Mike Murray wrote:
If perfect play results in a forced win for White (the advantage of
the move proves decisive) or for Black (the initial position turns
out to be zugzwang), then the side with the disadvantage could play
perfectly and still lose.

If one side (let's say White) can force a win with perfect play and
proceeds to do so, it doesn't really make sense to ask if Black was
playing perfectly. A strategy is perfect if it obtains the best
possible result from any position. If White has a forced win, the
best possible result for Black from any position that White will allow
to occur is a loss (for Black) so any move, including `Resigns', is
`perfect'.

You could argue that losing in a hundred moves is preferable to losing
in three but that's pretty subjective -- there's nothing in the rules
that says that a quick win is in any way preferable.


Dave.

--
David Richerby Evil Gnome (TM): it's like a smiling
www.chiark.greenend.org.uk/~davidr/ garden ornament but it's genuinely
evil!

"Larry Tapper" wrote in message
ups.com...
David Kane asks:

A simple question. List as many of the possible reasons
for draws between GMs you can think of *and* the
empirical evidence necessary to confirm *each* one.


Seems to me that the main reason for decisive games between GMs is the
fact that a win is better than a draw. Likewise the main reason for
draws between GMs is the fact that a draw is better than a loss.

I should expect empirical research to confirm this bold conjecture.
For example, given a choice between a drawing line and a losing line,
I'd wager that a GM would consistently pick the former. If he sees it,
which he often will.

LT


In my opinion, these conjectures are not useful. For example, many
games/sports have draws, but they don't occur as often as they
do in chess. In some games, e.g. Tic-Tac-Toe, they occur more
often.

The data to be explained is the following. GM play has a high draw
rate. Why? If you think about it, it's *not* an easy question to
resolve. I would be interested in your opinions on the matter.

To my mind, there can be a number of explanations, and they
must be resolved by evidence.

1. The "Tic-Tac-Toe theory". Players contest their games, but
perfect chess is a draw, and GMs play perfectly.
2. Players contest their games, play imperfectly, but play equally. Their
blunders
balance out and the difference in play is not large enough to overcome
chess' inherent winning threshold.
3. Players don't contest their games. They agree to draws without trying to
win,
collaboratively steer the game into drawish lines etc.

In case 3, there is an additional question of *why* they don't fully contest
their games. There could be a number of explanations as to why they
don't.

Of course, we can be pretty sure that all of these explanations account
for some GM draws. The goal is to figure out the exact proportions.

In practice, of course, the issue of whether game is truly contested
is not an easy yes or no. Suppose I play all out to win but find myself
in a position a pawn up in a complicated position but low on time.
I know that I'm terrible in time pressure and don't see a clear way to win.
My opponent doesn't know my weakness in time pressure, and
moreover, sees that I have a winning line. From each side's perspective
it may be rational to draw, even though perfect play is a win for me.
This drawn game is certainly not fully contested, but nor could it be
considered uncontested. It's somewhere in between.

Whether the decision to draw is rational depends on the
scoring system. If in the above example, I have the
Black pieces and am playing in a tournament
with BAP scoring, White would have no incentive to
draw and the game would be played out, i.e. more chess.
Alternative scoring proposals can be thought of as a way
to increase the amount of chess per game. It not only eliminates
the chessless "GM draw", which nobody but Mr. Houlsby likes,
but generally increases the amount of chess per game.

On Mar 16, 1:07 pm, "David Kane" wrote:
"Larry Tapper" wrote in message

ups.com...





David Kane asks:

A simple question. List as many of the possible reasons
for draws between GMs you can think of *and* the
empirical evidence necessary to confirm *each* one.

Seems to me that the main reason for decisive games between GMs is the
fact that a win is better than a draw. Likewise the main reason for
draws between GMs is the fact that a draw is better than a loss.

I should expect empirical research to confirm this bold conjecture.
For example, given a choice between a drawing line and a losing line,
I'd wager that a GM would consistently pick the former. If he sees it,
which he often will.

LT

In my opinion, these conjectures are not useful. For example, many
games/sports have draws, but they don't occur as often as they
do in chess. In some games, e.g. Tic-Tac-Toe, they occur more
often.

