Author: Mike S.
Date: 13:35:00 11/13/00
Go up one level in this thread
On November 13, 2000 at 15:39:50, Bob Durrett wrote: >(...) >Well, what I was thinking about when I wrote that was pretty much influenced by >Scheidl's comment "with more knowledge, possibly creating plans even, or more >selective." In my mind, at least, his words conjured up the idea that a more >human-like or interesting performance could be obtained. > >(...) then Scheidl's ideas >might make sense. I like to add, that I have only interpreted, or quoted more or less (in my own words), the ideas Jeff Lischer expressed in the first posting of this thread above. It seems to me that one conclusion of the discussion so far is, that a programmer will most likely want to apply the very same algorithm (which is the one he works on, improves constantly and is the one and only he trusts most) in each case, disregarding if the program has the move or ponders. But we have an example of an obviously very strong and successful program, which exists in two versions: Chess Tiger and Gambit Tiger. Maybe this can lead to an illustration of what Jeff may have meant probably: Assuming a programmer fears, a certain experimental, but promising version of his program might be too risky to use alone. In this case, he could have the more "conservative" algorithms do the search when the program has the move, while during the opponent's thinking time, the gambit version fills the hash tables with evaluations of positions and variations which occur from a different approach (which the standard algorithms then make use of, when it's the programs turn to move again). In other words, a kind of "Pinky and Brain" method. Regards, M.Scheidl
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.