Author: Rolf Tueschen
Date: 02:52:27 07/20/04
Go up one level in this thread
On July 19, 2004 at 22:32:10, Fernando Villegas wrote: >Seems to me that this is less a matter of dramatic or altogether different >concepts translated to code than a matter of delicate ponderation of common >variables. Probably engines share 99% of the tools already known that became >commodities since long ago, BUT a + or - of this or that in the scale of >decimal points to use and mix all of them makes huge diference after all. >Probably I can generalize this to any kind of high level perfomance, where >outputs depends at last not of big shunks, but of minute weights. >KInd of variation of the butterfly effect when and only when you are beyond >certain point of proficiency. >What do yo think? >Fernando Let's assume that all things related to engines is "similar", then for instance time management, book favorites and above all chess itself are enough variable to lead to results significantly different to 50%. Chess is a game of the importance of concrete details and exceptions of the general rules. In combination with the general blindness of machines that leads to enough variety of results. This must not be forgotten in the interpretation of tournaments and tests. Stats isn't crazy. Stats demands many many thousands of games for the exclusion of possible influences by chance. But no version lasts so long, so there you have the secret for our daily ballyhoo. It's a delusion, most of the time.
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.