murashka: I have no arguments agains Pascal itself. Itis just its reputation of language for schoolchildren and junior students. So when I see your bot the first thing that come in mind is 'hell, some yang man doing better than you!' 2011-12-16T13:24:29 yeah thats ok, am extremely happy with the learning I got here. sad that its going to end :( 2011-12-16T13:26:05 Third from Finland, what the hell. This makes me sad. 2011-12-16T13:26:11 Really, I understand that pascal is used by professional too sometimes so no offence. Just a stereotipe 2011-12-16T13:26:19 hasn't everybody moved from pascal to delphi? 2011-12-16T13:26:21 UncleVasya I'm far from young unfortunately :) Delphi just more comfortable for me. 2011-12-16T13:26:53 pairofdice: i would expect finland to do well. don't you all write demos every weekend? 2011-12-16T13:27:19 Yeah, sure. But I'm not a programmer . 2011-12-16T13:27:27 Murashka: btw, I've started my bot with Delphi previous challenge. It just wasn't supported so I moved to C++. 2011-12-16T13:27:48 you've done an awful lot of python programming for not being a programmer 2011-12-16T13:27:50 So the level of finnish programming where I'm third is a bit crazy 2011-12-16T13:28:05 wasn't supported by contest I mean 2011-12-16T13:28:07 oh. i see. 2011-12-16T13:28:09 The only thing I did before this challenge was some POroject euler problems 2011-12-16T13:28:14 Project* 2011-12-16T13:28:29 wow 2011-12-16T13:28:32 pairofdice: And tic-tac-toe :) 2011-12-16T13:29:14 UncleVasya FreePascal can compile Delphi code without any significant problems 2011-12-16T13:31:13 *** bretep has joined #aichallenge 2011-12-16T13:31:18 Murashka: Pascal wasn't supported. Some guy wrote a FreePascal/Delphi starterpack but tthere wasn't compiler on the server. 2011-12-16T13:31:25 sry for messing 2011-12-16T13:31:35 i only wrote finite difference codes and some python scripts before 2011-12-16T13:32:52 Murashka: Can you tell something about yourself? I'll keep the track of strong ukranian competitors. 2011-12-16T13:33:43 UncleVasya easy ;) 2011-12-16T13:34:36 *** dom7b5 has quit IRC (Ping timeout: 240 seconds) 2011-12-16T13:34:38 Are you waiting for specific questions or I can relax and read your upcoming text? 2011-12-16T13:35:50 she is married, has a boy and two girls, she is an artist btw, she paints, and loves to walk in the wind 2011-12-16T13:35:53 UncleVasya Kharkiv National University of Radio Electronics's \Computer Science\Intelligent Decision Support Systems 96. 2011-12-16T13:36:56 UncleVasya I was searching for university name to copy paste it ;) 2011-12-16T13:38:04 UncleVasya Primary Language: Delphi+OpenGL 2011-12-16T13:38:05 mcstar: And that it is _she_ you understand from my interest to this person? Because female gender is the only thing that can attract me in human? :) 2011-12-16T13:38:28 Sounds like gamedev 2011-12-16T13:38:38 *** mleise has joined #aichallenge 2011-12-16T13:38:47 you hoped for a she, dont tell me otherwise :D 2011-12-16T13:40:30 mcstar: I know the wisement: if girl can program than she cannot prepare tasty food. 2011-12-16T13:40:35 30ms 80 ants 2011-12-16T13:40:49 wisement :) 2011-12-16T13:40:53 :D 2011-12-16T13:40:55 UncleVasya not exactly. I have a dream to make 3D interface useful for common applications :) 2011-12-16T13:41:23 pairofdice: you have an excuisite dictionary 2011-12-16T13:41:38 esquisite damn 2011-12-16T13:41:46 exquisite 2011-12-16T13:41:49 It's all UncleVasya 2011-12-16T13:42:18 pairofdice: what do you do for a living? 2011-12-16T13:42:24 if you dont program... 2011-12-16T13:42:51 I worked at a retailstore, no job atm. 2011-12-16T13:43:21 good luck finding one 2011-12-16T13:43:25 or making one 2011-12-16T13:43:32 yeah ;_; 2011-12-16T13:44:07 Murashka: There is a russian contester bucash in the 11th place. Isn't your friend? :) 2011-12-16T13:44:24 mcstar and I'm not she ;))) 2011-12-16T13:44:49 Murashka it's a name of my bot :) 2011-12-16T13:47:58 UncleVasya nope. But i also found something similar in it. Anyway I was first;) 2011-12-16T13:48:46 Be careful Murashka, there is a tendention for middle-aged contestants: they get troubles with wife who doesn't let them to compete one more time. 2011-12-16T13:50:26 UncleVasya of course you understand that wife it's more important than any competition?:) 2011-12-16T13:51:00 it's just fun for me 2011-12-16T13:51:00 not until I don't have one :) 2011-12-16T13:51:09 *until I have 2011-12-16T13:51:58 cool bot for 'just fun'. Are you gonna release source after contest? 2011-12-16T13:52:19 UncleVasya cool bot it's a xathis 2011-12-16T13:53:21 UncleVasya it's wrote in drinken monkey code style. If someone will be interested in it ... 2011-12-16T13:53:40 i=u 2011-12-16T13:54:25 @later tell McLeopold could you or amstan get this https://github.com/aichallenge/aichallenge/issues/388 2011-12-16T13:54:25 janzert: OK 2011-12-16T13:54:35 Also I didn't counted sasha-s bot because my old version was higher than he is now. 2011-12-16T13:54:40 @later tell amstan could you or McLeopold get this https://github.com/aichallenge/aichallenge/issues/388 2011-12-16T13:54:40 janzert: Yes master! 2011-12-16T13:59:16 *** nha has quit IRC (Ping timeout: 240 seconds) 2011-12-16T14:05:01 @later tell amstan the FAQ answer regarding source files needing to be one language and in the same directory isn't correct for all languages anymore. Not sure where to change it, but probably just removing everything after the comma in the second sentence would be good. 2011-12-16T14:05:01 janzert: Yep. 2011-12-16T14:06:08 *** jstrong has quit IRC (Ping timeout: 240 seconds) 2011-12-16T14:17:22 *** AndrewBC has quit IRC (Read error: Connection reset by peer) 2011-12-16T14:18:43 *** AndrewBC has joined #aichallenge 2011-12-16T14:21:56 *** Surya has joined #aichallenge 2011-12-16T14:23:48 janzert: seems like the forum could use a general discussion area 2011-12-16T14:24:21 *** Boscop has joined #aichallenge 2011-12-16T14:24:45 :) talk to amstan and McLeopold, there was a push to split things up more a little while back 2011-12-16T14:25:37 I wanted to make a post to follow up on the resume question earlier 2011-12-16T14:25:55 I've noticed that when I "read new" I don't pay attention to where things are posted 2011-12-16T14:26:01 but when I post it never seems obvious where to go 2011-12-16T14:27:44 GreedyBot is at 2997. Is the tutorial bot better than greedy? 2011-12-16T14:28:01 is there a submission that is the tutorial result? 2011-12-16T14:28:20 I don't necessarily agree with the way it is currently organized, but I also don't have the time or motivation to work on changing it. 2011-12-16T14:28:43 not sure, I thought the result of the tutorial was a bot that timed out all the time :/ 2011-12-16T14:29:27 since it uses an extremely inefficient exploration algorithm 2011-12-16T14:29:42 I was just wondering if there was an easy way to draw a line in the results to say who had put in effort beyond the tutorial 2011-12-16T14:30:23 *** amstan__ has joined #aichallenge 2011-12-16T14:30:23 *** ChanServ sets mode: +o amstan__ 2011-12-16T14:31:48 http://aichallenge.org/profile.php?user=12359 2011-12-16T14:32:07 yeah I know about greedybot 2011-12-16T14:32:26 isn't that the line ? 2011-12-16T14:32:36 I don't know, I didn't really do the tutorial 2011-12-16T14:33:02 the tutorial should be worse than that 2011-12-16T14:33:21 all it does is move in the direction of food 2011-12-16T14:33:25 and not step on itself 2011-12-16T14:33:35 won't path around things unless by luck 2011-12-16T14:38:28 *** yoden_ has joined #aichallenge 2011-12-16T14:43:31 *** yoden_ is now known as yoden-cloud 2011-12-16T14:43:34 *** UncleVasya has quit IRC (Read error: Connection reset by peer) 2011-12-16T14:46:09 *** Antimony has joined #aichallenge 2011-12-16T14:47:13 *** Accoun has quit IRC () 2011-12-16T14:50:57 aichallenge: janzert epsilon * rede62a6 / website/api_game_result.php : Fix race between deleting matchup and updating player most recent game - http://git.io/CSzW0w 2011-12-16T14:54:36 rwest: looks like fluxid was doing some analysis earlier 2011-12-16T14:54:40 08:10 < Fluxid> ok, bots at page 24, skill 50, place around 2357 usually do something more 2011-12-16T14:55:19 yea he was just randomly clicking on bots as he went through the pages haha 2011-12-16T14:55:54 I hate whoever is making my fluxid games take like 3s per turn 2011-12-16T14:56:28 Yeah, rogue bots that don't care about turntime :) 2011-12-16T14:58:28 *** xathis has quit IRC () 2011-12-16T14:58:53 *** thestinger has joined #aichallenge 2011-12-16T14:58:59 let's see... 500ms of actual compute, 2.5s rtt, 1250ms one way, 1250ms * c = 375,000km.. they're on the moon 2011-12-16T14:59:01 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-12-16T14:59:27 hehe 2011-12-16T14:59:31 lol 2011-12-16T14:59:48 BenJackson: that's optimistic 2011-12-16T15:00:20 we just have to plot their turn times against the lunar calendar 2011-12-16T15:00:22 BenJackson: if you never check the timer, you might just use 2s a turn if you do something like A* per ant 2011-12-16T15:00:45 thestinger: You realize there's no need to actually refute my moon theory 2011-12-16T15:00:46 or sleep(2) 2011-12-16T15:00:53 but it's a valid theory 2011-12-16T15:00:58 somebody is taking like 30s to load their bot right now I swear 2011-12-16T15:01:11 at least the turns are instant now 2011-12-16T15:01:46 well my combat is better, but not where it would be if I had more time :-/ 2011-12-16T15:01:53 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-16T15:02:05 why couldn't I have found this challege like 2 months ago 2011-12-16T15:02:18 BenJackson: they could be broadcasting via sound waves from an underwater base to the surface 2011-12-16T15:03:06 *** Accoun has joined #aichallenge 2011-12-16T15:03:27 woo I won a 9p game with a bunch of good bots 2011-12-16T15:04:01 http://aichallenge.org/visualizer.php?game=251587&user=2255 yay for timeouts 2011-12-16T15:05:58 wow 2011-12-16T15:06:11 WOW 2011-12-16T15:06:11 who did you pay off to get ChrisH *and* pguillory to time out against you 2011-12-16T15:06:15 best timeout ever! 2011-12-16T15:06:36 pguillory timeouted right by two hills 2011-12-16T15:06:39 awesome 2011-12-16T15:06:53 my bot has magic voodoo powers 2011-12-16T15:07:20 hm... I really need to stop being lazy and implement detection of timed-out bots, if even top-10 guys time out occasionally 2011-12-16T15:07:50 I think you will find that the technology has applications even against non-timed-out bots 2011-12-16T15:08:04 after all, is there really a difference between a bot that's not moving and a timed out bot? 2011-12-16T15:08:07 true 2011-12-16T15:08:46 well, at least you don't need to predict movements of dead bot once you know he'll not move 2011-12-16T15:09:07 I need to merge working defense and combat scoring today 2011-12-16T15:09:10 Fluxid: that's the point, once my combat system knows an ant won't move it'll make short work of him 2011-12-16T15:10:26 hm, Hammerok seems to hit time limit too, started to choke at some point 2011-12-16T15:10:28 so I need to teach my bot to play dead? 2011-12-16T15:11:00 if (fighting_a_good_bot) { play_dead(); } 2011-12-16T15:11:22 i train on that map a lot 2011-12-16T15:11:46 pgulroy does what i want my bot to do 2011-12-16T15:11:50 what a coincidence 2011-12-16T15:12:51 olexs: do i remember correctly that you are writing your bot as part of your thesis? 