Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Great Result for Fritz and puts to rest some questions.

Author: Russell Reagan

Date: 17:56:59 10/19/02

Go up one level in this thread


On October 19, 2002 at 16:48:01, Alastair Scott wrote:

>Indeed, but endgame tablebases are a different matter; the winning sequences are
>often so far beyond the capabilities of the human mind, and are quite often  so
>counter-intuitive (as some of John Nunn's researches have found, for example), I
>would have real qualms about using them in a human-computer match.
>
>Your argument is good but I feel that, with tablebases, the issues of fairness
>raised with respect to opening books are so enormously magnified they can't be
>ignored.
>
>Alastair

This is EXACTLY what I'm talking about. You are saying we should not use
tablebases because computers happen to be better at generating and using this
sort of thing, which isn't "fair".

Let's try an example not related to chess. Let's take something like running a
foot race between humans. Along comes the invention of the motor, and now humans
can go faster than they could previously via the automobile. Maybe at first the
motor is primitive and not very powerful and doesn't go very fast, and humans
can still compete with motor powered entity in man vs. machine races.
Eventually, the motor powered machines become too fast for the humans. Now, what
do we do if we want to continue to run fair races between humans and the
machines to see who is faster? Do we (a) continue to run the races normally,
cranking up the machines to go as fast as possible, or (b) limit the machines
ability to go faster than our fastest humans? Today we have computer vs.
computer chess matches, car vs. car races, etc. How excited would you be if you
heard that there was going to be a race between the fastest man in the world and
a top fuel dragster that was limited to going 25 mph? You have this beast of a
racing machine, and you've reduced it to a child's go cart. Who cares if the
human wins?

Personally, I don't see the point in handicapping the computers just because
they happen to do something better than us. After we hold human vs. computer
with no opening book or tablebase matches, what will be next when the computer
is still competitive? Do we start calling for a limit on search depth or number
of nodes that are allowed to be searched?

As I see it, you only want to limit the computer because it has a huge advantage
in one area. If we limit the computer, then hold a match, and the human wins,
that's about as exciting as a human beating the top fuel dragster because we
gave the dragster square wheels, or how about if I go compete in the special
olympics. Would you think a lot of me when I came back with the gold medal? I
think beating a handicapped computer would hold about the same level of meaning.

You are essentially handicapping the computer, plain and simple. There is no
difference between banning opening books and limiting the search depth to 2 ply,
other than the level of the handicap you are giving the computer. How low do you
plan to lower your standards before you feel satisfied that humans are still
superior?

I think the fact that we can create such monsters shows our superiority. Even
when a computer program can write a computer chess program better than we can,
we'll still be superior because we created the program that created the computer
chess program. Whoever created us must be one heck of a guy :)

Russell



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.