36 | David Albert on Quantum Measurement and the Problems with Many-Worlds
What this covers
David Albert and Sean Carroll discuss the measurement problem in quantum mechanics and Many-Worlds Interpretation, the framework that takes the wave function and its deterministic evolution as fundamental without invoking collapse or adding extra rules for observation. Albert, a leading figure in the foundations of physics at Columbia, presents the philosophical case against Many-Worlds—not objections rooted in intuitive unease at branching universes, but serious technical problems with how the interpretation accounts for probability and chance in a fully deterministic system.
Albert's central concern is that Many-Worlds has failed to derive genuine probabilities from a branching determinism. He examines the two main attempts: the decision-theoretic approach, which he argues illegitimately assumes that preferences among non-branching futures can constrain preferences among branching ones; and approaches based on self-locating uncertainty, which require irreducibly indexical facts about which branch an observer finds themselves in. Since rival self-locating theories agree on all physical facts, Albert contends, no experiment can adjudicate between them—a problem that classical statistical mechanics never faced. He also critiques the collapse postulate of Copenhagen-style interpretations for resting on vague terms like "measurement," and explains John Bell's productive reframing of the problem, alongside the hidden variables approach of David Bohm. The conversation touches on the history of Hugh Everett's original formulation, how John Wheeler muddied Everett's thesis before publication, and the geographic concentration of Many-Worlds sympathy at Oxford around figures like David Deutsch and David Wallace.
Albert argues that while Many-Worlds takes the quantum equations seriously without ad hoc additions, it has not solved the problem of where probability and chance come from in a fully deterministic, branching universe, and that the leading attempts (decision-theoretic and self-locating/epistemic) both fail or invoke metaphysically dubious facts.
- The decision-theoretic derivation of Born-rule probabilities wrongly assumes non-branching preferences fully constrain branching preferences.
- Self-locating probability requires irreducibly indexical facts that cannot be empirically confirmed the way statistical-mechanical probabilities can.
- Because rival self-locating theories agree on all physical facts, no experiment can adjudicate between them, crippling physics in a way the classical case avoided.
Everett's deterministic branching world cannot ground genuine probability or explain our experience of definite outcomes.
- Everett's solution interprets the post-measurement superposition literally as describing multiple coexisting branches — in one branch the particle is at A and the observer sees A, in another the particle is at B and the observer sees B — and we are unaware of the splitting because each branch's observer perceives only one definite outcome.
“there were really two things there, there are, as it were, two internally coherent stories or worlds going on there. And it follows immediately that of course we wouldn't be aware of this kind of splitting because our mental state is either seeing it at A and not at B, or seeing it at B and not at A.”
- The central, coherence-level problem for Everett is that quantum mechanics is fundamentally a probabilistic theory whose empirical support rests on observed frequencies matching probabilistic predictions, yet Everett's deterministic, branching evolution leaves no room for genuine chance — so it is unclear how the theory can explain our experience of probabilistic outcomes at all.
“there's just a very prima facie tension between the claim that the laws of the evolution of the world are completely deterministic, and the appearance of these pervasive chancinesses or probabilities in our experience.”
- Just as Newtonian mechanics counter-intuitively predicts the Earth is moving rapidly yet also predicts that if it were, we wouldn't feel it, the linearity of the quantum equations both predicts the puzzling superposition of observers and predicts that if that were the case we would not notice it and would testify to the contrary — so the apparent falsity of the prediction is explained away by the theory itself.
“the very same laws that predict that the Earth is in motion, predict also that if that were true, you would think it wasn't... This same linearity that leads to this very puzzling and apparently false prediction of superposition... also predicts that if that were what was going on, you wouldn't know it, you would think otherwise.”
- Because the quantum equations of motion are linear, the condition of being superposed is 'infectious': any system that interacts with a superposed system becomes superposed too, which is why measuring devices and observers inevitably get entangled into the superposition.
“what the linearity of the quantum mechanical equations of motion entail is that this condition of being superposed in this way, turns out to be fantastically infectious. You touch it and you've got it too.”
