[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: computer-go: Applying Moore's Law to Computer Go

At 12:48 AM 11/14/99 -0600, you wrote:

...
Computer Go strength grew 1 rank each 20 months
during the last decade.  I reckon this trend will continue
in part for the foreseeable future.
maybe for a few years, but i suspect go strength is not linear, its
probably exponential.

  Particularly, let us
speculate growth to continue 1 rank each 18 to 30 months
for another 2 decades.  For example, in 2020 Computer
Go strength could reach 4k* to 1d* (IGS scale).
this would be about japanese amateur 1k-4d (i think). i seem to recall that
one program has an official japanese rank of amateur 3k. so this seems
reasonable (at least the 3k does). i am only american 1d, but it's real
hard to imagine a program that could beat me. i seem to recall that you are
a Dan player and a programmer also ( i think i met you at one of our
"cotsen open" tournaments?). can you imagine a program that could beat you?

... A combination of programming
efficiency and ever increasing colaboration may cover
the software side of Computer Go strength dependency.
i don't think so, the numbers are just too big. humans will still have a
*huge* advantage in the first third of the game.

thanks

Ray (will hack java for food) http://home.pacbell.net/rtayek/
hate spam? http://www.blighty.com/products/spade/