Under The Sun

Wednesday, March 05, 2003

NDPR: Making easy problems hard

Exhibit #6,374: Nick Bostrom's Anthropic Bias: Observation Selection Effects in Science and Philosophy, and Neil Manson's review thereof.

The basic question is simple: what may be reasonably inferred from the facts (1) life-as-we-know-it exists, (2) life-as-we-know-it depends crucially on very precise values for certain fundamental physical constants. Some people want to infer that very likely there is a reason why those constants have those values, e.g. Intelligent Design. Other people want to infer that very likely there are many universes with different constant-values, and therefore the specific set of constant-values we need is not unlikely to have been instantiated once. Which group has logic on their side?

Neither, of course; obviously we can't infer diddly-squat, because we simply don't have enough evidence. Both of the above theories put the "speculation" in "metaphysical speculation". But for several decades now, cosmologists and other interested parties have been debating this question, and the debate has crystallized around various competing forms of something called the Anthropic Principle. Now, at long last, a real philosopher (Bostrom) has arrived to set the record straight.

That the Anthropic Principle is multifariously untenable is not news; back in 1986, Martin Gardner distinguished in print between, among other contestants, the Weak Anthropic Principle (WAP), the Strong Anthropic Principle (SAP), and the Completely Ridiculous Anthropic Principle (CRAP). But Bostrom is made, I suppose, of sterner stuff than Gardner: he wades in where Gardner just held his nose and turned away.

Bostrom first considers "suprisingness", and rejects it as a useful criterion. This should be obvious: it's totally subjective, and therefore not good science. Yet Manson writes, "I don’t see why Bostrom abandons the well-established "surprisingness" analysis of need for explanation". Apparently it's not that obvious.

Then we get to the real meat: finding something that works better than the Anthropic Principle. Bostrom proposes the "Self-Sampling Assumption: One should reason as if one were a random sample from the set of all observers in one's reference class." But of course this won't quite work, and is quickly--er, after several chapters, rather--replaced with the "Strong Self-Sampling Assumption: One should reason as if one's present observer-moment were a random sample from the set of all observer-moments in its reference class."

This is true, of course--at least it's close enough for science--but it merely shifts the bump under the rug: what the heck is one's reference class? We have succesfully characterized the problem--not that it was unclear at the beginning; I stated it quite simply at the top--but we've made exactly zero progress toward a solution.

But to me, the reductio ad ha-ha of this whole project is that Manson actually feels the need to state as a postulate, at one point in his argument, that "one should take all relevant evidence into account". Why yes, yes, one should. Thank you for clarifying that.

This is the kind of CRAP that gives philosophy a bad name.

Comments: Post a Comment