From: Matt Gingell (mjg223@nyu.edu)
Date: Sun Sep 05 1999 - 08:08:06 MDT
From: Spike Jones <spike66@ibm.net>
>Ja. Cameron, Deep Blue changed the way we think about the task of
>playing chess. Im suggesting that *everything* we do can be calculated
>brute force style, even if we dont fully understand how our minds do it.
>This is a strong statement: that given sufficient computing power, computers
>can enjoy, can feel, can love, etc, as soon as we develop the algorithms.
>Chess is not a unique example, nor speech recognition. This is profound
>as all hell, Cam, since the faster computers are coming, like a speeding
>train. No stopping it.
I think this under appreciates the computational cost of exponentially hard
problems. Given a string encrypted with a 1-gigabyte key, you’d be hard pressed
to find enough energy in the galaxy to crack it by brute force. Computers can
handle chess because the branching factor averages around 15, which means we
only need a factor of (a bit less than) 15 improvement in machine speed before
we can look one more move into the future in the same time. Other games are much
harder though – Computer’s have yet to play a good game of Go, and it’s possible
this game will never succumb to brute force.
It wouldn’t be difficult to infer the grammar of the English language from
untagged text using some sort of minimum description length learning algorithm,
if we had a big enough computer. We probably never will though.
-matt
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:05:02 MST