Re: computer-go: Please stop this pointless and blasphemous discussion !!!

[Don Dailey <drd@xxxxxxxxxxxxxxxxxxxxx>]

Hi Roland,

First of all,  I agree with you  about using the God/Devil analogy, it
IS likely to  offend some people.  I'll change  my terminolgy from now
on, ok?

I am going to continue  to talk about the  subject matter as long as I
get interesting feedback.  This is a  discussion group and I feel that
I am free to keep  make my contributions.   I'm not obligated to prove
anything to earn the right to talk about things.

I really like  the point you made  about group B, the players claiming
that we  are still far  away from perfect  play.   Yes, if group  B is
correct, it's certainly easier to prove.  If group A is correct, group
B can continue to harass them, demaning a proof!!  However in Go, both
groups have their  work cut out for   them if either  expects to prove
anything!!!


   4) Maybe a more practical approach would be to achieve a mathematical
   method of calculating the perfect move in a given position without
   having to wade through the entire game-tree. In fact this has already
   be done for the very late GO endgame and also for some chess endgames.
   The problem is the beginning and middle game. So maybe we should try
   to answer the following questions:

   ...


Yes, I am aware of  these things and  you have reminded  me of a  good
point.  There   are  endgames  in   Chess   that  are known   to    be
incomprehensible  to   Grandmasters.  No  Grandmaster   can play  them
correctly and will  quickly draw a won position  without having a clue
about why  what they  did  was wrong.   These  are in  simple endgames
composed of  only  5 pieces!  Now  this  involved positions that  were
extremely uncommon, not likely to occur in real games.  However ...

A few years ago, an extremely common ending was  discovered to be much
more profound than any one imagined.   It was simply KQ  versus KR.  A
grandmaster was asked to win this and  couldn't.  It was thought to be
relatively simple.  After much study, the Grandmaster finally mastered
this position.  But my point is that  this ending is quite common, has
only 4 pieces on the  board, and yet  Grandmasters didn't really fully
appreciate its complexity.  It  certainly doesn't take too much effort
to imagine that it just might be possible that other positions are far
more complex.



Don




   Date: Thu, 21 Sep 2000 02:19:30 -0300
   From: "Roland G." <goprog@xxxxxxxxxxxxxxxxx>
   X-Mailer: The Bat! (v1.45) UNREG / CD5BF9353B3B7091
   X-Priority: 3 (Normal)
   Mime-Version: 1.0
   Content-Transfer-Encoding: 7bit
   Sender: owner-computer-go@xxxxxxxxxxxxxxxxx
   Precedence: bulk
   Reply-To: computer-go@xxxxxxxxxxxxxxxxx
   Content-Type: text/plain; charset=us-ascii
   Content-Length: 3068

   Hello computer-go,

     I'm following this discussion about strength of best players and how
   far we are from "perfect" play ...and would like to say the following:

   1) first keep GOD and the Devil out of it!!! We are not discussing any
   religious issue here(or are we in fact doing so?), and the way of
   talking of some people is blasphemous in my opinion. There have been
   people claiming that Kramnik would win from God if he has a good
   position and others claiming that they would win from God if they have
   a pawn advantage at the beginning(how do they dare to say this?). You
   must first know that God has more resources than playing perfect: he
   could also complicate a position to induce mistakes or manipulate your
   mind(even if you are a computer) to make you lose or maybe travel in
   time to change a lost position into a winning one... And to some
   people here it may be offensive if you talk about God in this way,
   especially when you don't really know how far away from perfect play
   you are...So instead of referring to God or the Devil why don't use
   another term, like an abstract "perfect player" or "perfect
   game-play"?

   2) Second, I think the discussion is pointless because basically none
   of the two sides can prove their statements. The two sides are the
   following:

   a) players claiming that we are very close to perfect play.

   b) players claiming that we are still far away from perfect play.

   If you belong to a) prove your statement. If you belong to b) prove
   your statement also. Otherwise we will just continue to exchange
   arguments for one side and the other without ever ending the
   discussion. To me it seems also that the group b) is simpler to prove,
   because to do it we just would need a computer program that plays far
   better than any of the top players. But then again the question would
   arise if that computer program is close to perfect play or not...

   3) Maybe we should change the whole question to "How far away are we
   from perfect play?" In this case the answer would not be a simple yes
   or no but a numeric value. But I don't think that this reformulation
   would help us either, because as we still don't know what is perfect
   play it will always be difficult to compare our current playing
   ability with perfect play.

   4) Maybe a more practical approach would be to achieve a mathematical
   method of calculating the perfect move in a given position without
   having to wade through the entire game-tree. In fact this has already
   be done for the very late GO endgame and also for some chess endgames.
   The problem is the beginning and middle game. So maybe we should try
   to answer the following questions:

   -how can we calculate the perfect move in a given position without
   having to move through the entire game tree(its possible for the
   endgame already).

   -can we prove that the above is not possible if we are not in the
   endgame? In this case the only way to achieve perfect play would be to
   produce a complete game tree which I think will always remain
   impossible(?).

   -- 
   Best regards,
    Roland                          mailto:goprog@xxxxxxxxxxxxxxxxx