Sunday, January 6, 2008

Occam's Razor

Occam's Razor is often considered to be one of the major tenets of skepticism. It has many forms and interpretations, ranging from "Keep it simple, stupid" to "Entities should not be multiplied beyond necessity". The problem is that Occam's Razor can be difficult to justify (depending on which interpretation you use).

The most common interpretation I see is that Occam's Razor is a rule of thumb that states, "The simplest explanation is the most likely one". This is useful against all kinds of conspiracy theories and pseudosciences. Which is a simpler explanation: physicists have to this date missed a fifth fundamental force that somehow acts only upon human brains, or that parapsychology studies occasionally have methodological flaws? Which is simpler: that our incompetent government succeeded in organizing a vast conspiracy to fabricate 9/11, or that our incompetent government failed to stop some terrorists?

But is it really true that the simplest explanation is most likely? It's not a logical contradiction to say otherwise. And what constitutes a "simple" explanation anyways?

One way to justify it is through empirical observation. We observe that simpler explanations are more likely to be successful than more complicated explanations. I'm not exactly sure how you would systematically test this fact, but in any case it's only a rule of thumb. On the other hand, I could point out various counterexamples. Confining myself to physics, I think Quantum Mechanics is not at all simple. Physicists are awfully fond of talking about how simple our fundamental laws are, but in the same breath they declare how counterintuitive Quantum Mechanics is. I don't think it can be both.

Or perhaps I not using the correct definition of "simple". Usually, in the context of Occam's razor, people define "simple" to mean involving the fewest entities ("entities" is not particularly well-defined). Furthermore, Occam's razor is not meant to trump empirical evidence. If there is evidence of many entities, we should posit these entities, but no more. Hence, "entities should not be multiplied beyond necessity". By this measure, Quantum Mechanics is definitely the simplest possible explanation of all the evidence. (An aside: The Many Worlds interpretation is sometimes said to violate Occam's Razor, since it has infinite parallel universes, but arguably, it only has one entity, the wavefunction of the multiverse.)

I think the empirical evidence suffices to justify this interpretation of Occam's Razor. Just keep in mind that it is only a rule of thumb, and it is known a posteriori, not a priori. Personally, I would be extremely careful to use Occam's Razor, and first try other kinds of arguments. Lastly, if you want to apply this to metaphysics, empirical evidence does not cover this region, and you will have to find other justification. As far as I can tell, there is no other justification, aside from the fact that simpler theories are more useful to humans. But the most "useful" theory isn't necessarily the true one, is it?

I wrote on a different interpretation of Occam's Razor here.