Hi all,

The other day I was analysing the game Popovic-Bagirov, Moscow 1989
(Chess Informant, Volume 47, game 159.) When I reached the position
White: Kg1 Bf5 Ph2,g5,e5 Black: Kf8, Ph7,g7,f6,b6, I switched on
Fritz 8.

Black, on move, is a piece down and struggling to draw.

After 1...fxg5 2.Bxh7 Kf7 3.Bf5! (safeguarding the e-pawn) Black
resigned.

If 1...g6 2.Bxg6! hxg6 3.gxf6, and White wins the pawn ending.

The main point of interest is what happens after 1...h6. In his
annotations Popovic gives 2.gxh6 (a little joke; 2.gxf6 comes to the
same thing) 2...gxh6 (or 2...fxe5 3.h7! and wins) 3.e6! With this
move White preserves the vital e-pawn and wins easily.

If instead 3.exf6? then Black draws with 3...Kf7! followed by
4...Kxf6. The point is that a rook pawn and a bishop cannot win
against a lone king if the bishop does not control the queening
square. This, of course, assumes that Black's king can safely arrive
there himself to set up a blockade. This is a fundamental piece of
endgame knowledge possessed by all serious players.

The question is, why doesn't Fritz "know" this? In the position after
2...gxh6 in the last variation above it insists on showing 3.exf6? as
a winning line. Should the program really have to analyse for many
moves and finally reinvent the wheel before changing its "mind" and
playing the correct move? How difficult is it to build in this sort
of endgame knowledge?

Cheers,
Dan

From: Dan Scoones
Date: 27/07/03 17:41 GMT Daylight Time
Message-id:

Hi all,

The other day I was analysing the game Popovic-Bagirov, Moscow 1989
(Chess Informant, Volume 47, game 159.) When I reached the position
White: Kg1 Bf5 Ph2,g5,e5 Black: Kf8, Ph7,g7,f6,b6, I switched on
Fritz 8.

Black, on move, is a piece down and struggling to draw.

After 1...fxg5 2.Bxh7 Kf7 3.Bf5! (safeguarding the e-pawn) Black
resigned.

If 1...g6 2.Bxg6! hxg6 3.gxf6, and White wins the pawn ending.

The main point of interest is what happens after 1...h6. In his
annotations Popovic gives 2.gxh6 (a little joke; 2.gxf6 comes to the
same thing) 2...gxh6 (or 2...fxe5 3.h7! and wins) 3.e6! With this
move White preserves the vital e-pawn and wins easily.

If instead 3.exf6? then Black draws with 3...Kf7! followed by
4...Kxf6. The point is that a rook pawn and a bishop cannot win
against a lone king if the bishop does not control the queening
square. This, of course, assumes that Black's king can safely arrive
there himself to set up a blockade. This is a fundamental piece of
endgame knowledge possessed by all serious players.

The question is, why doesn't Fritz "know" this? In the position after
2...gxh6 in the last variation above it insists on showing 3.exf6? as
a winning line. Should the program really have to analyse for many
moves and finally reinvent the wheel before changing its "mind" and
playing the correct move? How difficult is it to build in this sort
of endgame knowledge?

Cheers,
Dan


Generally speaking, programs are still useless at endgames which are not
tablebases. It's not so much that it's difficult to build tablebases. Rather,
because the complexity increases exponentially with each new piece that is
introduced, it demands a lot of time and storage space.

You may be interested to read the related thread "Huge EGTB Opportunities" (35
articles) started by Renze Steenhuisen, he

http://makeashorterlink.com/?J52332865

Best
Mark

"Mhoulsby" -remove- wrote

Generally speaking, programs are still useless at endgames which are not
tablebases.

That is absurd. Computers play most positions very well, including endgame
positions. There are of course some positions where the computer is
completley clueless, in all phases of the game (just like humans).

"Mhoulsby" -remove- wrote

You may be interested to read a conversation, between Gordon Rattray and
me,
which began with this post:

http://makeashorterlink.com/?I37126965

If, after reading this conversation, you still think my argument is
"absurd"
then feel free to make a case.

