Re: computer-go: GoeMate is now 1st in 13x13-CGoT

[Don Dailey <drd@xxxxxxxxxxxxxxxxxxxxx>]

   Date: Fri, 1 Feb 2002 09:12:55 +0000
   From: Nick Wedd <nick@xxxxxxxxxxxxxxxxx>
   References: <001901c1aa3f$0b55abb0$093ba8c0@xxxxxxxxxxxxxxxxx>
    <200202010539.AAA10030@xxxxxxxxxxxxxxxxx>
   MIME-Version: 1.0
   X-Mailer: Turnpike Integrated Version 5.01 M <IH17mcgxAnvOd8HDYCVHLMeIhr>
   Sender: owner-computer-go@xxxxxxxxxxxxxxxxx
   Precedence: bulk
   Reply-To: computer-go@xxxxxxxxxxxxxxxxx
   Content-Type: text
   Content-Length: 1151

   In message <200202010539.AAA10030@xxxxxxxxxxxxxxxxx>, Don Dailey
   <drd@xxxxxxxxxxxxxxxxx> writes

   < snip >
   >could prove my point.   The computer they didn't like  was known to be
   >the best at the time and dominated all  the other computers back then.
   >But it took a long time  to convince them otherwise,  they just had to
   >see several  wins  (the  better computer  rarely  lost but  they  kept
   >believing it was "lucky.")

   This suggests that the better computer understood something which the
   experts did not.

   Many backgammon players, playing for the first time against Snowie or
   Jellyfish, are convinced that the program is lucky with its dice;  some
   insist that it is cheating with the dice throws.  In fact, backgammon
   programs are better than most humans at "duplication": arranging the
   position so as to maximise the number of dice throws which will be good
   for themselves, and minimise the number which will be good for their
   opponents.

   In Go, I rather doubt that any experts, even 5-kyu experts, will be
   impressed by a Go program's "luck" at present.  But it will be
   interesting if they are.

But I would bet that even great go players could not look at the games
of several different computers and rank them with any accuracy.

They actually do try to rank amateur  tennis players using a system of
watching them hit a few  balls and filling out   a checklist of  which
kinds  of shots or strategy they  seem to have  a mastery of.  I don't
think they pretend to be accurate, but the idea is to get a rough idea
of how good an unknown player is.  I have seen  this system be way off
too because some players are truly ugly, but know how to win.



   Nick
   -- 
   Nick Wedd    nick@xxxxxxxxxxxxxxxxx


Your  observations  with  backgammon do  not   surprise  me.   Another
manifestation of this  phenomenon which I  have seen in chess  is when
strong players  lose due to a  blunder.  They invariably discount that
as a playing strength factor.  Even worse, they feel that pointing out
how silly  the error really was somehow  makes it clear  that it was a
completely freak occurrence and should have  no bearing on the outcome
of the game  (the blunder should  be taken back or  a new game started
without the game counting  simply because it's perfectly obvious  they
"should not" have made that error.)

The unknown factor in  the judgements of  playing strengths by experts
is that even though they themselves  may be incredibly strong players,
they are not consciously aware of all the factor that make them so.  I
have seen many examples where a strong player could not articulate why
a given move  is  great unless there  is  an obvious tactical  reason.
They simply know from years of experience.

Don