Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rebel vs Ferret on ICS

Author: Don Dailey

Date: 15:37:37 12/09/97

Go up one level in this thread


On December 09, 1997 at 14:55:27, Bruce Moreland wrote:

>You've got to take computer vs computer games on ICC with a grain of
>salt, since the manual operators can do things to try to influence the
>games.  Sometimes the things they are improvements, but other times they
>make the program play an unsound opening it wouldn't ordinarily play, or
>play an overly aggressive move it would otherwise avoid, or they tweak
>the program's tweakable parameters so it is playing in some odd fashion.
>
>And sometimes, when a program is in time pressure, they will simply turn
>the program off and start playing the game themselves.
>
>In my opinion this is not so much unethical as it is unscientific.  It's
>not really a computer program that you are playing, it is an account,
>and part of the account is the operator, who has his own distinct
>attitudes about what his role is, and there is no way to hold him
>accountable to anything other than his own sense of what is and isn't
>fair.
>
>The operator in this game told me he forced move 54 (I believe it was
>this move, from looking at the PGN with my eyeball, all of my computers
>are busy now), because "I wanted to see the mate", or words to this
>effect.  This disappointed me.
>
>Sometimes other things happen.  You get people who don't update their
>notes, so they are saying they are using one thing while using another,
>or their hardware is wrong, or they are an inexperienced operator and
>waste a lot of time, etc.  Other operators will change programs in the
>middle of a game, and may not remember what program they used three
>games ago, so it is very hard to know exactly what beat you.
>
>Ferret is capable of beating any other micro program, and who knows what
>else.  The reverse of course is also true.
>
>That is the second KNN vs KP my program has had in the last month or so.
>
>bruce

I never seriously considered playing my program Cilkchess on the
internet
for these very reasons.  It's probably a good thing if all you really
want is to get in some games in and find weaknesses, but I get very
uncomfortable with ambiguious results.

I feel the same way about "reports" I hear from people concerning
results
they get with program x vs program y.   Few pople realize that even a
small personal bias can go a long way toward distorting a result.  There
are too many ways this can happen.  Here are a few of them:

. Tending to end game as soon as your favorite machine has a good
position

. Tending to hold out forever  when your favorite machine is losing.

. Counting a drawn (or even lost) game as a win because it was a won
  position early and the rest of the game was "fluky."

. Finding trivial reasons to restart a badly played match or game. (I'm
  not sure I set the thing right) but only when favorite starts badly.

. Not counting the last few games of a match, just the first few that
  went well.

. Forgetting losses, remembering wins (people always do this where their
  own games are concerned too)

. Making ridiculous adjustments (in their favor) for hardware
mismatches.


Humans are extremely devious creatures!  And often extremely irrational.


-- Don









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.