| --- Log opened Sat Jan 15 00:00:51 2011 |
| 00:05 | | You're now known as TheWatcher[T-2] |
| 00:11 | | You're now known as TheWatcher[zZzZ] |
| 00:42 | | Attilla [Some.Dude@Nightstar-63ab3d46.threembb.co.uk] has quit [[NS] Quit: ] |
| 02:16 | | gnolam [lenin@Nightstar-38637aa0.priv.bahnhof.se] has quit [[NS] Quit: Z?] |
| 02:21 | | Tarinaky [Tarinaky@Nightstar-f349ca6d.plus.com] has quit [Connection closed] |
| 03:02 | | cpux[barf] is now known as shade_of_cpux |
| 03:08 | | Kindamoody is now known as Kindamoody[zZz] |
| 05:33 | | McMartin [mcmartin@Nightstar-5d6d0af3.pltn13.sbcglobal.net] has quit [Ping timeout: 121 seconds] |
| 05:33 | | McMartin [mcmartin@Nightstar-5d6d0af3.pltn13.sbcglobal.net] has joined #code |
| 05:33 | | mode/#code [+o McMartin] by Reiver |
| 05:34 | | * McMartin adds move logging to Hex Inverter. |
| 05:36 | | * McMartin then decides to put it under version control. |
| 05:47 | <@McMartin> | Hm |
| 05:48 | <@McMartin> | Maybe if I'm going to prototype minimax search I should do it for *real* Othello. |
| 07:36 | | ComradeBear [Stalker@3A600C.A966FF.5BF32D.8E7ABA] has quit [Ping timeout: 121 seconds] |
| 08:16 | | celticminstrel [celticminst@Nightstar-f8b608eb.cable.rogers.com] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 08:31 | | Derakon is now known as Derakon[AFK] |
| 08:36 | | AnnoDomini [annodomini@Nightstar-048bf7dd.adsl.tpnet.pl] has joined #code |
| 08:36 | | mode/#code [+o AnnoDomini] by Reiver |
| 09:43 | <@McMartin> | It's time for a real AI war! |
| 09:43 | | * McMartin sets his robots against one another. |
| 09:48 | <@McMartin> | Yeah, OK, that was pretty dramatic. |
| 09:52 | | You're now known as TheWatcher |
| 09:54 | <@McMartin> | I think we have our winner. |
| 09:54 | <@McMartin> | This guy only lost one game ever against any other of the AI candidates. |
| 09:54 | <@McMartin> | This is our Novice. |
| 09:54 | <@McMartin> | All of the other ones struggled to even beat random play. |
| 09:57 | | Vornicus is now known as Vornicus-Latens |
| 09:57 | <@Vornicus-Latens> | nice. |
| 09:57 | <@Vornicus-Latens> | So what are your various AIs? |
| 09:57 | <@McMartin> | Random Play, Maximum Grabby, and then zero- and one-ply minimax AIs that value hexes based on how hard they are to attack. |
| 09:58 | <@McMartin> | Based on how incredibly successful the one-ply was there I think this is a reasonable evaluation heuristic. |
| 09:58 | <@McMartin> | One-ply is kind of silly, though |
| 09:58 | <@McMartin> | But the zero-one-infinity rule is in full force here, so... |
| 09:58 | | * McMartin reorganizes his game resources and prepares to do another update. |
| 10:01 | <@McMartin> | Hex Inverter punishes thoughtless grabby so hard it only beats Random 60% or so of the time, apparently. |
| 10:02 | <@Vornicus-Latens> | Grabby lost to /antigrabby/ when I tried it. |
| 10:03 | <@Vornicus-Latens> | Literally: I had white playing "choose randomly among those moves that give the most immediate tiles" and black playing "choose randomly among those moves that give the /fewest/ immediate tiles" and black one. |
| 10:03 | <@Vornicus-Latens> | won* |
| 10:03 | <@Vornicus-Latens> | and making /that/ error means it's time for sleep. |