David,

Actually my earlier post, to which you are responding here, was meant
to be a bit of gentle needling. But irony doesn't register very well
in newsgroups without the liberal use of those awful emoticons. ;-!

Still, your outline of the issues below strikes me as reasonable.
Sometimes, though, I think your rhetoric flies too high. Those who are
satisfied with the status quo are not necessarily "lovers" or
"worshippers" of the bloodless draw. Speaking for myself, I don't
especially love draws, but I do think that it's logical and
appropriate for the number of points scored in a chess game to add up
to a constant, presently 1. So I don't like the BAP system at all.
This is not to discourage tournament organizers from experimenting
with BAP --- my guess is that if this were a commonly used format, its
defects would soon become apparent.

Consider Anand's winning performance at Morelia/ Linares. He played
very well --- his wins included an especially fine game against
Carlsen. Having built up a solid lead, he then coasted to first place,
taking fewer risks in the Linares leg of the tournament.

I have no complaints about that --- do you? To me it's just a part of
chess that high-level players are canny utility maximizers and don't
necessarily turn every game into a violent slugfest.

Larry T.




The data to be explained is the following. GM play has a high draw
rate. Why? If you think about it, it's *not* an easy question to
resolve. I would be interested in your opinions on the matter.

To my mind, there can be a number of explanations, and they
must be resolved by evidence.

1. The "Tic-Tac-Toe theory". Players contest their games, but
perfect chess is a draw, and GMs play perfectly.
2. Players contest their games, play imperfectly, but play equally. Their
blunders
balance out and the difference in play is not large enough to overcome
chess' inherent winning threshold.
3. Players don't contest their games. They agree to draws without trying to
win,
collaboratively steer the game into drawish lines etc.

In case 3, there is an additional question of *why* they don't fully contest
their games. There could be a number of explanations as to why they
don't.

Of course, we can be pretty sure that all of these explanations account
for some GM draws. The goal is to figure out the exact proportions.

In practice, of course, the issue of whether game is truly contested
is not an easy yes or no. Suppose I play all out to win but find myself
in a position a pawn up in a complicated position but low on time.
I know that I'm terrible in time pressure and don't see a clear way to win.
My opponent doesn't know my weakness in time pressure, and
moreover, sees that I have a winning line. From each side's perspective
it may be rational to draw, even though perfect play is a win for me.
This drawn game is certainly not fully contested, but nor could it be
considered uncontested. It's somewhere in between.

Whether the decision to draw is rational depends on the
scoring system. If in the above example, I have the
Black pieces and am playing in a tournament
with BAP scoring, White would have no incentive to
draw and the game would be played out, i.e. more chess.
Alternative scoring proposals can be thought of as a way
to increase the amount of chess per game. It not only eliminates
the chessless "GM draw", which nobody but Mr. Houlsby likes,
but generally increases the amount of chess per game.- Hide quoted text -

- Show quoted text -

On 16 Mar 2007 16:42:55 +0000 (GMT), David Richerby
wrote:

Mike Murray wrote:
If perfect play results in a forced win for White (the advantage of
the move proves decisive) or for Black (the initial position turns
out to be zugzwang), then the side with the disadvantage could play
perfectly and still lose.

If one side (let's say White) can force a win with perfect play and
proceeds to do so, it doesn't really make sense to ask if Black was
playing perfectly. A strategy is perfect if it obtains the best
possible result from any position. If White has a forced win, the
best possible result for Black from any position that White will allow
to occur is a loss (for Black) so any move, including `Resigns', is
`perfect'.

You could argue that losing in a hundred moves is preferable to losing
in three but that's pretty subjective -- there's nothing in the rules
that says that a quick win is in any way preferable.

In any position, most players would define "perfect" play in a lost
position as that posing the most problems for the opponent, giving the
opponent the most chances to go wrong. Your definition is fine if
one abstracts the element of competition from the problem, i.e., if
one assumes the opponent is capable of finding those perfect moves the
position demands, and can make these moves in the game within time
limits, etc. So, facing a choice of theoretically lost positions, a
GM forcing his opponent into one of the weird Queen endings for which
a forced mate in 144 (or some such thing) exists in the tablebases,
would be playing "better" than if he fell on his sword into a mate in
four, even though the outcome would be the same were he facing God or
Fritz XXX.

