[Computer-go] How many probes down the tree are necessary for a "good" bot?

Erik van der Werf erikvanderwerf at gmail.com
Fri Nov 15 02:36:17 PST 2013


On Fri, Nov 15, 2013 at 12:59 AM, Darren Cook <darren at dcook.org> wrote:
>>> ... I'm going by the "Measuring program strength" thread (Aug
>>> 3013) where people are getting 50% against GnuGo with 1K to 10K playouts.)
>>
>> Actually, in that thread Hiroshi was saying he only needs 350
>> playouts, and Detlef was at 700. So it seems you're off by roughly a
>> factor 10.
>
> Yes, I was surprised just how heavy the Aya (and Zen) playouts must be.
> We've almost come full loop, and soon the programs will be traditional
> computer go programs doing a single playout ;-)

You and Peter seem to focus quite a bit on the 'heavy' playouts while
it really isn't so much about that. The reason why strong programs
have to consider very few moves is knowledge applied in the tree
(selection phase). Simply stated, if you know a move is bad then you
don't have to run simulations on it.


> It is a shame the developer's skill is all dead knowledge (not open
> source, no papers). :-(

The developers aren't dead, so the knowledge isn't dead. Further, I
think most (if not all) of the required knowledge is already 'out
there'; you just need the skills to put things together in the right
way (and figure out which things don't work -- the typical academic
paper won't tell you that).

Erik



More information about the Computer-go mailing list