On Apr 25, 3:34 am, "David Kane" wrote:
"David Richerby" wrote in message

raylopez99 wrote:
http://www.chessbase.com/newsdetail.asp?newsid=3455

This was the article I was thinking of, per my earlier post, not the
Jeff Sonas article.

That article is, frankly, junk: I'm surprised it was ever accepted for
an academic conference.

They haven't determined the strongest champion of all time: they've
determined which World Champion plays most like a crippled version of
Crafty. That's better than working out which World Champion plays
most like me but not much better. See Soren Riis's rebuttal

http://www.chessbase.com/newsdetail.asp?newsid=3465

I don't think Riis or you understood the original article. The
researchers addressed in detail the objection that Crafty is not
the ultimate in determining the best move - obviously
we can find some specific positions where the version of
Crafty used in the analysis is wrong, but that is not a
fundamental objection.

There is much very interesting and original work
in the article - perhaps the Chessbase synopsis concentrates
excessively on the findings rather than on the methodology,
since it makes a better story. Certainly there were analyses
that they didn't do which should get done. That's just the normal
way that research advances. In any case, the approaches
investigated in the article are far preferable to the "historical
ELO" or "chessmetics" nonsense, which are *completely*
lacking in rigor of any kind.- Hide quoted text -



I tend to agree with you David Kane.

I find the rebuttal by Dr. Søren Riis, Oxford, UK unconvincing for a
number of reasons.

- it was clearly written with a popular audience in mind (witness the
exclamation point! It's been said that no serious article has ever
been written with an exclamation point! Unless the author did so
deliberately)

- it fails to understand the simple argument of 'normalization'. The
Matej Guid and Ivan Bratko original article pointed out that Crafty
was used since it was open source and could be modified; the stronger
programs are not, but in any event Crafty is hardly a weak tactics
program and the authors are looking for a standardized (normalized)
way of spotting blunders.

-The fact that Riis found positional sacrifices not evaluated by
Crafty is not convincing since: (1) such positional sacrifices are
rare--as computers have shown, chess is largely tactics; (2) everybody
will be judged equally by Crafty, so others pos sacs are also scored
'badly', so nobody will lose relative standing to one another, and
(3), as long as assumption (1) is valid, Crafty will find the most
"mistake free" chess player, or one that plays closest to being
"tactics mistake free", which is a very good way to determine a good
chess player IMO.

Now of course the surrebutter (rebuttal to the rebuttal) will be that
players like Tal will score poorly--and indeed they (he) did--but
let's face it, Tal was more of a shock player that relied on playing
the man rather than the board. In a match of coolheaded Karpov or
Kramnik versus Tal, all in their prime, the less emotional player is
likely to win (unless he loses his cool and loses...haha... think of
Topolov vs Kramnik). Also nobody ever became champion ignoring
tactics. That is the lesson of chess. Think of all the bogus moves
made by beginners, sacrificing knight for pawn, "to break up their
pawn chain", with no positional advantage. If you believe chess is
positional play more than tactics then such bogus moves should work
more often than they do. They do not.

So, understanding how chess works, and how chess playing computers
work, and having seen Crafty evaluate pretty good myself, I have to
side with the original article.

RL

raylopez99 wrote:
I find the rebuttal by Dr. S=F8ren Riis, Oxford, UK unconvincing for
a number of reasons.

- it was clearly written with a popular audience in mind (witness the
exclamation point!

Obviously, anything written with a popular audience in mind cannot
possibly be accurate.


- it fails to understand the simple argument of 'normalization'. The
Matej Guid and Ivan Bratko original article pointed out that Crafty
was used since it was open source and could be modified; the stronger
programs are not, but in any event Crafty is hardly a weak tactics
program and the authors are looking for a standardized (normalized)
way of spotting blunders.

Just because they used the same system for everyone doesn't mean the
system was good or useful. For example, they could declare that every
king move is a blunder. That's consistent across all the players but
would declare players who tend to win in the endgame (where the king
gets moved more) to be weaker than players who tend to win in the
middlegame. You need to apply the same *good* measure to everyone.


