Author: Fernando Villegas
Date: 16:46:43 09/06/98
Hi all: The long thread about what is best to defeat human seems to me lacking some previous analysis -if I missed a post with it, sorry- about what is best against us, what should happen to us in a game against a computer to make a differencce between best and no-so-best program. If we are sincere, there are scarce differences in whatever kind of programs with with respect its final results against average chess players, I mean, "A" and expert level class players like we could find in this site.I could bet my salary of half a year that we are just beaten the same, let us say, 85 to 95% of the time, the rest being draws and here and there a miracously got win. So, in terms of results -wins, loses,draws- I would say all top programs are equally "best" against us. I tend to think that, in this issue, we should, instead, just define what kind of defeat -victories are scarce- should be considered as a showing of "best-against-humans", not looking for an unexistent or unsignificant statistic of more or less victories from the human side. I suppose that Enrique is right when he say that more pasive programs give more time to lose; you lose the same, but it is a protracted death with a long agony first. On the contrary, agressive, even speculative programs get more scalps because they put pressure and so they push the human side to mistakes that are duly punished. So, best againts human is just the program that push harder and so kill you before move 30 or so. That's the reason CSTAL is so strong against human,but at the same time not so strong against computers because to be speculative and to create pressure means nothing if the oponent if another comp. At the same time, nobody except CSTAL is so biased towards just one side, so in fact differences in terms not only of number of wins but also in terms of how fast they win I would bet tend to be very similar. But, this means then that at the same time all comps are equally strong against others comp? Obviously they are not or we should have a SSDF list with 30 programs sharing the 2 top positions and is not the case. The answer maybe is this: what a program need to beat an average player is something totally different to what he need to defeat another program. The standard level of performance on tactical ground reached by ALL top programs is enough to get an amount of victories that anhilate any effort to see differences, but in fight between them another levels, above just tactical, normal proficience is needed. In fact, this is the same as happens between masters and between masters and amateurs; all of them defeat us almost the same in number and speed, but between them another kinds of abilities becomes important and makes the difference. Fernando
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.