new computer chess engines

this year (2005) a new engine came into the market,
started as an amateur engine, but being able to
beat some of the top commercial programs,
and quickly became commercial as well.

which it will do simply after some testing and, then adding
endgame database probing, some more tuning, and finally
a lot of brouhaha on discussion list such as www.icdchess.com

this latest hype was the socalled Rybka engine,
from obviously another gifted programmer (from MIT,
so no surprise that he's a top IT profession )
and.. het is an IM chess player !

Now programming a chess program at top level of
course is far from easy, but there is a lot of competition,
and basically the top programs are not so
different in strenght, compared to each other.

But in general the modern chess programs are still gaining
an amazing improvement every year, not only because
of some very clever programming innovations
w.r.t. to search techniques, but also adding chess knowledge
without slowing down the program (the latter must be some
modern type of secret invented by some top programmers)
And also simply because of law's Moore, namely the increase
in hardware speed, the programs get stronger every year.

As result of the inherent complexities in the chess middle
game, playing against a chess computer is becoming more
and more difficult for human chess players, and that includes even
the top grand masters such as Michael Adams who was crushed
this year by the hardware monster Hydra (software programmed
mainly by C.Donninger, programmer of former PC program Nimzo).

Where will this lead in a few decades ?
Nobody knows, there still are obvious weaknesses in many
chess programs, in middle game planning, and in endgame
strategy; they currently for example usually cannot foresee
socalled fortresses, even when a strong human chess
player can use such 'tricks' in trying to keep
a draw against these monsters..

maybe we should give up chess and go to Go ?
:) Well i don't know, but anyway,
Best wishes for the end of year holidays
and a happy new year 2006

New blitz rating record

well, my computer chess repertoire with e4 still seems to work,
achieved a blitz rating record with my Kec(C) account on Fics,
above 2600, and entered now also the ranks of the highest ten :)

a new update of my book now is available !

with a more appropriate title:
"How to improve your opening play" !

see selected pages at: www.superchess.com

and the winner is ..: kec(C)

Yes folks, we now have achieved nr one position in standard chess on FICS (the Free American Chess Server)
with my own repertoire (with Arena interface & Shredder engine)
So.. the repertoire as described in my book now has been verified
in ''real life'' as well.

while my blog only seems to receive attention from some ladies(?)
trying to invite me for some ''free'' dating sites, i'm
still planning to finalize the book ''a better repertoire'',
and then probably publish it via POD (printing on demand)
methods, and/or maybe via a publisher.

anyone interested in cooperating in this effort,
either as ghost writer, chess analyst, or publisher
is welcome to reply; yes and dating sites as well, but i will
not bother to use a credit card for such things, you know..
:-)

Jef

new chess book

well people, here's my new chess book;
sofar no hard criticism, maybe also
coz i've done quite some research on
the subject, and even have been a bit
self-critical, calling the concept of a better
chess opening repertoire controversial:

here are some selected pages:

http://home.planet.nl/~k.e.c/chessbk_sel_chapters.pdf

in the meantime we are going to look for a publisher,
and extend the book from about 60 to more than 150 pages.

the only opening book beginners need to have !

good luck with *real* chess i would say..
:)
Jef

cool

why computer chess is cool;
well that's simple, it's always different;
you can compile Craftty with DevC++,
update your repertoire with Bookbuilder,
and basically, just have fun.

In the meantime, in my repertoire my
ICS accounts have gone back to the Tarrasch
in the French, personally i'm still busy
trying to get a (C) account on FICS
(for my Djenghis account now all sanctions
already have been removed) and
still busy trying 'solving' chess, i.e.
trying to find a better opening repertoire..

Nope, not will succeed in my lifetime
but it's not impossible Nah, maybe
coz of the 50 move Fide endgame rule.
Let's say a knight is blocking a pawn.
And there are some more pieces.
Mate in 135. Bad luck for black.
Ah well, lets talk about that later
:)

opening repertoire, a little update

well after some more numbercrunching, my opening
repertoire/strategy has changed a bit for black;
not for white, i still believe in 1.e4 !

For myself, being a lousy Otb player, i just play
..e5 with black against e4. And against d4, well,
the Bogo-Indian might be a bit easier than
the Queens Indian. Advantage of the latter is
keeping the bishop pair in the endgame, but
muddling yourself through the middlegame
is a bit difficult, as usually a complicated positional
struggle, involving dominance in the center,
pawn advances, etc. are quite common.

For computers: well, against e4 i don't hesitate
to recommend the Sicilian, but because of the
Rossolimo (B5 against Nc6) i now recommend the
Najdorf; yep, very complicated, but a comp can do it.
And most GM's as well. And having looked at
some recent Linares games (2005) as well as
some more analysis, instead of the agressive
moves Bc4 and f4 against the Najdorf, the more
quiet Be2 or Be3 now seem a bit better imho.

