Author: Frederic Louguet
Date: 23:32:03 02/09/98
Go up one level in this thread
On February 09, 1998 at 08:41:01, Robert Hyatt wrote: >As I've mentioned in the past, this test is worthless to produce a >rating, if you want accuracy. The "formula" was derived by letting >programs >that are on the SSDF list solve the positions, then "fitting" the >solution >times/results to make the results predict SSDF ratings. As a result, >any >program on the SSDF will produce a result that is very close to its >current >SSDF ranking. Any program *not* on the SSDF can/will produce somewhat >random result since it is really independent of the SSDF list. I agree with you when you say that you cannot produce a valid rating for programs with LCT II results. No test can do that either, and enthusiastic users sometimes forget that. However, I totally disagree when you say that the LCT II was created by fitting the scores with SSDF results. I created this test, and I can tell you that it was made "on feeling", and not to follow SSDF results. In the past, the LCT II often proved to be a valuable tool to get an idea of the general strength of a chess program, and quickly, but of course it is only a test, and its results should be taken with serious precautions. Oh, and I consider the SSDF results also as - one - indication of the strength of a chess program. It says absolutely nothing about the strength against humans, and I believe this is the reason why Virtual Chess and Chess System Tal are not in the list : they generally do better against humans than against computers. Are they weaker ? Of course not. "weak" and "strong" are very relative notions these days.
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.