Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: chess computer ratings

Author: Thomas Mayer

Date: 18:52:27 09/23/01

Go up one level in this thread


Hi José,

>> Is this true regardless of what XXX is? To rephrase, while going from 350
>> megahertz to 700 might yield a 60-point elo bonus, would going from 1
>> gigahertz to 2 gigahertz also yield 60 points? And what about a hypothetical
>> 2 gigahertz to 4 gigahertz? Isn't there a horizon effect whereby the elo
>> gain diminishes?

> That's my opinion too. There must be a flattening of the curve, according to
> my intuition. In the extreme case going from 6000 plies to 6001 plies doesn't
> matter, because you are seeing the whole game. The same applies to time so, if
> in the end, the gain is zero, there must be a slow flattening, or a step. I
> believe it's a slow flattening.

theoretically I fully agree - but I believe that we are far away from any
flattening... Think about chess and what the solution might be... What is the
initial positon ? Mate in 500, Mate in 600, more ? Or is it a forced draw in
1000 ??? We do not know, but that means that are 1000-2000 plys... and we are
not near to that... when I am thinking about the curve we are not very wide
after the beginning, so I think the 60 ELO rule (which is quite conservative)
for doubling hardware speed in computer-computer matches is still correct -
maybe the next 10 years also... with humans it is different - strategics have
nothing to do with depth... So I believe that the beasts will get more dangerous
in future, but even the best computers as long they have the current technics
will loose from time to time a game against a human...

Greets, Thomas



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.