RYBKA wins the the Computer World Chess Championship 2007
with 9 wins and 2 draws.

Who here didn't see *THAT* one coming?

Round 1, The King - Rybka 0-1
Round 2, Micromax - Rybka 0-1
Round 3, Rybka - Zappa 1/2-1/2 [Note 1]
Round 4, Loop - Rybka 1/2-1/2
Round 5, Rybka - Jonny 1-0
Round 6, Isichess - Rybka 0-1
Round 7, Rybka - The Baron 1-0
Round 8, Sjeng - Rybka 0-1
Round 9, Rybka - Diep 1-0
Round 10, Gridchess - Rybka 0-1
Round 11, Rybka - Shredder 1-0

As for other potential competition:

Junior wet his pants and hid behind mommy's skirts. [Note 2]

Fritz screamed like a little girl and ran away. [Note 2]

Zappa (Zap! Chess) sat in his retirement home, marveling
at how good these nefangled whippersnappers are. [Note 3}

Hydra 64CL sat in the garage as his R&D team feverishly
tried to make him work. Previous Hydra versions slept
in, knowing that they couldn't beat the latest Rybka.
Look for an interesting matchup whenever Hydra 64CL is
completed. [Note 4]

NOTES:

Note 1: Rybka had a winning tablebase position but
did not have KRPknp tablebase installed.

Note 2: http://www.rybkachess.com/docs/ChallengeToFIDE.htm

Note 3: https://netfiles.uiuc.edu/acozzie2/www/zappa/

Note 4: http://www.hydrachess.com/
http://en.wikipedia.org/wiki/Hydra_(chess)


--
G o o g l e F o o d : http://www.guymacon.com http://www.guymacon.com/
Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon
Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon
Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon Guy Macon

On Jun 19, 6:44 pm, Guy Macon http://www.guymacon.com/ wrote:

RYBKA wins the the Computer World Chess Championship 2007
with 9 wins and 2 draws.

Who here didn't see *THAT* one coming?


Um, maybe the guy who entered Shredder?


Round 1, The King - Rybka 0-1
Round 2, Micromax - Rybka 0-1
Round 3, Rybka - Zappa 1/2-1/2 [Note 1]
Round 4, Loop - Rybka 1/2-1/2
Round 5, Rybka - Jonny 1-0
Round 6, Isichess - Rybka 0-1
Round 7, Rybka - The Baron 1-0
Round 8, Sjeng - Rybka 0-1
Round 9, Rybka - Diep 1-0
Round 10, Gridchess - Rybka 0-1
Round 11, Rybka - Shredder 1-0

As for other potential competition:

Junior wet his pants and hid behind mommy's skirts. [Note 2]

Fritz screamed like a little girl and ran away. [Note 2]


Maybe Junior and Fritz were exhausted from their
recent "ultimately" irrelevant match.


Zappa (Zap! Chess) sat in his retirement home, marveling
at how good these nefangled whippersnappers are. [Note 3}

Hydra 64CL sat in the garage as his R&D team feverishly
tried to make him work. Previous Hydra versions slept
in, knowing that they couldn't beat the latest Rybka.
Look for an interesting matchup whenever Hydra 64CL is
completed. [Note 4]

NOTES:

Note 1: Rybka had a winning tablebase position but
did not have KRPknp tablebase installed.


Yeah, but regardless of tablebases, a chess player
needs to learn a bit about endgames. Like maybe
how to play them without looking *really dumb*.

In both the games which Rybka drew, it seemed to
recklessly sacrifice pawns, aimlessly meander its
pieces about the board, and give pointless spite
checks. You well know what sort of (human) player
that describes. Hint: it starts with "duf" and ends
with "fer". :D

It is well known that in imbalanced endings, the
inferior side (i.e. Rybka's opponents, of course!)
wishes to trade the pawns off, while Rybby ought
to know better, since it is trying to win. Even in
some of the games it did win, its play in the
ending looks a tad fishy. For example, it will leave
its Rook en prise in order to (eventually) promote a
pawn, yet that pawn was going to promote anyway;
in effect it comes off looking like a Rook hang.

-- help bot

In article , Guy Macon
/ writes



RYBKA wins the the Computer World Chess Championship 2007
with 9 wins and 2 draws.

Who here didn't see *THAT* one coming?

