In some contexts I like to think of it as such: Gi...

2011-11-09T23:31:41.439-05:00

In some contexts I like to think of it as such: Given a phenomenon P, and a correct explanation E, it is always possible to come up with any number of alternative explanations that are more complicated but manage to include E under known conditions as a special case, or even appear to have a slightly better fit at the cost of great complexity. Thus, if we have two explanations that both appear to fit the data, there’s always the danger that the more complicated one is just such a complication of the simpler one. In fact, since there’s no end to the potential number of obfuscations—prefer the simplest (that works).

(When I arrive at this idea I’m thinking back to physics and statistics, and the admonition that the correct formula may be a quadratic even if a cubic or quantic might fit the data better, because they follow noise in the data. But their predictions outside a specific domain will tend to be inferior.)

Comments on Don't Feed the Animals: Occam Guides The Way

In some contexts I like to think of it as such: Gi...