Author: Vincent Diepeveen
Date: 17:01:28 12/22/04
Go up one level in this thread
On December 21, 2004 at 15:53:26, Dann Corbit wrote: >On December 21, 2004 at 08:49:39, Ernest Bonnem wrote: > >>Notice that Shredder 8 is quite unique , with a branching factor around 2 (on >>the average). >> >>Both Shredder 7.04 and Shredder Classic have a more "usual" branching factor, a >>little over 2.5 > >Branching factor is almost irrelevant unless the right moves are chosen. Please go stand in front of the mirror and repeat the sentence 10 times. If you still feel like that your above statement is clever then read on. >I can >trivially create a chess program with a branch factor of 2.0 or even 1.0. It >will perform terribly unless the right moves are chosen. I'm actually not observing a b.f. of 2.0 for Shredder8, but closer to 2.5 or something. Let's assume now that the position you tested you observed c * 2.0 ^ n. Please show a proof that it is impossible to search 1 ply deeper needing a b.f. 2.0 when just using nullmove (say R=3) and alfabeta and a very efficient huge hashtable and a very efficient qsearch, assuming a near to perfect move ordering. As an experiment just turn off all positional aspects and just search for material. You'll instantly hit 30 ply even in openings position as already 6 years ago proven. Later repeated also by Rudolf Huber using MTD. >Of course, the stupendous branching factor is one reason why such deep searches >are possible.
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.