Round 1, The King - Rybka 0-1
Round 2, Micromax - Rybka 0-1
Round 3, Rybka - Zappa 1/2-1/2 [Note 1]
Round 4, Loop - Rybka 1/2-1/2
Round 5, Rybka - Jonny 1-0
Round 6, Isichess - Rybka 0-1
Round 7, Rybka - The Baron 1-0
Round 8, Sjeng - Rybka 0-1
Round 9, Rybka - Diep 1-0
Round 10, Gridchess - Rybka 0-1
Round 11, Rybka - Shredder 1-0

As for other potential competition:

Junior wet his pants and hid behind mommy's skirts. [Note 2]

Fritz screamed like a little girl and ran away. [Note 2]

Zappa (Zap! Chess) sat in his retirement home, marveling
at how good these nefangled whippersnappers are. [Note 3}

Hydra 64CL sat in the garage as his R&D team feverishly
tried to make him work. Previous Hydra versions slept
in, knowing that they couldn't beat the latest Rybka.
Look for an interesting matchup whenever Hydra 64CL is
completed. [Note 4]

NOTES:

Note 1: Rybka had a winning tablebase position but
did not have KRPknp tablebase installed.

Note 2: http://www.rybkachess.com/docs/ChallengeToFIDE.htm

Note 3: https://netfiles.uiuc.edu/acozzie2/www/zappa/

Note 4: http://www.hydrachess.com/
http://en.wikipedia.org/wiki/Hydra_(chess)


If Zappa was sat in its retirement home, then which program of the same
name drew with Rybka in round 3 and came 2nd overall?

Although Rybka is clearly the best program there is at this time, and
fully deserves to win the WCCC at last as it is proven no.1 on all
rating lists and in other major tournaments, I think Zappa did well to
make 2nd in view of the fact that programmer Anthony Cozzie hasn't done
any work on it for (presumably) some 5 or 6 months.

Cheers - Eric
--
Eric Hallsworth, Countrywide Computers, Victoria House, 1 High Street,
Wilburton, Cambs CB6 3RB. Tel: 01353 740323
Website for Chess & Bridge Computers & Softwa
http://www.countrywidecomputers.co.uk

help bot wrote:

Guy Macon http://www.guymacon.com/ wrote:

Junior wet his pants and hid behind mommy's skirts. [Note 2]

Fritz screamed like a little girl and ran away. [Note 2]

Maybe Junior and Fritz were exhausted from their
recent "ultimately" irrelevant match.

They have their just reward. A Google search on
[ Junior Fritz ] turns up a page full of sites
discussing Rybka. See
http://www.google.com/search?q=Junior+Fritz

Yeah, but regardless of tablebases, a chess player
needs to learn a bit about endgames. Like maybe
how to play them without looking *really dumb*.

Why would a machine that has large enough tablebases to
always play perfectly in the endgame need endgame code
that is never executed?

In both the games which Rybka drew, it seemed to
recklessly sacrifice pawns, aimlessly meander its
pieces about the board, and give pointless spite
checks. You well know what sort of (human) player
that describes. Hint: it starts with "duf" and ends
with "fer". :D

It is well known that in imbalanced endings, the
inferior side (i.e. Rybka's opponents, of course!)
wishes to trade the pawns off, while Rybby ought
to know better, since it is trying to win. Even in
some of the games it did win, its play in the
ending looks a tad fishy. For example, it will leave
its Rook en prise in order to (eventually) promote a
pawn, yet that pawn was going to promote anyway;
in effect it comes off looking like a Rook hang.

It must be an interesting challenge to write a strong
program, see it turn stupid in certain situations, and
try to fix the problem without introducing stupidity
in other positions that it handled well before. This
becomes harder when the program is a much better player
than the programmer, and even harder when the program
is a much better player than the programmer and the
grandmaster who assists the programmer. A possible
way out of this bind: use genetic algorithms to optimize
the evaluation function.

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

Content-Transfer-Encoding: 8Bit


Eric Hallsworth wrote:

Guy Macon http://www.guymacon.com/ writes

Zappa (Zap! Chess) sat in his retirement home, marveling
at how good these nefangled whippersnappers are.

If Zappa was sat in its retirement home, then which program of the same
name drew with Rybka in round 3 and came 2nd overall?

Although Rybka is clearly the best program there is at this time, and
fully deserves to win the WCCC at last as it is proven no.1 on all
rating lists and in other major tournaments, I think Zappa did well to
make 2nd in view of the fact that programmer Anthony Cozzie hasn't done
any work on it for (presumably) some 5 or 6 months.

