Author: Bruce Moreland
Date: 19:06:34 10/22/00
Go up one level in this thread
On October 22, 2000 at 18:04:35, Thorsten Czub wrote: >direct the search. to the goal of the game: >that is in chess: mate the king. > >there is no other goal. >the bean counters tell us other stuff. they tell us: >win material. >and if you are lucky, there will be a mate soon. There are no bean counters other than people who don't know much about chess. There are a great many cases where the objectively best move involves a sacrifice of material. The difficulty where this concerns chess programs is that it is hard to tune an evaluation that values material less highly. Early on, there were people who very carefully clamped their evals so positional terms could never exceed a pawn, which effectively prevented any form of sacrifice. The first program I encountered that had positional terms obviously in excess of one pawn was MChess. There may have been more before that, but MChess had a gambity style even back in the early 90's. Perhaps you will tell me that there were others, but I'm not a student of that. I think we are well beyond sub-pawn positional evaluations now. With Chris we were talking about positional evaluations that seem to have been essentially limitless. But what I think was interesting about that program is that it achieved positions where its positional terms could be used. It is hard to create something that has a natural "pre-sacrifice" style. bruce
This page took 0.01 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.