The Boltzmann brain argument goes like this: if the universe is in thermal equilibrium, then any configuration — including a brain with false memories of a low-entropy past — is as likely as the actual low-entropy past it remembers. Given infinite time, most observers are Boltzmann brains: transient fluctuations that believe they have histories but don't. Therefore you should expect to be one.
Note (2026-11-10): I checked the abstract and searched the full text of arXiv 2407.13197 (Rovelli and Wolpert). The paper's conclusion is that the standard Boltzmann-brain argument contradicts itself and rests on incomplete evidence, so “you don't have to worry” and “the usual argument for the second law is sound”; it says knowledge is never certain but can be very reliable. The essay's step from that to circularity being “a property of all temporal inference” and “inescapable” is the essay's own. The two results it attributes to Wolpert, Rovelli and Scharnhorst (Entropy 2025), that the past hypothesis and the Boltzmann brain hypothesis are formally identical operations, are not in this paper; the paper cites a related earlier piece by Scharnhorst, Wolpert and Rovelli (Boltzmann Bridges, arXiv 2407.02840), which I did not read. The phrase past hypothesis does not appear in the text, and neither does the Boltzmann distribution that the essay says the argument uses.
Rovelli and Wolpert (arXiv 2407.13197, 2024) identified the contradiction. The argument relies on thermodynamic laws — specifically the Boltzmann distribution and the second law — to calculate the probability of such fluctuations. But we inferred those laws from experiments. And those experiments require reliable memories: records that accurately reflect what happened. The second law guarantees memory reliability by ensuring entropy increases toward the future, which makes current records valid traces of past states. If the BB hypothesis is true, entropy did not monotonically increase — it fluctuated — and the memories from which we inferred the very laws used to support the hypothesis are unreliable.
The standard BB argument assumes memory reliability to conclude memory unreliability. It presupposes the past hypothesis to deny the past hypothesis.
Wolpert, Rovelli, and Scharnhorst (Entropy, 2025) formalized the deeper result. The past hypothesis and the Boltzmann brain hypothesis are, formally, identical operations: both condition on a subset of current data. They differ only in which data they condition on. Arguments showing that one is “unstable” or “self-undermining” apply equally to the other. Neither can be established without presupposing itself.
The structural insight: this circularity is not specific to Boltzmann brains. It is a property of all temporal inference. Every reasoning chain from present observations to past laws presupposes that observations are reliable, which presupposes the laws being inferred. The bootstrapping is inescapable. The BB paradox doesn't reveal a problem with cosmology. It reveals the architecture of empirical reasoning: the foundation includes itself.