Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: engines and openings: can engines guide human play?

Author: Dann Corbit

Date: 14:37:30 02/27/06

Go up one level in this thread


On February 27, 2006 at 17:09:11, Joseph Ciarrochi wrote:

>But Dan, How much can you really trust the human experts? I mean, I have a video
>by andrew martin on the scandanavian, and I have become convinced that one line
>is just not good for black. The black side gets its but kicked again and again,
>regardless of engine, etc (position below). That passed white pawn is tough, and
>blacks so called "mobile pawn mass" on k-side can come under attack. (I have
>decided to ditrust a line, unless the IM or GM plays it himself at the top
>levels!!!!)
>
>ps I don't only use the computers evaluation of an opening. I let the comps play
>through the openings and see what happens.
>
>
>
>[d] 2kr3r/ppq2p2/1n1b2p1/B1pPpp2/2P4P/1B6/PP2QPP1/2KR3R w - - 0 19

I think it is good to be skeptical about everything.
Look at the lines in the MCO or NCO or whatever resource and see if they make
sense to you.
If a computer program has found a refutation, see if it has already been
considered in the human analysis.  Perhaps there is a rebuttle to the rebuttle.

Obviously, both the computer and the human analysis can be incorrect.
I think that the human analysis in the standard references like ECO, MCO, NCO is
probably better (on average) than computer analysis unless you let the computers
think for a very long time.  And in gambit openings, most computer programs are
woeful.  Rybka seems to be a notable exception from the few positions that I
have tried.




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.