[Computer-go] replacing dynamic komi with a scoring function

Don Dailey dailey.don at gmail.com
Mon Jan 9 06:32:28 PST 2012

On Mon, Jan 9, 2012 at 8:01 AM, terry mcintyre <terrymcintyre at yahoo.com>wrote:

> Just kicking out an idea: when you offer a handicap to a player, the
> presumption is that he or she or it is significantly weaker than yourself.
> Can that weakness be modeled in the playouts? A 5d player tends to know
> subtle things about shape and joseki which a 1d player does not; tends to
> read a little more deeply and accurately. Can a playout engine be tuned to
> represent a generic 1d player?

Your suggestion strikes at the heart of the matter.   If you could do
accurate opponent modeling you would get more realistic scores.   Realistic
in the sense that they would more accurately represent your chances of

I was thinking of a similar solution to yours but I wasn't thinking
directly in terms of opponent modeling but instead some modification to the
playouts that is more in line with how a player would maximize his chances.
    However I realize that we still have a lot to learn about playout
strategy as it is and so this may be a very tall order!     I doubt this
would be simple.

You asked if a playout could be tuned to represent a 1d player,  but I
don't think playouts are nearly as strong as a 1d player.   The playout
strategy is a black art as they say and I think it's been established that
the goal is not necessarily to make the playouts stronger but to make them
better predictors.

> Can such tuning be dynamically changed as the game progresses and more
> information is obtained about both the player and the game itself?
> _______________________________________________
> Computer-go mailing list
> Computer-go at dvandva.org
> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://computer-go.org/pipermail/computer-go/attachments/20120109/341c200b/attachment.html>

More information about the Computer-go mailing list