2011-12-16T15:13:02 mcstar: yes 2011-12-16T15:13:12 how is it coming along? 2011-12-16T15:13:15 *** nha has joined #aichallenge 2011-12-16T15:13:18 do you have some advanced ideas? 2011-12-16T15:13:53 it's actually less about advanced ideas and more about trying out a lot of "classic" techniques, evaluating and choosing what works best 2011-12-16T15:13:59 *** _flag <_flag!~flag@69-165-173-172.dsl.teksavvy.com> has joined #aichallenge 2011-12-16T15:14:24 i see 2011-12-16T15:14:56 do you spend much time with it? 2011-12-16T15:15:01 it seems like I lack the crucial know-how to get well into the top100 at the moment, the last version hovered about the 100 mark, the current one should be a bit better I think 2011-12-16T15:15:24 *** Antimony has joined #aichallenge 2011-12-16T15:16:02 it can be daunting to code under pressure 2011-12-16T15:16:12 http://ants.fluxid.pl/replay.37404 not sure how strcat_newdefend managed to survive this 2011-12-16T15:16:15 M is very aggressive 2011-12-16T15:16:16 *** Jacob_Strauss has joined #aichallenge 2011-12-16T15:16:25 quite a bit of time, though much of it goes into writing the paper, not quite as much into coding 2011-12-16T15:16:33 luckily the turn limit is 1500 :P 2011-12-16T15:16:44 M it's me :) 2011-12-16T15:17:53 and i'm very peaceful ;) 2011-12-16T15:18:40 socket.error: [Errno 110] Connection timed out 2011-12-16T15:18:44 stupid tcp... 2011-12-16T15:20:08 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-16T15:21:37 blame your isp, instead 2011-12-16T15:22:29 I blame TC pants 2011-12-16T15:22:41 never had that error on fluxid's server :P 2011-12-16T15:29:40 im gonna gonna watch this movie, theres a cop, coming from new york, for christmas, to california, to some party 2011-12-16T15:32:51 die hard 10? 2011-12-16T15:33:37 1 2011-12-16T15:34:09 *** Kurnevsky has quit IRC (Quit: Instantbird 1.0) 2011-12-16T15:34:28 *** Kurnevsky has joined #aichallenge 2011-12-16T15:35:58 *** Paradoxial has joined #aichallenge 2011-12-16T15:38:57 stupid gcc 2011-12-16T15:39:19 I was calling a function _one time_ and it didn't get inlined (it's also 3 lines long) 2011-12-16T15:39:47 going through every function atm and forcing them to be inlined if it makes the binary size smaller... 2011-12-16T15:40:38 thestinger: I'm pretty proud of this one: http://aichallenge.org/visualizer.php?game=252102&user=3473 2011-12-16T15:42:11 wow, you were penned inside a little area near your hill and somehow managed to push back and win :0 2011-12-16T15:43:21 blue was the weak spot ;-) 2011-12-16T15:44:18 BenJackson: does that version have the territory control stuff where you make sure you've walled off the enemy from your hill? 2011-12-16T15:44:50 penning trap 2011-12-16T15:45:12 I have been in a game on aichallenge for like 20 mins 2011-12-16T15:47:08 I still don't know if I have to rely aichallange that I'm in a game :p 2011-12-16T15:47:45 * avdg 's bot is playing a game since he checked the site again 2011-12-16T15:48:09 *** u_ has quit IRC (Ping timeout: 252 seconds) 2011-12-16T15:48:34 Yeah mine seems to be in a game for a very long time too 2011-12-16T15:49:59 *** praveen_ has quit IRC (Quit: Page closed) 2011-12-16T15:54:02 still in a game lol 2011-12-16T15:54:07 I think it lies 2011-12-16T15:54:21 *** TL_ has joined #aichallenge 2011-12-16T15:55:09 thestinger: yes, that's the current version 2011-12-16T15:55:21 fyi: http://aichallenge.org/server_stats.php 2011-12-16T15:55:33 game rate was/is very low 2011-12-16T15:55:42 owned 2011-12-16T15:55:47 meaning: takes long time for 1 game to finish 2011-12-16T15:55:54 thats what those numbers mean 2011-12-16T15:56:14 where are all the servers gone again, there were a million the last few days? 2011-12-16T15:56:23 money 2011-12-16T15:56:26 ec2 price spiking again? 2011-12-16T15:56:32 probably 2011-12-16T15:56:37 great 2011-12-16T15:56:40 i dont think its an "error" or something 2011-12-16T15:56:53 olexs: you dont play over tcp? 2011-12-16T15:57:22 mcstar: I do, but rankings on aichallege were nice and useful the last few days, with high game rates 2011-12-16T15:57:43 *** Palmik has quit IRC (Remote host closed the connection) 2011-12-16T15:58:53 *** ikaros has joined #aichallenge 2011-12-16T15:59:19 *** TL_ has quit IRC (Quit: Page closed) 2011-12-16T16:01:34 *** nplus has quit IRC (Read error: Connection reset by peer) 2011-12-16T16:02:29 *** nplus has joined #aichallenge 2011-12-16T16:03:38 oh wow, I didn't realize C++11 added a new type of enum 2011-12-16T16:03:42 ok, idle detection done, quick and actually rather clean... I like my codebase 2011-12-16T16:03:56 strongly typed, and you can give it an integral type other than int 2011-12-16T16:04:02 *** Antimony has joined #aichallenge 2011-12-16T16:04:58 *** u_ has joined #aichallenge 2011-12-16T16:07:10 you can even use forward declarations for them :) 2011-12-16T16:07:58 *** TL_ has joined #aichallenge 2011-12-16T16:08:33 *** Kurnevsky has quit IRC (Quit: Instantbird 1.0) 2011-12-16T16:08:54 Hmmm, seems like no more games 2011-12-16T16:09:35 "In Game: Playing in a game right now." 2011-12-16T16:09:36 lol 2011-12-16T16:09:44 like 40 minutes now 2011-12-16T16:11:01 me too 2011-12-16T16:11:12 0.0 games a minute 2011-12-16T16:11:28 oh there are only 3 servers left 2011-12-16T16:11:28 I'm seeing all the 18 workers idling atm 2011-12-16T16:11:44 I'm not sure if thats intentional by amstan__ or janzert or not.... 2011-12-16T16:11:57 not my doing 2011-12-16T16:11:59 *** amstan__ is now known as amstan 2011-12-16T16:12:21 do you have janzert on IM or something by any chance? 2011-12-16T16:12:41 i do, but he should be good in here too 2011-12-16T16:12:53 ah youre right 2011-12-16T16:13:46 yeah.. it seems like they're down 2011-12-16T16:13:48 sigh 2011-12-16T16:13:53 janzert is on it 2011-12-16T16:13:53 ug, I keep getting picked on at the start so I can never expand 2011-12-16T16:14:17 i can ssh into them, and do all that stuff, but there's a lot of them 2011-12-16T16:14:19 he made a bad chance somewhere but they should be running soon again :) 2011-12-16T16:14:36 hkraal_: how do you know he's on it? 2011-12-16T16:14:40 amstan well sorry for donating all those servers ;) 2011-12-16T16:14:42 amstan: it's the server side that's broken 2011-12-16T16:14:47 amstan: he replied on my PM :) 2011-12-16T16:14:48 janzert: hmm 2011-12-16T16:15:00 how come the other ones are up though? 2011-12-16T16:15:04 working on fixing some race conditions 2011-12-16T16:15:08 k 2011-12-16T16:15:11 still in their games I suppose? 2011-12-16T16:15:24 I reverted the code but apparently the others were still locked into the broken php 2011-12-16T16:15:34 trying an apache restart now 2011-12-16T16:17:21 I may need to go through and give 'em a kick but let me try and get this fix working first :} 2011-12-16T16:18:05 hehe, I'm glad it's not because of our servers ;) 2011-12-16T16:18:12 *** Scryer has joined #aichallenge 2011-12-16T16:19:02 nope all my fault :) 2011-12-16T16:19:22 always good to be able to point at someone ;) 2011-12-16T16:19:35 if you need anything let me know :) 2011-12-16T16:19:40 afk for a bit :) 2011-12-16T16:22:07 aichallenge: janzert epsilon * rf8cb099 / (website/api_get_task.php website/sql.php): Try and properly attain the lock on a matchup when assigning it to a worker - http://git.io/o-o_bw 2011-12-16T16:23:16 *sigh* 2011-12-16T16:23:32 i never ended up submitting anything but the starter bot 2011-12-16T16:24:12 BenJackson: http://tcpants.com/replay.2153 I think our bots formed an alliance in this game... 2011-12-16T16:24:16 Love the comment on that. "clobbering" and "persistent". Fixed that for ya. :) 2011-12-16T16:24:49 thestinger: aren't you a member of the Accords? 2011-12-16T16:25:06 janzert: fixed the wiki about the mybox stuff 2011-12-16T16:25:09 mybot* 2011-12-16T16:25:15 not sure about the timer 2011-12-16T16:25:26 the old code is all there 2011-12-16T16:26:26 thestinger: we were ideally positioned to mow everyone down 2011-12-16T16:26:34 I'm just not going to have time to implement the deadline timer so hoping someone else can get to it 2011-12-16T16:26:49 mcstar: http://forums.aichallenge.org/viewtopic.php?f=24&t=2139 2011-12-16T16:27:14 :D 2011-12-16T16:27:22 reading it right now 2011-12-16T16:28:04 a1k0n: dmj111: maybe you can chip in on that too 2011-12-16T16:28:15 i think there was a typo in my question, thank for correcting it 2011-12-16T16:28:21 np 2011-12-16T16:30:06 *** peerai has joined #aichallenge 2011-12-16T16:30:11 I wonder if I'm really playing in a game 2011-12-16T16:30:53 nice summary 2011-12-16T16:31:15 I've been "playing in a game" for about half an hour. I see the server count has gone into the crapper. 2011-12-16T16:32:07 * avdg is almost playing for an hour 2011-12-16T16:32:15 I'm in a game for more as an hour, so I think something chrashed 2011-12-16T16:32:29 why dont yall go and see a movie or something 2011-12-16T16:32:45 *** tmseiler has joined #aichallenge 2011-12-16T16:32:51 multitasking ain't that easy ;-) 2011-12-16T16:32:57 'Cause we want some live feedback before releasing our final tuning, mcstar? 2011-12-16T16:33:11 yes? 2011-12-16T16:33:25 hmm, did I break game result posting too? :( 2011-12-16T16:33:34 I'll check that in just a second 2011-12-16T16:33:39 Thanks! 2011-12-16T16:33:42 don't tell me aichallenge has gone the way of fluxid.pl 2011-12-16T16:34:03 actually no, it seems games are still posting fine 2011-12-16T16:34:14 my bot's been in a game for about 20 minutes now, I just uploaded a fresh one 2011-12-16T16:34:18 janzert: didn't you make some change to avoid saying "playing in a game" too early 2011-12-16T16:34:22 like at the matchup point? 2011-12-16T16:34:27 ah 2011-12-16T16:34:55 bitbucket really needs to run 'git gc' now and then, my git repo is up to 13MB on there 2011-12-16T16:35:04 700KB locally... 2011-12-16T16:35:18 yes, I fixed one thing with that the other day at least 2011-12-16T16:36:04 thestinger: the commands will do it every 14 days automatically 2011-12-16T16:37:42 I got really spoiled with the 20-25 server thang. :) 2011-12-16T16:38:29 oh, the extra servers are gone? 2011-12-16T16:39:58 It was at 3 when I said that, but they seem to be coming back on-line. (?) 2011-12-16T16:40:56 if i just could improve my inaccurate combat i'm going to be at least top 10. 2011-12-16T16:41:13 If I had wheels I'd be a roller coaster. 2011-12-16T16:41:18 BenJackson: I don't think bitbucket runs any of the commands that would cause it to run automatically 2011-12-16T16:41:21 http://tcpants.com/replay.2156 nice game here. saved by turn limit from strcat. :) 2011-12-16T16:42:19 BenJackson: oh, I guess it should be run by git-receive-pack... 2011-12-16T16:42:50 bleh, it takes a long time to kick all the workers now :} 2011-12-16T16:45:21 TL_: wow, I had over 500 ants in my hill at the end... that might be something I have to fix 2011-12-16T16:45:23 TL_: looks like GreenTea had a similar near-death experience vs xathis here: http://aichallenge.org/visualizer.php?game=252397&user=4513 2011-12-16T16:46:10 nvm, I think I was just gathering food too quickly - the hill isn't really ever blocked 2011-12-16T16:46:38 Yeah, you can only vomit out one ant epr turn 2011-12-16T16:46:50 per hill 2011-12-16T16:46:52 thestinger: that happens a lot on large single-hill maps, if a bot is any good at gathering food 2011-12-16T16:46:53 Need moar hillz! 