And against d4? Well i need to improve my
queens indian book; comps are not so good in positional
play, so with Arena (www.playwitharena.com) i also
choose to defend with the Bogo (named after Boguljyobov)
sometimes; even although it sometimes is costing a bishop.

Yet after a while probably the Queens Indian is more solid.
And Kings Indian ?
Nope, don't like it.
While i can't refute it, as it leads to some unbalanced endgames,
it doesn't seem to be rock solid to me, in a fundamental sense.
Same with the Gruenfeld.

Until next time,
Jef

engines and correspondence chess

ever wondered what the best program, aka 'engine' is
for correspondence chess ? Well with Chessbase engines
like Fritz 8, Shredder, Junior etc. it is possible to use
a correspondence mode, and let your comp krunch
the whole knight about a variant.

But what is it doing then ? Well, after a certain number
of moves, it simply continues, and makes a tree
for a number of subvariants (which you can choose
in the options). Now ideally such a tree would
be backsolved according to an alfa-beta algorithm,
whereby transpositions also are taken into account.

Is the Chessbase program doing that ?
No sir, it's not !

A whole big tree is calculated for all kinds of
variants, which can contain many transpositions,
and then a complete minimax is done. This
means *lots* of inferior variants are calculated into
much more depth than necessary; a waste of time !!

No, probably with an engine as DeepShredder, and a
powerful multiprocessor machine would be more
useful for the postal chess player, just by going
to the variants manually, scrolling back and
forth through the variants, and letting the hash
tables 'remember' the results (i.e. evaluation).

Still at higher levels it might help to have some
knowledge about endgame strategy, to find
the best lines during the middlegame.
Engines probably are just an aid, just like
a GM as J.Piket most likely recently has been
for the brandnew world champ correspondence
chess Van Oosterom.. ( a patzer anyway, as
he usually is opening with 1.d4?!).

the best chess repertoire

Although controversial (all normal chess opening variants are
considered to be equallystrong), with the program Bookbuilder
i found a 'best' repertoire: with white 1.e4!
No i'm not going to give all the subvariants, but some
important decisions have to be made also in later
moves to keep the white opening advantage; yes,
especially against the Sicilian; urgh, quite complicated indeed.

So can white also win against the Ruy Lopez Berlin defence ?
Well i don't know, probably yes, with proper play.
Computers are not so good in endgame strategy, some
books by eg. Shereshevsky demonstrate his obviously.

Against d4 Nf6 and if 2.c4 e6, and then:

Nimzo-Queens Indian hybrid system (after Nf3 b6! and after Nc3 Bb4!),
yes there are lots of variants in this complex.
No fancy Benoni, Benko, Kings Indian, though.

Against other white moves: ah well, not so important,
currently there's c5 against the English (symmetrical variant),
..d5 agains Reti, etc. (often transposing to the d4 variants).

Resultant improvement in playing strength:
well i don't know, i estimate its only about 50 points or so.
Comps are not afraid of gambits and unorthodox openings,
so such a repertoire would cost about 50 points on average.
But for humans, up to master level, with proper knowledge
an preparation, it could be an advantage. But then tactics
and middlegame planning are more important of course..

Back to FICS now again..
c.u. later.. :)

internet chess servers

ever wondered how computers can get 3600 blitz rating on ICC ?
well we all know of course, they don't play other computers,
have !computer in their formula, and play only humans.

And then the ratings on ICC are calculated differently
than those on FICS; the latter is using RD, which
leads to much more conservative values. And
also more realistic values for these stupid chips..

Any other experiences with comps; on
playchess.com or so ? just wondering..

well letz start

well lets start with this blog; blogging seems to be very cool,
so instead of making some elaborate project plan, its
just a matter of hitting the keyboard, en letz see..

now it would be cool to make a blog over computer chess
of course, spending so much time on it by making an
opening book for my bookbuilder account on the
internet chessclub (www.chessclub.com), but i guess
i first need to see how this thing works in order
to organize the various thoughts, blabla, etc.

in any case i want to keep the whole thing on topic,
that is, about computer chess; a wide field anyway,
that's for sure, and probably keeping me busy for
a while with this stuff. a nice distraction from
the engines, the servers, the bishops and knights,
the pgn, EPD's, and all the other stuff. Aargh
indeed, when the hash tables are overloaded
again, Arena is crashing with some access violation,
and my so carefully built-up (Comp)rating is going
down again; that's life i guess, sometimes i even
wonder why people takes these chips so seriously..