[Top][All Lists]

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

Re: [gnugo-devel] Re: global lookahead; distributed go; an experiment

From: bump
Subject: Re: [gnugo-devel] Re: global lookahead; distributed go; an experiment
Date: Mon, 17 May 2004 08:42:31 -0700

> ...which generates funny moves like a monkey jump inside your own territory
> on the side :)
> (as I saw in one of your posted games on the list)
> Apart from this, I propose the following experiment:
> A strong player plays against gnugo, but when it's his turn he is only
> allowed to choose among the
> Top 2,3,4,.. moves that gnugo would suggest for him.
> What is the maximum level of play (=num of handicaps) that is achievable for
> the human player?

Results from such an experiment a la Ingo Althofer were not impressive.

In that message I wrote:

> One problem was that if the top move is overvalued then that move that
> overvalued move will be proposed over and over again reducing the dreihirn's
> effective options to only one move. For much of the following game W
> wanted to play S5 before it was big enough.
> I can think of some strategies to avoid this.

What I had in mind is that if the top move is not selected, and if
on a subsequent move it is proposed with the same value as before,
that value could be deprecated.

> (perhaps implement a mode in gnugo, that when asked to generate a move also
> sends the top10 of a possible answer to the client (as a comment), along
> with the executed move)

In GTP the top_moves commands generate the top ten moves. The
metamachine tries to select among these. See


reply via email to

[Prev in Thread] Current Thread [Next in Thread]