2011-12-16T16:47:24 *** ChrisH_ has joined #aichallenge 2011-12-16T16:47:36 and 30 more turns to finish off TL_ :) 2011-12-16T16:48:13 aichallenge: janzert epsilon * r928fdad / website/api_get_task.php : Fix lock check - http://git.io/OXg03w 2011-12-16T16:48:24 I guess that's another thing I need to add, if a hill is completely surrounded there's no point in sending more ants to it 2011-12-16T16:49:02 thestinger: you we're busy with that single hill while i take down other hills. :) 2011-12-16T16:49:13 were 2011-12-16T16:49:27 yeah, my weighting for hill attacking sucks 2011-12-16T16:49:40 TL_: Beware tcpants has the wrong turn limit (750 vs 1000) 2011-12-16T16:50:01 red was really close to me, so defense was also kicking in 2011-12-16T16:50:01 @later tell BenJackson re: "get a load of this map: http://ants.fluxid.pl/replay.35878" wow, don't think i've had a game on that one yet, what's the map name? 2011-12-16T16:50:01 ChrisH_: Yes master! 2011-12-16T16:50:23 doh, didn't see you there BenJackson 2011-12-16T16:50:34 ChrisH_: I figured you could get it out of fluxid's game list but maybe it's erased now 2011-12-16T16:50:45 yeah, it has scrolled past page 9 now 2011-12-16T16:50:50 tcpants uses a lot of maps that aren't in the official rotation too 2011-12-16T16:51:13 Oh, boy, let's do another one on tutorial.map! 2011-12-16T16:51:13 ahh, but changing the url to p10 works... 2011-12-16T16:51:39 no, just shows page 0 2011-12-16T16:51:44 I did lose (or do poorly) in an aichallenge zillion hill map and was wrestling with the idea of adding code like yours 2011-12-16T16:51:47 what's with the bots on aichallenge still stuck in "playing a game right now"? the workers seem to be working again, but my bot still does nothing 2011-12-16T16:52:04 trying to justify the idea that I'd only enable it if I detected the multi-hill low-food case 2011-12-16T16:52:06 TL_: you destroyed me in our next game :) 2011-12-16T16:52:08 to reduce risk 2011-12-16T16:52:10 I think janzert is messing with the code that does that. 2011-12-16T16:52:33 ChrisH_: also check out my game list, I played 10 in 15 minutes to get to 63 or something overnight 2011-12-16T16:52:40 ok 2011-12-16T16:52:51 I was quite pleased with the matchup error in my favor 2011-12-16T16:53:12 got it: maze_p08_07.map 2011-12-16T16:53:14 I played 4 games with Speedy_Consoles 2011-12-16T16:53:40 I was wondering whether you could fake out the servers by taking all your time per move, and queue up more games on the other servers who don't realize you're already in a game. Or doesn't it work that way? 2011-12-16T16:53:48 at 5:08, 09, 11, 18 2011-12-16T16:54:13 Scryer: plenty of people just straight up run multiple bots on tcp 2011-12-16T16:54:46 Oh, thought you meant on "live". 2011-12-16T16:54:57 oh, I did 2011-12-16T16:55:03 the matchup thing got confused I guess 2011-12-16T16:55:27 BenJackson: I replied to your forum post about job hunting. Great post BTW. 2011-12-16T16:56:06 thanks 2011-12-16T16:56:07 I think I'd set the bar a little higher, though - 2000 is still danged weak. 2011-12-16T16:56:49 ok, it's going to take a while to work through the extraneous "playing in game" messages but it should be working itself out now 2011-12-16T16:57:00 Thanks, janzert. 2011-12-16T16:57:21 I may (or may not) be job hunting soon, and you better believe I will use my experience in this contest to my advantage. 2011-12-16T16:57:25 and by "a while" my best guess is just under 1.5 hours :/ 2011-12-16T16:57:36 *** nha has quit IRC (Ping timeout: 240 seconds) 2011-12-16T16:58:01 And here is a point that I didn't make in my response, but if I interview someplace and they are not impressed with doing well in this contest I don't want a job there. 2011-12-16T16:58:59 ^^ 2011-12-16T16:59:49 My current employer bragged about my #24 finish in PW on the company FB page. :) 2011-12-16T17:00:41 my personal goal is top 100 before this is over, it's like exponential toughness as you climb this ladder 2011-12-16T17:01:29 tmseiler: once you get in top 100 you will want top 50. ;) 2011-12-16T17:02:07 My current TM wished me luck. He encourages such challenges 2011-12-16T17:02:09 I know, it's true 2011-12-16T17:02:27 Dlayne: that's great! 2011-12-16T17:03:06 i was in top 100... then people implemented combat logic 2011-12-16T17:03:07 :> 2011-12-16T17:03:36 I'm adhering as strictly as possible to emergent techniques 2011-12-16T17:03:42 so combat is really the weakness 2011-12-16T17:03:56 I'm kind of resigned to that limitation 2011-12-16T17:04:47 some of these bots handle combat as well as an expert RTS player controlling units manually 2011-12-16T17:05:01 My early versions got to about 250 without ever looking at where enemy ants were. Then things got tougher. 2011-12-16T17:05:20 yeah, in fact I noticed blind aggression actually having advantages over those overly cautious guys 2011-12-16T17:05:32 "right this way, sir" 2011-12-16T17:05:51 *** g0llum has quit IRC (Read error: Connection reset by peer) 2011-12-16T17:07:02 *** Jak_o_Shadows has joined #aichallenge 2011-12-16T17:10:21 *** amstan has quit IRC (Read error: Operation timed out) 2011-12-16T17:10:38 the competition has definitely gotten a lot stronger since the beginning, there are some bots I remember seeing in the top 20 that are now down around 200 2011-12-16T17:11:43 I wonder how many had AI backgrounds before entering 2011-12-16T17:12:43 ok, after realizing that the time to play out all the extra matches by playing them was going to take closer to 3 hours I've just manually cleared them out now. so in game display should be returning to normal now. 2011-12-16T17:14:22 BenJackson: so, tcpants appears to have died 2011-12-16T17:14:22 ZumZoom was #14 on Nov 4th and hasn't resubmitted, now that bot is #215. 2011-12-16T17:14:43 tcpants website is down again. 2011-12-16T17:14:55 momobot did never resubmit and is still in top 20 2011-12-16T17:15:05 I think the tcp servers keep the web server portion on extremely low priority or something 2011-12-16T17:15:59 ikaros: yes, he was #1 or #2 for a long time. In fact he was the one to beat before xathis took over #1. 2011-12-16T17:16:27 *** kire has quit IRC (Remote host closed the connection) 2011-12-16T17:17:09 yea 2011-12-16T17:17:27 one of the tough beta guys :) 2011-12-16T17:18:01 tcpants is working for right now, BTW. 2011-12-16T17:18:10 but the webserver isn't always responsive 2011-12-16T17:18:21 *** bergmark has quit IRC (Ping timeout: 252 seconds) 2011-12-16T17:18:37 looks like the servers of aichallenge are going back as well 2011-12-16T17:19:00 thanks. yes it works. 2011-12-16T17:19:56 I like replaying games as soon as they finish to "improve" my bot. 2011-12-16T17:20:48 fluxid's server is kind of crazy 2011-12-16T17:20:55 sometimes I'll clearly win a game as #1, but there will be one tiny fight where I screw up, and I view the game as a wash 2011-12-16T17:21:19 I think the rankings on it are totally screwed up now... 2011-12-16T17:21:31 holy shit, i want it to end 2011-12-16T17:21:37 two more days 2011-12-16T17:21:47 strcat is stuck between rank 30 and 40, strcat_next between 10 and 20 2011-12-16T17:21:55 both are the same code atm :P 2011-12-16T17:22:01 lol 2011-12-16T17:22:46 * avdg wonders how the final high scores would be 2011-12-16T17:22:49 *** arscan has joined #aichallenge 2011-12-16T17:23:00 will we be able to watch each game as it is played during finals ? 2011-12-16T17:23:30 I don't mean live but the way it is carried out now. 2011-12-16T17:23:35 thestinger: well it's running trueskill without proper matchmaking, and some people got bots with hundreds of games idling in rankings... 2011-12-16T17:23:37 live feeds in a stadium maybe 2011-12-16T17:23:53 let's meet in a bar and watch ant games 2011-12-16T17:23:58 besh: afaik finals will be exactly the same as normal games, only with no new submissions 2011-12-16T17:24:14 Fluxid: Excellent plan - I'm in! Diet Coke for me, please. 2011-12-16T17:24:23 *** codetiger has joined #aichallenge 2011-12-16T17:24:29 >coke 2011-12-16T17:24:31 oh come on 2011-12-16T17:24:37 BEER! MORE BEER! 2011-12-16T17:24:37 With lime. 2011-12-16T17:24:42 I should run replays with my computer hooked up to a TV :) 2011-12-16T17:24:45 This statement from official webpage "Upon completion the contest winner will be announced and all results will be publically available." 2011-12-16T17:24:46 or maybe you're underage 2011-12-16T17:25:44 besh: i think we will be able to watch games as the finals unfold, that's how it worked last time for PW. 2011-12-16T17:25:47 besh: I think that means that there will be an official announcement, in addition to just a ranking table where it is now 2011-12-16T17:26:05 janzert says we'll likely see the games as usual: http://forums.aichallenge.org/viewtopic.php?f=27&t=2070 2011-12-16T17:26:07 besh: also the rankings were continuously updated last time as well 2011-12-16T17:26:17 Ok thanks guys. 2011-12-16T17:26:38 besh: we have in the past not updated the rankings for the last part of finals to leave the final result a bit of a suspense but will probably not do that this time 2011-12-16T17:26:47 i.e. the rankings will stay live throughout 2011-12-16T17:27:24 gn 2011-12-16T17:27:26 *** mcstar has quit IRC (Quit: WeeChat 0.3.6) 2011-12-16T17:28:19 *** tmseiler has left #aichallenge 2011-12-16T17:29:29 *** valydo has joined #aichallenge 2011-12-16T17:31:00 *** Redgis has joined #aichallenge 2011-12-16T17:35:58 *** valydo has quit IRC () 2011-12-16T17:36:48 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-12-16T17:41:07 *** Dlayne has quit IRC (Ping timeout: 240 seconds) 2011-12-16T17:45:39 *** GeorgeJ has quit IRC (Ping timeout: 255 seconds) 2011-12-16T17:49:45 *** rintaun has quit IRC (Ping timeout: 268 seconds) 2011-12-16T17:50:29 *** rintaun has joined #aichallenge 2011-12-16T17:52:17 my latest v23 always beats my v19 which is on the aichallenge.org with rank #292 in 29 games. Does that mean I'll make it to 100? 2011-12-16T17:52:20 Grr - I was chugging along at 3-4ms per move on TCPants, and timed out. Fmpph. 2011-12-16T17:53:08 It could be taking advantage of something specific you fixed, codetiger. It'll probably be better than 292, though. 2011-12-16T17:53:56 my aim is to reach top 100, but that seems to have become too hard now 2011-12-16T17:55:00 I think my only change to get there in 48 hours is to implement brilliant combat tactics. 2011-12-16T17:55:05 *chance 2011-12-16T17:55:17 at this point of time I feel too bad to see the aichallenge server has only 5 gpm 2011-12-16T17:55:38 Wish we had another 2 or 3 months. I found it late. 2011-12-16T17:55:52 use Memetix's combat method. You can implement it in couple of hours 2011-12-16T17:56:14 * Scryer digs out the forum posting again. 2011-12-16T17:56:38 another 2 months will definitly impact my happy family condition 2011-12-16T17:57:12 My wife's very understanding... gives her more time to do Killer Sudokus. :) 2011-12-16T17:59:06 *** Antimony has joined #aichallenge 2011-12-16T18:00:22 of, its a trade-off then 2011-12-16T18:03:34 after a month's work, my bot now can take down a few games on fluxid and tcpants 2011-12-16T18:05:28 good to know it is not only my life that is affected by ants :) 2011-12-16T18:07:01 codetiger: i am also trying to get to 100. You never know what happens when ratings are reset 2011-12-16T18:07:07 *** Surya has quit IRC (Ping timeout: 258 seconds) 2011-12-16T18:07:46 oh, rank resseting is mostly favourable? 