Re[2]: Go program strength

[rbrown@xxxxxxxxxxxxxxx]

Nick Wedd <nick@xxxxxxxxxxxxxxxxx> writes:

> Tim Hunt <T.J.Hunt@xxxxxxxxxxxxxxxxx> writes
> 
> >There is also the theory that since the Computer Ing Cup and simmilar
> >events are the main testing ground for go programs, the programs are
> >designed to perform well there, rather than against humans. What do people
> >here think about that? Is it just slander, or is there some truth in it?
> 
> I think that there is truth in this.

Me too.

Tim, I have to ask, why did you choose the word "slander"?  Do you think that
programs should not be tuned to perform well against specific opponents?  Do
you regard this as being somehow immoral?  If it is true that programs in a
competetive event are in fact tuned specifically to perform well in that event,
do you think there is something shady about that approach?

I would regard this capability a desirable feature of a go-playing program.

I know that IBM's "Deep Blue" team was accused of doing this sort of tinkering
before chess games against Kasparov.  (It was discussed on this mailing list --
Is there an archive?)  Apparently, Kasparov took it quite personally.  At the
time, I could only say, "So what?"  I remember thinking that Kasparov must be
quite the poor sport to make a big deal out of this.  And I still think that.

Anybody here remember the sort of computer game where you were _required_ to
create a specific persona before you could play, and the program kept track of
your progress, your strengths and weaknesses?

What's wrong with a go program that knows who it's playing against?  If someone
said that about my program, I wouldn't call it slander, I'd call it praise!

Rich Brown
-- 
Richard L. Brown              Office of Information Services
rbrown@xxxxxxxxxxxxxxxxx        University of Wisconsin System Administration
rlbrown6@xxxxxxxxxxxxxxxxx    780 Regent St., Rm. 246  /  Madison, WI  53715