What is interesting to me is how recent Occaam's razor is (800 years) and why it was introduced (to understand the Trinity in western Christianity).
It's been a successful heuristic, but for understanding the things that can be understood simply - we shouldn't believe that nature "wants" to be simple though...
Occam's razor in a lot of ways is a proxy for Bayes' theorem. The theorem "punishes" more complex models by shrinking the prior term and increasing the marginal term, while the likelihood is pegged at P = 1.0 at most.
You can come up with a model that perfectly explains the outcome ( P(e|H) = 1 ) but you have to offset the insane improbability of pegging all those free parameters at the value they are at ( P(H) ).
It's been a successful heuristic, but for understanding the things that can be understood simply - we shouldn't believe that nature "wants" to be simple though...