2011-12-16T18:08:01 13:56 < janzert> ok, it's going to take a while to work through the extraneous "playing in game" messages but it should be working itself out now 2011-12-16T18:08:02 depends on matchmaking I guess... 2011-12-16T18:08:06 janzert: what's a while? I'm still stuck 2011-12-16T18:08:12 it takes ~8 games for me to climb back to the top 100 2011-12-16T18:08:24 Didn't he say he manually cleared them 2011-12-16T18:08:29 and then ~30 to get into the top 25 again (where I get stuck at around rank 15-20) 2011-12-16T18:08:46 aichallenge: janzert epsilon * r1307221 / sql/2_generate_matchup.sql : Correct seed selection to look at all players in upcoming matches not just seeds - http://git.io/faudUg 2011-12-16T18:08:50 so there isn't really time left for new bots to get to their stable high ranking imo 2011-12-16T18:08:53 BenJackson: can you please remove bj_v5 from bhickey server and add a new version. may be v9 2011-12-16T18:09:20 BenJackson: you're not stuck for what I can see, just showing your real place in the queue 2011-12-16T18:09:30 78 players ahead 2011-12-16T18:10:02 *** dom7b5 has joined #aichallenge 2011-12-16T18:10:40 if you read down from that line you quoted you'll see I realized 'a while' was probably 3 hours so just cleared them out instead of letting them run through 2011-12-16T18:10:49 I think my problem on tc pants is that I'm always pink. Why do I have to be Mr. Pink? I feel like Steve Buscemi in Rservoir Dogs. 2011-12-16T18:10:56 janzert: ah, it finally quit saying "playing in a game" 2011-12-16T18:11:00 although I did not play 2011-12-16T18:11:06 right 2011-12-16T18:11:14 thestinger: took 48 games to get to #10 for my V11 2011-12-16T18:11:32 took 17 to get to #42 this time 2011-12-16T18:11:37 I got much better matchups 2011-12-16T18:12:01 my mu might actually be close to correct now 2011-12-16T18:12:02 my current upload took 16 games to get to rank 55 2011-12-16T18:12:05 still high sigma 2011-12-16T18:12:33 mu is 91.66 which is almost what it got to last time, and I know it's a slight improvement - so it has a bit more to go 2011-12-16T18:13:54 BenJackson: please plz remove bj_v5 from bhickey servers. Its just playing with your v7 always and not letting me to get that oppurtunity 2011-12-16T18:15:26 Where do you see your mu and sigma? 2011-12-16T18:16:41 *** arscan has quit IRC (Quit: Page closed) 2011-12-16T18:18:36 roll over your skill to see your mu and sd 2011-12-16T18:19:02 Ah, thanks! 2011-12-16T18:21:57 http://ants.fluxid.pl/replay.37791 my bot razed hills like crazy here. 19 hills! 2011-12-16T18:23:19 yeah you can get crazy scores on multi-hill mazes 2011-12-16T18:23:50 TL_: you version bots like chrome 2011-12-16T18:23:54 v56! 2011-12-16T18:23:58 codetiger: going beyond V7 is only really V10. V8, 9 were shortlived and had silly bugs 2011-12-16T18:24:07 V10 is pretty strong for bmh's server 2011-12-16T18:24:28 V10 was about 13th overall when it was retired 2011-12-16T18:24:31 but bmh has only 2 users now 2011-12-16T18:24:35 ok 2011-12-16T18:24:41 anyway 2011-12-16T18:24:44 *** foRei has quit IRC (Read error: Connection reset by peer) 2011-12-16T18:24:55 codetiger: I'll do it when I get home 2011-12-16T18:24:59 plz atleast remove v5 so I get chance to play with v7 2011-12-16T18:25:06 I was thinking earlier in the contest of writing reverse-tcp-server code 2011-12-16T18:25:07 Next game should be within 124 minutes. 2011-12-16T18:25:11 wtf happened 2011-12-16T18:25:22 basically providing a service where you could use "telnet me someport" and I would run an ant remotely 2011-12-16T18:25:25 that you could play locally 2011-12-16T18:25:44 Fluxid: I'm back to "playing in a game" but I don't know if I believe it these days :) 2011-12-16T18:26:09 BenJackson: i didn't see so high numbers for a while 2011-12-16T18:26:18 and there are more srvers 2011-12-16T18:27:14 games must be completing near me because my rank slipped a couple while I'm in the game 2011-12-16T18:28:17 *** Redgis has quit IRC (Quit: ... mains libres) 2011-12-16T18:29:31 Fluxid: yeah. i increment version number every improvement made, though my improvements seems worth only 0.5 skill rating. 40 versions to go and i'll be on top. :) 2011-12-16T18:30:19 TL_: say your game that took 19 hills. Wow 2011-12-16T18:31:26 *** ikaros has quit IRC (Quit: Ex-Chat) 2011-12-16T18:32:56 *** dom7b5 has quit IRC (Ping timeout: 240 seconds) 2011-12-16T18:33:48 codetiger: i've razed 19 out of 28 enemy hills. got some improvements with my hill attacks by fixing a bug where it doesn't see the hill when covered by an enemy ant. 2011-12-16T18:34:11 oh ok 2011-12-16T18:34:29 Fluxid: is your web server down? 2011-12-16T18:35:08 codetiger: up for me 2011-12-16T18:35:11 codetiger: it's up here, must be on your end 2011-12-16T18:35:40 http://ants.fluxid.pl/player/codetiger does not load for me. 2011-12-16T18:36:02 it loads 2011-12-16T18:36:12 *** ikaros has joined #aichallenge 2011-12-16T18:36:34 works well 2011-12-16T18:36:40 http://fluxid.pl/misc/2011-12-17-003125_1337x812_scrot.png 2011-12-16T18:37:26 thanks for the help... problem at my end may ba 2011-12-16T18:39:17 *** Jacob_Strauss has quit IRC (Quit: Page closed) 2011-12-16T18:39:25 *** pairofdice has quit IRC (Quit: When cryptography is outlawed, bayl bhgynjf jvyy unir cevinpl.) 2011-12-16T18:39:41 http://ants.fluxid.pl/replay.37703 2011-12-16T18:39:49 500 turn fight with chu did me in there lol 2011-12-16T18:41:00 lone survivor ftw tho 2011-12-16T18:41:21 *** sigh has joined #aichallenge 2011-12-16T18:41:26 haha pguillory's gf works for google 2011-12-16T18:42:31 arrgh, finally got my game, lost it 2011-12-16T18:42:38 in very good company 2011-12-16T18:42:54 truskill should note I lost with #6, 13, 20 2011-12-16T18:43:03 so clearly I should increase in skill :) 2011-12-16T18:43:14 lol 2011-12-16T18:44:05 *** keith_ has joined #aichallenge 2011-12-16T18:46:14 even my tcp failed for fluxid... :( My bot was atleast wining some game after months of work 2011-12-16T18:48:42 Fluxid: just tried fluxid.pl and it works. but not ants.fluxid.pl 2011-12-16T18:50:24 I barely have anything left to code and I might finish it all today 2011-12-16T18:50:43 I am trying to stay rational and not change anything 2011-12-16T18:50:47 I know of some minor tweaks 2011-12-16T18:50:54 and I'm inspired by ChrisH's low food/hill work 2011-12-16T18:51:16 http://paste.pocoo.org/show/knPXxGvim0CqTG6hXS4g/ <- that's my todo list 2011-12-16T18:51:16 damn tcpants and the 750 turn limit 2011-12-16T18:51:19 and I think that even a bad implementation of that might have won the game I just lost 2011-12-16T18:51:28 I basically just have stuff left to merge, but it's not tested enough so I doubt I will 2011-12-16T18:52:40 If I could reliably detect the "chaos" games 2011-12-16T18:52:48 *** bluegaspode has joined #aichallenge 2011-12-16T18:52:52 eg multi hill, high player count, low food 2011-12-16T18:53:02 then I'd be tempted to throw in some really dumb "stand on hill" code 2011-12-16T18:53:16 because I stand a decent chance of losing anyway 2011-12-16T18:53:21 and it wouldn't affect other maps 2011-12-16T18:59:02 My term for those are "clusterf..." maps, and I don't try to detect them either. 2011-12-16T18:59:23 Scryer: ChrisH_ seems to have solved them 2011-12-16T19:00:25 I think if I just turn on defense for 1 hill it will fix those a lot, but picking that hill is tricky 2011-12-16T19:00:38 *** flowenol has quit IRC (Ping timeout: 252 seconds) 2011-12-16T19:00:51 *** hkraal_ is now known as hkraal^afk 2011-12-16T19:01:28 BJ - Interesting - he seems to have "smart" stand-on-hill code: he stands there unless he's ready to spawn one. 2011-12-16T19:03:50 yes 2011-12-16T19:04:51 I don't even know that it would have helped in my last game 2011-12-16T19:06:44 With a month to go that level of detail might exceed my attention cutoff. :) 2011-12-16T19:08:06 yeah, your last game looks like the biggest problem is you let yourself get distracted from expansion to easily and soon 2011-12-16T19:08:22 BJ: how did your v7 do on aichallenge server? 2011-12-16T19:08:27 *** peerai has quit IRC (Ping timeout: 258 seconds) 2011-12-16T19:08:30 it's more than just waiting until I have ants ready to spawn, it also includes chosing where to spawn them to suit my strategy 2011-12-16T19:08:41 codetiger: you can see it on my profile, it's the version 7 bot 2011-12-16T19:08:50 Now my bot can take it down, if not everytime, atleast most of the time 2011-12-16T19:09:10 janzert: I didn't measure that game but I do have some constants based on the rules for minimum hill distance 2011-12-16T19:09:15 and there are maps in the rotation that violate them 2011-12-16T19:10:00 janzert: I think in that case parapsprites (sp) got off to a good start and I was first on the menu 2011-12-16T19:10:15 that game was lost when I didn't find closer food and got behind in ant count 2011-12-16T19:10:30 being a maze the consequences of random decisions are more significant 2011-12-16T19:11:17 but maybe if I'd focused on one hill or the other I could have kept one 2011-12-16T19:11:44 if my other one had not ALSO been right next to the same guy it might have worked out that way automatically 2011-12-16T19:12:00 I'd say you sent to many ants south east then let a few ants dance around your south hill and combined that put you too far behind to have any chance 2011-12-16T19:12:04 I posted a suggestion about evaluating maps based on how accurately their outcomes match the true-skill of the participants. I'm not a big fan of luck-of-the-draw. 2011-12-16T19:12:25 Scryer: that would be interesting to see 2011-12-16T19:12:39 My bot is much better now, on multi hill mazes with very close neighbor. My last version has a bug that solved it 2011-12-16T19:12:57 but don't expect it to make it into use for the contest this time 2011-12-16T19:13:13 Heh. We'll start in March for the next contest. :) 2011-12-16T19:13:38 Scryer: I was thinking along the same lines 2011-12-16T19:13:50 however that just tends to reward people who are already good 2011-12-16T19:14:05 there was talk of doing that for this contest very early on but no one ever actually did it 2011-12-16T19:14:09 and credit to ChrisH_ for providing a practical example of how someone could actually find advantage in a map that looks liek crap to the rest of us 2011-12-16T19:14:24 if you screened those maps because xathis wasn't good at them it would just reward xathis 2011-12-16T19:14:50 BenJackson: yeah, you'd have to be very careful on what the actual criteria for defining a good map is 2011-12-16T19:15:02 Well, the point is to *find* who the best players are. Is xathis the best? Fine. Could we make the maps random enough to make him lose more? Probably, but what does that prove? 