-The fact that Riis found positional sacrifices not evaluated by
Crafty is not convincing since: (1) such positional sacrifices are
rare--as computers have shown, chess is largely tactics; (2) everybody
will be judged equally by Crafty, so others pos sacs are also scored
'badly', so nobody will lose relative standing to one another

No. A player who plays more positional sacrifices will be penalized
for playing moves that crafty doesn't understand.


and (3), as long as assumption (1) is valid, Crafty will find the
most "mistake free" chess player, or one that plays closest to being
"tactics mistake free", which is a very good way to determine a good
chess player IMO.

But World Champions make very few tactical mistakes.


Now of course the surrebutter (rebuttal to the rebuttal) will be
that players like Tal will score poorly--and indeed they (he)
did--but let's face it, Tal was more of a shock player that relied
on playing the man rather than the board.

I'm not convinced by that assertion. Tal played games that were sound
enough that they were very hard to defeat over the board. I don't
think that counts as playing the man rather than the board.


In a match of coolheaded Karpov or Kramnik versus Tal, all in their
prime, the less emotional player is likely to win

Hmm... The two Botvinnik-Tal matches between them were only won by
Botvinnik +12-11=19. Hardly a convincing victory for the cool head.


Think of all the bogus moves made by beginners, sacrificing knight
for pawn, "to break up their pawn chain", with no positional
advantage. If you believe chess is positional play more than
tactics then such bogus moves should work more often than they do.
They do not.

This argument is bogus. Sacrificing a knight against one's opponent's
pawn structure is hardly a prime example of `positional chess'. You
might as well say that all the bogus tactical shots attempted by
beginners to `win material' or `checkmate the king' show that tactics
play a small role in chess.


Dave.

--
David Richerby Frozen Erotic Gerbil (TM): it's like
www.chiark.greenend.org.uk/~davidr/ a children's pet but it's genuinely
erotic and frozen in a block of ice!

In article ,
David Richerby wrote:

Now of course the surrebutter (rebuttal to the rebuttal) will be
that players like Tal will score poorly--and indeed they (he)
did--but let's face it, Tal was more of a shock player that relied
on playing the man rather than the board.

I'm not convinced by that assertion. Tal played games that were sound
enough that they were very hard to defeat over the board. I don't
think that counts as playing the man rather than the board.

The whole idea of judging a player by his "error rate" presumes that the
way to win at chess is to commit no errors.

But a quick look at players like Lasker, Tal, and Bronstein shows that
there's another way: make an error in order to induce your opponent to
make a bigger error.

Many of Tal's sacrifices would be considered errors by a chess program
(and that's just counting the ones where you could expect a program to
see it through to the end, in all variations, in however much time you
gave it - and if you're only giving even a top program ten minutes a
move, you're not getting there on a lot of sacrifices) but Tal wasn't
trying to play perfect chess. He was trying to win games.

And judging by his results (a world championship; the longest undefeated
streak in tournament games) he did so incredibly well.

To say, therefore, that he was making errors strikes me as somewhat
absurd.

If the "error" was never intended to be an irrefutable move, and it
leads directly to victory against a top player, how can you call it an
error?

-Ron

"Ron" wrote in message
...
In article ,
David Richerby wrote:

Now of course the surrebutter (rebuttal to the rebuttal) will be
that players like Tal will score poorly--and indeed they (he)
did--but let's face it, Tal was more of a shock player that relied
on playing the man rather than the board.

I'm not convinced by that assertion. Tal played games that were sound
enough that they were very hard to defeat over the board. I don't
think that counts as playing the man rather than the board.

The whole idea of judging a player by his "error rate" presumes that the
way to win at chess is to commit no errors.

But a quick look at players like Lasker, Tal, and Bronstein shows that
there's another way: make an error in order to induce your opponent to
make a bigger error.

Many of Tal's sacrifices would be considered errors by a chess program
(and that's just counting the ones where you could expect a program to
see it through to the end, in all variations, in however much time you
gave it - and if you're only giving even a top program ten minutes a
move, you're not getting there on a lot of sacrifices) but Tal wasn't
trying to play perfect chess.


