Re: The NABT controversy

Loren Haarsma (
Mon, 16 Feb 1998 17:21:44 -0500 (EST)

[Note: this was posted to NTSE and SCICHR lists on the 12th, and
forwarded by Burgy to Massimo Pigluicci at my request. Since
Burgy posted the Pigluicci's reply to this list, I figured I should
post the original.]

Regarding the recent change in wording of the NABT statement on teaching
evolution, Massimo Pigliucci wrote:

>> That evolution is "unguided and unsupervised" is simply another -
>> stronger - way to say that it is "natural and random" (the latter still
>> in the NABT statement).

I am going to present an argument that "natural and random" is not
necessarily the same thing as "unguided and unsupervised." The
distinction is subtle but, I believe, important.

Short form of the argument: Under certain circumstances, it is
impossible for an observer to distinguish an unguided random process
from a supervised, guided pseudo-random process. If the distinction
cannot be observed, then it is unnecessary --- when making an empirical
model of the process --- to specify such a distinction.

Long form of the argument:
The theory of evolution attempts to explain biological history as a
result of "natural and random" processes. The word "random" has
different connotations in different settings. If we're using the word
"random" in a scientific theory, we should restrict the meaning to what
is actually necessary for the theory.

What is meant by "random" in an empirical theory? (1) That the final
state of the system cannot be completely specified in terms of the
(known) initial conditions; (2) That the cumulative behavior of the
"random" process shows no detectable pattern (other than conforming
probabilistically to the ensemble probabilities which can be calculated
from other considerations).

That, as far as I can tell, is all that is required of "random" in an
empirical theory. As I will try to show, this leaves open the question
of guidance and supervision.

[[Side note: I do not intend to debate whether or not biological
history is best described by a "natural and random" theory. Rather,
given that evolution is a "natural and random" theory, I argue that
the question of guidance and supervision is still open.]]

An analogy may help. Consider a computer program which uses a
mathematical algorithm and a very large seed number (more than 10^9
bits) to generate pictures. Each time the program loops, it generates a
new picture. With access to the seed numbers, you soon notice this
pattern: the seed number changes by one bit-flip each loop. After it
does this long enough, you conclude that the program operates by
randomly choosing each loop which bit to flip.

Suppose someone tells you that the programmer has two ways to guide and
supervise this program. First, the programmer can completely rewrite
the seed number. Second, the programmer can override the random number
generator and choose which bit is flipped on any given loop. How would
you be able to tell, simply by looking at the history of the seed
numbers, whether the programmer had been guiding and supervising the

I can think of two ways. First, if the seed number ever changed more
dramatically than a single bit-flip, you would know instantly that the
programmer intervened ("instantaneous evidence"). Second, if the bit-
flipping showed a distinctive pattern over a period of time, that would
be "cumulative evidence" of guidance. An example of cumulative evidence
could be: if you could determine, from knowledge of the mathematical
algorithm, that most seed numbers would generate simple and boring
pictures, and only a tiny subset of seed numbers generate complex and
interesting pictures, and over the course of watching the program
execute you notice that the program hits these "interesting" pictures
much more often than it should.

[[Side note: If you did observe this non-random behavior, you might
suspect that the program has certain subroutines beyond random bit-
flipping which allow it either to radically re-write the seed number,
or to guide it into "interesting" pathways, without the intervention
of a programmer. That may be the case, but that is not relevant to my
argument. I am not arguing that "instantaneous" or "cumulatively
suspicious" changes in the seed number prove that a programmer is
intervening. Rather I am arguing that, *even in the absence of these
events*, the programmer *might* still be guiding and supervising the

Suppose, instead, that you determined that a very large fraction of all
possible seed numbers generate interesting pictures. If all you could
do was monitor the seed numbers as they changed, it is possible that the
programmer, without your knowledge, might be occasionally (or
continually) selecting which bit is flipped in order to guide the
program onto a particular pathway to produce particular pictures. In
this case, the changes in the seed number would be --- as far as you
could determine --- random, yet they also could be guided and


The term "random" in a physics theory means simply this: The final state of a system cannot be completely specified in terms of its initial conditions, either in principle (e.g. the results of a "quantum measurement"), or in practice. In quantum mechanics, the element of chance is formally built into the theory; the outcomes of quantum measurements can only be specified probabilistically. In classical mechanics, the final state of "chaotic" systems depend so sensitively upon the initial conditions that, in practice, it is impossible to specify all the variables precisely enough to predict the final state. In these systems, based upon experience and certain general considerations, ensembles of final states can be assigned certain probabilities of occurring.

Evolutionary biology uses the term "random" in this second, classical sense. A "random" event is simply an event which is not caused by the organism itself, and which we could not have predicted given our limited knowledge of the initial conditions, which affects the organism's survival (e.g. a natural disaster) or its genetic information (e.g. a mutation).

If the development of life and biological complexity via "natural and random" processes was, in fact, highly probable on this planet, then how might we determine whether or not the *particular* pathway which biological history followed was guided and supervised? We cannot.

"natural and random." To add "unguided and unsupervised" goes beyond the requirements of the scientific theory.

Loren Haarsma