That's what I mean. The Jensen inequality applies to any distribution, but the intuition presented in the post is only good for simple distributions and all exa

That's what I mean. The Jensen inequality applies to any distribution, but the intuition presented in the post is only good for simple distributions and all examples are gaussian/binomial-like. It would be difficult to raise the same points with something multimodal or arbitrary.