A new paper by Wolpert and Rovelli is out claiming to help resolve the Boltzmann Brain (BB). In short, what if your memories were not real, but just randomness with no actual prior events?

Are your perceptions, memories and observations merely a statistical fluctuation arising from the thermal equilibrium of the universe, bearing no correlation to the actual past state of the universe?

Immediately I was struck how the Plato’s cave metaphor does something this paper doesn’t. It makes the stakes ethical, more than just epistemic. Plato’s prisoners who glimpse the outside have an obligation to return and tell the others, even knowing they’ll be disbelieved or killed. That’s an important function.

The BB framing drops ethical aspects and treats the question as a puzzle about priors rather than a situation with consequences for how you act toward other apparent minds. Which is strange, because if you’re a BB, so is everyone you’re arguing with, and the ethics of that situation are genuinely weird. Imagine everyone thinking they are the queen and their memory is the only valid one, now bow down. (Hat tip to Wollstonecraft)

Or imagine watching Blade Runner where everyone can claim to be the Blade Runner, instead of everyone wondering if they are a replicant.

Descartes saw the ethical aspect and tried to escape with God as an external guarantor. Kant gave up on the outside and relocated the problem to the structure of cognition itself, which is what this paper also does. Their “construction rule” is basically a Kantian move: here are the rules any coherent reasoner must follow, given that we can’t step outside reasoning to check it. Or as Hume would probably say, we are creatures of our habits far more than our justifications.

Modern physics gets stuck like this. There’s only inside, no outside. The faculties you’d use to check whether you’re in a cave are produced by the same physics being checked. Wolpert and Rovelli are Plato’s prisoners who’ve proven rigorously that they can’t tell if they’re in a cave, and then they just stopped and hit publish.

They say the answer depends on your prior and leave it there, as if prior selection is a separate problem someone else can solve. But prior selection has the same structure as the original question. You pick a prior using judgment, which depends on memory and inference, which depends on the second law, which depends on the prior. The circularity doesn’t get resolved by being pushed to the meta-level. It just relocates.

That’s a cheat. They aren’t admitting it’s inherited, yet placing it somewhere you would inherit it from.

Plato’s escape was that something outside the cave (the Forms, the sun, reason properly trained) gives you leverage the cave can’t corrupt. That’s the authentic inherited-rights move. External source, not controlled by the thing being questioned. It’s law and order, which physicists are disciplined to deny has an outside.