EZoto wrote:
Give me Tigran Petrosian

I wonder what it would be like if Petrosian in his prime were alive
today to play these machines. It would have been very interesting.

Hard to say. I like Petrosian a lot but his style can be caricatured
as `Wait 'til the other guy makes a mistake, then kill him. If he
doesn't make a mistake, agree a draw.' I'm not sure computers make
enough mistakes for this strategy to be effective.


Dave.

--
David Richerby Unholy Priest (TM): it's like a man
www.chiark.greenend.org.uk/~davidr/ of the cloth but it's also a crime
against nature!

On Mar 7, 6:15 am, David Richerby
wrote:

I wonder what it would be like if Petrosian in his prime were alive
today to play these machines. It would have been very interesting.

Hard to say. I like Petrosian a lot but his style can be caricatured
as `Wait 'til the other guy makes a mistake, then kill him. If he
doesn't make a mistake, agree a draw.' I'm not sure computers make
enough mistakes for this strategy to be effective.


I don't even know how they handle the problem
of setting a contempt factor for this match.

Every game has Rybka -- listed as being rated
3100 on one site -- starting off down a different
pawn, as Black. Its opponent is a competent
GM, but he is listed as 25xx. What do you do:
Set no contempt factor, thinking the pawn and
move offsets the difference in strength? Or go
hog wild and set the program to believe it is
invincible? Or perhaps set a small negative
hero-worship factor so the program will not be
forced to self-destruct, then kick back and
watch the human self-destruct?

Tigran Petrosian's style was shown to have
serious flaws in his matches with GM Fischer;
but that was /after/ his peak. Writers would
often describe TP's style as anticipating every
conceivable attack, and thwarting it before it
could even begin. But the truth is, he agreed
to many, many draws, so how do they know
this?

To my thinking, the best way to handle a
computer is to play so well that it cannot
handle *you*. Take the Deep Thought match
with Gary Kasparov, for instance. Afterward,
people were writing things like "the computer
was unrecognizable in this game" -- just like
they do when a great human player gets
clobbered!

But I do believe there are plenty of stronger
GMs around who could give Rybka a tougher
match with these odds. Joel Benjamin, for
instance, got the upperhand (to say the least)
in the openings, game after game, but he
crumbled in the middlegame. I have yet to
see GM Dzindzi make much of his odds in
the opening, and half the time his extra pawn
gets doubled or becomes useless, due to his
own clumsy play or to wild King assaults in
which the pawn-count is rendered moot.

I will say that he has demonstrated a
mastery of K & p endings, and has also
shown himself to be competent in R & p
endings-- unlike Rybka! :D


-- help bot

On Mar 7, 3:42*pm, help bot wrote:
* But I do believe there are plenty of stronger
GMs around who could give Rybka a tougher
match with these odds. *Joel Benjamin, for
instance, got the upperhand (to say the least)
in the openings, game after game, but he
crumbled in the middlegame. *I have yet to
see GM Dzindzi make much of his odds in
the opening, and half the time his extra pawn
gets doubled or becomes useless, due to his
own clumsy play or to wild King assaults in
which the pawn-count is rendered moot.


Bear in mind that Dzindzi is 63 years old.

Even now I would bet serious money that he would mop the floor with
you giving you 5-1 blitz odds. In his prime he used to be able to do
that with IMs, spotting them several cognacs to boot. Ask anyone who
saw him play in NYC in the 80s.

LT

On Mar 7, 4:12 pm, Larry Tapper wrote:
On Mar 7, 3:42 pm, help bot wrote:

But I do believe there are plenty of stronger
GMs around who could give Rybka a tougher
match with these odds. Joel Benjamin, for
instance, got the upperhand (to say the least)
in the openings, game after game, but he
crumbled in the middlegame. I have yet to
see GM Dzindzi make much of his odds in
the opening, and half the time his extra pawn
gets doubled or becomes useless, due to his
own clumsy play or to wild King assaults in
which the pawn-count is rendered moot.

Bear in mind that Dzindzi is 63 years old.

Which is a perfect example of what I said:
there are plenty of stronger GMs around, if
for no other reason than this: GM Dzindzi is
well past his peak.

The things to consider are twofold:

1) Rybka is the strongest program in the
world.

2) It is at its all-time peak, right now!


Even now I would bet serious money that he would mop the floor with
you giving you 5-1 blitz odds.


A shallow blitz player, eh? Reminds me
of Skip Repa-- although he was even
shallower, being a bullet-chess player.


In his prime he used to be able to do
that with IMs, spotting them several cognacs to boot. Ask anyone who
saw him play in NYC in the 80s.


You seem to be under the impression that
I am a newbie, unfamiliar with RD; in fact, I
still recall his picture appearing on covers of
Chess Lies magazine, after winning many
events. Those were regular chess tourneys,
not just blitz!


