Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Famous games in the Computer Age ;)

Author: Paul Richards

Date: 17:05:51 06/11/99

Go up one level in this thread


On June 11, 1999 at 16:16:50, Dann Corbit wrote:

>Here is an idea for an "Anti-Human" setting for computers.  Search through a
>database and look for positions/opportunities like this.  They seem to be
>almost irresistable to humans (I would have taken the piece too, *no doubt
>about it*). You would not want to play them against computers, but they might
>be a very good idea against humans.

That could certainly be done with your CAP database.  In general a program
might exhibit more "human" play if it deliberately tried to set up positions
like this.  Instead of immediately going for a slight advantage, it could
try to set up a deep but completely winning combination.  It would lose
to a computer on such a setting, but it would make for a very interesting
opponent just because we expect computers to grab every minute advantage.
Computers don't normally make moves with lesser immediacy in order to
set up bigger attacks.  It would be quite a shock to the human opponent,
who would assume that the position is stable, then all of a sudden it
explodes in one move. :)  Eventually computers will be able to slowly
crush everyone, so their ability to play in an interesting manner and
produce "immortal" games in a flashier human style will begin to distinguish
different programs.

Speaking of immortal games, 18. Bd6 again looks like is really isn't
best.  The game from the position given went 18. Bd6 Bxg1 19. e5 Qxa1+
20. Ke2 Na6.  Working backwards, 20...Na6? is a mistake, falling into
a forced mate in 3. 20...Ba6 is rather amazing since it refutes the
Nxg7 mating attack by virtue of opening both c8 and b7 for the king to
run to.  Bb7 isn't enough to do this, but that's not easy to calculate
unless of course you're a machine.  This is a great example of how
machine calculation spoils a lot of good fun. Playing Ba6 earlier at
move 19 is also good.

However 18...Qxa1+ appears to totally refute everything: 18...Qxa1+
19. Ke2 Qb2! and black is way ahead.  Unless someone can find a
winning continuation for white, 18. Nc7+ really does look better.
It just doesn't lead to an immortal game, but then an immortal game
inevitably requires a more cooperative opponent. ;)



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.