That's total rubbish "Ron". You're obviously someone who doesn't know much
about the game of chess. Tal didn't set out to make errors, with the
lamebrain idea that this would somehow cause his opponents to make bigger
errors. Tal set out to create COMPLICATIONS for his opponents. Obviously Tal
desired for all of his sacrifices to be sound and forcing, but no human can
calculate everything to the end, so computer analysis has shown flaws in
many of his games. This is meaningless, because he wasn't playing against
computers.

Your comment is similiar to a common theme of beginner (or patzer) level
thinking, ie: "I know this move is bad, but if he doesn't see Bxf7+ then it
will be very good for me.

JMR

In article c27Yh.131489$6m4.63198@pd7urf1no,
"Chess Sadist" wrote:

That's total rubbish "Ron". You're obviously someone who doesn't know much
about the game of chess. Tal didn't set out to make errors, with the
lamebrain idea that this would somehow cause his opponents to make bigger
errors. Tal set out to create COMPLICATIONS for his opponents. Obviously Tal
desired for all of his sacrifices to be sound and forcing, but no human can
calculate everything to the end, so computer analysis has shown flaws in
many of his games. This is meaningless, because he wasn't playing against
computers.

Have you read Tal's books?

I have. There are many time when he says things like, "It's clear 36. f4
was stronger," (Tal-Gligoric, Zagreb 59), or see his note to 5. ... Qc7
in Tal-Olaffson, Bled 1961 (a move he describes as "bad" - but that he
clearly made intentionally) or, say, 10. a3 in Tal-Bagirov,
Dnepropetrosk, 1970, which he describes as "in no way stronger than the
approved Re1."

(I found these notes by basically opening "The Life and Games of Mikhail
Tal" at random. Stuff like this is all over that book. You should try
reading it sometime, before you talk about what Tal was, or wasn't,
thinking. His book on his match with Botvinnik goes into even more depth
on his thinking, again, and does a good job explaining the emphasis Tal
put of psychology over soundness. And what is psychology, in chess,
other than playing an inferior move which you think your opponent will
respond badly too. In particular, I'd point you to his discussion of his
12th move of game 17.)

It's clear from his notes that he doesn't care if his sacrifices were
"correct" or not. He made a move - which he knew could well be unsound -
with the expectation that in the resulting position his opponents would
play incorrectly.

That's pretty much the definition of "making an error to induce your
opponent into making a bigger one."

-"Ron"

On Apr 26, 11:00 am, David Richerby
wrote:
raylopez99 wrote:

I find the rebuttal by Dr. S=F8ren Riis, Oxford, UK unconvincing for
a number of reasons.

- it was clearly written with a popular audience in mind (witness the
exclamation point!

Obviously, anything written with a popular audience in mind cannot
possibly be accurate.

No, but popular means not as accurate as a journal paper, which the
original paper was. Otherwise it's like saying whoever wins this
Usenet thread is right moreso than two chess researchers debating.


- it fails to understand the simple argument of 'normalization'. The
Matej Guid and Ivan Bratko original article pointed out that Crafty
was used since it was open source and could be modified; the stronger
programs are not, but in any event Crafty is hardly a weak tactics
program and the authors are looking for a standardized (normalized)
way of spotting blunders.

Just because they used the same system for everyone doesn't mean the
system was good or useful. For example, they could declare that every
king move is a blunder. That's consistent across all the players but
would declare players who tend to win in the endgame (where the king
gets moved more) to be weaker than players who tend to win in the
middlegame. You need to apply the same *good* measure to everyone.

That is the ideal, but my point stands--equally bad is not so bad.
And BTW using your example, a player who wins in the middlegame is
indeed probably stronger than one who wins in the endgame (it's
tougher to win a short game--think of winning a chess brilliancy
against equally matched opposition--than to grind out a win in the
endgame. In fact, a standard technique I use to draw against my much
more powerful chess playing computer is to reduce to the endgame and
go for the draw).



-The fact that Riis found positional sacrifices not evaluated by
Crafty is not convincing since: (1) such positional sacrifices are
rare--as computers have shown, chess is largely tactics; (2) everybody
will be judged equally by Crafty, so others pos sacs are also scored
'badly', so nobody will lose relative standing to one another

No. A player who plays more positional sacrifices will be penalized
for playing moves that crafty doesn't understand.

