Author: Uri Blass
Date: 09:21:23 10/25/01
Go up one level in this thread
On October 25, 2001 at 11:56:57, Robert Hyatt wrote: >On October 25, 2001 at 11:41:28, Uri Blass wrote: > >>How much time does your program need to avoid Nxg3? >> >>Here is Junior7's analysis on p200 after some learning >>It has no problem to avoid Ng3 after learning 18...Nxg3 19.hxg3 Bxg3 20.f4 Qh4 >>21.Rf3 Qh2+ 22.Kf1. > > >I'm not sure how this helps at all. What good is it to have to _first_ >search the position beyond where you are and "learn" before you can >solve it? If we use that testing methodology, Crafty should be able to >solve _all_ tactical positions with its "position learning". But that won't >help it one bit in similar positions in real games... The first thing that we need to test is if a test position is correct I was not sure that the test position is correct because I suspected that maybe shredder lost because of another mistake. The fact that Junior did not like the sacrifice after learning supports the conjecture that the sacrifice was wrong. Junior7 without learning wants to play the wrong move 18...Nxg3 and I guess that it may need hours to avoid this mistake. Uri
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.