Rybka is shown as having a rating of "3100"
on the site linked to from its home page; to
me, this clearly indicates that we humans
need to send our "007 James Bond" types
to defend the honor of humankind, not just
the GM Joel Benjamins or GM Roman
Dzindzichashvilis; I want Gata... Vishy...
Gary... Bobby (oops, too late!). These
old-timer American players are making the
Siliconoids look good. Can you remember
back when Gary Kasparov made DT look
like a fish? I can.


-- help bot

On Mar 7, 4:12 pm, Larry Tapper wrote:


But I do believe there are plenty of stronger
GMs around who could give Rybka a tougher
match with these odds. Joel Benjamin, for
instance, got the upperhand (to say the least)
in the openings, game after game, but he
crumbled in the middlegame. I have yet to
see GM Dzindzi make much of his odds in
the opening, and half the time his extra pawn
gets doubled or becomes useless, due to his
own clumsy play or to wild King assaults in
which the pawn-count is rendered moot.

Bear in mind that Dzindzi is 63 years old.

Even now I would bet serious money that he would mop the floor with
you giving you 5-1 blitz odds.


Interesting that you are eager to give me *time*
odds, not material (like say, two Knights). :D

I also notice you avoided the issues I raised
above. One issue was the fact that GM RD
kept hanging his extra pawn, or else getting it
doubled. What I assume is the final game was
a perfect example; in that game, RD just hung
his h-pawn due to carelessness, then the
operator generously "gave" him a draw where
there was still some play left for Rybka; I've
seen it win such positions many times, when
facing other GM-strength chess programs
(but not Zappa!).

In sharp contrast, GM Benjamin normally
obtained a clearly winning advantage in his
odds match, then was squashed in the
middlegame, or else squeaked out a draw.


In his prime he used to be able to do
that with IMs, spotting them several cognacs to boot. Ask anyone who
saw him play in NYC in the 80s.


In 1978, GM Dzindzi ranked ahead of Gary
Kasparov, according to Chessmetrics' data.
GMs Karpov and Kortchnoi were at the top
of the lists -- which brings us back to the
fact that these guys are far beyond their
best days.

I did a little reading, and it turns out that
this match was played at a very fast pace:
just 45 minutes per side, plus ten lousy
seconds per move! No wonder the quality
of play was unimpressive. This compares
well to some of my recent experiences at
the total-patzer level, where the time control
was game/61 minutes, with some games
using a time-delay clock, and some not.
Some games finish without any apparent
affect from time-pressure, while others
enter a phase in which the clock is *the*
decisive factor.

My guess is that the quick time controls
are intended to make "less work" for the
grandmasters who have agreed to play, but
this has a detrimental effect on the quality
of play, as does the fact that neither player
is "booked up" once the pawn is removed.

The closest thing I saw to a "book" line
was the game where Black offered the
f-pawn, and then played the old line we've
seen from the days of Paul Morphy (...e6).
I wanted to scream: "Advance Variation!"
when I saw 1 e4 e6, 2. d4 d5, since it
is impossible for Black to play the French
Defense properly without the pawn-break
....f6 (the missing pawn!).


-- help bot

On Mar 7, 4:12 pm, Larry Tapper wrote:
On Mar 7, 3:42 pm, help bot wrote:

But I do believe there are plenty of stronger
GMs around who could give Rybka a tougher
match with these odds. Joel Benjamin, for
instance, got the upperhand (to say the least)
in the openings, game after game, but he
crumbled in the middlegame. I have yet to
see GM Dzindzi make much of his odds in
the opening, and half the time his extra pawn
gets doubled or becomes useless, due to his
own clumsy play or to wild King assaults in
which the pawn-count is rendered moot.


What are the mysterious "positional errors"
I mentioned earlier in some other thread? If
I am not mistaken, the idea I discuss below
happened in two different games, though
there were plenty of other errors worth
examining.

One example was a game where Black (in
this match, this is always Rybka, who is a
pawn down) advanced its a-pawn. Now, I
could hardly count the number of top-level
games I have seen where it was automatic
to reply with p-a4, but GM Dzindzi ignored
the "threat" (to gain space, etc.) and next
came ...a4!, to which he felt obliged to
reply a3 (otherwise Black would play ...a3,
establishing a thorn in his Queenside).

White's two Queenside pawns became
completely immobilized by a single Black
pawn (yuck). So, what was Dzindzi's "plan"
here? He next went after Black's a-pawn
with B-b5, to which Rybka replied ...Bd7,
and they traded Bishops. Note how Black
traded its (active) bad Bishop for White's
good Bishop (yuck again).

Time after time, while GM Dzindzi did a
good job of not hanging pieces, he let his
extra pawn become doubled or else hung
it outright (in fact, one of the games he
*won* resulted from this same problem).

Apparently, he drew the match, but one
of the games he lost was practically given
away via reckless play in the opening, so
it is not inconceivable that he could have
won the match-- making good on claims
that his actual rating was not reflective of
his anti-computer prowess, and so forth.
In fact, just by drawing the match, he has
given those fans justification, since all the
other GMs *lost* their odds matches.