Of course, if one makes an opening book consisting entirely of white wins
in
one's own repertoire, in which black blunders a piece on move 7, and one
plays
white against a program under these conditions, one's chances of winning
are
increased.

Note, however, that I gave *specific examples* to support my argument.

If you can give *specific examples* which refute it, then I'll be most
interested to see them...

I read your discussion and saw nothing new.

First, many tests have been done to show that there is approximately zero
difference in playing strength between a chess program that uses endgame
tablebases and one that does not. This tells me that either A) computers
play most endgame positions very well on their own (since playing
"perfectly" doesn't show any improvement), or B) the number of practical
endgame positions that computers can't play well are very small and don't
often arise in a real game.

Are you aware that there are around 10^40 chess positions (give or take)?
You gave four positions as examples. Even if you gave a billion positions
that computers play horribly, I'd still have
99.99999999999999999999999999999% (yes, I calculated this) of the positions
as "*specific examples* to support my argument" that "Computers play most
positions very well." I'd say 99.99999999999999999999999999999% is "most",
wouldn't you? I believe the burden of proof is on you, and you've got quite
a job to do, since I doubt you have a billion positions as supporting
evidence. Even if you did, I'm still winning by a "few"
(99.99999999999999999999999999999%) :-)

"Mhoulsby" -remove- wrote

Interesting. So why bother with tablebases at all?

There are times to use them. For instance, if the current position on the
board is a tablebase position, then it is probably a good idea to use them
(especially with pawn endings). The debate over their usefullness starts
when people start using tablebase results during a search (while the
computer is "thinking"). For instance, let's say that the computer is close
to the endgame, and it is looking 15 half moves ahead. Maybe the first 10
half moves are NOT in the tablebases, but the last 5 moves are. This means
that the program is going to do a _lot_ of reading from the hard drive, and
reading from the hard drive is much slower than reading from the computer's
main memory. The problem is that those scores from the tablebases are 10
half moves away, so many of them won't be used. So there is an advantage of
using tablebases in the search (you have "perfect" knowledge), and there are
disadvantages (it is slower). The result is that using tablebases ends up
giving the same results overall, in the long run. There have been many
people experiment with this. I would suggest searching the Google newsgroup
archives or the Comptuer Chess Club archives for this information if you
want to know more.


Substantiate not just with near 100% statistics but with *practical
examples*
of programs playing endgames practically as well without tablebases as
they do
with tablebases...

You seem to miss the point. There _are_ practical examples that will give
support for and against tablebases, but that is beside the point. There are
two kinds of positions we can talk about here. In positions where you are
not actually in a tablebase position (but you're close), then what I
described above happens, and you get better "knowledge", but you search
slower, so the net effect is that the program doesn't play any stronger. The
other case is that you are in a position that _is_ in the tablebases, and
then it is a good idea to use the tablebases I think.

Most decent chess engines will play very close to the tablebases. If they
don't choose the "best" move, they still choose a winning move, and that is
enough. I think you will find relatively few cases where a computer
completely has no idea, and causes the game's result to change. What I mean
is, I can give you a position where the pawns are completly locked and it is
a clear draw because neither side can make any progress. You can give one
side an extra rook, and the computer will evaluate it as approximately +5
for the side with the rook, and people like you will go around waving their
hands shouting about how horrible computers are at the endgame. But, think
about that situation. The computer may misevaluate the position, but it is
still going to get a draw, just like the human, and tablebases aren't going
to help that. The number of positions where tablebases will change the
result of a game are very few. I'm sure they exist, and if you post a
hundred of them that doesn't mean computers play the endgame horribly.

Of all the phases of the game, computers play the opening the worst.
Fortunately most commercial programs have very good opening books :-)

From: "Russell Reagan"
Date: 29/07/03 03:59 GMT Daylight Time
Message-id: CqlVa.3137$Oz4.39@rwcrnsc54

"Mhoulsby" -remove- wrote

Interesting. So why bother with tablebases at all?

