Subject: Re: adb & genetic algorithms
To: None <rhawkins@iastate.edu>
From: Ken and Masami Nakata <masami@fa2.so-net.or.jp>
List: port-mac68k
Date: 05/07/1997 07:03:00
On Tue, 06 May 97 13:42:34 -600,
rhawkins@iastate.edu (Rick Hawkins) wrote:
> I can pull up a few, but here's the rough descriptin.  They're usually 
> fairly quick to write.  I'll assume here that
> a) bsd has loaded enough of itself to run the program
> b) the search space is known
> c) that there is a way of measuring how successful the attempt was, with 
> intermediate states between success & fail.

This is an interesting idea for reverse engineering, but unfortunately
the search space is infinitely large in concept: i.e. you may have to
write or read any bit any times at any order.  When hardware I/O space
is N bit in size, you can't even assume the search space is 2 * N! 
because you may have to repeatedly read or write one bit many times.

IOW, even though we are interfacing the ADB chip with a small number
of bits, the state machine behind the bits may be any one of
infinitely many of such machines.

Ken