randomness and design

I've not yet posted a lot on this blog concerning the philosophy of science, or for that matter just science straight up—which is a little odd, considering my vocation. But tonight Donald Sensing has provided inspiration enough.

He writes:

The Intelligent Design debate continues. I have searched for a site that dispassionately defines the idea of ID but all I have found are polemic sites on one side or the other. So allow me to very briefly define what ID is, as I understand it.

ID is the proposal that the complexity of the universe and of earth's creatures cannot be explained by random processes. Hence, IDers (as ID's proponents are sometimes called), say that it is reasonable to posit that creation was designed by a power outside nature.

Now, I happen to believe that, but I also know that ID is not science. At best, Intelligent Design is a conclusion from science. The postulate of a Creator of nature is a non-scientific postulate.

First off: I might humbly recommend my own site for a non-polemical treatment of Intelligent Design.

Second: By “continues”, Rev. Sensing is referring to his earlier post on the subject here, and to Rand Simberg's response to Hugh Hewitt. I do not want the scope of this post to become unmanageably broad, so in précis: I largely agree with Simberg's take, and moreover consider lawyerly arguments in favor of ID to be (for the most part) both misguided and unhelpful.

At present I want to examine Rev. Sensing's definition of ID, and explain why it is incomplete. Once again:

ID is the proposal that the complexity of the universe and of earth's creatures cannot be explained by random processes. Hence, IDers (as ID's proponents are sometimes called), say that it is reasonable to posit that creation was designed by a power outside nature.

Randomness is not really the issue. Instead, what ID proponents reject is the self-sufficiency of natural law—and by extension that of natural processes, some of which are irreducibly random. More on that a little later.

The larger issue with the definition as given is that it is not specific enough. Indeed, it can also describe some flavors of theistic evolution: a problem, as ID advocates are wont to consider theistic evolutionists as tragically misguided (even those in the latter camp who regard the world as in some sense designed).

Intelligent Design adds two postulates to the basic claim of design: (1) extra-natural action has played a fundamental role in the development of life; and (2) the fingerprints of that action can be objectively identified. As I wrote in the introduction of my dissertation proposal:

ID theorists postulate that explicit divine action has been integral to the history of life. This commitment is in itself nothing new; what distinguishes ID from its forebears is the additional claim that the signature of design is both detectable and quantifiable, and hence is a legitimate subject for scientific inquiry in spite of its extra-natural origin. But one striking aspect of this doctrine is an imposed limit on allowed divine action: the possibility that God might have front-loaded all necessary design into the laws of nature, thus rendering redundant further action beyond the purview of those laws, is excluded. The motive for such explicit rejection is a continuing antipathy towards biological evolution—not just the neo-Darwinian theory, but also what are perceived as the philosophical, cultural, and theological corollaries of that theory.

Without these additional postulates there is no Intelligent Design program.

If Sensing's definition is incomplete, a portion of what follows is unfortunately just wrong. He writes:

There cannot be a science of randomness, for science depends on repeatability. The conclusion that randomness explains the beginning and history of life is not really a scientific conclusion. It is one thing, and a properly scientific thing, to say that here are processes that seem to explain the evolution of species. But it is not science to say with finality that no intentionality was involved. The exclusion of intentionality is not a scientific conclusion, but an ideological one. (Emphasis added.)

Sensing is not (as he makes clear) an anti-evolutionist. He is nonetheless employing a shorthand common in anti-evolution circles: natural selection acting upon variation in a population becames equated with randomness. True enough, the source of that variation is in a sense random, perhaps even irreducibly so because of effects on the quantum level. But the terms “evolution by natural selection” and “random causes” are by no means interchangeable.

I'm not going to quibble any further on that, as Sensing is clearly intending to provide only a thumbnail sketch. What I will focus on instead is his claim that There cannot be a science of randomness. On the contrary: much of modern physics, for instance, deals with the probabilistic nature of the world.

Here is what I posted in the comments, with links added (and lightly edited for clarity).

Your last commenter is (partly) correct in regards to “a science of randomness”. On the microlevel chance is ineliminable—that is, there are no “hidden variables” that might allow us to completely control (or even predict) outcomes of repeated trials, except statistically. The various No Hidden Variables theorems are not empirical, but are instead derived from the mathematical foundations of quantum theory itself.

On the macro level “randomness” means something else entirely. In systems involving a large number of particles (that is, anything in the visible world), an experimenter cannot control every degree of freedom within the system, and hence the initial conditions are never fixed precisely. Repeated experimental runs will thus give different results; and if the system is well-behaved, those results will cluster about a central value.

The difference between the two cases is that randomness on the quantum level is ineliminable in principle. On the macro level, the limitation is instead a practical one. Consider a mechanical lottery machine (say of the Powerball painted ping-pong ball type). The output of such a device is not random in any fundamental sense—if the system could be sufficiently isolated, and the initial conditions precisely enough known, then the outcome could be predicted.

Of course, the micro/macro dichotomy I've portrayed is an oversimplification. Countless experiments, both thought and actual, have been devised for bringing quantum effects to the macro level (most famously Schrödinger's Cat). And some theorists argue that the very notion of a precise initial state is ill-conceived.

Quantum mechanics might well be described as the science of ineliminable randomness—and it is a very successful science, at that, even if the question of what it all means is far from settled. Classical statistical mechanics is concerned with the probabilistic behavior of bulk matter. Population genetics predicts changes in allele frequencies in a mathematically rigorous fashion. And each of these can be construed as a science of randomness.

Still, it may be that Sensing means that there can be no science of contingency. If so, I am inclined to agree. We can, for instance, only speculate about what life's history might have been in the absence of the various extinction events; and there is no way to predict exactly which variations will arise in a population. Without a God's-eye view we can only take contingencies as given. The study of counterfactuals may be an interesting problem in logic, but it is not properly speaking a part of science.

_________________________

UPDATE. Discussions of ID seem to break out quasiperiodically in the right- and libertarian-leaning parts of the blogosphere (on the left it isn't typically discussion so much as snarky dismissal). The most recent is between the aforementioned Rand Simberg and John Mark Reynolds. I haven't read the posts yet, as it is now very late and I just now stumbled across Simberg's latest.

(Even later: here's a related post by David Mobley.)

For those interested in the topic of randomness and science, Ian Hacking's The Taming of Chance is likely a good place to begin. I've not read this particular book, although Hacking's Representing and Intervening—a robust defense of pragmatic realism—remains my single favorite text in the philosophy of science.

UPDATE 2. There is another problem with the premise that science must be repeatable—namely, that it renders all historical sciences suspect: not just evolutionary biology, but geology, archaeology, and physical anthropology as well. As I have noted elsewhere, some ID arguments against evolution would, were they to be successful, undermine all forms of historical inference.

Also, for sake of completeness I should note that there are ways to circumvent the results of No Hidden Variables theorems in quantum mechanics. But these alternate approaches have their own issues.