Naive Bayes

Awesome art is riddled with uncertainty.

One possibility is just chance.

If you need a statistician to know whether art matters, then art doesn't matter. Looking for dung where the cow never browsed, statistical mistakes are widespread.

Paul Klee

Bayes theorem is more often about "how" than about "why".

"How" uses algebra as medium but does not give a clue about the actual aim.

"Why" refers to the target : adjusting risks (probabilities) to actual hazards; it's like drawing only what you have before your eyes. Conditional probabilities are easy to represent with algebraic symbols but very difficult to represent as physical entities (for our brains).

Bayesian cat

To understand the above picture, one may start from a vague prior idea of what the girl is doing, but intuitively we have to take into account what we actually see : cat's paws had to be cleaned before the statisticians had begun sharpening their pencils.

The theory can represent arbitrarily fine-grained graphic degrees of plausibility intermediate between good art and bad art because drawing is thinking on paper.

bayesian elephant shark

Clearly, you have a higher risk of dying from elephant hazards than being eaten by a shark.

These worldwide statistics are true, but not useful. One needs to adjust risks to actual life hazards.

See also "posterior probabilities"

The negative is just as important as the positive

Ellsworth Kelly