At any rate, since we were arguing against Houlsby's claim that "Every
lost position results from a decisive mistake. No decisive mistake, no
lost position, drawn game", my paragraph cold be rewritten as

"If perfect play results in a forced win..., then the side with the
disadvantage could make no errors and still lose."

which seems to finesse your objection.


Dave.

"Larry Tapper" wrote in message
oups.com...
On Mar 16, 1:07 pm, "David Kane" wrote:
"Larry Tapper" wrote in message

ups.com...





David Kane asks:

A simple question. List as many of the possible reasons
for draws between GMs you can think of *and* the
empirical evidence necessary to confirm *each* one.

Seems to me that the main reason for decisive games between GMs is the
fact that a win is better than a draw. Likewise the main reason for
draws between GMs is the fact that a draw is better than a loss.

I should expect empirical research to confirm this bold conjecture.
For example, given a choice between a drawing line and a losing line,
I'd wager that a GM would consistently pick the former. If he sees it,
which he often will.

LT

In my opinion, these conjectures are not useful. For example, many
games/sports have draws, but they don't occur as often as they
do in chess. In some games, e.g. Tic-Tac-Toe, they occur more
often.

David,

Actually my earlier post, to which you are responding here, was meant
to be a bit of gentle needling. But irony doesn't register very well
in newsgroups without the liberal use of those awful emoticons. ;-!

I did detect the irony.

Still, your outline of the issues below strikes me as reasonable.
Sometimes, though, I think your rhetoric flies too high. Those who are
satisfied with the status quo are not necessarily "lovers" or
"worshippers" of the bloodless draw. Speaking for myself, I don't
especially love draws, but I do think that it's logical and
appropriate for the number of points scored in a chess game to add up
to a constant, presently 1. So I don't like the BAP system at all.
This is not to discourage tournament organizers from experimenting
with BAP --- my guess is that if this were a commonly used format, its
defects would soon become apparent.

I actually do *not* consider it *logical* to count draws as 0.5. It's
traditional, and it would be a reasonable default assumption if we
knew nothing about chess. But it also leads to some horrible
situations in practice.

And I don't dispute that BAP or its equivalent could well have
problems that we don't expect, as well as the ones that we know
that it does have. The question in my mind is, on the whole, which
makes chess better. The only way that I can think of to find out
is to try.

Consider Anand's winning performance at Morelia/ Linares. He played
very well --- his wins included an especially fine game against
Carlsen. Having built up a solid lead, he then coasted to first place,
taking fewer risks in the Linares leg of the tournament.

I have no complaints about that --- do you? To me it's just a part of
chess that high-level players are canny utility maximizers and don't
necessarily turn every game into a violent slugfest.

My complaint is that it isn't very interesting. Why wouldn't the same
high level players also be canny utility maximizers if the scoring
encouraged more chess per game? Wouldn't it be more interesting
if something were actually on the line when the titans face each other?
You seem to suggest that there is something natural about employing
a strategy consisting of "coasting to victory by a string of draws"
to win tournaments. What I'd like to get you to consider is that
there is nothing "natural" about it - it's the organizers' (dubious
from a marketing perspective) *decision* to make that a
viable strategy.


The data to be explained is the following. GM play has a high draw
rate. Why? If you think about it, it's *not* an easy question to
resolve. I would be interested in your opinions on the matter.

To my mind, there can be a number of explanations, and they
must be resolved by evidence.

1. The "Tic-Tac-Toe theory". Players contest their games, but
perfect chess is a draw, and GMs play perfectly.
2. Players contest their games, play imperfectly, but play equally. Their
blunders
balance out and the difference in play is not large enough to overcome
chess' inherent winning threshold.
3. Players don't contest their games. They agree to draws without trying to
win,
collaboratively steer the game into drawish lines etc.

