Author: Ed Schröder
Date: 22:58:12 09/29/99
Go up one level in this thread
>Posted by Joshua Lee on September 29, 1999 at 14:43:02: > >>Some good ideas take some time :) > >Thanks Ed what your doing,having GM's Play your software is just what every >other programmer should do.I would think that someone ought to be able to >make a better list of ratings than the SSDF with your results and if more programmers >follow suit then maybe it's a good idea that the SSDF people find a new >line of work. Human vs computer is more relevant than computer vs computer. >Thanks again very much Mr Shroder No doubt that human vs computer is more relevant than computer vs computer and certainly at 40/120. In the early days of chess when computers played on a 1700 elo level it was a lot easier to conclude which program was the best against humans. These days when a chess program loses from an IM you have done very badly, when it draws the IM has done very well and finally when the program wins nobody is going to be excited about it because nobody expected otherwise. Nowadays you have to play GM's only. I formulate the impression I have got from postings here and elsewhere: a loss is accepted as long as the number of wins is the same or higher. A draw against a GM is nothing to be excited about but the computer did very well. A win is what people hope for and more or less expect these days. Competing on this high level has the big disadvantage that the number of available data (games) will remain very low because GM's don't play for free and it will be very difficult to judge the state of art in the ongoing battle between Man versus Machine with the few games we have until now. Computer versus Computer is a world of its own. There is a massive number of data because almost every program supports a device for automatic play. Next, comp-comp is very popular and big fun to do. It's logical people create rating-lists from the data like the SSDF people for instance. There is nothing wrong with the SSDF list as long as you know how to interpret the list. These interpretations are discussed here from time to time and IMO they keep the balance regarding the value of the SSDF list in a good way. Ed Schroder
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.