Author: stuart taylor
Date: 17:00:42 11/05/00
Go up one level in this thread
On November 05, 2000 at 13:06:16, stuart taylor wrote: >On November 05, 2000 at 01:10:41, r.c. richards wrote: > >>On November 04, 2000 at 21:56:28, stuart taylor wrote: >>>I wish a few people would have read my post from about a week ago, or more, >>>about great idea for chess championship/tournament. >>> I think it would test many things in a most economical and accurate way, and >>>give room for great insights into the programs. >>> I had very little response. Was everyone on vacation? And I see it WAS a >>>subject that interests people. >>>I could summarize it again. >> >>Summarize it or repost it. > >1.Firstly, it should be all play all, atleast, say, 8 times. Once a KP opening, >once a QP opening, once natural opening without books, and once fischerandom. >These 4 types of opening should be done for each program as black and as white >=8 games, all play all. >2. Now, even better, repeat it all at rapid time control, and even a third time >at blitz time control. (Origional was tournament time control). >3.Even better still wouuld be to include all programs which are of interest, >esp. all the new ones. And also, in many cases, their previous versions, and if >interesting, the version before that too.(That will also indicate if there were >serious improvements). >S.Taylor This is a more directed testing, and there are many aspects which could be studied and understood with many less random test games. This should be much more revealing, and much quicker, for many reasons. I don't mind if it is improved upon, but these are ideas from which much more can be learned. S.Taylor
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.