Author: Ulrich Tuerke
Date: 09:56:38 03/17/98
Go up one level in this thread
> ... > >Also, the Fritz5 SSDF rating is based upon 200 or so games. Where are >your 200 hand-tested games, for a start? > IMO, one should not overestimate the relevance of the 200 SSDF games. I have convinced myself, that there are numerous doubles (and even n-fold repetitions) involved, where games had been reproduced exactly. I am a layman in statistics, but I cannot imagine that counting 5 repetitions of one and the same game in the same way as 5 different games is a reasonable thing to do. If I am right (I'm pretty sure that I am), then the statistical uncertainty for the Fritz5 games as published by SSDF is far too optimistic. I am convinced that time will show the true strength of Fritz5. With an increasing spectrum of opponents for Fritz5 in future, statistical evaluation will be on a sounder basis. The result will correct itself without any additional action required. The same has already happened for other programs before (e.g. HIARCS). Just wait and see. >...
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.