Author: Uri Blass
Date: 09:03:51 08/11/04
Go up one level in this thread
On August 11, 2004 at 09:04:03, Paul Clarke wrote: >On August 11, 2004 at 06:23:32, Derek Paquette wrote: > >>On August 10, 2004 at 16:59:55, Ricardo Rego wrote: >> >>>Hi >>> >>>I heard from a plenty number of guys at Go schools (Nihon Ki-in) & Go Forums >>>that "chess has been more or less 'solved' as a programming problem". Indeed, >>>they provided me the following link, where a gentlemen explains why computers >>>can't play Go: http://www.cns.nyu.edu/~mechner/compgo/sciences/ >>> >>>Someone says "Go is totally intuition, on the other hand, chess is totally >>>tactics"... and so on and so for. > >This is a considerable exaggeration. Calculation is less important in Go than in >chess, but it's still important. > >>>Ok, I am a Chess lover and I like to play Go, as well, but I cannot see how >>>programming Go programs can be more dificult than Chess. >>> > >There are two main problems (1) As someone has already mentioned, it's hard to >write even half-decent evaluation function for Go. At least with chess you can >start with material and piece-square tables, and go on from there. (2) The >branching factor for Go on the full-size (19x19) board is large: over 300 in the >opening and still 100 or so in typical endgame. You can get round (2) by playing >9x9 Go, but that still leaves (1). > >>>What is your opinion about that ? > > >>>TIA, Ricardo >> >>When I think about this question, >>how long has someone or some people's really trying to code a really good Go >>program compared to the amount of time people have been playing and studying it >>over in asia and around the world. >> >>Now consider how long they have been programming for it, and look at the >>programs, already master level, > >Go programs aren't at master level: they're not even at my level (amateur 2-dan >- at a very rough guess that might correspond to about an 1800 chess rating). Hard to compare but let try to do it. There are childs under 10 in the world who are better than 1800 and the best player in the european championship has more than 2000. see http://www.euroyouth2004.com/rounds/10b_RatingReport.HTM Are there childs under 10 who are amateur 2-dan at go or better? > >>give it another 10 years, and you'll see the same affect the programs are having >>in chess >> >>masters running scared :) > >If programmers make good progress I might be running scared in 10 years time, >but I think the professional players will be safe for a while. I see no way to give good estimate about improvement in 10 years. Everything that humans can do good programmers can do better so I am sure that no human is going to have chances in 10 years and not only masters if the right programmer decide to work on the problem. I have no idea if the right programmer will decide to work on the problem in the next 10 years so I can say nothing about what is going to happen in 10 years. Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.