Occam’s Razor is Wrong (Sort of)

The danger in relying on Occam?s Razor in decision making.

I can’t think of a single model that is so widely used by many intelligent people, yet is so often abused. The truth is our reliance on Occam’s razor is a textbook example of the availability cascade phenomenon: It has become a self-reinforcing heuristic, that has advanced to legendary status, simply because no one questions it.

There are two main issues with the common application of Occam’s razor: The first is that it is very often mis-applied. Occam’s razor is often quoted as “the simplest explanation is the best explanation,” but a more accurate translation is “all else equal, the simplest explanation is often preferable.” The problem is all else equal virtually never happens. Even is lab experiments, assumptions and varying degrees of veracity of underlying information mean all else is never equal. And in real-world situations, the explanatory power of each alternative is virtually never the same [Source]. In these cases, Occam’s razor has no useful guidance and is not supposed to be applied.

The second problem is Occam’s razor doesn’t make any intrinsic claims of veracity. It was not intended as a way to determine what is right or wrong. Instead it is more of a guideline for the evaluation of competing hypotheses. The simple explanation is a better one to accept, not because it is true; in fact it is likely to be wrong based on prior probability: But a simple explanation combats over-fitting and puts us on better footing to expand our understanding of the subject area.

There is evidence and rational arguments for simplicity and parsimony: These are bounded by constraints like equal explanatory power. These modern definitions have also evolved far beyond what William of Ockham (spellings vary) said or didn’t say in the 1300’s. And really it doesn’t matter if our primary concern is finding good explanations. Let’s throw out Occam’s Razor in favor of good old (or new) parsimony!

Hellblade: Senua’s Sacrifice