“My point was not to blindly confuse marginal and conditional coverage”

No one is confusing it, as I have stated no framework can guarantee conditional coverage anyway. So why does it matter if CP can’t do it?

“Not sure how bayesianism slipped into the conversation.”

Aren’t you a Bayesian? It says “bayesian with a cause”

What “cause” might that be?

And that’s exactly why you are raising this irrelevant for the introductory article issue. Because you are sore that suddenly there is a much better framework than your “bayesianism with a cause.”

But again as I have said bayesianism can’t guarantee anything, even marginal validity let alone conditional.

And that’s the gist of the issue.

Let me give you just some reasons why your sore bayesianism is very flawed

⚠️Bayesian methods in a nutshell 🚨 : ❌ Lacking theoretical backing ❌ Based on questionable priors ❌ Yield beliefs, not probabilities ❌ Often misplace posteriors without warning ❌ No validity guarantees ❌ Uncalibrated results ❌ Built on artificial assumptions ❌ Prone to instability ❌ Disrupt base models ❌ Tough to build and maintain