Re: computer-go: minimax and go

[Don Dailey <drd@xxxxxxxxxxxxxxxxxxxxx>]

   From: "Clay ChipSmith" <weiqi@xxxxxxxxxxxxxxxxx>

   >A brute force program will "work."   Give me infinite computing power,
   >or enough computing power  to search the  entire game tree and I  will
   >write a perfect Go program.

   Oh, you are so right.

   >Short of that, I can  write a Go program
   >that plays better with every hardware doubling of power.  It might not
   >play much better,  it might even  be hard to  measure but it will play
   >better on the average  with each doubling, until  it reaches the point
   >where it can search the entire game tree.


   An uninteresting argument at best.  Doubling every 2 years would require
   over 60 years to catch a similar search depth as chess.

So what?  I didn't make any claims about how long  it would take, only
that it was scalable.   

Maybe you have a better idea?   How would you write  a program on that
computer we will have in 60 years that doesn't do any kind of search?

Don