In a message dated 7/22/99 2:23:48 PM Mountain Daylight Time,
bharper@postbox.acs.ohio-state.edu writes:
> Whether Davies is using the word random correctly or not is
> not really very interesting, IMHO. The point really is that
> the nonrandomness (technical sense) of the process is what
> guarantees small information content. To see this consider the
> extreme case of a completely nonrandom (completely deterministic)
> process wherein there is a single outcome. In Shannon information
> theory the information content is reflected by the number of messages
> in an ensemble of messages. If there is only one outcome, then the
> number of messages in the ensemble is one and the information
> content is zero. Yes, it's counter-intuitive, but this is why random
> sequences have the highest information content.
>
According to Daniel Brooks, Edward Wiley and John Collier, communication
theory, as examplified by Shannon/Weaver theory, is inappropriate to account
for biological information, including molecular self-assembly. Their theory,
as described in _Evolution as Entropy_ by Brooks and Wiley, demonstrates that
one can have an increase in order and complexity simultaneous with an
increase in information and entropy, and that this process is driven at
higher levels of organization by reductions in randomness rather than
increases in randomness. So it is in fact theoretically possible for more
information to be created by a nonrandom process, provided that the kind of
information being created is physical array information controled by
physiochemical laws, which is what biological information is.
>
> We also arive at the same general conclusion within the context
> of algorithmic information theory (AIT). Here, the information
> content is the length of the shortest algorithm which
> performs some task. Since natural laws are generally short
> and concise, they can be described by short algorithms and
> thus have small information content.
>
Our **description** of natural laws may be short and concise, but I would
dispute that the natural laws themselves are. In fact, what we usually call
a natural law is actually part of a larger phenonemon, which usually requires
a theory or a more complex law to unite all the seemingly disparit parts into
one comprehensive whole.
>
> Kevin:==
> >The problem with this claim is that it assumes life has always been
> >"information - rich"; it is more likely it started out "information -
poor"
> >and evolved to its present level of "richness", which is what the
> >experimental results in fact demonstrate.
>
> The only "complaint" I might have about this statement is the
> degree to which experimental results demonstrate the evolution
> from poor--->rich.
>
If you agree that a modern cell is information rich, then by comparison a
protocell is information poor. What is interesting, though, is that a
protocell is able to do as much as a modern cell or more with its small
amount of information than a modern cell can do with its greater amount of
information.
>
> One way of looking at this is that the results of info-theory
> may give an indication of the appropriate direction for further
> research. Let's suppose that one is convinced that the first
> stages of the origin of life are deterministic. Not everyone
> believes this, but Fox certainly did. Let's suppose also that
> one thinks that information theory is telling us something
> important. That deterministic laws are information-poor.
>
Brook/Wiley/Collier theory, itself a form of information theory, states just
the opposite: deterministic laws are information-rich, because they provide
the instructions by which precursor biomolecules can nonrandomly
self-assemble into macromolecules.
>
> This would then lead one in the direction that you indicate.
> A search for evolutionary mechanisms that would increase the
> information content. Eigen's hypercycles are an example of
> such a search, though I believe that search fell short of the
> mark :).
>
Brooks/Wiley/Collier theory predicts that these mechanisms are simply
emergent features of biological information as a hierarchical physical array
operating according to the known physiochemical laws.
>
> But this general notion in no way negates what Davies said,
> unless, of course, one views the process of evolution itself
> as being deterministic.
>
At the risk of beating a dead horse, your general notion may not negate what
Davies said, but the experimental results do. Which should be the entire
point.
>
> Kevin:
> >The way thermal copolymerization works is that the shape and chemical
> >nature of the amino acids determines which amino acids will bind together.
> >This is controlled by the known physiochemical laws. I tend to doubt
that
> >information theory really proves that the physiochemical laws cannot
create
> >life, but if it does then there is research that refutes this proof.
>
> Ah yes, very good. There is a lot of misunderstanding about what
> this result from information theory really means. To use info
> theory terminology, it means that life is undecidable given only
> the natural laws. Creationists have misunderstood the result to
> mean that life requires input from an intelligent agent. The
> information cannot arise, so the argument goes, from the action
> of natural laws, so it must be added from an external source.
> Actually, this seems to assume some type of conservation law
> for information. Anyway, what some fail to understand is that
> undecidable means undecidable :).
>
Again, that may be the conclusion of some forms of information theory, but
the experimental results demonstrate otherwise, and Brooks/Wiley/Collier
theory provides the theoretical framework for explaining the experimental
results.
>
> Since information theory deals with information directly
> and only indirectly with life, let me rephrase what you
> say above. What we should say is that it is uncertain whether
> the physiochemical laws, acting alone, can create information.
>
Since according to Brooks/Wiley/Collier theory biological information is
physical both in the sense of being material (ie, real, not abstract like
Shannon information) and an actual physical property of an organism or a
species, there must be a hierarchical physical array made up of units of
specific structural and chemical nature, which vary from one level to the
next. The physiochemical laws work through these units, so in that sense the
physiochemical laws cannot create information by themselves. However, since
new biological information is created solely by the action of the
physiochemical laws through these units -- ie, nothing else is needed -- in
that sense the physiochemical laws acting alone can indeed create
information. And again, this is demonstrated by experimental results.
>
> "It is not certain that all is uncertain, to the glory of
> skepticism." -- Pascal ;-)
>
> Undecidability is not nearly so mysterious as it sounds.
> Chaitin has shown, for example, that there is undecidability
> even in pure mathematics. If there, why not in biology? :)
>
It might help if you could precisely define "undecidability". ;-)
Kevin L. O'Brien