Author: Graham Laight
Date: 16:24:22 01/10/99
Go up one level in this thread
On January 09, 1999 at 08:59:53, James T. Walker wrote: >On January 09, 1999 at 05:55:25, Graham Laight wrote: > >>As I was sitting eating my breakfast just now, it occured to me that there are >>basically 3 items that, between them, will influence how close an evaluation of >>a chess position is to how good that position really is: >> >>1. The number of pieces of knowledge the evaluation function can call upon >> >>2. The quality of those pieces of knowledge >> >>3. The accuracy of selecting the right pieces of knowledge (and their >>appropriate weightings) for the position at hand >> >> >>Does anybody have any thoughts about this? > >Yes, but I'm not an expert so take it with a grain of salt. :-) The knowledge >you mention seems correct but would be almost useless without a deep search so >it seems to me the search depth and knowledge are intertwined and as everything >else in life, must be a comprimise. Unless you have deep blue hardware and can >afford both. >Jim Walker I'm not really an expert either - but I enjoy visits to this discussion group because I find it stimulates my thinking about AI in general. If you say that to play chess well you must do a big search, how do you account for the fact that GM humans can play the game so well? Because GMs can beat most computers without a deep search, they must be doing things differently to the computer. They are probably saying things like, "I know how to break this position". Then starting from that, they can work out what things can be done, what things need to be achieved first, and so on. You could say that this equates to doing a great deal of evaluation. They are doing a lot of computing as well. Although one can only think about one thing at a time consciously, one's brain is actually working on masses of things in parallel fashion at the unconscious level. Most of this work, I suspect, relates to finding patterns in the position which the player has seen previously. The conscious mind becomes alerted when something significant is found. This, I believe, explains why GMs can't explain their thinking - they simply don't know about most of it. Anyway, there is always a balance to be found between deep searching and heavy evaluation. Until a few years ago, the fast searchers always seemed to do best. Searching 5 ply deep lightly is more effective thatn searching 4 ply deep with heavy evaluation. However, as search depths get deeper (as computers get faster), there seems to be a diminishing return - and successful programmers are having to work at putting more knowledge into their evaluations (and search path selection). Therefore, with the passage of time, search techniques may become relatively less important, and "accurate" evaluation may become more important. In the past, I have written about the need to organise knowledge in the evaluation in a more systematic way. Graham
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.