Two Types of Methodological Naturalism

It is often claimed nowadays that "science", by definition, can only invoke "natural" causes as explanations, where "natural" causes are restricted to entities or laws that will presumably find inclusion in a completed form of physics or to things that are either ontologically reducible to or strongly supervenient on such entities and laws. In short, according to this view "science" has to approach the world as a causally closed physical system. Let's call this view strong metaphysical naturalism (SMN).

Now, I think this is the wrong way to define "science" for the reasons that I give here. But many people are apt to think that there is something right about SMN. After all, if we don't place any restrictions on what types of explanations are permissible, then don't we open the floodgates for all sorts of pseudo-scientific nonsense to masquerade as genuine science? How else could we exclude from science 'magical', paranormal, and miraculous explanations? Certainly a rational person ought to be very cautious if not downright skeptical about such explanations.

I'm very sympathetic to such concerns, but SMN is not the answer. So far as I can see, there's nothing intrinsically wrong with magical, paranormal, or miraculous explanations. I certainly don't want to rule out such things a priori on the off-chance that such explanations just might, on occasion, be correct. The problem with pseudo-scientists is not the types of explanations they propose but the fact that they invoke such explanations way too soon, before properly looking for and excluding the plausible alternatives. This leads to what I call weak methodological naturalism (WMN), which says that in science natural explanations are to be preferred over non-natural explanations unless there are no plausible natural explanations and there is good reason to think that no plausible natural explanation will be forthcoming.

The basic idea behind WMN has been captured by ID-theorist William Dembski in what he calls the 'explanatory filter': "Roughly speaking the filter asks three questions and in the following order: (1) Does a law explain it? (2) Does chance explain it? (3) Does design explain it?" Thus, explanations in terms of deterministic or stochastic natural laws need to be ruled out (i.e., shown to be implausible) before it is reasonable to appeal to chance or luck, and chance needs to be ruled out before it is reasonable to appeal to design. And, extending this further, it seems to me that appeals to little 'd' designers ought to be ruled out before we are justified in appealing to a transcendent Designer-God. In short, there's an appropriate epistemological order of explanation that good scientific methodology will respect. Miracle-type hypotheses are not ruled out a priori, but they do face a stiff burden of proof.