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

Re: computer-go: Two ways to program a GO-engine



----- Original Message -----
From: "Heikki Levanto" <heikki@xxxxxxxxxxxxxxxxx>
> Vlad Dumitrescu <vladdu@xxxxxxxxxxxxxxxxx> wrote:
> >> >   1. Try to make a program that achieves perfect or near perfect
play.
> >> >   2. Try to make a program that plays like the best human players.
> >>
> I, Heikki, proposed:
> >>  3. Try to make a program that beats more and more of its opponents.
> >>  4. Try  to make a program that learns to play go.

I am sorry, I didn't express myself well enough. What I meant was that the
main purpose for developing a Go program is one of the two: either to
"solve" the game, or to study the human mind and its processes.

"Solving" the game is in my opinion a search for the perfect player - of
course, one has to settle for what one manages to get and fight to improve
on that result, but the aim is for a player as good as possible (= perfect).
To this goal, any method of solving is acceptable - including those that we
can't easily reverse engineer to extract knowledge to be used in human play
(like neural networks or genetically evolved algorithms).

The other option is for those more interested in learning techniques, human
perception and behaviour, abstraction layers, linguistic theory and such.
For those, Go is just a very good study case, it could have been anything
else sufficiently complicated.

Of course, the distinction line is fuzzy not only when it comes to practical
implementations. Myself for example I place myself in the gray zone, being
interested in both areas.

/Vlad