No. See my point above. And chess is 99% tactics (famous quote).


and (3), as long as assumption (1) is valid, Crafty will find the
most "mistake free" chess player, or one that plays closest to being
"tactics mistake free", which is a very good way to determine a good
chess player IMO.

But World Champions make very few tactical mistakes.

Not true. Nearly all games are full of tactical mistakes, except
perhaps at the correspondence chess level. I was reading a book by
John Nunn ("Chess explained move by move") that makes this point in
the preface--Nunn had a hard time finding 20 OTB games that were
'mistake free' for his book, after searching 1000s of games.


Now of course the surrebutter (rebuttal to the rebuttal) will be
that players like Tal will score poorly--and indeed they (he)
did--but let's face it, Tal was more of a shock player that relied
on playing the man rather than the board.

I'm not convinced by that assertion. Tal played games that were sound
enough that they were very hard to defeat over the board. I don't
think that counts as playing the man rather than the board.

But on balance Tal was a shock player. Deny that and you become a
chess revisionist.


In a match of coolheaded Karpov or Kramnik versus Tal, all in their
prime, the less emotional player is likely to win

Hmm... The two Botvinnik-Tal matches between them were only won by
Botvinnik +12-11=19. Hardly a convincing victory for the cool head.

Pace Karpov's lifetime record against Tal, which is way positive. Of
course it was a young Karpov against an older, sick Tal, but the point
stands.


Think of all the bogus moves made by beginners, sacrificing knight
for pawn, "to break up their pawn chain", with no positional
advantage. If you believe chess is positional play more than
tactics then such bogus moves should work more often than they do.
They do not.

This argument is bogus. Sacrificing a knight against one's opponent's
pawn structure is hardly a prime example of `positional chess'. You
might as well say that all the bogus tactical shots attempted by
beginners to `win material' or `checkmate the king' show that tactics
play a small role in chess.

Positional chess SACRIFICE was my point. A positional chess sacrifice
is rare in chess is my point (goes to chess being 99% tactics). A
positional chess sacrifice is one where you do indeed exchange knight
for two pawns, so you're down a pawn, with no immeadiate hope of
recapturing your lost material. But the positional gain will help you
20 moves from now. This is common in GO but not in chess.

Ray

On Apr 25, 10:26 pm, help bot wrote:

You know, if you took the games of a typical (1300) rated
player and checked them with a dumbed-down-Crafty (1500),
you might get some useful information, but not nearly as
much as hoped for. But when you take the games of the
world champions and check them with a program which is
short of 2800, you get mainly garbage, combined with many
instances where a tactical oversight is correctly pinpointed.

But chess is 99% tactics help bot.


You also penalize those players who *deliberately* chose
to play what they knew to be sub-optimal moves, for
whatever reason. I just did this myself at RedHotPawn,
choosing to grab a Knight rather than leap in with another
piece to set up a 95%-certain mating net. Why? Because
while the mating net was around 95% certain, the capture
of the free piece was 100% certain (unless I have lost my
mind)! When I spot another mating net, things should be
simple enough for me to get the 100% certainty I desire,
and having captured yet another piece, this is all but
inevitable, barring my opponent's resignation.

But you risk the chance of letting your opponent escape--remember the
maxim: "always check, since the next move may be mate". Just recently
I did not follow this move and instead of winning a pawn against my PC
I drifted and eventually lost.


Another item which these statistical analyses overlook
is the deliberate gift of, say, a half-point. These have
been known to occur in world championship level play,
and of course the "nice guys" will be penalized for not
being "tough players", despite clinching the match
with their action.

Keep in mind this was not a statistical analysis of the kind Sonas is
famous for, but a different kind. Also over time the "nice guys"
penalty will statistically average out.