In playing against the relatively weak
computer at GetClub, I am constantly
faced with positions where I need to avoid
getting my pawns doubled, or hanging my
one, hard-earned extra pawn. Recently,
I had a number of far-advanced passed
pawns and decided to "trade" Queens by
sac'ing mine for a Rook, then forking his
with my Rook. Unfortunately, the GC
program played the one move which
refuted this idea and I have had to
"resign" several drawn games in order to
get on with the show. Things are getting
tougher in the middlegame-- especially if
I take the program lightly.

Yet I believe I have played a number of
games there where the problems which
plagued GM Dzindzi were carefully
avoided; I do not routinely let my extra
material get taken away, nor rendered
moot, nor crippled such that it is useless.
Trouble is, I am nowhere near as good as
RD at *not* dropping pieces! :D


-- help bot

* Rybka is shown as having a rating of "3100"
on the site linked to from its home page; to
me, this clearly indicates that we humans
need to send our "007 James Bond" types
to defend the honor of humankind, not just

Or just pay GetClub 1 Million Dollar And GetClub will beat the Rybka.
As I have stopped further improvement due to lack of finance.

Now Advance Level plays as good As Rybka.

Play Chess at: http://www.GetClub.com/Chess.html

Advance Level thinks 256 times longer. To make it play as fast as
Rybka I need to put the code in Native C using Assembly level Language
which are 20 times faster than an Applet.

And rest 20 times fast can be achieved using Parallel Processing.

So incase the above 2 things are done GetClub Beginner Level will play
as good as Advance Level and give tough competition to Rybka.

I feel the above hypothysis is correct but it still needs to be
tested.

Has anyone having Rybka. Just play a Match between Master Level &
Rybka and tell me how it played.

2 things to be done.

1. Rybka should not think in opponents time.
2. Rybka do not use processing power of multiple cores.

Then I think (Master/Advance vs Rybka) will play simmilar game.

Bye
Sanny

Play Chess at: http://www.GetClub.com/Chess.html

Even now I would bet serious money that he would mop the floor with
you giving you 5-1 blitz odds. In his prime he used to be able to do
that with IMs, spotting them several cognacs to boot. Ask anyone who
saw him play in NYC in the 80s.

LT

I remember those days. He also was a chain smoker and could beat you
with those odds smoking a cigarette at the same time. A lot of people
don't realize how good Dzindzi really was in those days. After he got
married and had a daughter he literally became a different man and
even played first board for the olympiad and thrashed Beliavsky in
there encounter.

EZoto

On Mar 8, 1:46 am, Sanny wrote:

Rybka is shown as having a rating of "3100"
on the site linked to from its home page; to
me, this clearly indicates that we humans
need to send our "007 James Bond" types
to defend the honor of humankind, not just

Or just pay GetClub 1 Million Dollar And GetClub will beat the Rybka.
As I have stopped further improvement due to lack of finance.

Whew! And just when you were getting soooo
close.


Advance Level thinks 256 times longer. To make it play as fast as
Rybka I need to put the code in Native C using Assembly level Language
which are 20 times faster than an Applet.

Even so, what would happen in the endgame,
when GetClub drops to zero and Rybka kicks
in with some table-bases? (Scary thought, huh?)


1. Rybka should not think in opponents time.

Useless, as GetClub will always choose a
"surprise move", emptying the hash tables.


2. Rybka do not use processing power of multiple cores.

I think a good match would be QN odds,
then as above: no thinking on opponent's
time and single core. Something should
e done with the contempt factor, so Rybka
won't go after a quick draw.


-- help bot

On Mar 8, 3:50 pm, EZoto wrote:

I remember those days. He also was a chain smoker and could beat you
with those odds smoking a cigarette at the same time. A lot of people
don't realize how good Dzindzi really was in those days. After he got
married and had a daughter he literally became a different man and
even played first board for the olympiad and thrashed Beliavsky in
there encounter.


In the old days, every time a Russian grandmaster
emigrated to the USA he would soon appear on the
cover of Chess Lies magazine, having won a big
tourney or two. Lev Alburt, Boris Gulko, Roman
Dzindzichashvili, and so on down the line. Finally,
there were so many "Russian" GMs here that new
ones could no longer expect such instant success.
Today, it seems that the big-money events have
grown larger, and often end in multi-way ties.


But I wish they could afford to pit the current top
players against Rybka, at *slower* time controls.
Some people don't like the idea of material odds,
probably because they live and breathe a detailed
knowledge (read: memory) of the chess openings.
This doesn't bother me, but a part of the odds
package could be time odds, for instance. I have
grown accustomed to post-game analysis, where
the computer is given oodles of time to think over
each position before rendering a judgment; so
when I see these fast games -- what was once
considered "action chess" -- I cringe at the crude
time-pressure-induced mistakes.

Even Rybka, at this fast time control, is hardly
immune from error. One Rook ending ought to
have been drawn rather than lost, and in the
game where GM Dzindzi lost the exchage,
Rybka's "defense" reminded me of GetClub.


-- help bot