Author: Tord Romstad
Date: 12:23:33 04/13/04
Go up one level in this thread
On April 13, 2004 at 14:20:08, Anthony Cozzie wrote: >My point was that it is a lot harder to get it working, because every term in >the eval must be tuned to every other term. If you increase mobility, you may >have to tweak the passed pawn bonus, etc. Hence there is a rough "square" in >total tuning decisions vs number of eval terms. Which is why often times the >simple but well tuned program does as well as the more complex evaluator. I think this is, to a great extent, a matter of skills, ambitions, and preference. There is no "right way" which works for everybody. If I just tried to write a fast and simple bitboard engine with a PVS search, recursive null move pruning and a simple capture-only qsearch, I would just end up with a boring and vastly inferior version of Crafty. So many people have followed the same approach in the past, and most of them are much better than me. My only hope of ever being able to compete is to try to do something completely different. Tord
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.