In case 3, there is an additional question of *why* they don't fully contest
their games. There could be a number of explanations as to why they
don't.

Of course, we can be pretty sure that all of these explanations account
for some GM draws. The goal is to figure out the exact proportions.

In practice, of course, the issue of whether game is truly contested
is not an easy yes or no. Suppose I play all out to win but find myself
in a position a pawn up in a complicated position but low on time.
I know that I'm terrible in time pressure and don't see a clear way to win.
My opponent doesn't know my weakness in time pressure, and
moreover, sees that I have a winning line. From each side's perspective
it may be rational to draw, even though perfect play is a win for me.
This drawn game is certainly not fully contested, but nor could it be
considered uncontested. It's somewhere in between.

Whether the decision to draw is rational depends on the
scoring system. If in the above example, I have the
Black pieces and am playing in a tournament
with BAP scoring, White would have no incentive to
draw and the game would be played out, i.e. more chess.
Alternative scoring proposals can be thought of as a way
to increase the amount of chess per game. It not only eliminates
the chessless "GM draw", which nobody but Mr. Houlsby likes,
but generally increases the amount of chess per game.- Hide quoted text -

- Show quoted text -

Mike Murray wrote:
In any position, most players would define "perfect" play in a lost
position as that posing the most problems for the opponent, giving
the opponent the most chances to go wrong.

That's not `perfect' play but `best' play. The point about perfect
play is that my opponent doesn't have any problems: he knows exactly
how to checkmate me given any possible move I could make.


Your definition is fine if one abstracts the element of competition
from the problem, i.e., if one assumes the opponent is capable of
finding those perfect moves the position demands, and can make these
moves in the game within time limits, etc.

Any such discussion of perfect play is almost forced to make these
assumptions: by assuming that the opponent plays perfectly, we are
necessarily assuming that he is able to make the moves. If he doesn't
make the perfect moves, he isn't playing perfect chess!

As such, discussions about perfect play are inherently theoretical.
Your comments are entirely practical and they're absolutely correct as
far as practical chess is concerned.


At any rate, since we were arguing against Houlsby's claim that
"Every lost position results from a decisive mistake. No decisive
mistake, no lost position, drawn game", my paragraph cold be
rewritten as

"If perfect play results in a forced win..., then the side with the
disadvantage could make no errors and still lose."

which seems to finesse your objection.

Agreed. (Though one could say that the error was bothering to play a
known-lost game. ;-) )


Dave.

--
David Richerby Edible Devil Shack (TM): it's like a
www.chiark.greenend.org.uk/~davidr/ house in the woods that's possessed
by Satan but you can eat it!

On 16 Mar, 17:07, "David Kane" wrote:
"Larry Tapper" wrote in message

ups.com...





David Kane asks:

A simple question. List as many of the possible reasons
for draws between GMs you can think of *and* the
empirical evidence necessary to confirm *each* one.

Seems to me that the main reason for decisive games between GMs is the
fact that a win is better than a draw. Likewise the main reason for
draws between GMs is the fact that a draw is better than a loss.

I should expect empirical research to confirm this bold conjecture.
For example, given a choice between a drawing line and a losing line,
I'd wager that a GM would consistently pick the former. If he sees it,
which he often will.

LT

In my opinion, these conjectures are not useful. For example, many
games/sports have draws, but they don't occur as often as they
do in chess. In some games, e.g. Tic-Tac-Toe, they occur more
often.

The data to be explained is the following. GM play has a high draw
rate. Why? If you think about it, it's *not* an easy question to
resolve.

Sure it is. Chess is inherently a draw, so "draw" is a good result to
obtain, all things being equal.

I would be interested in your opinions on the matter.

To my mind, there can be a number of explanations, and they
must be resolved by evidence.


Read an Informator.

1. The "Tic-Tac-Toe theory". Players contest their games, but
perfect chess is a draw, and GMs play perfectly.

Ummm... GMs *don't* play perfectly. They do draw though. Unless they
lose.

2. Players contest their games, play imperfectly, but play equally. Their
blunders
balance out and the difference in play is not large enough to overcome
chess' inherent winning threshold.