And it appears that he has no plans to do any work on it in the
future.

"As of the summer of 2006, I am retiring as an active computer
chess programmer in order to concentrate on my PhD work. This
means that if you are waiting for the next version of Zappa
(Zap! Chess) you will be waiting for a very long time ... I
will not rule out completely a Jordanesque return to computer
chess, but I have not even had the Zappa source code on my
machine for the past 6 months."

"Erdo came to me before Leiden and asked to participate. Since
Erdo has done a lot for me, I basically said that he can do
whatever he wants, as long as he doesn't expect me to provide
him with updated versions of the program. He may play in future
tournaments or he may not; it is entirely up to him."

Source: https://netfiles.uiuc.edu/acozzie2/www/zappa/

In other words, Anthony Cozzie -- the author of Zappa -- retired
it and likely won't be making any further improvements, but gave
Erdogan Günes permission to enter it into tournaments.

Perhaps I can do a better description:

Zappa (Zap! Chess) sat on the porch of his retirement home,
marveling at how good some of these nefangled whippersnappers
are. Every so often one of the nefangled whippersnappers cuts
across Zappa's lawn and the old fellow gives him a good
thrashing.

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

On Jun 20, 9:20 am, Guy Macon http://www.guymacon.com/ wrote:

Yeah, but regardless of tablebases, a chess player
needs to learn a bit about endgames. Like maybe
how to play them without looking *really dumb*.

Why would a machine that has large enough tablebases to
always play perfectly in the endgame need endgame code
that is never executed?


It wouldn't. However, most endings are sufficiently
complex that the endgame tablebases do not save
their bacon, and the glaring weaknesses are plainly
evident.


In both the games which Rybka drew, it seemed to
recklessly sacrifice pawns, aimlessly meander its
pieces about the board, and give pointless spite
checks. You well know what sort of (human) player
that describes. Hint: it starts with "duf" and ends
with "fer". :D

It is well known that in imbalanced endings, the
inferior side (i.e. Rybka's opponents, of course!)
wishes to trade the pawns off, while Rybby ought
to know better, since it is trying to win. Even in
some of the games it did win, its play in the
ending looks a tad fishy. For example, it will leave
its Rook en prise in order to (eventually) promote a
pawn, yet that pawn was going to promote anyway;
in effect it comes off looking like a Rook hang.


It must be an interesting challenge to write a strong
program, see it turn stupid in certain situations, and
try to fix the problem without introducing stupidity
in other positions that it handled well before. This
becomes harder when the program is a much better player
than the programmer, and even harder when the program
is a much better player than the programmer and the
grandmaster who assists the programmer. A possible
way out of this bind: use genetic algorithms to optimize
the evaluation function.


I'm not sure the problem lies in the program being all
that much stronger than humans.

For example, look at the game where world champion
GM Kramnik was beating Deep-something; he first let
go of the win, and then several moves later, overlooked
a one-move-mate on himself. Does the fact that the
program *saw* the mate-in-one mean that humans can
never expect to teach it anything? Naah.

Once you take into account the vast tactical superiority
of computers, their innate inability to hang pieces out of
blindness to obvious threats, the difference between a
top computer and the top human players vanishes into
thin air. Apart from perfect memory in both the opening
and endgame, and the perfect sight of the board as to
short-range tactics, things are fairly evenly matched. If
anything, humans still have the edge if given blunder
takebacks, because of their far superior grasp of long
range strategy.

One example would be the game where Rybka only
drew a clearly won ending in the championship. I
could hardly believe my eyes when it just sat there,
and then placed nearly every pawn on the "wrong"
color.

This strategy, that of limiting the mobility of the
opponent's Bishop by placing one's pawns on the
"wrong" color, is okay early in the game or where the
pawns are not far advanced, and can later be shifted
as needed, but for gosh sakes, what was Rybka
thinking? Even with SSDF ratings approaching the
3000 mark, there is clearly much room left for
improvement in the endgame.

One thing which should be mentioned is that in
the championship tournament, one of the key games
was decided entirely in the "book openings" phase.
Shredder simply lost to a programmed-in book line,
so the real superiority of Rybka was never tested.
Several of the contestants were fairly weak programs,
and so the win here was not so impressive as it
might otherwise have been. What does impress is
the fact that all the other Chessbase programs stayed
away. :D