| 10:04 | <@Vornicus-Latens> | The final outcome was like 70/30 all told. |
| 10:05 | <@Vornicus-Latens> | (or its equivalent in the right number of tiles) |
| 10:09 | <@McMartin> | Yeah, OK |
| 10:09 | <@McMartin> | My first human game against the novice AI and I only managed to eke out a win by 2 points. |
| 10:15 | <@McMartin> | http://hkn.eecs.berkeley.edu/~mcmartin/GM8/inverter.exe |
| 10:59 | | Kindamoody[zZz] is now known as Kindamoody |
| 12:49 | | Rhamphoryncus [rhamph@Nightstar-473f8685.abhsia.telus.net] has quit [Client exited] |
| 12:55 | | gnolam [lenin@Nightstar-38637aa0.priv.bahnhof.se] has joined #code |
| 13:23 | | ComradeBear [Stalker@3A600C.A966FF.5BF32D.8E7ABA] has joined #code |
| 14:35 | | Kindamoody is now known as Kindamoody|out |
| 15:48 | | Tarinaky [Tarinaky@Nightstar-f349ca6d.plus.com] has joined #code |
| 16:49 | | celticminstrel [celticminstre@Nightstar-f8b608eb.cable.rogers.com] has joined #code |
| 17:18 | | shade_of_cpux is now known as cpux[barf] |
| 17:36 | <@ToxicFrog> | McMartin: "zero-one-infinity"? |
| 17:50 | | Derakon[AFK] is now known as Derakon |
| 18:25 | <@Vornicus-Latens> | Zero, One, and Infinity are the only three quantities that matter. |
| 18:57 | | Vornicus-Latens is now known as Vornicus |
| 20:10 | | Kindamoody|out is now known as Kindamoody |
| 20:37 | | Rhamphoryncus [rhamph@Nightstar-473f8685.abhsia.telus.net] has joined #code |
| 21:13 | | Attilla [Some.Dude@Nightstar-3416c3e3.threembb.co.uk] has joined #code |
| 21:13 | | mode/#code [+o Attilla] by Reiver |
| 21:44 | | Tarinaky [Tarinaky@Nightstar-f349ca6d.plus.com] has quit [Client closed the connection] |
| 21:55 | | RichardBarrell [mycatverbs@Nightstar-689c9c54.cable.virginmedia.com] has joined #code |
| 22:26 | | Rhamphoryncus [rhamph@Nightstar-473f8685.abhsia.telus.net] has quit [Client exited] |
| 22:35 | | Serah [Stalker@3A600C.A966FF.5BF32D.8E7ABA] has joined #code |
| 22:35 | | ComradeBear [Stalker@3A600C.A966FF.5BF32D.8E7ABA] has quit [Connection reset by peer] |
| 23:15 | <@McMartin> | In particular, a zero-ply AI is just "pick the max value of this evaluation function", a one-ply AI is "pick the max value of your evaluation function after simulating the other guy running your zero-ply AI", and anything 2+ involves producing an actual search algorithm and building up a tree and such. |
| 23:32 | | Tarinaky [Tarinaky@Nightstar-f349ca6d.plus.com] has joined #code |
| 23:37 | < RichardBarrell> | The definition is inductive, isn't it? An n-ply AI is, "maximise this function, after simulating that the other guy ran the (n-ply) algorithm and then made a move." and a 0-ply AI is, "maximise this function." |
| 23:40 | <@McMartin> | Yes, but 0 and 1 are both nonrecursive and you can bang them out in 10 minutes. |
| 23:49 | < RichardBarrell> | Good plan. |
| 23:52 | <@McMartin> | I did that last night |
| 23:52 | <@McMartin> | Now I have to figure out what the fanout actually is and how it's going to interact with my nonthreaded windowing-system event pump. |
| 23:55 | <@McMartin> | Also, it isn't really until you hit 2-ply that things like alpha-beta pruning actually have an effect. |
| --- Log closed Sun Jan 16 00:00:52 2011 |