Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: True Ratings

Author: Ricardo Gibert

Date: 12:02:58 01/05/04

Go up one level in this thread


On January 05, 2004 at 08:51:39, Ted Summers wrote:

>If a program, any program (like X3D Fritz) plays some positions like a 2700+
>player and other positions like a 1600 or 1700 player what would you say it's
>true rating should be? I am asking because of the 3rd game of the Kasparov vs.
>X3D Fritz match. In human games I don't think that their is such a huge gap in
>how a human would play a open position verus how they would play a closed
>position in terms of chess strength for example. However in computer chess this
>seems to be very normal. So I don't see how we can say that a program is 2700+
>when it plays some positions and as such weak player in other positions. Granted
>in tactics they are a GrandMaster, however in strategy they are maybe a grade A
>(1900) player. These are my thoughts, interested in hearing yours.
>
>Thanks
>Ted Summers

Consider the following classification of positions:
    Type 1: Easy for Human, Easy for Computer
    Type 2: Easy for Human, Hard for Computer
    Type 3: Hard for Human, Easy for Computer
    Type 4: Hard for Human, Hard for Computer

What you have done is look at Type 2 positions and decided that:
    computers play Type 2 positions like a "1600 or 1700" human

However, from the computer perspective:
    humans play Type 3 positions like a "1600 or 1700" computer

The situation is really symmetrical  :)

As you can see, we are as justified in ranking computers at the 2700+ level as
we are in ranking humans at the 2700+ level, because The criticism you apply to
comps can be applied to humans as well.



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.