Author: Enrique Irazoqui
Date: 16:35:02 12/13/97
Go up one level in this thread
On December 13, 1997 at 19:02:33, Don Dailey wrote: > >Thanks for the feedback, I'll try the test on Cilkchess. > >I am hoping to build a pretty large set, hopefully at least >100 positions but I don't know how it will work out. With your >permission I may append your set to the 1000 >positions I start with, presumably most of them will survive >the "consent" test and be in the final set. > >I chose 30 minutes (which of course is subject to change) to >allow the set to grow a little with time. I'm hoping that a >year or two down the road, it will still take modern programs >a few minutes to solve the harder problem. > >As far as scoring is concerned, I would like time to solution >to be part of the scoring. We can build a formula that makes >sense with regard to what is known about speed vs rating improvement >in chess programs. It does make scoring a little more >complicated but should be more precise. I'll try to keep it >as simple as possible. Of course I'm still listening for ideas >from people out there. > >It looks like the basic idea was originally your idea. There seems >to be almost no difference in the two, except your positions were >hand picked first. Anyway, the basic concept may be pretty sound. Particularly if the test is not published. Test positions tend to end up cooked and then the set is not valid anymore. :( As for the 1,000 positions, I think that a test that includes many positions may be more accurate, but also too time consuming. A criteria I followed when building my test was to find positions as unambiguous as possible, with one good answer and only one. Still, programs hesitate, at times going back and forth. Enrique >-- Don
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.