In short, what can be learned is who was least prone
to tactical blunders, and apparently, whose style leans
most toward a sizable gap between what the program
sees as the #1 optimal move, and #2 -- something I
think may be termed the sharpness of play. For one
example, I am playing a game at RedHot now where
I had to decide whether to develop my QB "normally"
via ...d6 and then B-moves somewhere, or fianchetto
via ...b6 and B-b7. It was a toss-up, since it makes
no difference whatever to the outcome. I expect a
computer would see both moves as being nearly
equal, weighing them in such a way as to slightly
favor the move which gives the Bishop immediate
control of squares, though this immediacy is quite
irrelevant to the true value of the moves.

Again, over time this will "wash out" or "average out". In general
sharp play is better than just pushing yourself into a passive
position, don't you think? That's what Crafty is looking for--sharp
play. Sharp play = sharp mind bot!


I wonder just how much time, and to what depth
the moves were analyzed before scoring them. I
recall that often a player's move may be scored poorly,
but if executed and stepped forward, a program may
change its mind completely about this, suddenly
realizing it had overlooked something.


No, you're talking about "move on opponent's time" feature. The way
the study was done was to analyze each move for a fixed time, so no
"changing of mind", and even if so, each player had the same scoring
applied, so it doesn't really matter (over time). Besides, have you
noticed that _MOST_ of the time (not always) the best move found by
Fritz or Crafty in the first five seconds is also the best move found
after 60 seconds? Because chess is 99% tactics, and often the tactics
are no more than 4 moves deep (most of the time).

RL (a 1950 Elo player, so I can speak with some authority).

On Apr 26, 6:26 am, raylopez99 wrote:

- it fails to understand the simple argument of 'normalization'. The
Matej Guid and Ivan Bratko original article pointed out that Crafty
was used since it was open source and could be modified; the stronger
programs are not, but in any event Crafty is hardly a weak tactics
program and the authors are looking for a standardized (normalized)
way of spotting blunders.

But the test crippled Crafty by cutting off the search
at only 12 plys. In a game between two patzers, this
might be a minor flaw, but at the world championship
level, things are not always so simple.

Although this cutting off at a specific ply makes it
possible to duplicate the test on any computer, it may
have been more useful to use a fixed time instead
(provided the time is equal to or greater than the average
time to complete 12 plys).


-The fact that Riis found positional sacrifices not evaluated by
Crafty is not convincing since: (1) such positional sacrifices are
rare--as computers have shown,

No, they aren't. Only in games between patzers is
the intentional sacrifice of material for position "rare".


chess is largely tactics;

True. But not all tactics are visible at a depth of
only 12 plys. Tactics can flow from positional
advantage, with virtually no limit as to depth.


(2) everybody
will be judged equally by Crafty,

Misjudged would be more accurate.



so others pos sacs are also scored
'badly', so nobody will lose relative standing to one another,

Except at random, due to all the errors.



and (3), as long as assumption (1) is valid, Crafty will find the most
"mistake free" chess player, or one that plays closest to being
"tactics mistake free", which is a very good way to determine a good
chess player IMO.

But not good enough for these guys.

As all the world champions were good at tactics, it
requires a bit of subtlety to differentiate between them.


Now of course the surrebutter (rebuttal to the rebuttal) will be that
players like Tal will score poorly--and indeed they (he) did--but
let's face it, Tal was more of a shock player that relied on playing
the man rather than the board. In a match of coolheaded Karpov or
Kramnik versus Tal, all in their prime, the less emotional player is
likely to win

A silly statement. As we saw, the wild, attacking
style of GK gave GM Karpov a very hard time, except
for their very first match. How was GM Tal, in his
prime, all that different from GM Kasparov?

Another example was the cool, calm, collected
Bobby Fischer, who was overwhelmed by GM Tal in
his prime, and who calmly observed after the fact
that GM Tal's hyper-aggressive play was "unsound".



(unless he loses his cool and loses...haha... think of
Topolov vs Kramnik). Also nobody ever became champion ignoring
tactics.
That is the lesson of chess. Think of all the bogus moves
made by beginners, sacrificing knight for pawn, "to break up their
pawn chain", with no positional advantage. If you believe chess is
positional play more than tactics then such bogus moves should work
more often than they do. They do not.

It's not this simple. The world champions are all
competent at tactics, so the differences between
them are more subtle than just "who was the best
tactician".


