Algorithms, evolution, and Dennett

[From Peter Cariani (960312.1300 EST)]

I haven't been closely following the discussion about Dennet's book and
algorithms, so I hope this doesn't repeat. Algorithms, as Bill P. points
out, are external, symbolic constructs that operate on the basis of rules:

Bill Powers 960311.1600 MST:

It's a _proposed_ algorithmic process, proposed by a human mind. I
presume that there is some actual process going on; however, what makes
it run is not an algorithm, but whatever interactions exist among the
elements of reality that comprise the system. We are not privileged to
experience reality at that level. Instead, we try to make something
orderly out of our basic intensity inputs, and one of the ways we do
that is to invent algorithms, expressed in terms of rules for
manipulating symbols (example: "integration"). Unless we have been
unthinkably lucky in our choice of perceptions and our inventions of
symbol-systems and rules, there is no reason to think that the natural
world operates in the same way our algorithms operate. It is not
necessary, for algorithms to be useful, that Nature work in the same way
our algorithms work; viz. epicycles, an algorithm that works just fine
to explain planetary movements, but which we no longer think has any
counterparts in reality.

If one examines Natural Selection and evolutionary processes, there is
nothing that looks remotely like an algorithm (a general heuristic of
reproduction-selection-mutation-reproduction maybe, but this is
not an algorithm). As Howard Pattee has pointed out over and over,
even the process of folding up a protein involves complex
"analog" dynamics that are not readily (and perhaps cannot be)
described in terms of rules operating on discrete symbols.
Natural selection is even more ill-defined than protein folding
(where are the "rules" that govern the interaction of
a phenotype with its environment such that
its survival/reproductionis determined?
is there even a well-defined set of "states" that characterizes a
"phenotype"?).

The hallmark of an "algorithm" or "computation" or
"effective procedure" is that it can be exhaustively specified such that
another observer-participant, armed with the rules and the symbols (tokens)
can reliably get the same (symbolic, formal) result.

I was really appalled when I heard that Dennett was taking the position
that evolution is an "algorithmic process". Without defining algorithm,
so that one can decide whether a natural process is "algorithmic" or
"non-algorithmic", the distinction is meaningless and the position is
vacuous. Yet more appalling, perhaps, is the approval of more serious
evolutionary biologists like Maynard-Smith, who a few months ago,
praised the book in the New York Review of Books. Too many people are
uncritically seduced by the cachet of the "computer metaphor" to stop and
think about what it really means. Clarity and criticality count for so
little in this race for popularity and media attention.....

I do agree with Martin Taylor [3/12/96] that there
can be "computational" or "algorithmic" processes
in nature, once it is made clear what the
"operational states" of the system in question are and what rules
are being implemented. Many sequence-based operations on DNA and
RNA look this way, and many neural informational processes that
involve all-or-nothing or discrete response alternatives can also
be seen in this way given the right observables.

-- Peter Cariani

[Martin Taylor 960312 14:50]

Peter Cariani (960312.1300 EST)

Algorithms, as Bill P. points
out, are external, symbolic constructs that operate on the basis of rules
...
even the process of folding up a protein involves complex
"analog" dynamics that are not readily (and perhaps cannot be)
described in terms of rules operating on discrete symbols.
...
there
can be "computational" or "algorithmic" processes
in nature, once it is made clear what the
"operational states" of the system in question are
...
many neural informational processes that
involve all-or-nothing or discrete response alternatives can also
be seen in this way given the right observables.

I hope you don't mean your intervention in the way it seems to read.

You seem to suggest that a process that produces as its result the
continuous sum of two continuous variables is _not_ an algorithmic process.
You don't mean that, do you?

Martin