There are times to use them. For instance, if the current position on the
board is a tablebase position, then it is probably a good idea to use them
(especially with pawn endings). The debate over their usefullness starts
when people start using tablebase results during a search (while the
computer is "thinking"). For instance, let's say that the computer is close
to the endgame, and it is looking 15 half moves ahead. Maybe the first 10
half moves are NOT in the tablebases, but the last 5 moves are. This means
that the program is going to do a _lot_ of reading from the hard drive, and
reading from the hard drive is much slower than reading from the computer's
main memory. The problem is that those scores from the tablebases are 10
half moves away, so many of them won't be used. So there is an advantage of
using tablebases in the search (you have "perfect" knowledge), and there are
disadvantages (it is slower). The result is that using tablebases ends up
giving the same results overall, in the long run. There have been many
people experiment with this. I would suggest searching the Google newsgroup
archives or the Comptuer Chess Club archives for this information if you
want to know more.


I agree entirely with this, and it seems to support precisely what I was
arguing in the subthread to which my link was directed.

This is (part of the reason) *why* programs play badly in endgames - they are
in a quandary, do they go for a TB (if it's an absolutely forcing line which
leads to at least a draw [assuming the position to be roughly level] then the
answer, clearly, would be "yes"). In complex endgames, however, it's rare for
circumstances to be so clear-cut - which explains the quandary which you
describe.


Substantiate not just with near 100% statistics but with *practical
examples*
of programs playing endgames practically as well without tablebases as
they do
with tablebases...

You seem to miss the point. There _are_ practical examples that will give
support for and against tablebases, but that is beside the point.

No, it isn't. It's precisely the point.

If there were 32-man tablebases, then there would be no problem.


There are
two kinds of positions we can talk about here. In positions where you are
not actually in a tablebase position (but you're close), then what I
described above happens, and you get better "knowledge", but you search
slower, so the net effect is that the program doesn't play any stronger. The
other case is that you are in a position that _is_ in the tablebases, and
then it is a good idea to use the tablebases I think.


Clearly.

Most decent chess engines will play very close to the tablebases.

Really? I hadn't noticed (which may simply be my ignorance - if so, please post
examples).

If they
don't choose the "best" move, they still choose a winning move, and that is
enough. I think you will find relatively few cases where a computer
completely has no idea, and causes the game's result to change.

I posted a few in the subthread. There are more....


What I mean
is, I can give you a position where the pawns are completly locked and it is
a clear draw because neither side can make any progress. You can give one
side an extra rook, and the computer will evaluate it as approximately +5
for the side with the rook, and people like you will go around waving their
hands shouting about how horrible computers are at the endgame.

"People like me"? What does that mean. The above is self-contradictory. Here's
why: we have all 5-man and some 6-man tablebases. Now, I'm not sure which 6-man
tablebases exist, because I don't have them loaded on my HD. Let's assume,
however, that we *do* have the tablebase for, say: KPPkpp (which we don't,
since that would imply our having *all* 6-man TBs) How, then, might one
construct a position to which this tablebase pertains in which the pawns are
"completely locked" and neither side may make progress?

If one side has "an extra rook", can you construct *any* position from, say,
KRPkp in which the pawns are locked and the rook does not have the freedom of
the board.

This, I suspect, may be the point which *you* are missing...

But, think
about that situation. The computer may misevaluate the position, but it is
still going to get a draw, just like the human, and tablebases aren't going
to help that.

They would if they existed. That's the point. They don't.


The number of positions where tablebases will change the
result of a game are very few.

The number of positions with respect to which tablebases will change the result
of a game is indeed very few, precisely *because* (as yet) there are *so few
tablebases*. Ergo programs play badly in known endgames positions which are not
tablebases....

I'm sure they exist, and if you post a
hundred of them that doesn't mean computers play the endgame horribly.

Of all the phases of the game, computers play the opening the worst.

Absolutely.

Fortunately most commercial programs have very good opening books :-)


Yup.

