Author: Howard Exner
Date: 12:26:10 01/23/00
Go up one level in this thread
On January 23, 2000 at 10:42:23, Fernando Villegas wrote: >You are right. Computer tend to play as if the rival is always the perfect or >almost perfect machine they are, so play the very best instead, many times, to >play the one that will be more lethal to certain kind of player. The fun of >CSTAl is precisely hat it seems does not intend to play the best, but the most >tricky. That's reason it is a very strong againts human and not that strong >against computers. I believe this also, that strong vs another computer does not necessarily mean strong vs humans. Not that a program can be both, but the notion of predicting success in the human realm based on success in the computer arena may not transfer. Again, an adequate data pool of games would tell us more. >Maybe what a program could do to decide between to play safe >and "perfect" or to play risky and tricky could be just to evaluate the quality >of the oponent in terms of, say, the number of times he play the very best or >not. I mean, if after 20 moves or so the oponent has played always or almost >always the best openning move or middle game move, he can consider he is playing >a strong player that deserves high consideration; if not, then he can play >riskier in order to kill at once the poor bastard. This is just a rude example; >maybe the evaluation system of who-is-the-guy in the board could be refined a >lot more. Yes I follow. That would be a nice feature, where the computer keeps a running log of your moves and then alters its play accordingly.
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.