On Sep 3, 9:55 am, Sanny wrote:

In order to get a feel for the program's strength, you need
to look beyond the raw score and play through the games;
in many cases, it is miniaturized (or plays on in what ought
to have been a resigned miniature). For the record, I do not
get miniaturized like that except once in a blue moon. If I
wanted to play a dull, 100 move game I could be very careful
and win every game on technique, but the program is so
slow I would die of old age before finishing more than just a
few games. I do know enough to run it out of book in some
main line Caro-Kann or Sicilians, but I wanted to test it sans
book, and so I played some passive moves to accomplish
that. When I did play carefully, against the master level, I
won fairly easily, just like before.

Oh, you were testing, I saw you just gave your Rook for Bishop But
later fork his Queen. But Computer was able to play a Queen move with
Mate in 1.


You are of course referring to the game where I had
White and played a sort of perverse Larsen Attack, but
lost my QB. In that game I deliberately chose to muck
up my pawns in order to launch a Reti-like attack on g7,
expecting to later offer up my Queen for a routine
windmill (White Bishop on b2: Rxg7!, Black captures
the White Queen, White plays Rg6 mate or sweeps
the entire seventh rank with discovered checks).
I lost, not because I made so many small errors, but
because after ...Qxf3 and Rg8, I thought I could play
Rg3 -- missing that it would be pinned at that point.
From then on, there was no way to put up a fight. I
underestimated the program; took it very lightly.

Just look at the move times and note what is going
on: while I am in many games (not that one, I suspect)
playing like ten minute blitz chess, the program is
taking a long time in the middle game, so I switch to
looking at something else. Where it may appear I
have taken ten+ minutes to move, that indicates a
disconnect, of which there are many. Were it not for
the loud signal when the program finally moves, I
simply could not play there at all.

I find it amusing that Sanny keeps bragging about
these wins while ignoring all the games where I have
miniaturized the program. A miniature is a game
ending in fewer than say, twenty or twenty-five moves.
Obviously, the program can play on until mate-in-one,
but losing a piece or more in the first twenty moves
is not the hallmark of a great chess program.

At ChessWorld.net I am playing several newcomers
like myself, who started off with a rating of 1400.
Several have dropped to 1000 or even lower already,
due to that site's massive K-factor, or weighting of
each game versus one's current rating. These guys
lose in much the same way as Sanny's program
does: very quickly once they are out of their book
knowledge. One fellow played a Pirc Defense, and
before he could even get his "counterattack" off the
ground, was simply going to be checkmated. Even
the game where I was *busted* right in the opening
(a Latvian Gambit), I now have the upper hand.

So, what is the difference? Sanny's program will play
moves like B-d2 or B-d7, essentially doing nothing,
while human players will often self-destruct, and this
makes his program tougher to beat. The program
appears to see a couple-three moves ahead tactically,
except when itself being mated.


-- help bot