I'm no expert on this matter like Mhoulsby and Russell Reagan seem to be but
I just think a computer has to calculate everything whereas in some
positions, for humans, it is 'obvious'.

For example a 16 move mate in some positions might be easier for a human to
calculate, indeed the computer might not calculate that far, it may only be
programmed to go 8 or 10 moves ahead.

So how can a human calculate a 16 move mate better than a computer?
Imagine this position, white king on e1, black king on e8, white pawns on a2
and h2.

Now we know that black can't stop both the pawns, he goes after one, the
other queens then king and queen mate the king. This is dead easy for us
humans, but a computer, unless it has this position programmed into it, has
to work it all out.

The example you quoted, where black ends up with a rooks pawn and wrong
coloured bishop to queen, well it seems to me unless this knowledge was
programmed into the machine - just a simple bit of logic to say bishop has
to be same colour as queening square of rooks pawn - then the computer has
to sit and work it out and the drawn ending you speak of may be 20 moves
down the line, the computer hasn't calculated that far.



"Dan Scoones" wrote in message
...
Hi all,

The other day I was analysing the game Popovic-Bagirov, Moscow 1989
(Chess Informant, Volume 47, game 159.) When I reached the position
White: Kg1 Bf5 Ph2,g5,e5 Black: Kf8, Ph7,g7,f6,b6, I switched on
Fritz 8.

Black, on move, is a piece down and struggling to draw.

After 1...fxg5 2.Bxh7 Kf7 3.Bf5! (safeguarding the e-pawn) Black
resigned.

If 1...g6 2.Bxg6! hxg6 3.gxf6, and White wins the pawn ending.

The main point of interest is what happens after 1...h6. In his
annotations Popovic gives 2.gxh6 (a little joke; 2.gxf6 comes to the
same thing) 2...gxh6 (or 2...fxe5 3.h7! and wins) 3.e6! With this
move White preserves the vital e-pawn and wins easily.

If instead 3.exf6? then Black draws with 3...Kf7! followed by
4...Kxf6. The point is that a rook pawn and a bishop cannot win
against a lone king if the bishop does not control the queening
square. This, of course, assumes that Black's king can safely arrive
there himself to set up a blockade. This is a fundamental piece of
endgame knowledge possessed by all serious players.

The question is, why doesn't Fritz "know" this? In the position after
2...gxh6 in the last variation above it insists on showing 3.exf6? as
a winning line. Should the program really have to analyse for many
moves and finally reinvent the wheel before changing its "mind" and
playing the correct move? How difficult is it to build in this sort
of endgame knowledge?

Cheers,
Dan

From: "Neil Coward"
Date: 29/07/03 09:43 GMT Daylight Time
Message-id:

I'm no expert on this matter like Mhoulsby and Russell Reagan seem to be but
I just think a computer has to calculate everything whereas in some
positions, for humans, it is 'obvious'.


I'm sorry but I must disagree with your generous assessment that I seem to be
an expert. I'm not. I couldn't program my way out of a paper bag.

For example a 16 move mate in some positions might be easier for a human to
calculate, indeed the computer might not calculate that far, it may only be
programmed to go 8 or 10 moves ahead.

So how can a human calculate a 16 move mate better than a computer?
Imagine this position, white king on e1, black king on e8, white pawns on a2
and h2.


This is a tablebase position. Therefore no program which uses tablebases would
*need* to calculate it at all.

The side without the pawns would resign (or its programmers would resign on its
behalf).

Now we know that black can't stop both the pawns, he goes after one, the
other queens then king and queen mate the king. This is dead easy for us
humans, but a computer, unless it has this position programmed into it, has
to work it all out.

The example you quoted, where black ends up with a rooks pawn and wrong
coloured bishop to queen, well it seems to me unless this knowledge was
programmed into the machine - just a simple bit of logic to say bishop has
to be same colour as queening square of rooks pawn - then the computer has
to sit and work it out and the drawn ending you speak of may be 20 moves
down the line, the computer hasn't calculated that far.