So, understanding how chess works, and how chess playing computers
work, and having seen Crafty evaluate pretty good myself, I have to
side with the original article.

If you mean the one I think, it was horribly
skewered by a whole slew of critics under
"Reader's Feedback", in addition to all the
points made by the various critics who had
their articles published.

The primary issue is not that computers are
incapable of ranking the world champions by
accuracy, it is that attempting to do this with
a crippled Crafty and just the games from the
world championships is a poor method.

I would have preferred a deeper analysis by
a stronger program of all their important games,
in conjunction with a side-by-side subjective
analysis of the same games by a human GM
who, instead of tweaking the program to suit his
whims/preconceptions, simply comments on
where he thinks the program went astray.

The ideal might be the HAL9000 computer
"discussing" the games and results in plain
English, and giving "his" considered opinion
on the strengths and weaknesses of each of
the world champions, as seen by a program
rated (in the future) 9000 USCF. :D

-- help bot

On Apr 26, 2:00 pm, David Richerby
wrote:

In a match of coolheaded Karpov or Kramnik versus Tal, all in their
prime, the less emotional player is likely to win

Hmm... The two Botvinnik-Tal matches between them were only won by
Botvinnik +12-11=19. Hardly a convincing victory for the cool head.

In the first match, won by GM Tal, he often stood
worse out of the opening but maintained a cool head,
realizing the only chance was to complicate, apply
pressure to the opponent's King, to randomize the
position a bit. It was GM Botvinnik who choked,
rather than maintaining his coolness.

In the second match, the annoying attacks were
fended off in part by a switch to different openings
which were less conducive to GM Tal's wild,
attacking style. It should not be assumed that
GM Tal was a hothead, while his victims were all
coolheaded.

-- help bot

On Apr 26, 5:09 pm, Ron wrote:

That's total rubbish "Ron". You're obviously someone who doesn't know much
about the game of chess. Tal didn't set out to make errors, with the
lamebrain idea that this would somehow cause his opponents to make bigger
errors. Tal set out to create COMPLICATIONS for his opponents. Obviously Tal
desired for all of his sacrifices to be sound and forcing, but no human can
calculate everything to the end, so computer analysis has shown flaws in
many of his games. This is meaningless, because he wasn't playing against
computers.

Have you read Tal's books?

I have. There are many time when he says things like, "It's clear 36. f4
was stronger," (Tal-Gligoric, Zagreb 59), or see his note to 5. ... Qc7
in Tal-Olaffson, Bled 1961 (a move he describes as "bad" - but that he
clearly made intentionally) or, say, 10. a3 in Tal-Bagirov,
Dnepropetrosk, 1970, which he describes as "in no way stronger than the
approved Re1."

(I found these notes by basically opening "The Life and Games of Mikhail
Tal" at random. Stuff like this is all over that book. You should try
reading it sometime, before you talk about what Tal was, or wasn't,
thinking. His book on his match with Botvinnik goes into even more depth
on his thinking, again, and does a good job explaining the emphasis Tal
put of psychology over soundness. And what is psychology, in chess,
other than playing an inferior move which you think your opponent will
respond badly too. In particular, I'd point you to his discussion of his
12th move of game 17.)

It's clear from his notes that he doesn't care if his sacrifices were
"correct" or not. He made a move - which he knew could well be unsound -
with the expectation that in the resulting position his opponents would
play incorrectly.

Mr. Mitchell has made a serious error here in
equating game commentary after the fact with
what a player may have been thinking at the
time.

All of these games have been annotated --
often after looking at notations by others -- by
such players as GM Tal. This in no way means
that if, say, GM Kortchnoi said move x was
better and then GM Tal wrote in his book that
move x was better (since he agreed), that at
the time the game was played GM Tal *saw*
that move x was better, but deliberately chose
to play a stupid move instead!

On the contrary, in match one against GM
Botvinnik, GM Tal often points out that he had
to choose a different line because his old
choice had failed the last time out. In effect,
he readily admits when his openings were poor,
despite having won the game anyway.

Besides, those books on GM Tal, by GM Tal,
were all written by someone we now know to be
a horrible patzer, thanks to crippled-Crafty! :D

-- help bot