The semantic body

Clay Shirky
Tue, 28 Aug 2001 11:32:30 -0400 (EDT)

This is right, I think, and I will retract the neural analogy, for
the reasons you list below.

The problem with groping towards something via analogies is that you
are never sure if the differences between the systems being
compared can be abstracted away for the sake of comparison, or if the
very things being abstracted away are critical.

So in pursuing the idea of IPC as a basic building block of biological
organization, I will stick with systems where the units are mobile, so
that the 'hardwiring' problem is lessened, as any two random units can
encounter one another, and the communications channels better model
the many-to-many environment of the net.

Eric Bonabeau's work creating virtual ants that do Postman's problem
or resource optimizations are an example, and I'll try to write up a
little something that is more rigorous in its comparisons, so that it
can be clean enough to actually criticize, which the neural analogy


> (Random thoughts, after an exhausting weekend.)
> I know it is unfair to pick too much at an example
> that perhaps it intended more as analogy than 
> illustrative case in point. And I mean no offense
> to Clay Shirky by this. But the more I think about
> the biological example, the more convinced I am
> (1) that it is nothing like the internet, and (2) that
> one of the points pressed from the analogy is 
> wrong.
> It is true that the complexity of the cell is greater
> than the complexity of its membrane interface, and
> so on with tissues, and organs. But a tremendous
> difference between biological systems and the
> internet is that the former are hardwired. Nerve
> cells do not communicate with arbitrary other
> nerve cells, but with specific other nerve cells,
> muscles, and sensory organs. The blood stream
> provides a broadcast messaging system, of
> course, but it it filtered quite a bit in the different
> organs. To then point out that there are relatively
> few biochemical intermediaries overlooks the fact 
> that each of these may have different meaning
> -- different "semantics" -- in each of the places
> where it is meaningful. Nitric oxide in the 
> amygdala may carry some emotional meaning,
> in the hippocampus, it may affect memory, in
> some other part of the brain, yet something 
> else, and in the corpus spongiosum, it causes a 
> boner. One can no more say the semantics
> is simple because of the small number of 
> biochemical intermediaries than one can say
> the semantics of the world's digital protocols
> and file formats are simple, because they all 
> use a relatively small number of byte values,
> 0 to 255. (And in the first case, "relatively 
> small" is at least a few thousand.)
> Bottom line: The body works through a
> hard-wired architecture involving tens of 
> thousands of specialized protocols. These 
> carry deep and subtle semantics that are still
> causing biochemists and physiologists to 
> scratch their head, despite the *relatively* 
> small number of biochemical "symbols" that
> are used and reused in all of these. I'm not 
> sure what one should conclude from this 
> about the internet. 
> Russell

Clay Shirky - Writer and Consultant
Internet Ecology and Emerging Technologies