Re: computer-go: What makes a ladder

[Heikki Levanto <heikki@xxxxxx>]

Don Dailey <drd@xxxxxxxxxxxxxxxxx> wrote:

> But a  rule is still far cheaper,  if you can  find one  that actually
> works.

Yes, but that is a big "if".

Actually, the cost of reading all possible variations in a ladderis not that
high, as the branching factor is so low (1-2), and the terminal conditions
are so easy to detect: The runnig stones got captured, or got three
liberties or more.

> Most of the  programmers  have complained that  evaluation  is
> really expensive, so   I  don't see where  it  hurts  trying  to  find
> something cheaper.  

Of course it depends what you are trying to do. If you are trying to build
a function that can be evaluated a million times a second, and that will
give an acceptable result more often than not, then some sort of "simple"
rule may make sense.  But if you want to know if a stone in a ladder can be
captured or not, I fear there is no cheaper way than to read it out. After
all, it is not much more expensive than just following the path of the
ladder and seeing what it will meet.

> Even if you don't succeed, you are likely to learn something.

Yes, of course.  If you do, let us know.  But don't be too disappointed if
we know it already, or do not find it easy to believe it without further
explanation.

Best of luck, anyway

  -H


-- 
Heikki Levanto     LSD Levanto Software Development   heikki@xxxxxxxxxxxxxxxxx
               "In Murphy we Turst"