Author: Ratko V Tomic
Date: 11:02:08 04/28/01
Go up one level in this thread
> I meant to diminishing returns in rating and not to diminishing > returns in p. If the chess programs are to play for money, the measure of "returns" would be "p". For this type of _tangible_ returns, increasing depth (by adding processors) clearly wastes money beyond certain point. Since one can define arbitrary non-linear monotonic function f(p) to be, by convention, a measure (of its own kind) of "returns-f" the issue of diminishing returns of such f(p) is a matter of definitions i.e. there are infinitely many functions f(p) for which returns-f diminish and infinitely many for which returns-f increase with depth. ELO is merely one such conceivable convention of "strength" measure and thus the question whether the returns-ELO are diminishing is of the same kind as the question whether, say, the size of the Egyptian pyramids measured in US inches has some numeric relations to the age of universe in 10^6 years divided by the number of bytes in Windows 2000 Advanced Server kernel32.dll, or some such.
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.