2011-12-16T19:15:19 Scryer: but that's just it 2011-12-16T19:15:26 I would have said multi-hill maps were not good at determining skill 2011-12-16T19:15:35 it just turns out that the aspect of skill they required hadn't been discovered yet 2011-12-16T19:15:52 if a _lot_ of games are played in the finals, I don't think it will matter 2011-12-16T19:15:54 you could make maps with 1-wide mazes everywhere 2011-12-16T19:15:59 it sounds pointless, but maybe there's a strategy there 2011-12-16T19:16:25 thestinger: ...and that is why I try to ignore my losses 2011-12-16T19:16:31 but I think you really need to do tons and tons of games to eliminate the noise 2011-12-16T19:16:32 I have not written a "win every game" bot 2011-12-16T19:16:35 but it will win some 2011-12-16T19:16:37 Ah. So you're suggesting that if it were all the clusterfun maps, ChrisH_ might well end up the winner, so he'd be "best" on that set. I see your point. One needs a wide range. 2011-12-16T19:16:37 and against good players 2011-12-16T19:17:02 Scryer: there are lots of maps were two great players can be screwed over because they started next to each other 2011-12-16T19:17:22 Scryer: right. at this poitn I have no doubt that if the finals were all multi-hill that ChrisH would have a shot at #1 2011-12-16T19:17:40 but there *are* still highly asymmetrical maps 2011-12-16T19:17:41 my bot can beat xathis/pguillory if they had a really bad starting position :) 2011-12-16T19:17:44 I think you could rule those out 2011-12-16T19:17:48 I can also imagine that somebody optimizes for 4-player maps and is unbeatable at them, but when it comes down to a 50-game finals of 2-player matches he's lost. 2011-12-16T19:17:59 "when" 2011-12-16T19:18:06 I don't think that's the plan, is it? 2011-12-16T19:18:14 no 2011-12-16T19:18:28 I would expect *some* winnowing of the "go-north" bots. 2011-12-16T19:18:49 if I give my bot a direction bias it does a lot better 2011-12-16T19:18:57 I'm questioning your 2-player assumption 2011-12-16T19:19:05 it might winnow to the top N, but still withthe normal map rotation 2011-12-16T19:19:13 and I found that if I randomly change the bias now and then, it's a huge improvement over no bias at all :P 2011-12-16T19:19:34 It was a 2-player hypothesis, not an assumption. :) 2011-12-16T19:19:36 thestinger: hah. while I worked hard to make sure I selected influences that distributed my ants evenly 2011-12-16T19:19:53 my latest bot's main innovation is trying to predict trouble and go there 2011-12-16T19:20:03 like a "danger" scent? 2011-12-16T19:20:08 I've tried stuff like that 2011-12-16T19:20:10 something like that 2011-12-16T19:20:15 Something more predictive than dead ants? :) 2011-12-16T19:20:21 like if a bot is keeping me out of an area 2011-12-16T19:20:29 Scryer: as in going to empty areas beause I think an enemy will show up there 2011-12-16T19:20:41 if there's an area I haven't explored by the end of the game, there's probably a really damn good bot there 2011-12-16T19:20:42 do you use the dead ants infor for that ? 2011-12-16T19:20:45 if you watch my current version you can see it intercepting ants it can't even see sometimes 2011-12-16T19:20:48 which is pretty neat 2011-12-16T19:21:00 my bot tries to pick off weak hills and get points, not win the long-term game 2011-12-16T19:21:09 that's a good idea 2011-12-16T19:21:26 I try to win the long term game 2011-12-16T19:21:29 if the finals were done with an elimination tournament, I would do terribly :P 2011-12-16T19:21:41 in fact even moreso now my bot's goal is to stay alive and create a buffer of safety 2011-12-16T19:21:46 it has almost zero idea of defense now 2011-12-16T19:21:50 if it has to defend it messed up earlier 2011-12-16T19:22:05 I aimed for doing something like that, but I didn't get it finished 2011-12-16T19:22:07 if you *see* my hill I messed up 2011-12-16T19:22:10 so I just go for points 2011-12-16T19:22:16 "If you find yourself in a fair fight you've failed the planning phase." 2011-12-16T19:22:26 that's why I'm proud of that one game 2011-12-16T19:22:27 heh, isn't the stated goal purely defense? 2011-12-16T19:22:30 ;) 2011-12-16T19:22:37 where I got tied up, but pushed out and killed everything 2011-12-16T19:24:00 I have *got* to write down that url 2011-12-16T19:24:11 For some reason, I keep losing, and I can't figure out why 2011-12-16T19:24:20 I don't see any obvious flaws in my bot 2011-12-16T19:24:35 I see lots of huge flaws in my bot but I can't fix them or I ruin it 2011-12-16T19:24:50 it does a lot of risk taking and "all in" stuff 2011-12-16T19:25:11 I have an estimate of the number of enemies and the scores they have 2011-12-16T19:25:24 cool 2011-12-16T19:25:26 and all I do is try to get the easiest 2 points :P 2011-12-16T19:25:52 http://aichallenge.org/visualizer.php?game=252102&user=3473 that one! 2011-12-16T19:25:56 controlling territory is beyond the intelligence of my current bot 2011-12-16T19:25:58 ok time to put that in a bookmark 2011-12-16T19:26:06 yeah, that was a great game 2011-12-16T19:26:13 you didn't even have extra ants in your hill 2011-12-16T19:26:19 that's a game I totally would have lost in earlier versions 2011-12-16T19:26:21 I have a lot of trouble getting my bot smarter than "Here is Grond. There is gate. I go smash gate." 2011-12-16T19:26:30 Scryer: that's my bot! :P 2011-12-16T19:26:45 thestinger: you totally got me with that strat in some fluxid (?) game 2011-12-16T19:26:48 *** Jak_o_Shadows has quit IRC (Remote host closed the connection) 2011-12-16T19:26:48 Strcat the Barbarian 2011-12-16T19:26:54 Yeah - and I've been the gate very often on Fluxid for you. :) 2011-12-16T19:26:58 where a snake of your ants just dove on my hill early 2011-12-16T19:27:11 I think current version would have intercepted that 2011-12-16T19:27:16 maybe 2011-12-16T19:27:24 you would probably lose anyway, you would just screw me over too 2011-12-16T19:27:33 acceptable 2011-12-16T19:27:34 like, you would kill me 2011-12-16T19:27:39 but you probably wouldn't recover :P 2011-12-16T19:28:02 I only have to hold you until your other neighbor kills you 2011-12-16T19:28:05 my bot has a lot of games with 1st place, and a lot with 0 points 2011-12-16T19:28:09 you see that all the time when a bot has a lopsided start 2011-12-16T19:28:18 BenJackson: the idea is that it attacks the weak neighbour first :) 2011-12-16T19:28:23 thestinger: THAT is exactly what I'm trying to address with this version 2011-12-16T19:28:26 more mid-finishes 2011-12-16T19:28:39 thestinger: that game you went so early you hadn't seen me 2011-12-16T19:28:58 like you said, it was suicide or it was easy points 2011-12-16T19:29:01 in that case it was easy points 2011-12-16T19:29:16 aichallenge: janzert epsilon * rec95e83 / sql/opponent.sql : Update opponent.sql seed selection to match actual matchup generation - http://git.io/8iB9Uw 2011-12-16T19:29:18 http://tcpants.com/replay.2288 I hate these maps 2011-12-16T19:29:28 I really have no idea what is going on in those, even after watching them 20 times 2011-12-16T19:29:46 my new technique is to try to note the order of hill loses 2011-12-16T19:29:49 losses 2011-12-16T19:29:50 I can't figure out _why_ my bot does stuff because the enemies are so close that the weight I give to ants near my hills totally changes them 2011-12-16T19:29:55 so I can focus in the right area 2011-12-16T19:29:57 without watching 20 times 2011-12-16T19:30:35 thestinger: I've found I have to be careful with metrics that are "close to my hill" or "close to enemy" and make sure they get some meaningful answer for "close to both" 2011-12-16T19:30:36 ? 2011-12-16T19:30:46 BenJackson: yeah 2011-12-16T19:30:46 by the way, do you have information on ChrisH's multihill strategy? 2011-12-16T19:31:02 not really 2011-12-16T19:31:10 I think the biggest "leak" is showing that it can work 2011-12-16T19:31:12 *** dom7b5 has joined #aichallenge 2011-12-16T19:31:16 my bot is smarter than I am most of the time though 2011-12-16T19:31:17 making it work is not rocket science 2011-12-16T19:31:27 just risky this late 2011-12-16T19:31:46 Antimony: along those lines we were discussing my secret symmetry detection the other day 2011-12-16T19:31:53 McLeopold guessed when I hinted 2011-12-16T19:32:00 which makes sense because he's sort of responsible 2011-12-16T19:32:06 using the food spawns? 2011-12-16T19:32:08 yes 2011-12-16T19:32:17 yeah, that's very neat 2011-12-16T19:32:18 Antimony: the game engine leaks symmetry information in the food spawning 2011-12-16T19:32:28 oh 2011-12-16T19:32:38 I thought about using food for symmetry but I haven't implemented that 2011-12-16T19:32:45 I didn't think it would help enough to be worth it 2011-12-16T19:32:53 I spent most of my time learning C++ well, rather than implementing cool things :P 2011-12-16T19:33:09 I only knew enough C to submit a patch or two to some projects 2011-12-16T19:33:24 never actually coded anything in C (and I hadn't touched C++) 2011-12-16T19:33:38 By the way, is there supposed to be a link in this post? http://forums.aichallenge.org/viewtopic.php?f=21&t=2056 2011-12-16T19:33:50 Antimony: (almost) every set of food you see appear simultaneously is symmetric around the hills 2011-12-16T19:34:12 so if you see two food you know that the relationship between the food is identical to the relationshipo between two hills 2011-12-16T19:34:23 and you can pick your hill arbitrarily to be one of them 2011-12-16T19:34:32 only if the 2 food are conjuagates 2011-12-16T19:34:38 so basically every time you see simultaneous food you discover a possible hill 2011-12-16T19:34:43 there are several things that mess it up, so it'd have to be probablistic 2011-12-16T19:34:46 my method didn't find water at all 2011-12-16T19:34:49 like dmj111 2011-12-16T19:35:02 right, if you generate a possible hill and you can see it's not you skip it 2011-12-16T19:35:09 and you have to track known dead hills 2011-12-16T19:35:14 oh 2011-12-16T19:35:29 well the brute force method is more likely to give results 2011-12-16T19:35:34 I have something pretty cool for symmetry detection. Its very fast and really definitely not worth it. 2011-12-16T19:35:38 you don't even have to see the hills to confirm 2011-12-16T19:36:08 I guess doing probabilistic reasoning would help 2011-12-16T19:36:12 bugnuts: ditto 2011-12-16T19:36:30 Is there supposed to be a link in this post? http://forums.aichallenge.org/viewtopic.php?f=21&t=2056 2011-12-16T19:37:02 I was a lot of fun to write though. 2011-12-16T19:37:42 yeah, I wrote a lot of neat stuff I'm not going to be using :P 2011-12-16T19:38:00 the successful improvements were all really simple things 2011-12-16T19:39:06 ChrisH talks about a game, but there's no game in the post 2011-12-16T19:41:35 sorry, just did an edit and put the link in 2011-12-16T19:42:10 *** Murashka has quit IRC (Quit: Page closed) 2011-12-16T19:42:27 *** kire has joined #aichallenge 2011-12-16T19:43:38 I reckon to win this I just need one more thing -- a montage. 2011-12-16T19:44:26 I need a better random number generator that always gives the "right" numbers :) 2011-12-16T19:45:07 Yeah, and a catchy backing beat would be helpful... getting the crowd onside is a big boost 2011-12-16T19:45:26 how do you decide which hill to spawn at? 2011-12-16T19:45:44 Ant: you never spawn at a covered hill 2011-12-16T19:45:58 I know 2011-12-16T19:46:04 how does Chris decide which hill to uncover? 