Author: Robert Hyatt
Date: 07:46:27 07/18/00
Go up one level in this thread
On July 17, 2000 at 23:13:03, Will Singleton wrote: >On July 17, 2000 at 22:29:11, Robert Hyatt wrote: > >>On July 17, 2000 at 20:26:44, Will Singleton wrote: >> >>>These are from an interesting game/30 on ICC. >>> >>>21...Qxa2 the best? >>>5rr1/p5pp/1pp2k2/q3p1NQ/4P3/1RbP4/P4PPP/5K1R b - - >>> >>>22.Nxh7+ (how long to avoid?) >>>5rr1/p5pp/1pp2k2/4p1NQ/4P3/1RbP4/q4PPP/5K1R w - - >>> >>>36...Qe6!? (anyone play this?) >>>8/3k1r2/1pp4P/p3p2K/3bP1Q1/5P2/q5P1/7R b - - >>> >>>after 38.Kg6 (can the black pawns be stopped? if not, what's the proper eval >>>score here?) >>>8/5r2/1pp1k1KP/p3p3/3bP3/5P2/6P1/7R b - - >>> >>> >> >> >>These are crossover positions. There is interaction between wanting to reach >>an endgame with a queen-side majority, and trying to mate your opponent with >>his king way up the h file and queens still on. If you want to play >>karpov-like, trading queens is the right idea. If you want to play kasparov- >>like, keeping queens on is the right idea... >> >>all a matter of 'style' here... >> > >Hmmm.. so, you're saying that these positions (the last two) are too deep to be >evaluated to a conclusive understanding by human or computer. Style implies >intuition. Would be interesting to show otherwise, with some confidence. \ I am not sure about "too deep". Just that if a program favors complexities, then it will keep queens on unless it sees a _sure_ win. Remember that an outside passed pawn _usually_ wins. But _not_ always... I'd probably rather stay in a position where I think I am stronger than my opponent (tactics in the case of the computer) rather than entering a position where my 'hunch' might be wrong. And in endings, humans can actually out-calculate computers by many plies. So if you screw up and enter a losing king and pawn ending, you are really probably going to lose.
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.