From: hal@finney.org
Date: Tue Feb 22 2000 - 14:24:25 MST
Billy Brown, <bbrown@transcient.com>, writes:
> The fact that chemical reactions tend to increase entropy in the long run
> does not necessarily imply that they destroy information in this sense. If
> you know what the reaction is, and you know what the reactants and the end
> products look like, you can just look at the end result and say "oh, this is
> decay reaction X, so this protein must have stated out like so..." Doing
> the same thing when you have lots of different reactions going on is more
> complicated, but that just increases the computational requirements. As
> long as the end state is uniquely determined by the initial state and the
> know history of the system, you can in principle compute the initial state.
Sure, in principle, but then you are back to Eugene's super-gods who can
reconstruct everyone from atmospheric molecule motions. Once information
has degraded into heat, there is no practical technology which is going
to be able to re-create that information. And such degradation is exactly
what happens in chemical reactions that increase entropy. Distinguishable
states transform into indistinguishable states. Macro scale information
(at least at the molecular and structural level) changes into micro
scale information (meaning heat).
If you grant the impossibility of a technology which can analyze random
molecular thermal motion and "play it back" to recreate the entire past
in microscopic detail, then you must accept that an increase in entropy
is a loss of information. From the final state there are multiple initial
states which could have produced it, and the only way to distinguish them
is in the micro-information of the thermal motions. Put another way,
from decay products alone you cannot construct a unique set of initial
compounds which would have decayed to form these results, in general.
I maintain that given our current state of knowledge, we simply don't
know whether there is sufficient information in frozen tissue to
reconstruct its initial state to any particular degree of precision.
Hal
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:26:58 MST