>> Note that after tournament we can check "what would be the score" if
>> there would be no truncation.
> 
> You can check afterwards, but the programs should be adapting their
> behaviour to the rules, so they will play for a safe +50 rather than a
> risky +75.

This is a very good point. It may have the opposite of the desired
effect, meaning it may encourage bots to go with a dynamic komi
approach, rather than the more revolutionary (and therefore more
interesting) min-max approach.

As for convention, there are two established similar systems that I know of:

  1. BankNeki (used by gamblers, mainly in Korea)
   http://senseis.xmp.net/?BangNeki
    Here the cut off is 91 points.

  2. Hahn, as used on Little Golem
   http://www.littlegolem.net/jsp/games/gamedetail.jsp?gtid=go19&page=rules
   Here the cut off is 41 points.

Notice that both use a step system, rather than the actual points. (E.g.
in both cases a win of 1 point is worth as much as a win of 10 points.)

At first glance that seems to spoil the whole point; we want to motivate
our bots to quibble over the last point. But the real reason for these
rules, for humans at least, is to encourage more life/death fights.

Darren


-- 
Darren Cook, Software Researcher/Developer

http://dcook.org/work/ (About me and my work)
http://dcook.org/blogs.html (My blogs and articles)


_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go

Reply via email to