Chess doesn't have an inherent winning threshold, unless the one side
or the other makes a decisive mistake, so it can't be that.

3. Players don't contest their games. They agree to draws without trying to
win,
collaboratively steer the game into drawish lines etc.


Right. They do this, in large measure, because chess is a draw,
inherently. Of course they also do it to split prize money, catch an
early train, whatever.

In case 3, there is an additional question of *why* they don't fully contest
their games. There could be a number of explanations as to why they
don't.


There could, and indeed there are. I've listed the most important
ones. The fact that chess is a draw is the most important of all.

Of course, we can be pretty sure that all of these explanations account
for some GM draws.

No, we can't...entirely, since 2) is hogwash.

The goal is to figure out the exact proportions.


Is it? Why would anyone want to do that?

In practice, of course, the issue of whether game is truly contested
is not an easy yes or no. Suppose I play all out to win but find myself
in a position a pawn up in a complicated position but low on time.
I know that I'm terrible in time pressure and don't see a clear way to win.
My opponent doesn't know my weakness in time pressure, and
moreover, sees that I have a winning line. From each side's perspective
it may be rational to draw, even though perfect play is a win for me.

Perfect play *after a decisive mistake*, yes.

This drawn game is certainly not fully contested, but nor could it be
considered uncontested. It's somewhere in between.


In the sense that no opening variation has been fully played out, no
draw is "fully contested", by this definition.

Whether the decision to draw is rational depends on the
scoring system.

Not *only* the scoring system.

If in the above example, I have the
Black pieces and am playing in a tournament
with BAP scoring, White would have no incentive to
draw and the game would be played out, i.e. more chess.

More bad chess, yes.

Alternative scoring proposals can be thought of as a way
to increase the amount of chess per game. It not only eliminates
the chessless "GM draw", which nobody but Mr. Houlsby likes,

Where did I indicate that I like such draws? You really need to learn
to read.

but generally increases the amount of chess per game.

The amount of bad chess, yes.

On 16 Mar, 19:22, "David Kane" wrote:
"Larry Tapper" wrote in message

oups.com...





On Mar 16, 1:07 pm, "David Kane" wrote:
"Larry Tapper" wrote in message

roups.com...

David Kane asks:

A simple question. List as many of the possible reasons
for draws between GMs you can think of *and* the
empirical evidence necessary to confirm *each* one.

Seems to me that the main reason for decisive games between GMs is the
fact that a win is better than a draw. Likewise the main reason for
draws between GMs is the fact that a draw is better than a loss.

I should expect empirical research to confirm this bold conjecture.
For example, given a choice between a drawing line and a losing line,
I'd wager that a GM would consistently pick the former. If he sees it,
which he often will.

LT

In my opinion, these conjectures are not useful. For example, many
games/sports have draws, but they don't occur as often as they
do in chess. In some games, e.g. Tic-Tac-Toe, they occur more
often.

David,

Actually my earlier post, to which you are responding here, was meant
to be a bit of gentle needling. But irony doesn't register very well
in newsgroups without the liberal use of those awful emoticons. ;-!

I did detect the irony.

Still, your outline of the issues below strikes me as reasonable.
Sometimes, though, I think your rhetoric flies too high. Those who are
satisfied with the status quo are not necessarily "lovers" or
"worshippers" of the bloodless draw. Speaking for myself, I don't
especially love draws, but I do think that it's logical and
appropriate for the number of points scored in a chess game to add up
to a constant, presently 1. So I don't like the BAP system at all.
This is not to discourage tournament organizers from experimenting
with BAP --- my guess is that if this were a commonly used format, its
defects would soon become apparent.

I actually do *not* consider it *logical* to count draws as 0.5.

In that case, you're being *illogical*, since, clearly, chess is a
draw.

It's
traditional, and it would be a reasonable default assumption if we
knew nothing about chess.

It's even more reasonable for those who *do* understand chess--GMs--
evidently.

But it also leads to some horrible
situations in practice.


Says who?

