Re: Reversible Computation and Experience

From: hal@finney.org
Date: Thu May 10 2001 - 11:20:01 MDT


I'm sorry I don't have time to organize my thoughts better, but let me
note that although the Brownian computer is not the only possible model
of reversible computing, it does work in principle (and some biological
processing of DNA can be modelled rather well as a reversible Brownian
computation).

Hence the puzzles that I posed regarding the nature of conscious
experience on such a computer (which may step backwards as well as
forwards) are still valid and must be addressed. After all it is possible
that our universe is being simulated on such a computer right now. Do we
notice? Maybe you un-read and re-read this message half a dozen times.
What impact would this have on our experience of the universe?

Some people on Wei Dai's everything-list propose that each possible
observer-moment (for all possible conscious observers) makes up a set
and that we randomly experience observer-moments chosen from that set.
Hence we are more likely to experience observer-moments with a larger
measure. The regularity of the universe is explained by the fact that
the programs to generate lawful observer-moments similar to what we
see are much shorter and hence more probable than programs which would
generate random, chaotic observer-moments.

A consequence of this view is that observer-moments which execute
more often would be more probable. Hence we are more likely to be
experiencing observer moments that have been executed multiple times.
This might suggest that Brownian implementations would be preferred and
make it more likely than not that we are running on a Brownian computer.

(Wei Dai suggested that another consequence is that observer-moments
running on large computers would have larger measure than those running on
small computers. This addresses the paradox that running a calculation
multiple times in parallel can be the same as running the calculation
a single time on a larger computer.)

Hal



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:07:35 MST