Richard wrote:
On Jul 24, 10:53 am, "
wrote:
There it is. How on earth can it be possible to
solve 21 out of 156 checkers openings instead of
156 out of 156 UNLESS they have not really
solved checkers completely? This also would
seem to explain the strange comments posted
here earlier, which suggested the same thing by
misusing the term solved to mean partly-solved.
This is a misinterpretation of the results. The Chinook team proved
checkers is a draw from *the starting position* or so called "go as
you please" play. The 156 checker openings are ballots used in the 3-
move restriction style of play, where the first three plies (2 red
moves, one white move) are pre-determined. The Chinook team is not
saying that they have solved all of these, but they don't need to in
order to claim that unrestricted checkers is a draw.

Exactly. To put it in chess terms, this would be like a computer
program that proves that it can always win or draw at chess with the
white pieces. Every game, it starts with 1. e4, and it never loses,
then help bot comes along and says "But it didn't prove that chess is
solved, because it's not proven for games starting with 1. d4."

--Richard



Not quite. It's like a computer program that proves that it can always
draw with 1. e4 as White AND CAN ALSO always draw with Black (after 1.
any)...and then help bot comes along and (correctly) says "this means
that chess is 'weakly solved', but not yet 'solved'".


--
Kenneth Sloan
Computer and Information Sciences +1-205-932-2213
University of Alabama at Birmingham FAX +1-205-934-5473
Birmingham, AL 35294-1170 http://www.cis.uab.edu/sloan/

"help bot" wrote in message
oups.com...
On Jul 24, 12:59 pm, "David Kane" wrote:

|The new research proves that Chinook is invincible in traditional
|checkers. In most tournament play, however, a match now starts with
|three moves chosen at random. In solving the traditional game, the
|researchers have also solved 21 of the 156 three-move openings,
|leaving some hope for humans.

There it is. How on earth can it be possible to
solve 21 out of 156 checkers openings instead of
156 out of 156 UNLESS they have not really
solved checkers completely?

Because you can force a draw without reaching
the positions that they haven't calculated.


But this only proves that Chinook is invincible, which
is not the same as "solving" checkers; if you mean
only solving for the result (win, loss, draw), fine. But
to me, solving is more than just that. In chess, one
could claim to have calculated the best line of play in
every conceivable opening, and that it is a draw, but
this leaves much to be desired; what is desired is to
know the best move in every position AND the result
with best play AND maybe also the distance to
conversion. If we must sacrifice something, let it be
only the massive move trees, which understandably,
cannot be kept watered and fertilized and free of
insects forever.


Here's what makes me question all these accounts:
every one which didn't merely summarize, but instead
included direct quotes, inserted all sorts of qualifiers
which pulled back from the headline (Checkers is solved!).
Real science requires no such waffling of this sort. In
real science, a headline which reads "Gold from lead!"
would read simply as a process in which lead is
converted into gold (without need to add platinum or
diamonds or even silver).


seem to explain the strange comments posted
here earlier, which suggested the same thing by
misusing the term solved to mean partly-solved.

In traditional checkers, also called
Go As You Please (GAYP), any
legal move is possible. The version of
the game that starts from one of 156
positions selected at random and has
not (yet) been solved. In fact, the
openings contained in the proof are
not the best in practical play.


Same with chess!


I think the conclusion is that these guys are
confident that no human or computer will ever
beat their machine from now on, and are moving
on, satisfied with less than perfection.

No. Checkers has been understood to
be a draw for at least a century


So has chess, but this was merely an assumption,
not a fact.


and computers have been unbeatable for ~10.


You must mean unbeaten. Obviously, any
program which is not perfect can be beaten, if
you simply write a superior program.

This is not correct. If the weaker
program can always find the drawing line, then it
will never lose. In fact, Chinook does *not*
play "perfectly". If its' opponent plays
an inferior move which allows a win, Chinook
may *not* find it. That's not part of what
has been proven.


Another
way is to have the program play itself (thinking
on opponent's time = off), and give one side
much more time than the other.


What has been
done here is to document the path from starting
position to draw. So GAYP is a proven draw.


Okay. The link posted earlier went to a checkers
program where the user was invited to test his skill
against the improved Chinook program; this was not
very convincing, and reminded me of somebody
setting up a chess Web site and inviting all comers
to try and beat him (i.e. Rybka). Obviously, anyone
could do that and no one alive could even hope to
win a single game.

That's not what the Chinook website claims to be
doing, even if it superficially resembles that example.
It's documenting the drawing path.