2011-12-16T19:47:06 Presumably he knows where he is going and sends the ant on the hill out to other places 2011-12-16T19:50:31 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-12-16T19:54:58 Antimony: A full explanation would require explaining my whole pathfinding system, but the basic idea is to spawn ants at the hill closest to desirable targets 2011-12-16T19:55:16 doh, he just quit 2011-12-16T19:55:19 oh well 2011-12-16T19:57:34 *** olexs has quit IRC (Quit: Leaving.) 2011-12-16T20:01:11 is euiclidean distance better than manahattan for ants ? 2011-12-16T20:01:44 those are the kinds of changes I want to make right now 2011-12-16T20:02:16 *** dom7b5 has quit IRC (Ping timeout: 240 seconds) 2011-12-16T20:02:40 I guess eucildean may be better for flocking since it follows the "natural" path. But i use manahttan since it is faster 2011-12-16T20:03:11 I'm not using manhattan distance for anything 2011-12-16T20:03:32 I use the radius around certain things to mark off information at the start of the turn, but I don't really use euclidean distance for anything 2011-12-16T20:03:53 my A* uses manahattan and I know that is faster for ants (NWSE) 2011-12-16T20:04:25 *** bretep has quit IRC (Ping timeout: 258 seconds) 2011-12-16T20:04:32 what do you use there ? 2011-12-16T20:04:57 I don't use A* anymore 2011-12-16T20:05:04 *** kire has quit IRC (Quit: Leaving) 2011-12-16T20:05:05 I only used it at the start 2011-12-16T20:05:19 Fluxid told me a better way of doing it early on and I switched to that 2011-12-16T20:05:30 bfs ? 2011-12-16T20:05:37 similar 2011-12-16T20:05:45 sort of like a BFS from a target, but I make distance maps 2011-12-16T20:05:54 so I know all the true distances and paths to a target for every tile 2011-12-16T20:05:59 but there are lots of hacks and weights now 2011-12-16T20:06:02 so it's not BFS 2011-12-16T20:06:15 and I combine some of the maps 2011-12-16T20:06:21 anyway 2011-12-16T20:06:32 with just the vanilla BFS from a target, it was way faster than A* 2011-12-16T20:06:49 even for 1 ant 2011-12-16T20:07:11 your BFS doesn't have to allocate any memory 2011-12-16T20:07:29 you can statically allocate the distance map (a 1D or 2D array) and use that to avoid revisiting tiles 2011-12-16T20:07:41 and the FIFO queue has a known max size 2011-12-16T20:08:37 Locations are just hardware ints (row * cols + col), so expanding 4 nodes just means sticking 4 ints onto the queue 2011-12-16T20:08:50 that is interesting. I still do A* since I do only one pass. 2011-12-16T20:08:54 not expensive with a struct either 2011-12-16T20:09:39 A* is a bit easier to tweak than this, because you can use a weighted 2d array to get costs 2011-12-16T20:10:06 but I was able to get really good food searching with this by getting nice spreading behavior to emerge 2011-12-16T20:11:08 i am afraid my ants are one dimensional most of the time. 2011-12-16T20:11:20 i go either for attacking or exploring. there is no balance 2011-12-16T20:11:47 my individual ants are pretty narrow minded 2011-12-16T20:11:59 but I start fresh every turn, I don't have assigned roles 2011-12-16T20:12:15 besh: I figure that since you have to walk by the Manhattan distance, might as well calculate in walking steps 2011-12-16T20:13:34 Samvara: yes that is why i do manahattan. But when attacking a hill f.i the ants start to converge earlier with euclidean better (more flocking) 2011-12-16T20:13:47 thestinger: my bfs is similar 2011-12-16T20:14:05 but it's funny that I had put some key ones (like distance to any enemy, any hill, etc) in Bot so they didn't get reallocated 2011-12-16T20:14:14 then I discovered that it was actually faster to alloc new ones 2011-12-16T20:14:32 what if you just use static local vars? 2011-12-16T20:14:44 I haven't speed tested either so I'm just reallocating most things 2011-12-16T20:14:52 my grid always allocs 2011-12-16T20:15:10 I did implement grid.swap() which was actually handy in several places 2011-12-16T20:15:38 *** bluegaspode has quit IRC (Quit: Page closed) 2011-12-16T20:16:37 In java i just reserve 600 tiles for the priority queue. Allocation/deallocation should be fast. 2011-12-16T20:17:02 *** Boscop has quit IRC (Ping timeout: 252 seconds) 2011-12-16T20:18:27 i use a distance array which avoids use of closed/open sets. 2011-12-16T20:18:54 FWIW, my code is probably simiilar to thestinger, but I actually use Djikstra's algorithm and use node traversal costs to help guide ants away from places that they aren't needed or would have a hard time getting through but since they are just higher cost the ants will still go that way if there are no cheaper alternatives. 2011-12-16T20:20:00 I have a cost of 1 for normal tiles, but I do have different costs 2011-12-16T20:20:06 *** ikaros has quit IRC (Remote host closed the connection) 2011-12-16T20:20:17 like 3 when going through a tile with one of my ants (which causes them to fan out into lines) 2011-12-16T20:20:52 *** treeform has quit IRC (Remote host closed the connection) 2011-12-16T20:21:14 ChrisH_: How do you get the frontier combat formation in your bot if you don't mind ? 2011-12-16T20:21:45 similar to what thestinger just said 2011-12-16T20:21:57 I hope to do a writeup on the forums next week. 2011-12-16T20:22:52 I do what thestinger does i.e penalize path through tiles with 2 but ants still accumulate on choke points. But yours and pigullary's seem to handle that much better I think. 2011-12-16T20:23:07 ChrisH_: does a wall of your ants basically stop ants behind it from choosing that as a goal? 2011-12-16T20:23:23 no 2011-12-16T20:24:05 but in the case of battle it will have that type of effect, but it doesn't stop them 2011-12-16T20:24:27 it will tend to push my ants toward the flank though, which is good 2011-12-16T20:24:53 but if there are walls at the flanks already then that's impossible obviously 2011-12-16T20:25:27 so a few ants might stack up behind the front line, which can help strengthen the line, and it also makes other ants even less likely to go that way 2011-12-16T20:26:28 but if there is no other choice, then they will just keep piling up, which is why I'm experimenting with some logjam breaking code now 2011-12-16T20:27:41 It is excellent that you are able to get that formation with this small change. I thought of using influnce maps to send ants where ants are needed most. 2011-12-16T20:28:47 but this is where one of the really cool things happens with my spawn control code on multiple hills: if a hill gets surrounded by the enemy and I build a wall they can't get through, my spawn control will interact with the extra cost of going through the wall to prevent more ants from spawning behind the wall and let them go out another hill to do useful stuff, maybe even attack the enemy wall from behind. :evil grin: 2011-12-16T20:31:22 I have increased upto +6 and it seems that I still get pile of ants 2011-12-16T20:31:29 hi, how is the contest going 2011-12-16T20:31:34 BenJackson: Nice write up. Not sure I can chime in on much... because my opinion mostly matches yours 2011-12-16T20:31:41 you can see that happen in this game: http://aichallenge.org/visualizer.php?game=248555&user=10 2011-12-16T20:32:49 I see you leave an ant on the hill. 2011-12-16T20:33:23 dmj111: thanks 2011-12-16T20:33:53 ChrisH_: synergies like that are really interesting 2011-12-16T20:34:02 how do you decide to keep the formation or go for the kill ? My bot is generally adventorous and will break its formation even when being close to its own hill. 2011-12-16T20:34:40 A few enemy ants pass through and my formation is gone. Do you have weights to adjust behaviour when being close to enemy hill / our own hill ? 2011-12-16T20:34:51 yeah, my bot's logic has actually gotten simpler over the last few weeks because I have found more of those synergies and made the code more general purpose and less special cases. 2011-12-16T20:35:47 besh: I don't have weights exactly for that, but there are some thresholds related to hill distances that trigger adjustments to behavior 2011-12-16T20:37:31 besh: my formations are completely emergent based on how my pathing, target selection, and the enemy behavior interact 2011-12-16T20:37:45 I don't try to create any specific fomations 2011-12-16T20:37:48 my bot got more complex because I think I'm approaching the problem backwards now 2011-12-16T20:38:27 i don't look for choke points (which at the beginning of this contest I spent a lot of time thinking about), controlling choke points turns out to be emergent as well 2011-12-16T20:38:51 which is surprising, but cool 2011-12-16T20:39:15 I thought I would need anti-chokepoint avoidance too 2011-12-16T20:39:42 but in my latest code once I cover a chokepoint the combat code doesn't call for help and the "make sure my base isn't jumped" code ignores that path 2011-12-16T20:39:57 huh, i always thought the opposite, try to jam the choke points as a means to defend with less ants 2011-12-16T20:40:01 Even with the emergent behaviour , you have to guide it somehow to prefer to standoff when being close to your hill. You have f.i a double layered emergent formation at Move 498. My bot would completely break that I think.. 2011-12-16T20:40:03 hey I just beat pguillory and a1k0n on aichallenge 2011-12-16T20:40:14 *** Gnome has quit IRC (Quit: Gnome) 2011-12-16T20:41:18 ChrisH_: Thanks for the explanation. I have to go now. 2011-12-16T20:41:24 besh: once my ants get that close to the enemy then their moves are completely controlled by combat logic 2011-12-16T20:41:25 *** besh has quit IRC (Quit: Page closed) 2011-12-16T20:41:44 BenJackson: nice job 2011-12-16T20:42:49 BenJackson: I'm in a game now, I wonder who against. 2011-12-16T20:43:41 janzert: quick make a top mu list ;-) 2011-12-16T20:44:40 wow greentea is at 15 with only 18 games 2011-12-16T20:45:20 I reuploaded again :P 2011-12-16T20:45:37 ah helps he beat xathis on a p10 map 2011-12-16T20:45:50 http://pastebin.com/KC30AM8R 2011-12-16T20:46:36 http://aichallenge.org/visualizer.php?game=254738&user=2255 this map is terrifying for some reason 2011-12-16T20:47:01 wow my mu is in a very busy area 2011-12-16T20:47:25 someone with mu 94.5 would be all alone in 5th pretty much 2011-12-16T20:47:56 wow, xathis is way ahead now 2011-12-16T20:48:15 my game was with pguillory and cheeser, finished a three way stalemate 2011-12-16T20:48:35 ChrisH, pguillory, a1k0n and GreenTea seem to have good control of 2-5 2011-12-16T20:48:50 but right after that it's wide open for about 10 places 2011-12-16T20:48:57 well, _flag will be back up there 2011-12-16T20:49:16 oh I was misreading rank as games 2011-12-16T20:49:29 janzert: you should modify that query to have games @ current version 2011-12-16T20:50:08 I was at 91.75 two days ago, but I only gave my last version 1 day to play games 2011-12-16T20:50:12 *** kire has joined #aichallenge 2011-12-16T20:50:16 hopefully I get back up there :) 2011-12-16T20:50:19 *** bmh has joined #aichallenge 2011-12-16T20:50:30 * BenJackson decides too much time has passed for any matchup glitch miracles to occur for this round of the queue 2011-12-16T20:50:57 chrish: you're the front page game 2011-12-16T20:51:07 I probably was just before that but I forgot to look 2011-12-16T20:51:11 yeah, that's not surprsing 2011-12-16T20:51:12 before I start in on combat, I should probably submit my bot to make sure it compiles 2011-12-16T20:51:32 wow pguillory mowed down your first defensive line 2011-12-16T20:51:38 for combat, do you imagine, all enemy neighbors or only passable ones? 2011-12-16T20:51:41 I'm amazed you lived 2011-12-16T20:51:55 *** kire has quit IRC (Client Quit) 2011-12-16T20:53:22 hmmm, my logjam breaking code has some potential. 