"Eliezer S. Yudkowsky" <firstname.lastname@example.org> writes:
> Hal Finney wrote:
> > This would seem to indicate that the early universe, despite being
> > exceedingly hot and dense by our standards, was not a good environment
> > for evolution.
> > Without energy differences there is no way for life to exist.
> I don't believe this is true. Our kind of life requires energy
> differences, yes. But the laws of thermodynamics aren't really laws;
> they're statistical guidelines. There is no physical reason why you
> can't take tap water and produce electricity and ice cubes; it's just
> very very improbable. The laws of physics are time-symmetrical (except
> for state-vector reduction); if the atoms are in the right places with
> the right velocities, a glass of water can leap from the floor and unshatter.
The problem is the probabilities. Life based on improbable events will be wiped out by the far more probable noise; there are lots of orders of magnitude between them.
One way of seeing why life in a state with no arrow of time is unlikely is that it cannot evolve, since evolution includes an irreversible selection step where the unfit are weeded out. In a reversible world individuals could un-die and and there would be no evolution. In the same way memory and learning would be impossible, since forgetting would be just as likely as learning.
> In fact, information is directly proportional to
> entropy. The more information it takes to encode a system, the more
> mathematically random ("normal") that system is, the more entropy that
> system has. Ergo, maximally efficient information-processing takes
> place at uniform temperature.
Be careful about the word "information", it has different meanings in different fields. The information in information theory is rather different from the information concept used in daily life and not necessarily even identical to negentropy.
> The question we have to ask is not, "Are there visible computations?",
> but "Are there computations?" How much computing power would it take to
> accurately and precisely simulate the Big Bang? If the answer is
> "infinity", then any computation which has a finite chance of taking
> place ought to occur. Actually, there are some caveats to that, since
> there has to be an infinite number of interactions between particles;
> moreover, there has to be an infinitely long linear sequence of
> interactions, with each interaction affecting the next interaction.
But remember that there were particle horizons; at time t only particles within each others past lightcone could have affected each other.
> Anyway, the question isn't whether there are macroscopic computations
> (of the sort we're used to) taking place, but whether arbitrary
> computations can be encoded in the physical process of the Big Bang.
That is a different questions. There are arbitrary computations encoded in the thermal vibrations in my desk. Somewhere it is running my old ZX81 fractal program...
-- ----------------------------------------------------------------------- Anders Sandberg Towards Ascension! email@example.com http://www.nada.kth.se/~asa/ GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y