Don Dailey wrote:
The utility of CGOS wouldn't change except for the better. The way it
works is this:
After each game:
1. The ill-behaved bot gets rated by the full formula.
2. The opponent of the ill-behaved bot gets a smaller adjustment
than usual.
Excellent, that makes a lot more sense than what I'd imagined, that the
ratings weight were symmetrical.
The adjustment for the opponent would approach zero if the ill-behaved
bot had a serious problem. The utility of CGOS for such a bot isn't
very high anyway since the results for it have little meaning. So
the rating adjustment for the opponent should be correspondingly small,
reflecting that the results is almost meaningless for the opponent.
That seems only fair.
Quite.
Go ahead, play mirror go if you wish. What is your answer to 1. e5
?
p15! It would only really make sense on a large board, where the opening
is less tactical, and more strategic. On 9x9 it will almost always make
sense to break out at the start.
Or what is your first move if you are black?
Black is easiest to play mirror go with, since you start at tengen. If
you're white, your opponent can at any time break out of mirror go by
playing there. But I doubt many computer players would do this. Which is
where I think some people would object; they might feel pressured into
putting mirror-go-detecting code into their bots, which might be a break
with purity/elegance.
> I don't think you
have much to work with but you are welcome to try it. CGOS is open to
every bot - as long as you are not malicious or buggy in an annoying
way. I would prefer that bot's that play random moves do not
play, but that is a request not a rule.
Ah, but I'm sure some would see mirror go as malicious. Some even see it
that way in human-vs-human play. But I'm glad to hear that experimenting
with such things isn't discouraged :)
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