2011-12-16T20:53:34 *** amstan has joined #aichallenge 2011-12-16T20:53:34 *** ChanServ sets mode: +o amstan 2011-12-16T20:54:36 BenJackson: yeah, he really outflanked me there and you would expect that to be the end, i'm not sure how i survived either 2011-12-16T20:55:00 he got you again when he managed to anchor on a bit of water 2011-12-16T20:55:01 but time ran out 2011-12-16T20:56:53 looking at the game a little bit more I think there is an interesting dynamic between pguillory and my bot. we both form very similar battle lines, and when we are next to each other in games you always see a big straight line of our ants between our hills like that. 2011-12-16T20:57:30 ChrisH_: Are you counting the portion of the board you control and making your bot more aggressive if you're getting squeezed? 2011-12-16T20:57:44 but i've noticed pguillory tends to put less emphasis on food gathering and tries to push as many ants as possible to the perimeter, so he has a few more to flank me, but then I have more in my interior to recover with 2011-12-16T20:57:52 bmh: no 2011-12-16T20:58:14 i tried things like that in my V6 and it didn't work very well 2011-12-16T20:59:02 my bot gets more aggressive if it has local numbers, if it's close to taking or losing a hill, and if it has too many total ants 2011-12-16T20:59:10 i think i had a larger feeding ground at the time he flanked my first line, but his maneuver had the effect of crippling me to a degree and had the game gone on longer he would have won 2011-12-16T20:59:12 and also randomly in stalemates 2011-12-16T20:59:13 BenJackson: http://www.youtube.com/watch?v=O2IuJPh6h_A&ob=av3e 2011-12-16T20:59:34 mleise: not available in my country 2011-12-16T21:00:00 same for me ^^ 2011-12-16T21:00:06 wow, greentea just had a great win 2011-12-16T21:00:21 *** Antimony has joined #aichallenge 2011-12-16T21:00:24 mleise: I love the Dr. Strangelove War Room in that video. 2011-12-16T21:00:24 mleise: what was the url about then? 2011-12-16T21:00:27 (blocked in my country) 2011-12-16T21:00:41 he took down #1 #2 and #3 all in the same game 2011-12-16T21:00:53 BenJackson: about running out of time 2011-12-16T21:01:23 mleise: I'm *starting* combat tonight :-D 2011-12-16T21:01:40 Do you all like my visualizer? :D 2011-12-16T21:01:46 ChrisH - Step 1. Jettison a couple of extraneous GreenTea hills. :) 2011-12-16T21:01:48 mleise: link? 2011-12-16T21:02:02 Scryer: that often works 2011-12-16T21:02:05 bmh: it is the official onw ;) 2011-12-16T21:02:10 oh 2011-12-16T21:02:11 itsTime is running out song video 2011-12-16T21:02:25 its "Time is running out" song video 2011-12-16T21:02:26 hence I'm considering ChrisH_'s idea of standing on hills because even if it's a poor implementation it may work anyway 2011-12-16T21:02:48 Today was a strange evening at the pub. 2011-12-16T21:03:05 I'm really really trying to consider myself finished at this point though 2011-12-16T21:03:19 worst case is resubmitting any time on sunday or even saturday 2011-12-16T21:03:20 BenJackson: one side benefit of standing on hills in the tightly clustered multi-hill maps is that your hills can't get ambushed by a single enemy coming in through the back door 2011-12-16T21:03:25 yep 2011-12-16T21:03:53 but even if it works and I place higher I don't think I'll be too excited if it's just because I stole your idea 2011-12-16T21:03:56 and once you see them you can spawn some ants to drive him off, assuming there is food to be had 2011-12-16T21:04:18 we all steal from each other in the end 2011-12-16T21:04:31 it'd be funny to time it perfectly so that as their 1 ant sneaks up you are suddenly 2 2011-12-16T21:04:40 ooh hills should be wormholes for your ants 2011-12-16T21:04:47 that would be really interesting 2011-12-16T21:05:06 yes, let's change the rules now 2011-12-16T21:05:07 step on your own hill while it's occupied, 1 ant goes back in your hive 2011-12-16T21:05:55 mleise: that's how ICFP runs 2011-12-16T21:06:12 really? I thought ICFP was just 3 days long 2011-12-16T21:06:19 but I haven't participated in over 10years 2011-12-16T21:06:39 must be some lisp thing 2011-12-16T21:06:51 BenJackson: they change the spec 2011-12-16T21:12:54 *** bmh has quit IRC (Quit: Page closed) 2011-12-16T21:17:47 *** mleise has quit IRC (Ping timeout: 240 seconds) 2011-12-16T21:21:14 *** amstan_ has joined #aichallenge 2011-12-16T21:21:14 *** ChanServ sets mode: +o amstan_ 2011-12-16T21:25:22 *** treeform has joined #aichallenge 2011-12-16T21:28:35 Benjackson: early in my analysis I was sure that more boots on the ground must always be better -- you see more, you have more chance to collect food and probably better odds in a fight. I'm still finding it hard to believe that standing on holes is a good plan. 2011-12-16T21:31:23 Samvara: standing on a whole may let you put more boots on the ground faster elsewhere on the map. 2011-12-16T21:31:30 s/whole/hole 2011-12-16T21:32:14 so more boots is good; more boots in the right place is better 2011-12-16T21:32:24 Yeah, I've noticed that the multi-hole games are getting more common in recent rounds. Check game 254825 (if you haven't already) your fronts get pushed back a bit easily 2011-12-16T21:32:46 I can't figure out how to improve my bot. There aren't any obvious flaws 2011-12-16T21:32:56 You are getting nice neat fronts though... I've not been able to get mine as nice as that 2011-12-16T21:33:23 Samvara: pguillory is a tough opponent, his battle code is better than mine 2011-12-16T21:33:42 at least at the moment 2011-12-16T21:33:52 don't know if i will ever match his though 2011-12-16T21:34:00 not with two days left 2011-12-16T21:34:26 but i do hope to have the edge in multi-hill maps 2011-12-16T21:34:34 I also did all my optimising against single-hole maps... and roughly mid-size maps. I'm getting whacked in the "cage match" small maps 2011-12-16T21:35:17 yeah, i was getting very annoyed by those maps too, so I decided to do something about that, i am a bit surprised how well it has worked actually 2011-12-16T21:35:32 i'm working on my bots movement out from hills 2011-12-16T21:35:36 at the moment that is just awful 2011-12-16T21:35:37 it was an experiment at first, but if it works you run with it 2011-12-16T21:36:24 now I'm trying to make an improvement to how I handle stalemates of combat lines 2011-12-16T21:36:39 not sure if it will work, but the early experiments have had good results 2011-12-16T21:37:09 i'm probably going to resubmit in a before my next match so I get a full day of games with my latest code before the finals start 2011-12-16T21:37:54 *** kevlar has joined #aichallenge 2011-12-16T21:38:16 I'm wondering if it's worth running a massive optimization run against small maps and have a bunch of brain-transplant decisions early in the game 2011-12-16T21:39:18 depends on your style. I prefer one size fits all approaches usually 2011-12-16T21:42:11 I've done OK on the midsize maps that I optimised against (e.g. 249305 or 252994) but yeah my front lines are a bit messy 2011-12-16T21:47:44 I only heard about the comp a couple of weeks ago and I couldn't get time off work so I've been missing heaps of sleep... and I've also been on the oncall list so I've been missing even more sleep. Things are starting to get a bit weird. 2011-12-16T21:48:15 i'm with chris, i like it to work generally 2011-12-16T21:48:20 within reason 2011-12-16T21:48:27 There's this instinct to say, "well that's quite a reasonable effort, better give up now", but it's hard to see sense when you know you could do better 2011-12-16T21:48:36 you could even vary the attack radius etc. for my bot and it should still work pretty well 2011-12-16T21:48:43 although i haven't tested that 2011-12-16T21:49:03 I should probably wear the colours and stuff :-) 2011-12-16T21:49:08 *** Samvara is now known as Kettling 2011-12-16T21:49:44 *** Paradoxial has quit IRC (Quit: :) 2011-12-16T21:50:15 Zannick, are equidistant food sets possible? The specification and comments in the engine say yes, but it looks like they are filtered out in the engine 2011-12-16T21:50:33 i'm not really the one to ask here 2011-12-16T21:51:14 oh 2011-12-16T21:51:16 who is? 2011-12-16T21:53:40 what do you mean by equidistant food sets? 2011-12-16T21:53:43 possibly me 2011-12-16T21:54:57 A...*...B 2011-12-16T21:56:00 *** ChrisH_ has quit IRC (Quit: Page closed) 2011-12-16T21:57:02 hmmm not sure if the engine does them or not 2011-12-16T21:57:04 probably ask mcleopold 2011-12-16T21:57:12 it should 2011-12-16T22:03:57 has anyone done searching from targets to sources and not allowing ants to stay still? 2011-12-16T22:04:01 i'm finding that surprisingly hard 2011-12-16T22:07:32 why would you prevent ants from staying still? 2011-12-16T22:07:42 because otherwise they can pile up back to your hill 2011-12-16T22:07:46 and you sort of get stuck 2011-12-16T22:08:01 well I have an anticlustering tendency 2011-12-16T22:08:02 or they stall when they could move closer towards their target cause they're stuck by a wall of ants 2011-12-16T22:08:06 so that doesn't happen to me 2011-12-16T22:08:16 but you want them to stall when encountering enemies 2011-12-16T22:08:36 i'm going to randomise my path finding after i've fixed this so ants sort of spray along the shortest paths 2011-12-16T22:08:44 but i want to get this streamlining properly first 2011-12-16T22:09:15 you can still have choke points that ants have to pass for the shortest path though 2011-12-16T22:09:18 Anyway, I'm wondering about lines 1321-1326 of ants.py 2011-12-16T22:09:24 and sometimes it's better to just get some ants to take a slightly suboptimal path 2011-12-16T22:09:31 I do that too 2011-12-16T22:09:46 distances are increased if they go through ants 2011-12-16T22:09:59 if the blockage is big and dense enouhg they'll go around 2011-12-16T22:10:22 is ChrisH here? 2011-12-16T22:10:27 *** amstan_ has quit IRC (Ping timeout: 240 seconds) 2011-12-16T22:12:03 i think i got it 2011-12-16T22:15:44 Winning strategy seems to depend heavily on the map. In a tight corridoor map you want to focus on exploration and food finding and then just block off a corridoor if you can't easily push through. 2011-12-16T22:16:34 In a big open map good battle tactics seem to win the day and being able to push back your fronts efficiently (gain ground, without losing too many guys) 2011-12-16T22:20:57 In a tight cage match just throw everything into food collecting and aggression, even if you die grabbing a food at least the other guy didn't get that food, and there's no chance to collect it later 2011-12-16T22:27:04 Kettling: a good bot will attempt to do all of these things 2011-12-16T22:27:36 Yeah well the emphasis makes a difference 2011-12-16T22:28:26 *** JorgeB has quit IRC (Quit: Computer has gone to sleep.) 