And I don't dispute that BAP or its equivalent could well have
problems that we don't expect, as well as the ones that we know
that it does have. The question in my mind is, on the whole, which
makes chess better.

Better play makes chess better, clearly. In order to produce their
best in a tournament, GMs make short draws sometimes. To do otherwise
would lower the quality of the games. Not that anyone who doesn't
understand an Informator annotation would notice....


The only way that I can think of to find out
is to try.


Good idea! Try to become a GM, then you'll find out that chess is a
draw.

Consider Anand's winning performance at Morelia/ Linares. He played
very well --- his wins included an especially fine game against
Carlsen. Having built up a solid lead, he then coasted to first place,
taking fewer risks in the Linares leg of the tournament.

I have no complaints about that --- do you? To me it's just a part of
chess that high-level players are canny utility maximizers and don't
necessarily turn every game into a violent slugfest.

My complaint is that it isn't very interesting.

So go do something else. Chess isn't for you. You're too dumb.

Why wouldn't the same
high level players also be canny utility maximizers if the scoring
encouraged more chess per game?

It's because chess is *inherently* a draw. Become a GM, and you'll
find out why.

Wouldn't it be more interesting
if something were actually on the line when the titans face each other?

That's right, it wouldn't be. More exciting for morons, maybe, but
definitely not more interesting that top-level chess.

You seem to suggest that there is something natural about employing
a strategy consisting of "coasting to victory by a string of draws"
to win tournaments.

There *is* something natural about it. Chess *is* a draw. Read
Nimzowitsch's essay. Read an Informator. Become a GM and you'll find
out why.

What I'd like to get you to consider is that
there is nothing "natural" about it -

Of course there is, idiot. Does *nothing* sink in with you? You're too
ignorant and too dumb to be expressing such opinions in public.
Really.

it's the organizers' (dubious
from a marketing perspective) *decision* to make that a
viable strategy.


No, it's the *fact* that chess is a *draw*.

On 16 Mar, 13:41, David Richerby
wrote:
Mark Houlsby wrote:
I said no such thing, you illiterate moron. I indicated that *with
perfect play chess is a draw* and that, knowing this, GMs are
content to draw. That's all. No mention of perfect play *by
GMs*. Idiot.

GMs do not *know* that chess is drawn with perfect play.

Sure they do.

They may
well strongly suspect it and the suspicion may well be true. But, at
the moment, nobody knows whether chess is a draw, a win for White or a
win for Black with perfect play.


Sure they do. Read an Informator.

I have to say, you look rather foolish calling somebody an `illiterate
moron' and an `idiot', when you can't even get the facts straight that
are central to your argument.


Ummm... I have to say that *you* have a habit of not getting *your*
facts straight which makes *you* look like an idiot, which you do,
again, now.

Mark.

Dave.

--
David Richerby Cyber-Monk (TM): it's like a man ofwww.chiark.greenend.org.uk/~davidr/ God that exists only in your computer!

On 16 Mar, 15:43, Mike Murray wrote:
On 15 Mar 2007 05:40:35 -0700, "Mark Houlsby"

wrote:
Read an Informator. Every lost position results from a decisive
mistake. No decisive mistake, no lost position, drawn game. It's very
simple.

Finding the best move is in principle a matter of analysis, not
statistics, although the latter can be useful in determining practical
chances.

Correct.

A move employed hundreds of times in GM praxis will be
ash-canned if someone finds a refutation.

Correct. Until someone refutes the refutation.

So, Informator is not the
critical resource for this problem.


No, that's a non-sequitur. Informator indicates that games containing
decisive mistakes tend to be lost, and that games not containing
decisive mistakes tend to be drawn, is the point.

If perfect play results in a forced win for White (the advantage of
the move proves decisive)

....which is impossible...

or for Black (the initial position turns out
to be zugzwang),

....which is also impossible...

then the side with the disadvantage could play
perfectly and still lose.


....except that this is impossibility owing to the impossibility of the
conditions attached to it.


Therefore, the claim that every lost position results from a decisive
mistake begs the question as to whether perfect play leads to a draw.

Another non-sequitur. Perfect play *does* lead to a draw.

Discuss.