This Bradford commment following a blog entry by Mike Gene at Telic Thoughts is the focal point for this post:
When you absorb information through the symbols you see on your computer screen you know the sender's pet cat or infant toddler cannot be implicated. That type of chance event is ruled out by probability. But what about selection? If a mutation in a unicellular organism can generate an adaptive response then why could a similar process not generate an initial genome? The reason lies in the properties of functional genomes. In contrast to "emergent properties" these ones are identifiable. Selection is linked to replication. If a self-replicating molecule is the object of study the only valid assumption that can be made, from a selection POV, is that changes disrupting replication will not be passed on. One cannot even assume that changes that make a replicator a more efficient one will be selected without knowledge of the supply of relevant nucleotides or AAs. A theoretical change that enhances efficiency might exhaust a natural supply source. In any case there is still no directional indicator pointing toward a cell. There is no theoretical support for von Neumann's insight when "information" and its replicating mechanism are one and the same. The type of materialist preconceptions represented by self-replicating molecules are inconsistent with what we actually observe. We observe encoding conventions and symbolic representations and accurately attribute their cause to intelligence. We are more justified in logically linking codes to intelligence than the materialist is in linking them to unknown forces of nature. The intelligence linkage needs to be the base assumption. When it is, ID arguments for an independence of reason from brain matter easily flow. Intelligence predates brains, stunney and the design you are looking for, bFast, is found in biology's first von Neumann transcriptor.
The point is made that natural selection indicates no directional flow to a self-replicating molecule model. For example, why should we assume selection directs an RNA self-replicator to a cell? Based on experimental evidence? Yet there is more to this. A self-replicating molecule is the wrong model. Cells are best described as having numerous biological sub-systems with different functions; each one having many molecules. Organized biological systems replicate. If von Neumann were to theorize about a self-replicating robot, having many interacting parts, why would he use a single nut or bolt around which to fashion a model? Why would he even use a computer chip if the chip contains no programming information and there are no protocols yet existing?