[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: computer-go: Good Play (was FPGA)



My guess (and  it's only a guess)  is that we  vastly overestimate the
strength of human go players compared to GoGod or GoDevil.

I'm pretty new at go, but I have observed that the literature seems to
make GODS out of the top go players.

>From my knowledge of the top chess players I see  that they tend to be
quite cocky with  respect to their own  play.  I believe that if chess
players are  quite   far from Godly  play,  then  Go players are  even
farther away.  I am sure even Kasparov is quite far from perfect play.
I  don't know how  to translate ELO   rating points to stones however!
Maybe 2 stones is a huge amount!

Take one more step down to checkers.  The top players in this game are
even  pretty far removed  from  perfect play  and this game  is vastly
simpler than chess.   In checkers,  the top  players draw  much of the
time, and   the  difference is  measured  by the  occasional  win over
another player.  And yet this differentiates them signficantly.

Currently, the  best player seems to  be a  computer, Chinook.  If you
read  the authors book  or talk to him,  you will see that he believes
Chinook has a long way to go!


Don



   Date: Fri, 22 Sep 2000 22:26:32 +0200
   From: Robert Jasiek <jasiek@xxxxxxxxxxxxxxxxx>
   Organization: jasiek
   X-Mailer: Mozilla 3.01Gold (Win95; I)
   MIME-Version: 1.0
   References: <a4.9396707.26eb0ac8@xxxxxxxxxxxxxxxxx>
   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: 816

   Compgo123@xxxxxxxxxxxxxxxxx wrote:
   > FYI. Some time ago there was a very good post on this list concerning how
   > perfect the top Go players play. The common point of view from the top Go
   > players is 1 to 2 stones from the perfect play.

   Although no proof either, consider a position with n kos of different
   shapes in it using the superko rule. You won't know what to do even if
   getting n/4 handicap stones. It is theoretically possible to convert a
   starting position into such a complex ko position. For this play need
   not be perfect since you also do not see all such possible whole
   board positions with many kos that could occur. Thus you cannot prevent
   all of them. All GoDevil needs to do is to create some such position
   with many kos. You would need enough handicap stones to prevent most kos
   at all.

   --
   robert jasiek