Many Worlds and the Problem of Believing It

The strangest thing about the Many Worlds Interpretation isn't that it posits an unimaginable number of parallel universes branching off with every quantum event. It's that one of the sharpest objections to it isn't "that's too weird" — it's "you can't rationally believe it even if it's true."

That's the fault line running through a recent conversation on Curt Jaimungal's Theories of Everything podcast, where David Wallace, professor of philosophy of science at the University of Pittsburgh and one of the leading philosophical defenders of Everettian quantum mechanics, sat down with Emily Adlam, his former Oxford student and now a philosopher of physics in her own right. The setup is unusual: a teacher and a student who have clearly spent years talking this through, now doing it publicly, with the intellectual rapport of people who genuinely enjoy disagreeing.

The disagreement is worth understanding carefully, because it's not the kind that gets resolved by more data. It lives in the territory between physics and epistemology — and both of them know it.

The self-undermining problem

Start with what Adlam is actually claiming. She isn't saying the Many Worlds Interpretation (MWI) is false. She's making a more vertiginous argument: if MWI is correct, then the probabilistic reasoning we've used to confirm quantum mechanics might not be reliable in the way we need it to be.

The core of quantum mechanics is supported by empirical evidence — experiments yield outcomes with frequencies that match the theory's probability predictions. But in an Everettian universe, every outcome happens somewhere across the branching structure. The question of why we should interpret our particular branch's frequencies as confirming the theory becomes philosophically fraught. If you can't ground probability in the usual way, you've potentially sawed off the epistemic branch you're sitting on.

As Adlam puts it: "It doesn't make sense to believe an interpretation of quantum mechanics which says we shouldn't believe quantum mechanics. So the whole thing looks a bit self-undermining if the probability problem can't be resolved."

That's a tight argument. It doesn't say MWI is physically wrong. It says MWI might be epistemically incoherent — a theory that, if accepted, gives you reason to doubt the evidence that led you to accept it.

Wallace's response is instructive, and it reveals something important about how the two of them weight different kinds of intellectual risk. He doesn't dismiss the probability problem — he's written extensively about it, and spent a significant portion of his career working on decision-theoretic approaches to Everettian probability. What he resists is the inference that a philosophical puzzle about epistemology should be strong enough to overturn a robustly successful scientific framework.

"I think if there is some kind of clash between deeply held principles of epistemology and the framework of quantum mechanics, I think it's much more likely we'll learn something about epistemology from resolving that than learn something about physics."

This is a genuinely interesting methodological position, not just stubbornness. Wallace is pointing out that analytic philosophy doesn't have an especially strong track record of successfully overturning well-established science on a priori grounds. His implicit claim: when philosophy and successful physics conflict, update your philosophy.

Adlam doesn't fully disagree. She's not abandoning quantum mechanics. Her ideal, she admits with some self-aware humor, would be "an interpretation that preserved quantum mechanics as is and yet was not the Everett interpretation" — while acknowledging she doesn't currently have one. What she's really holding out for is the possibility that something we don't yet understand, perhaps at the intersection of quantum mechanics and gravity, might open a different interpretive route.

The relational dead end

A significant portion of their conversation explores whether relational quantum mechanics (RQM) — the view associated with Carlo Rovelli, which holds that quantum states are always defined relative to a system rather than absolutely — offers a genuine alternative.

It's an attractive idea: maybe the reason quantum mechanics is so strange is that we've been trying to describe things from a "view from nowhere" that doesn't exist. States are always states-relative-to-something, measurements always happen relative to a reference frame, and a universal wave function describing everything from outside is simply not a coherent notion.

Adlam is drawn to this, but cautious. She thinks the relational core is promising while the specific formulations have serious problems. The trouble, she and Wallace eventually agree, is structural: any version of RQM that preserves the full quantum formalism without modification ends up collapsing back into something that looks a lot like Everett.

When Adlam discussed recent work with Rovelli and collaborators on how to define quantum events within RQM, the conclusion was essentially that the only way to make it precise was to look at entanglement from the outside — which is more or less what Everett says. As Adlam puts it: "There is no relational approach that's significantly distinct from Everett which preserves the entire formalism of quantum mechanics and adds nothing to it."

Wallace's summary is characteristically pointed: commit fully to relationalism and you probably just get Everett, complete with the probability problems Adlam wants to avoid. Try to avoid those problems by holding a hybrid view and you get internal inconsistencies. It's a genuine dilemma, not a rhetorical one.

Rationality under branching

There's a question that Wallace says he gets from students constantly: if every possible outcome happens in some branch, what does it matter what you do?

His answer is that there's a meaningful distinction between low-probability quantum branches — yes, there's almost certainly some vanishingly rare branch in which a random neurological malfunction causes someone to do something completely out of character — and the ordinary exercise of rational agency, which tracks the high-weight branches where your decisions actually matter in the relevant sense. Conditional on accepting Everett at all, you're committed to not caring much about what happens in branches with negligible amplitude.

Adlam introduces a more pointed version of the problem via what she calls "Costanza observers" — named, apparently, for a Seinfeld character she's never actually seen. The idea, developed by philosopher Jacob Barandes, is that in any Everettian universe there must exist some very-low-weight branches populated by observers who reason in thoroughly irrational ways and nonetheless succeed wildly. If you can't rule out being such an observer, why should rational constraints bind you?

Wallace finds this less threatening than Adlam does. His response is roughly that any sufficiently large collection of observers — in any possible world — will include some who are irrational and lucky. The question is what you should do given your uncertainty, and the low branch weights still do the work of making such branches negligible for practical purposes. Adlam thinks the argument is somewhat circular: you're using probability-style reasoning to dismiss the problem with Everettian probability.

The QBism detour

Both Wallace and Adlam share a skepticism of QBism — the view that quantum states are just agents' personal probability assignments, not descriptions of anything objective. Wallace's objection is blunt: an interpretation should explain all of known physics, including why helium-4 and helium-3 become superfluid at different temperatures. QBism has nothing to say about that. It handles a stylized abstract formalism and leaves the actual physical phenomena unexplained.

Adlam agrees QBism pushes a genuinely interesting insight too far. The observation that quantum mechanics was developed from a specific standpoint and incorporates aspects of that standpoint is worth taking seriously. But concluding that quantum states are therefore just personal credences, with no fact of the matter about which assignment is correct, severs the connection between the formalism and the observations that justified developing it in the first place.

What the disagreement is really about

Strip away the technical machinery and this is a conversation about where the burden of proof sits in philosophy of science. Wallace thinks a spectacularly successful theoretical framework earns a very strong presumption in its favor — strong enough that clean philosophical arguments against it should prompt us to revise the arguments, not abandon the framework. Adlam thinks epistemic coherence is a genuine constraint that a physical theory has to satisfy, not an optional extra, and that the Everettian probability problem hasn't been solved to her satisfaction.

Neither of them is being unreasonable. They're applying different priors about which is more likely to be wrong: the physics or the philosophical principles used to evaluate it.

What's striking, watching two people who clearly respect each other work through this, is how much genuine agreement there is underneath the headline disagreement. Both think MWI is the most technically workable interpretation. Both think the probability problem is real and worth taking seriously. Both are skeptical of approaches that require large modifications to the formalism. Both are waiting to see what quantum gravity eventually forces on the table.

The actual disagreement is about what to do right now, given current evidence: commit to the best available framework despite unresolved epistemological questions, or hold out for something we don't yet have.

That's not a question physics alone can settle. Which may be exactly why they're both philosophers of physics rather than physicists.

---

— Nadia Marchetti, Unexplained Phenomena Correspondent