The closest thing I have seen to a proof is one
statement that 10 men were solved, and an engine
of some sort cranked away from the opening while
accessing this database... yet they did not claim
to have tackled the whole analysis tree, and this
fell right in line with all the other accounts which
pulled away from the headline.


In one of the earliest linked-to articles, the word
solved was used interchangeably with partly-solved,
or as we in the chess world would understand it, not
really solved at all... .

Your misunderstanding is not common to all
chessplayers.



-- help bot

help bot wrote:

But this only proves that Chinook is invincible, which
is not the same as "solving" checkers; if you mean
only solving for the result (win, loss, draw), fine. But
to me, solving is more than just that.

What the word "solved" means to you is of no importance.

The accepted definition is as follows:

"Ultra-strongly solved" or means that there is an algorithm
which leads to a win for one player or a draw or a game that
goes on forever, against any possible moves by the opponent,
starting from any position, even ones that cannot be reached
from the initial position.

"Strongly solved" or "completely solved" means that there
is an algorithm which leads to a win for one player or a
draw or a game that goes on forever, against any possible
moves by the opponent, starting from any position that can
be reached from the initial position, even if one or both
players has already made one or more mistakes.

"Weakly solved" means that there is an algorithm which leads
to a win for one player or a draw or a game that goes on
forever, against any possible moves by the opponent, starting
from the initial position only.

"Ultra-weakly solved" means that it has been proven whether
the first player will win, lose, draw or play on forever,
against any possible moves by the opponent, starting from
the initial position only, but there is no algorithm telling
us how to do it. Typical proofs of this class are strategy
stealing arguments -- proving that if player two has a winning
strategy player one can always steal it. For example, change
chess to allow white to "pass" (make no move) or to swap
colors as his first move and chess is ultra-weakly solved:
If there is a forced win or draw for white, the first player
plays it. If there is a forced win for black, the first player
passes or swaps colors. Thus we know that the first player
can always draw or win, but we don't know how.

When the word "solved" is used without a qualifier, the
default assumption is "weakly solved."

I hope this helps.

...but I have a feeling that it won't.


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

Kenneth Sloan wrote:

It's like a computer program that proves that it can always
draw with 1. e4 as White AND CAN ALSO always draw with Black
(after 1. any)...and then help bot comes along and (correctly)
says "this means that chess is 'weakly solved', but not yet
'solved'".

In general, when a game is said to be solved, the default meaning
is usually assumed to be weakly solved.

Guy Macon wrote:
Kenneth Sloan wrote:

It's like a computer program that proves that it can always
draw with 1. e4 as White AND CAN ALSO always draw with Black
(after 1. any)...and then help bot comes along and (correctly)
says "this means that chess is 'weakly solved', but not yet
'solved'".

In general, when a game is said to be solved, the default meaning
is usually assumed to be weakly solved.




A matter of taste - which I do not agree with. This is a relatively
recent trend, in my opinion.


--
Kenneth Sloan
Computer and Information Sciences +1-205-932-2213
University of Alabama at Birmingham FAX +1-205-934-5473
Birmingham, AL 35294-1170 http://www.cis.uab.edu/sloan/

Guy Macon wrote:
Richard wrote:

Exactly. To put it in chess terms, this would be like a computer
program that proves that it can always win or draw at chess with the
white pieces. Every game, it starts with 1. e4, and it never loses,
then help bot comes along and says "But it didn't prove that chess is
solved, because it's not proven for games starting with 1. d4."

You have posted a clear and simple explanation. So of course
help bot won't understand it...


I don't understand it either. Where in the checkers proof does it say
that only ONE SIDE is assured of a draw?



--
Kenneth Sloan
Computer and Information Sciences +1-205-932-2213
University of Alabama at Birmingham FAX +1-205-934-5473
Birmingham, AL 35294-1170 http://www.cis.uab.edu/sloan/

Kenneth Sloan wrote:

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

Richard wrote:

Exactly. To put it in chess terms, this would be like a computer
program that proves that it can always win or draw at chess with the
white pieces. Every game, it starts with 1. e4, and it never loses,
then help bot comes along and says "But it didn't prove that chess is
solved, because it's not proven for games starting with 1. d4."

You have posted a clear and simple explanation. So of course
help bot won't understand it...


I don't understand it either. Where in the checkers proof does it say
that only ONE SIDE is assured of a draw?

When I read it, I assumed that everyone reading it would see the
obvious other half, that the second player has several of the single
best moves described above -- one for every possible initial move by
the first player.

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