Author: Otello Gnaramori
Date: 13:38:17 07/07/01
Go up one level in this thread
On July 07, 2001 at 10:39:25, David Blackman wrote: >On July 06, 2001 at 20:10:03, Joshua Lee wrote: > >>Computers are Tactically as Strong as GMs? > >You have to realise that for a GM it is very hard to seperate tactics from other >aspects such as positional play, endgame play, and swindles. In a difficult >tactical position, a GM, especially if in time trouble, will not try to >calculate every line to evaluate what is the best move. Instead they rely on >their positional understanding of the kind of position, to decide what moves are >likely to work, then calculate just a few sample lines to see how they turn out >and confirm their positional judgement, then in the end they make an educated >guess based on all this. > >So it is often the case that a GM can make a correct move without fully >understanding all the tactical sequences of moves from the position. (They can >also totally stuff up where a computer could instantly make the right move. Look >at some of the GM games that are hidden inside the tournament bulletins, rather >than the ones that make lead articles in the magazines.) > >So if you get a sharp position that has positional motifs that the GM is >reasonably familiar with, the GM will sometimes find the right move where a >computer will fail to in any reasonable time. > >On the other hand, in a mixed up position where all that matters is shear >calculating power because there are no obvious positional ideas, the computer >will find the right move (or a good enough move) more often than a GM. That's right. This comes from a process of "pattern recognition" referring past situations to present positions on the board. But sometimes you have to calculate,calculate and calculate to avoid bad surprises. You can't rely only on "heuristics" feeling. > >My feeling is that most of the top programs today have a slight edge in tactics >against most GMs. There are exceptional positions where the GMs get it right and >the computers get it wrong. There are also just a few GMs who are probably >similar strength (or maybe even better) than the top programs tactically. I disagree since computers are tactically with no reasonable doubt leaders and moreover I think that CHESS is for 90% tactics game. (This >doesn't count Deep Blue, which is (a) retired, (b) of unknown strength really, >and (c) quite likely a real tactical monster.) > >And of course, in many games there are important positions which are not about >tactics at all.
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.