Main
Date: 15 Nov 2007 09:01:58
From:
Subject: Why Published Theory Differs From Computer Analysis!?
Greetings, Friends

In tune with the times I often use computer programs
to study textual material. While I understand how older
prints occasionally fail silicon surveying, I wonder about
the shortcomings shown in more recent ones? After all,
don't today's publishers use computerized editing? Or is
it just that the feedback I get is based on suspect
parameters?

I'm also pondering the kaleidoscope of output arising
when different programs explain the very same text?
Since chess is so rich, maybe each engine is fantasizing
variously?

Thank you all for any insights. Have a good day and a
better tomorrow.


thechesscoachexperienceDOTwordpressDOTcom




 
Date: 20 Nov 2007 00:26:40
From: help bot
Subject: Re: Why Published Theory Differs From Computer Analysis!?
On Nov 15, 12:01 pm, [email protected] wrote:

> Greetings, Friends
>
> In tune with the times I often use computer programs
> to study textual material. While I understand how older
> prints occasionally fail silicon surveying, I wonder about
> the shortcomings shown in more recent ones? After all,
> don't today's publishers use computerized editing? Or is
> it just that the feedback I get is based on suspect
> parameters?
>
> I'm also pondering the kaleidoscope of output arising
> when different programs explain the very same text?
> Since chess is so rich, maybe each engine is fantasizing
> variously?
>
> Thank you all for any insights. Have a good day and a
> better tomorrow.


I often use my computer to play over games, and
because there is no openings book installed, the
results can sometimes be strange and surprising.

For instance, in reviewing just a few of the games
published recently in Chess Life, I have found many
cases where the writer just throws out bad moves
(given by my computer in a shallow search) or
else, clearly *not* using a computer, misses big
mistakes which refute his/her entire thesis. An
example of the latter was a game presented as
strategical excellence, but where in fact the
winner had thrown away most of his advantage
near the end! Fortunately for him, his opponent
erred and his blunder slipped by.

One caveat is that in a shallow search, even the
best computers are ignorant in the openings (if as
with me, you have no by-rote moves to reference).
For example, several like to block in their own
pieces, leading to funky Knight forays later on to
relieve the self-inflicted cramp by hopping about
just to get out from in front of their own obstructed
pawn.

As for text auto-produced by chess programs,
I don't think they have reached a level where it
can be trusted -- although in many cases it may
be spot-on. I remember back when I had the
CD version of Fritz, and it talked all day long --
but often as not, the talk was mostly nonsense.

The main use for computers is in analyzing
very complex positions, or looking at tactics.
While most humans will go astray here, a
computer may find a forcing line or defensive
resource overlooked by even the strongest
players in the world. One game I went over
recently had been analyzed by several world
champs, and they all went wrong somewhere
in their analysis (Em. Lasker least of all
though he was still bested by my computer).

In that game, Jose Capablanca mucked up
the tactics in a won position and all the
others tried to claim they could do better, but
in each case (except GM Lasker) the analysis
was flawed, and it didn't take long for my
computer to find big improvements.


-- help bot






  
Date: 20 Nov 2007 10:26:53
From: Guy Macon
Subject: Re: Why Published Theory Differs From Computer Analysis!?



help bot wrote:

>As for text auto-produced by chess programs,
>I don't think they have reached a level where it
>can be trusted -- although in many cases it may
>be spot-on. I remember back when I had the
>CD version of Fritz, and it talked all day long --
>but often as not, the talk was mostly nonsense.

We have Usenet newsgroups for that.

--
Guy Macon
<http://www.guymacon.com/ >