Ockham's Razor
A Wee Gripe

First off, it's not Ockham's. Not really, anyway. The same basic principle shows up earlier in Scotus (and used to be known as "Scotus's rule"), and, in fact, it shows up in Aristotle, too. I gripe about this because I have run across about a hundred scientists who, eager to show that they're philosophically literate, mumble some pedantic nonsense about Occam being the first to articulate the principle. Quit it. You're scientists. You can admit some ignorance of philosophy. Nobody will care.

Second, the principle is widely misunderstood. According to the most common misunderstanding, the principle says, roughly, that the simplest explanation is the most likely to be true. Gah. Now ignoring the fact that this formulation is conentious on other grounds, let me just gripe in particular about the fact that an important bit has been left out here. It is not true that simpler explanations are more likely to be true; if anything, what is true in this vicinity is something more like: among competing explanations that all account for the same phenomena, the simpler hypothesis is in some sense preferable (perhaps because it is more likely to be true; perhaps for some other reason. We're not sure).

One problem here is that there's no uncontroversial sense of 'simpler.' This and other problems have driven me to take Peirce's decidedly non-standard view seriously. Peirce says that 'simpler' ought to be interpreted as meaning more natural. And that means something like: more likely to suggest itself to the inquiring mind. Now, you've got to buy Peirce's conception of abductive inference for that to make much sense to you. But if you think that hypothetical inferences are abductions, and if you think that abductions are guesses, and if you think that abductive inference works for us because we are good guessers, then an attempt to explain simplicity in terms of naturalness will be of interest to you. On such a view "simpler" (i.e. more natural) hypotheses are something like very, very, very slightly more likely to be true, and hence are the hypotheses that should be tested first. On this view, the emphasis on simplicity is misplaced. What scientists really value is something more like elegance. Some mistakenly think that talk of elegance is slightly confused talk about simplicity, whereas Peirce seems to think that talk of elegance is slightly confused talk about naturalness.

One part of this package is that appeals to Ockam's razor don't count for as much as many contemporary philosophers and scientists think they do. On Peirce's view, all the principle helps us do is rank hypotheses according to which should be tested earlier and which later. That doesn't do very much for us; it's part of Peirce's theory of the economics of research. Many contemporary scientists and philosophers--in particular "explanationists" and those who think hypothetical inference is a matter of "inference to the best explanation"--seem to think that the razor can do far more for us. Those folks think, roughly, that we do our experiments or observations first, and then use canons of theory preference--such as the razor--to pick out the best theory from among the ones that are equally well-confirmed. At any rate: inferences to the best explanation are supposed to be very powerful, whereas abductions are very weak--in valid abductive inferences, the conclusions only follow with "plausibility" from the premises, and that means only that they are hypotheses worthy of testing. Some advocates of the IBX approach (e.g. Gilbert Harmon) think that IBX is at least as powerful as induction, as they construe inductions as types of IBXs. Peirce, on the other hand, construes inductive inference as probabilisitc, and, hence, abductions as very, very much weaker.