There is a view that everything that works must be an approximation of the ideal Bayesian method.
You can reformulate many problems in the Bayesian framework. This does not mean that everything is an approximation of Bayesianism—just like the ability to translate a novel into French does not mean that each novel is an approximation of a French roman.
It’s deeper than that. Bayesian probability theory is a mathematical law. Anything method that works must be computing an approximation of it. Just like Newtonian mechanics is a very close approximation of relativity. But they are not equivalent.
Bayesian probability theory is a mathematical law.
That is not true. The Bayes equation is mathematically correct. A theory is much wider—for example, Bayesians interpret probability as a degree of belief—is that also a mathematical law? You need a prior to start—what does the “mathematical law” say about priors?
You can reformulate many problems in the Bayesian framework. This does not mean that everything is an approximation of Bayesianism—just like the ability to translate a novel into French does not mean that each novel is an approximation of a French roman.
It’s deeper than that. Bayesian probability theory is a mathematical law. Anything method that works must be computing an approximation of it. Just like Newtonian mechanics is a very close approximation of relativity. But they are not equivalent.
That is not true. The Bayes equation is mathematically correct. A theory is much wider—for example, Bayesians interpret probability as a degree of belief—is that also a mathematical law? You need a prior to start—what does the “mathematical law” say about priors?