Author: Ratko V Tomic
Date: 15:27:31 09/29/99
Does anyone have any concrete info about algorithms used by Hiarcs? Having played with it for couple years (since version 6, through the latest 7.32), it is my impression that it understands LONG TERM PIECE MOBILITY (control), going well beyond the simple checks of that type (such as good-bad bishop, knight outposts, pawn blocades, etc) which other programs probably use. Hiarcs seems to aim to maximize its long term mobility in a concrete way for a given position (and of course, to minimize the opponent's, which you can notice when you find yourself without any good move, with board still full, and wonder how did it do it without me noticing what he was up to), which suggests it does some kind of preprocessing where it computes approximately the effects of each root move on the overall long term mobility. From other programs I don't get any such impression. The only published algorithms (or sketches for such) that look at this kind of evaluations using multimove single piece paths and accountng (approximately) for possible blockades, I know of are from the Botvinnik's book "Computers in Chess" (Springer-Verlag 1984, ISBN 0-387-90869-2). Does anyone know whether Hiarcs have implemented any of such algorithms (perhaps even the Botvinnik version), or perhaps other methods which have a side-effect of improving long term mobility? Could a large collection of rules-of-thumb (from various chess tutorials) used in root choice preprocessing produce the same side-effects?
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.