Seconded. There are a lot of clever ideas that I haven’t seen anywhere in other probability books: A_p distributions, group invariance, the derivation of an ignorance prior as a multi-agents problem, etc.
The only lacking (due to obsolescence) chapter is the one about quantum mechanics. Jaynes advocates (although implicitly) a hidden variables theory, but so far Bell’s and Kochen-Specher’s theorems imposed heavy constraints on those.
You seem to imply that Jaynes was writing before Bell. That is not true by many decades. I suppose it is possible that the chapter is based on a paper he wrote before Bell, but he had half a century to revise it.
Jaynes thought he had found an error in Bell’s theorem, but he was wrong. (I wrote a comment somewhere on LW about this before; I’ll link to it as soon as I find it.)
I’m under the impression that he was so committed to the idea that there are no probabilities due to intrinsic indeterminacy of nature rather than our ignorance that he got mind-killed. (I wonder whether he had ever heard (and seriously thought) about the MWI.)
That is a remarkable error, actually. As far as I can tell, it’s basically denying that conditional independence is possible in Nature (!?)
The existence of Bell’s inequality is basically a theorem about marginals of Bayesian networks with hidden variables. If you get an independence in the underlying Bayes net, you sometimes get an inequality in the marginal. This is not about causality at all, or about physics at all, this is a logical consequence of a conditional independence structure. It does not matter if it is causal or physical or not. Bell’s theorem is about this graph: A → B ← H → C ← D, where we marginalize out H. My “friend in Jesus” Robin Evans has some general conditions on graphs for when this sort of thing happens.
Jaynes was aware of MWI. Jaynes and Everett corresponded with one another, and Jaynes read a short version of Everett’s Ph.D. dissertation (in which MWI was first proposed and defended) and wrote a letter commenting on it. You can read the letter here. He seems to have been very impressed by the theory, describing it as “the logical completion of quantum mechanics, in exactly the same sense that relativity was the logical completion of classical theory”. Not entirely sure what he meant by that.
I don’t think it’s mind-killed. It’s possible to reject the premise of the Bell inequality by rejecting counterfactual definiteness, and this is a small but substantial minority view. MWI then takes this a step further and reject factual definiteness, but this is not the standard way in which it’s presented, so someone who has issues with the notion of “Alice makes a decision ‘of her own free choice’, unaffected by events in her past light cone” but has never encountered the descriptions of MWI which mention factual and counterfactual definiteness, can justifiably believe that contrary to appearances, some hidden-variable or superdeterminist theory must be true.
I speak from personal experience, here. Up until about a year ago, I held two beliefs that I recognized were in defiance of the standard scientific conclusions, both on logical grounds. One was belief in hidden variable theories of quantum physics; the other was belief that the Big Crunch theory must be correct, rather than the Big Chill (on counter-anthropic grounds; a Big Chill universe would be the last of all universes, and that we should happen to live in the last universe, which happens to be well-tuned for life, strains credulity). Upon realizing that MWI solved the problems that led me to hidden-variable theories, and also removed the necessity for an infinite succession of universes, thus reconciling the logical non-exceptionalist argument and the Big Chill data, I switched to believing in MWI.
Seconded. There are a lot of clever ideas that I haven’t seen anywhere in other probability books: A_p distributions, group invariance, the derivation of an ignorance prior as a multi-agents problem, etc.
The only lacking (due to obsolescence) chapter is the one about quantum mechanics. Jaynes advocates (although implicitly) a hidden variables theory, but so far Bell’s and Kochen-Specher’s theorems imposed heavy constraints on those.
You seem to imply that Jaynes was writing before Bell. That is not true by many decades. I suppose it is possible that the chapter is based on a paper he wrote before Bell, but he had half a century to revise it.
Jaynes thought he had found an error in Bell’s theorem, but he was wrong. (I wrote a comment somewhere on LW about this before; I’ll link to it as soon as I find it.)
I’m under the impression that he was so committed to the idea that there are no probabilities due to intrinsic indeterminacy of nature rather than our ignorance that he got mind-killed. (I wonder whether he had ever heard (and seriously thought) about the MWI.)
http://arxiv.org/pdf/physics/0411057.pdf
That is a remarkable error, actually. As far as I can tell, it’s basically denying that conditional independence is possible in Nature (!?)
The existence of Bell’s inequality is basically a theorem about marginals of Bayesian networks with hidden variables. If you get an independence in the underlying Bayes net, you sometimes get an inequality in the marginal. This is not about causality at all, or about physics at all, this is a logical consequence of a conditional independence structure. It does not matter if it is causal or physical or not. Bell’s theorem is about this graph: A → B ← H → C ← D, where we marginalize out H. My “friend in Jesus” Robin Evans has some general conditions on graphs for when this sort of thing happens.
Jaynes was aware of MWI. Jaynes and Everett corresponded with one another, and Jaynes read a short version of Everett’s Ph.D. dissertation (in which MWI was first proposed and defended) and wrote a letter commenting on it. You can read the letter here. He seems to have been very impressed by the theory, describing it as “the logical completion of quantum mechanics, in exactly the same sense that relativity was the logical completion of classical theory”. Not entirely sure what he meant by that.
I don’t think it’s mind-killed. It’s possible to reject the premise of the Bell inequality by rejecting counterfactual definiteness, and this is a small but substantial minority view. MWI then takes this a step further and reject factual definiteness, but this is not the standard way in which it’s presented, so someone who has issues with the notion of “Alice makes a decision ‘of her own free choice’, unaffected by events in her past light cone” but has never encountered the descriptions of MWI which mention factual and counterfactual definiteness, can justifiably believe that contrary to appearances, some hidden-variable or superdeterminist theory must be true.
I speak from personal experience, here. Up until about a year ago, I held two beliefs that I recognized were in defiance of the standard scientific conclusions, both on logical grounds. One was belief in hidden variable theories of quantum physics; the other was belief that the Big Crunch theory must be correct, rather than the Big Chill (on counter-anthropic grounds; a Big Chill universe would be the last of all universes, and that we should happen to live in the last universe, which happens to be well-tuned for life, strains credulity). Upon realizing that MWI solved the problems that led me to hidden-variable theories, and also removed the necessity for an infinite succession of universes, thus reconciling the logical non-exceptionalist argument and the Big Chill data, I switched to believing in MWI.