2011-12-16T22:28:29 * antimatroid goes back to trying to improve his bot 2011-12-16T22:29:04 *** jab_bott has joined #aichallenge 2011-12-16T22:29:39 just uploaded my first bot with battle capabilities - though not very advanced (or good) I can't wait to see how it does 2011-12-16T22:30:55 it'll probably be my final one though anyway, because I have AI-class study and exam to do, and a whole load of other stuff 2011-12-16T22:38:56 http://ants.fluxid.pl/replay.37960 2011-12-16T22:39:00 must be some bugs in the end there 2011-12-16T22:39:08 not sure it's worth fixing but 2011-12-16T22:39:18 *** Antimony has quit IRC (Ping timeout: 252 seconds) 2011-12-16T22:39:26 *** Antimony has joined #aichallenge 2011-12-16T22:44:11 *** Antimony has quit IRC (Read error: Connection reset by peer) 2011-12-16T22:45:01 *** Antimony has joined #aichallenge 2011-12-16T22:45:28 crazy game! 2011-12-16T22:45:47 you could have cleaned up 2011-12-16T22:46:20 hmm, login was broken for a few minutes 2011-12-16T22:46:46 grr - I hate it when you play against some loserbot who just times out and wastes your go 2011-12-16T22:47:38 http://aichallenge.org/visualizer.php?game=255441&user=3473 2011-12-16T22:47:41 that's kind of funny 2011-12-16T22:47:46 fourmidable is hard to get rid of 2011-12-16T22:47:57 look at us escorting ants across the entire map 2011-12-16T22:48:10 my bot got a load worse when I started writing battle code - I think I got something half-decent in the end though, will have to see how it does against some of the better bots tomorrow 2011-12-16T22:48:11 jab_bott: I hate bots who can mass produce at the start, go to the first hill they find and then time out (because they don't clear their cache) 2011-12-16T22:48:59 *** dragonfyre13 has joined #aichallenge 2011-12-16T22:49:33 bah 2011-12-16T22:49:48 BenJackson: have you ever tried doing pathfinding without letting ants sit still? 2011-12-16T22:49:57 what now? 2011-12-16T22:49:58 ok, quick question (hopefully). Is there something out there to how I need to create/name my setup_ext file so it's run on the server? Using cython for a module. 2011-12-16T22:51:06 BenJackson: http://pastebin.com/bNX5CBav 2011-12-16T22:51:30 what about that? 2011-12-16T22:51:30 one 'a' is trapped, and another 'a' shouldn't wait, but walk around 2011-12-16T22:51:47 you want me to run my bot against that? 2011-12-16T22:51:51 no 2011-12-16T22:52:08 i want to path find from targets to sources and move them as i search 2011-12-16T22:52:15 but i want to for example get that 'a' to walk around 2011-12-16T22:52:29 i think i can then get the trapped 'a' out by randomising path finding 2011-12-16T22:52:32 there are a lot of a 2011-12-16T22:52:43 yes, that's why i want to move them as i search 2011-12-16T22:52:57 antimatroid: why not have the ant join the end of the queue, they all move along one and the front one becomes the one you want to get there? 2011-12-16T22:53:17 *** SMJ has quit IRC (Read error: Connection reset by peer) 2011-12-16T22:53:30 jab_bott: wouldn't it be better to have 1 ant move out of the line and walk around the outside? 2011-12-16T22:53:43 locally I'm running "python setup.py build_ext --inplace", just to get the module available. So have a setup.py file, not sure what else I need to do. tmc seems to indicate it's as easy as that here: http://forums.aichallenge.org/viewtopic.php?f=26&t=1830&p=11024&hilit=cython#p11024 2011-12-16T22:53:59 Tried renaming setup.py as "setup_ext", but that didn't work :P 2011-12-16T22:54:02 and the 'a' on the right is going to be waiting ages to jump on the end of the line 2011-12-16T22:54:08 http://ants.fluxid.pl/replay.37195?turn=1050 2011-12-16T22:54:09 he could be out into the map way before then 2011-12-16T22:54:12 that game is kind of screwed up 2011-12-16T22:54:18 antimatroid: well you could get an ant to your target very quick - like short-circuit path finding... join a solid line of ants and you can effectively teleport to anywhere along the line 2011-12-16T22:54:25 antimatroid: but watch purple from that point forward 2011-12-16T22:54:35 watch the multi-ant columns of ants navigate the maze 2011-12-16T22:55:35 *** SMJ has joined #aichallenge 2011-12-16T22:57:21 heh - my bot stands no chance up against the likes of you! 2011-12-16T22:59:07 night :) 2011-12-16T22:59:12 *** jab_bott has quit IRC (Quit: Page closed) 2011-12-16T23:02:22 * antimatroid gives up 2011-12-16T23:02:27 ants are going to just wait in line 2011-12-16T23:03:10 hmm, is it just me, or does greentea look like almost always winning from xathis? 2011-12-16T23:03:40 quite a lot 2011-12-16T23:04:01 perfect, because I beat greentea 2011-12-16T23:04:45 xathis has a flaw of taking down the first enemy, even when there is a gap to explore the map (before it gets closed by enemies) 2011-12-16T23:05:20 at least if the gap is close to enemies 2011-12-16T23:05:39 BenJackson: what do you do when path finding out though? 2011-12-16T23:05:43 now you've just got to exploit that flaw 2011-12-16T23:05:54 antimatroid: general throngs of ants like that aren't pathfinding 2011-12-16T23:06:04 ie. if you have a choke point on the shortest path routes, do you make ants line up for the shortest path? or find a less optimal path for other ants? 2011-12-16T23:06:08 *** Antimony has quit IRC (Ping timeout: 240 seconds) 2011-12-16T23:06:10 they're all together in a soup and I'm maximizing the best result for all of them 2011-12-16T23:06:28 I don't pay attention to other ants when moving the big group 2011-12-16T23:06:34 * avdg doesn't have a good bot because his language was slow and the other one didn't develop well 2011-12-16T23:06:34 I considered it but I worried about weird bugs 2011-12-16T23:08:04 *** dom7b5 has joined #aichallenge 2011-12-16T23:08:09 avdg: can you cope with C programming? 2011-12-16T23:08:28 codetiger: you've probably noticed, but v10 on bhickey 2011-12-16T23:08:36 Kettling: I was trying c++ :p 2011-12-16T23:09:44 well, at least, I found out I had -or- spend my times with api's of other libraries -or- spending time writing code I need to find information about the language 2011-12-16T23:10:23 Do the comp organizers come down hard on core sharing? I'll send you my code (with the neural-nets wiped) if you like. Get you started, it's fast :-) 2011-12-16T23:10:25 and having good data structures is just damn important 2011-12-16T23:10:55 avdg: C++ has everything except algebraic data types/tagged unions 2011-12-16T23:11:42 thestinger: I'm probably a bit too blind fooled by c++11 2011-12-16T23:11:55 well, what data types are in JS that you didn't have? 2011-12-16T23:12:12 javascript has hash tables and dynamic arrays built on hash tables 2011-12-16T23:12:14 js is just too slow :p 2011-12-16T23:12:31 and the data structures aren't that efficient enough 2011-12-16T23:12:44 std::vector is a dynamic array, and std::unordered_map is a hash table like a javascript object 2011-12-16T23:13:00 yeah 2011-12-16T23:13:12 Someone put in a javascript entry? Imagine winning the comp with a javascript entry, how cool would that be? 2011-12-16T23:13:17 BenJackson: thanks for that... Just noticed 2011-12-16T23:13:20 or tr1/unordered_map if you aren't using C++11 2011-12-16T23:13:34 heh, I'm wondering what the position was of my previous js bot 2011-12-16T23:13:39 Kettling: winning with tcl would be awesome :) 2011-12-16T23:13:42 I've just submitted a new one 2011-12-16T23:13:52 *** bgs100 has joined #aichallenge 2011-12-16T23:13:52 (bug fix) 2011-12-16T23:14:02 codetiger: funny that I watch v10 and think 'nooo don't do that' and then 'ahh, not the current version' 2011-12-16T23:14:21 I would be 13th in js rank 2011-12-16T23:14:37 but the ranks are close together 2011-12-16T23:15:16 being 1st doesn't seem to be enough for xathis 2011-12-16T23:15:59 it looks like that, but I think he just got a good way to develop a good bot quickly 2011-12-16T23:16:34 how's that? 2011-12-16T23:17:02 he submitted his bot quite early (but maybe he wrote some code while the site was beta) 2011-12-16T23:17:22 and I don't think its bot was really much improved after v1 2011-12-16T23:17:36 though I could be wrong 2011-12-16T23:17:52 avdg: beta was going on for months before the contest started in october 2011-12-16T23:17:57 but they added hills 2011-12-16T23:18:07 xathis v2 is way better than v1 2011-12-16T23:18:08 did xathis v1 actually know about hills? 2011-12-16T23:18:23 which is scary because v1 was good even though you could see obvious bugs 2011-12-16T23:18:43 maybe I'm blind fooled by the bugs in v2 2011-12-16T23:20:03 is his combat code just exceptionally good? 2011-12-16T23:20:13 no, he has amazing strategy 2011-12-16T23:20:19 his ant efficiency is very good 2011-12-16T23:20:24 and his strategy is very good 2011-12-16T23:20:27 all those top bots are pretty indistinguishable when it comes to combat 2011-12-16T23:20:35 well, his strategy is fine 2011-12-16T23:20:39 except mine is probably worse 2011-12-16T23:20:50 he sends his ants to exactly where they're needed 2011-12-16T23:21:08 and he has a perfect spread of ants behind the borders for gathering food 2011-12-16T23:21:32 I think his spread was one of the basic things of his bot (by my view of it) 2011-12-16T23:22:01 well, he doesn't waste ants 2011-12-16T23:22:22 he does, by sending 10 ants where 4 would do the job perfectly 2011-12-16T23:22:39 and his bot isn't optimized to work well with less ants 2011-12-16T23:22:47 but he doesn't know how many ants are behind the fog of war 2011-12-16T23:22:52 thestinger: http://pastebin.com/HHtaZVx1 do you have any suggestions for this example? I want to do one search from targets to sources moving all sources, i want the bottom right 'a' to walk around the line of ants leading away from the hill, and i want an 'a' walking away from the hill to step out of line and walk around to let the trapped 'a' in 2011-12-16T23:22:53 too 2011-12-16T23:23:27 but sometimes its obvious there couldn't be any more enemies behind 2011-12-16T23:24:17 antimatroid: not sure 2011-12-16T23:24:27 *** Antimony has joined #aichallenge 2011-12-16T23:24:38 some of my ants would go to the lower row 2011-12-16T23:24:48 (well, I might be very critical, but I won't write a bot as good unless I have a way to put my thinking in code on a better way) 2011-12-16T23:24:49 until both rows had equal numbers 2011-12-16T23:24:52 sort of 2011-12-16T23:25:06 not really sure how to solve it in a better way 2011-12-16T23:25:53 *** codetiger has quit IRC (Remote host closed the connection) 2011-12-16T23:26:15 has anyone tried playing around with max flow type stuff? 2011-12-16T23:27:45 yep, I wrote a max matching solver 2011-12-16T23:27:47 and didn't use it 2011-12-16T23:28:15 *** bmh has joined #aichallenge 2011-12-16T23:29:06 the idea was you'd put ants down the side and goals across the top and fill it in with values 2011-12-16T23:29:10 and it could tell you which ant for which goal 2011-12-16T23:29:31 goals don't really work out that well though 2011-12-16T23:31:28 gn 2011-12-16T23:33:54 Hm. Will my code build on the server? 2011-12-16T23:35:16 *** avdg has quit IRC (Quit: Leaving.) 2011-12-16T23:36:39 *** huisho has joined #aichallenge 2011-12-16T23:36:40 bmh: btw codetiger asked for a bigger challenge on your server 2011-12-16T23:36:49 so we're up to v10 2011-12-16T23:37:01 they're getting perilously close to my state of the art ;-) 2011-12-16T23:37:15 BenJackson: ha. I should really get the maps up... 2011-12-16T23:37:32 more like.. I should finish my bot 2011-12-16T23:46:31 bmh: i'm working on mine still :P 2011-12-16T23:54:59 *** TheLinker has quit IRC (Quit: Bye) 2011-12-16T23:55:57 *** Scryer has quit IRC (Quit: Page closed)