Author: Roberto Nerici
Date: 01:18:14 02/18/04
Go up one level in this thread
On February 17, 2004 at 18:28:24, Will Singleton wrote: >On February 17, 2004 at 06:02:10, Roberto Nerici wrote: > >>>why should a computer emulate the human approach to chess? not everything in >>>nature is perfect... are our aeroplanes equipped with big feathery flapping >>>wings? if we find a better solution, we should use it. chess computers can do >>>many things the best human players are quite uncapable of, and sacrificing this >>>kind of power seems wrong to me (and you would be sacrificing it if you wanted >>>to build in sophisticated pattern recognition in a program, slowing it down by >>>many powers of 10...). >> >>I agree with the analogy but remember that there is more than one possible goal >>being aimed for when writing a chess program. Sticking with your feathery >>analogy, it would be very interesting to build a robot with flappy wings and >>feathers and no doubt you would learn a lot. In that sense, it would be "worth" >>doing. What it wouldn't be, is a good way to build a machine for flying around >>doing something useful. >> >>Similarly with the chess engine. I'm skeptical that Symbolic will ever be a very >>strong engine, so if his only goal is to write a strong engine, I think he's >>going about it the wrong way. However, he's said that that isn't the case and he >>may well achieve his other goals; it is (to me at least) an interesting project, >>and if he can produce a reasonable program which can describe its plans that it >>be a good tool for many people. >> >>However, I do agree with your argument about the weakness of emulating human >>play so it is worth expanding on it. I think (as you seem to) it is a false >>assumption that to play chess well you must play it as humans do. I think this >>was obvious 20+ years ago; now it is blindingly obvious :-) Chess is an >>interesting human game because humans find it difficult. Humans are basically >>bad at chess; it isn't something human brains are particularly suited for. >>Making one type of hardware stuggle to behave like a very different type of >>hardware struggling to do a task it is unsuited for is, well, we can't expect it >>to be very good at it. You have to do it for reasons other than pure strength. >> >>Roberto/. >I diasgree with all of your points (with the possible exception of Symbolic). Oh well, can't please everyone all the time! >Humans bad at chess? With our best computers *still* unable to play better than >humans? You have to think about it for a minute. The best humans and the best computer systems are now very close but the comment "humans bad at chess" wasn't a about the comparison in strength between humans and computers. I tried to state that so sorry if I wasn't clear enough. (If in 10 years time computers were regularly beating the very best human chess players, would you then claim that humans are bad at chess when previously they weren't? ;-)) Instead, I was making the (not original) observation that chess is not the kind of task that human brains are optimised for. Human brains are evolved to cope with things like social interaction and scheming, learning and problem solving, lots of fuzzy things like that. >We are generally using, with >one or two improvements, the same algorithm that Shannon came up with 54 years >ago. It happens to work very well. But the likelihood of it being the best way >is pretty small. I don't claim that the way computers play chess now is the best way. What I was saying was that the way humans play is not the best way for computers to play, and also that the way computers play now is better for computers than the way humans play. (Obviously playing the way computers do is a non-starter for humans). Roberto/.
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.