On Feb 19, 11:12 am, Mike Murray wrote:

channel the game away from anti-computer strategies -- human analysis
is of too low quality to help the silicon monsters these days.

Whoa there, fella! You may speak for yourself, but
a few of us happen to know better. Computers with
no openings book still founder in the opening.

For example, many a book has been written in
which it is carefully explained how blocking the
c-pawn early with N-c3 or N-c6 constricts one's
own pieces; yet even the very best programs
will do exactly this, if you disable their openings
books. Now, while weak players may argue
that perhaps these monstrosities "know better"
than the authors of those books, real chess
players understand that it is merely an
indication that they have been programmed
poorly in this area of the game.

Think of the famous game in which DeepBlue,
or some other killer program, was stomping all
over GM Karpov, UNTIL it had to choose
between grabbing lots of pawns, or keeping the
opponent from getting far-advanced connected
passers which could not be stopped; the
clueless, materialist program grabbed the pawns
as a matter of course, judging it obtained a hefty
material advantage, only to lose the game like a
carrot.

Don't deceive yourself into thinking that these
new programs are near to perfection; they are
better than us, but that is no reason to give
them very much credit at all. It is still apparent
that they suffer from the dreaded horizon effect,
that they will play for spite checks in order to
push the fact that they are making zero head-
way beyond their own sight-lines, much like an
ostrich which buries its head in the sand.

I believe that human analysis can still help
these programs, but it needs to be of the
highest quality-- not the type of stuff typically
found published in Chess Lies magazine,
which I find is easily refuted by these very
programs, sans books and sans table-bases!


-- help bot