Author: José Carlos
Date: 11:03:00 05/22/01
Go up one level in this thread
On May 22, 2001 at 13:28:26, Dann Corbit wrote:
>On May 22, 2001 at 03:58:16, Ratko V Tomic wrote:
>
>>> Actually, everything is a coin toss {probability function},
>>> including such mudane things as:
>>> "Will I hit the 'post' button?"
>>
>>While it is true that there is some non-zero uncertainty in
>>most things (but not all, e.g. what is 2+2), the information
>>you gain about the coin or dice bias from the outcome of single
>>toss is much smaller than what you gain from a single game.
>>
>>That is, even if one accepts that there is uncertianty in the
>>chess move selection, one can extract great number of tosses
>>out of a single game, not just a single toss (as when
>>picking the final result, and ignoring everything else).
>>
>>As Uri mentioned, by looking the program's analysis during the
>>game, you can find out and often eliminate superficial effects
>>on the outcome of a single/few games an opening trap/bad opening,
>>you can see how far they look ahead etc.
>
>True. And if a program or person is dominantly superior (e.g. 1000 ELO) to the
>other, observing a few games will probably give the same determination as just
>counting up the scores. But consider the recent match between Junior and Fritz
>to determine the Kraminik opponent. If you saw the first 5 games, you would say
>that Junior was (without question) the better, because Fritz was simply
>outplayed. If you observed the last five games, you would reach the opposite
>conclusion. Therefore, if the programs are close, there is little or nothing to
>be gained by examination of the individual moves of a game. And if there is a
>wild difference between the ratings, it will show up quickly as a lopsided
>result set. Therefore, where is the value in observing the moves to decide who
>is stronger?
"Stronger" is a vague concept for most chess players. A mathematician, like I
think you are, will define it clearly, I guess: the better results, the
stronger. From that point of view, no doubt. Only results give the answer of
who's better, and by what margin.
But if you ask a GM who is the best player of the history, he will not even
consider mathematic results. Someone would say "Tahl, for his attacking talent"
or "Botvinnik for his scientific method" or "Kasparov for his enormous energy".
Those are not measurable concepts.
I haven't looked at Junior/Fritz games, but I'm sure one is better in
something and the other in something else, so depending on the shape of the
games, anyone can win 5 games in a row.
Difficult to say which one is stronger in one of those "not measurable ways"
only by looking at results.
José C.
[snip]
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.