Whenever there are 5 pieces or fewer (i.e. 4 white, 1 black; 3 white, 2 black;
2 white, 3 black or 1 white, 4 black) *including the two kings* any program
using tablebases doesn't need to calculate at all. The *perfect* response to
*every possible move* is stored on the HD, enabling the program to play
perfectly and quickly.

In other words, "this knowledge" *is* "programmed into the machine" as you put
it.

It's the positions with 6 (for configurations of pieces without corresponding
tablebases) or more than 6 pieces with which programs have problems, since they
*are* difficult to calculate.

Cheers
Mark


"Dan Scoones" wrote in message
.. .
Hi all,

The other day I was analysing the game Popovic-Bagirov, Moscow 1989
(Chess Informant, Volume 47, game 159.) When I reached the position
White: Kg1 Bf5 Ph2,g5,e5 Black: Kf8, Ph7,g7,f6,b6, I switched on
Fritz 8.

Black, on move, is a piece down and struggling to draw.

After 1...fxg5 2.Bxh7 Kf7 3.Bf5! (safeguarding the e-pawn) Black
resigned.

If 1...g6 2.Bxg6! hxg6 3.gxf6, and White wins the pawn ending.

The main point of interest is what happens after 1...h6. In his
annotations Popovic gives 2.gxh6 (a little joke; 2.gxf6 comes to the
same thing) 2...gxh6 (or 2...fxe5 3.h7! and wins) 3.e6! With this
move White preserves the vital e-pawn and wins easily.

If instead 3.exf6? then Black draws with 3...Kf7! followed by
4...Kxf6. The point is that a rook pawn and a bishop cannot win
against a lone king if the bishop does not control the queening
square. This, of course, assumes that Black's king can safely arrive
there himself to set up a blockade. This is a fundamental piece of
endgame knowledge possessed by all serious players.

The question is, why doesn't Fritz "know" this? In the position after
2...gxh6 in the last variation above it insists on showing 3.exf6? as
a winning line. Should the program really have to analyse for many
moves and finally reinvent the wheel before changing its "mind" and
playing the correct move? How difficult is it to build in this sort
of endgame knowledge?

Cheers,
Dan

There does seem to be an occasional evaluation bug in Fritz8- at least
in the infinite analysis mode.

A number of times, Fritz has given an obviously bad move a (= 0.00)
rating, independently of how long I wait. But if I play the move, it
immediately shows it to be worth as low as -6. This is always
characterized by the display not showing any moves for this particular
move - just the first move. Clearly it is not kept as part of the
analysis. This has always happens as much as I can remember when the
position being analyzed was losing.

Unfortunately I did not save the positions, but it has happened more
than once or twice. It is not so serious in analysis, since it is so
obvious, but if Fritz does this while playing, then it could affect its
performance (unless it only happens when already losing).

Henri

"henri Arsenault" wrote

There does seem to be an occasional evaluation bug in Fritz8- at least
in the infinite analysis mode.

A number of times, Fritz has given an obviously bad move a (= 0.00)
rating, independently of how long I wait. But if I play the move, it
immediately shows it to be worth as low as -6. This is always
characterized by the display not showing any moves for this particular
move - just the first move. Clearly it is not kept as part of the
analysis. This has always happens as much as I can remember when the
position being analyzed was losing.

Unfortunately I did not save the positions, but it has happened more
than once or twice. It is not so serious in analysis, since it is so
obvious, but if Fritz does this while playing, then it could affect its
performance (unless it only happens when already losing).

This might be a situation where a computer uses kind of lazy repetition
detection. Many programs do this. Basically they don't check for 3-fold
repetition. Instead, they only check for 2-fold repetition, and if a
position occurs twice, the computer assumes that it will occur a third time.
The computer works in a way such that it assumes both players are playing
perfectly (it's not really true, but that is how the computer "thinks" about
its move). To learn more about why this might have occured, read Bruce
Moreland's webpage:
http://www.brucemo.com/compchess/pro...repetition.htm