Author: Bruce Moreland
Date: 16:25:21 06/02/98
Go up one level in this thread
On June 02, 1998 at 16:48:17, Komputer Korner wrote: >It is not a requirement of alpha beta that every move be analyzed. Alpha >beta is actually a rigorous mathematically sound rule that certain moves >do not have to be analyzed because mathematically they have been shown >to be inferior without having to search their move trees. You are >confusing alpha beta and intuition. They ARE NOT exact opposites of each >other despite what you think. >Chess players are intuitive even Kasparov. This is necessary because of >poor human calculation speed. Chess computer programs can be either >selective search or full width. Either way they all use alpha beta which >is really only a mathematically pruned minimax. Even humans use alpha >beta. Your argument is with intuition, knowledge and full width >searching. What you call knowledge is stuff that hasn't been programmed >into chess computers yet. It is very difficult to teach a robot to pick >up a glass off of a table. You accomplish that task very easily. The >same thing goes for teaching chess knowledge to a program. Don't confuse >a mathematical principle like Alpha beta with chess knowledge. I think >it is time for Bob Hyatt's yearly alpha beta lesson because it is clear >you do not understand what alpha beta is. Alpha-beta is just an optimization of min-max, which is just an exhaustive "if I go here, he goes here" kind of strategy, to some fixed depth. Sure, there must be ways to know when to extend promising lines, so that some lines are analyzed more deeply, or prune bad-looking lines, so that some others are analyzed less deeply. This in cooperation with min-max, or perhaps in cooperation with something else, who knows. It doesn't take a lot to see the truth of the above. bruce
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.