support our further discussions. A more complete
presentation would include considerations of the neuron and the role played by
evolutionary mechanisms in the construction of the brain (Gerald Edelman’s Neural Darwinism). We will bravely
assume that we’ve transmitted enough of the message that we can now turn to
looking at computer systems, weaving their description with that of biological
systems, even at the risk of touching material that we have not properly
introduced.
Evolutionary
processes for organic life can derive from mutational events coupled to a form
of feedback process control effected through the modulation of genetic
adaptation by natural selection; in essence, a crap shoot coupled to a positive
feedback loop. Note that we mention a positive feedback loop. Such a feedback
mechanism has the tendency to reinforce the change that caused it. Within an
electrical circuit, such a feedback mechanism will tend to force a circuit into
unstable saturation; within living organisms, it will tend to enhance the
propagation of newly developed traits.
Certainly all of
this discussion has an a-religious ring to it; but again, the devil is in the
details. As was observed in the previous section, the vast majority of
mutations prove to be either innocuous or fatal to the resulting individual
organism. Certainly, the probability of a chance mutation becoming a
contributing facet of the evolutionary design of a species is very low indeed.
On the other hand, low probabilities applied over a sufficiently long time
period can yield non-zero results. So, perhaps it is useful to consider our epigraph
for this chapter. If an event of exceedingly low probably actually does occur,
it is difficult to name its true origin; was it chance, or was it due to very
subtle design selections? Or more radically, is chance simply an abdication of
design? For example, Stephen Wolfram in A New Science has shown that a
sequence of numbers fulfilling all known tests for randomness could be generated
from a deterministic automaton. The concept of causality arises as a
point of interest because of its subsequent impact on the mechanics of policy.
From a policy consideration standpoint, causality is important because it can
be related to trust; and trust, we will contend, is the basis of all policy
mechanisms, including social organizations.
So, now let us
consider the progression of computer systems. Does a change mechanism similar
to the mutation of DNA structure exist for computer systems? In fact it does.
It can be characterized as directed or serendipitous basic research.
Subsequent fine tuning of new characteristics first introduced through such
mutational change occur through an analogue to genetic selection that is
generally termed applied research. As with organic design and
feedback loops through the DNA molecule, research based mutation and refinement
are completely grounded in physical law. These two forms of fundamental change
show characteristics more like a roulette wheel than of the pair of dice in a
crap game. That is, they may result in a profound, beneficial development along
the lines of hitting en plein (a straight bet on the ball landing
on a specific number which pays 35 to 1 odds); but, a more modest enhancement
may result, corresponding perhaps to hitting a dozen bet (a wager that the ball will land on one of
twelve selected numbers which pays 2 to 1 odds).
Basic research
is best characterized as an undirected search for answers to previously unknown
questions, while applied research is aimed at finding new approaches (e.g. new
technologies) to solve known problems in more effective ways. As we said, one
might at least qualitatively compare basic research to a mutation and applied
research to genetic adaptation. Indeed, just as the typical organic mutation
results in either a benign variant or in a catastrophe, so is the result of
basic research most often a dead-end, if not a catastrophe in its own right
when evaluated by the actions or reactions of the market. The useful discoveries
of basic research tend to be
|