- If quantum mechanical laws govern everything including measuring devices and brains, then because the equations are linear, interacting with an electron in a superposition of positions deterministically yields a state in which there fails to be a fact of the matter about which way the pointer points or which outcome the observer perceives — directly at odds with our definite empirical experience.
“the very same laws that predict that the Earth is in motion, predict also that if that were true, you would think it wasn't... This same linearity that leads to this very puzzling and apparently false prediction of superposition... also predicts that if that were what was going on, you wouldn't know it, you would think otherwise.”
- An electron in a superposition of positions is not in some definite location we are merely ignorant of; rather there is no fact of the matter about its position at all, analogous to asking about the marital status of the number 5 — a category mistake rather than missing information.
“asking about the spatial location of the electron would be analogous to asking about the marital status of the number 5, or about the length in meters of Catholicism, or something like that, something that philosophers sometimes refer to as a category mistake”
Self-locating probability in Everett cannot be empirically tested since all theories predict identical physical configurations.
- The crucial disanalogy that cripples Everettian probability is that self-locating probabilities attach not to physical configurations of the world but to where one finds oneself among identical branches; since all candidate probability theories agree completely on the physical configuration, no experiment can confirm or disconfirm them universally, leaving only an a priori argument that, if rejected, leaves physics with no empirical recourse.
“since the probabilities do not attach to claims about the physical configuration of the world, there isn't gonna be an empirical access to them.”
- Sidney Coleman and others argued that the deterministic equations imply that as the number of repeated identical experiments goes to infinity, the world approaches a state where, although there is no fact about any individual outcome, there is a definite fact that the proportion of up results approaches exactly one half — but this fails because before the infinite limit the theory says there are no facts about anything, and we never perform infinitely many experiments.
“you can prove a theorem very easily that is the number of times, call it n, that you've repeated that experiment goes to infinity. The world approaches a funny state in which even though there is no fact of the matter about how any individual one of those experiments came out... there is a perfectly definite fact of the matter about the proportion of them that came out up”
- When a branching experiment is repeated, there will always be a real branch whose observer saw frequencies confirming any given probabilistic theory; the only way an Everettian can claim a theory is confirmed is in a purely, irreducibly indexical sense ('confirmed for me'), and taking such irreducibly indexical facts into fundamental physics is metaphysically puzzling and undesirable since other interpretations solve the measurement problem without them.
“The sense in which one of these probabilistic theories is confirmed or disconfirmed by such experiments, is a purely indexical sense, an irreducibly indexical sense, that is, it is not possible, it doesn't make any sense to ask, did this sequence of experiments confirm or disconfirm the theory?”
Decision-theoretic arguments fail to show rationality requires treating Everett branches as chancy rather than merely consistent with doing so.
- Albert's core objection to the decision-theoretic derivation is that it is simply mistaken to think one's preferences among non-branching futures can constrain, on pain of irrationality, one's preferences among branching futures; branching opens an entirely new set of options (e.g. preferring to be richer in the branch where one is fat), and there is no reason the limited non-branching preferences should determine these new preferences.
“It's just nuts to think that your preferences among non-branching futures could in any way constrain on pain of straightforward irrationality your preferences among branching futures. The branching futures are a whole new set of flavors available to you”
- The Sebens-Carroll approach proposes that during the interval after a measurement is completed but before the observer looks, the observer knows the physical state of the world completely yet is genuinely uncertain which branch they are now in, providing self-locating uncertainty into which ordinary chance can get a foothold.
“I know that I'm now already either in the electron was at point A world or I'm in the electron that was in point B world. But I don't know which. So, here it seems as if we have the sort of thing we wanted. We have some genuine uncertainty.”
- In response to such objections, David Wallace's claims about what the decision-theoretic argument establishes have become more modest, now framed as showing the consistency of treating the branching world as chancy rather than the rational necessity of doing so.
“David Wallace's claims about what this kind of argument establishes are now a little bit more modest than they used to be, that they're clearly showing the consistency of such a thing, but not necessarily the rational necessity of such a thing.”
- The Deutsch-Wallace (Oxford) program concedes there are no genuine probabilities in Everett, then argues that a rational agent who knows she will deterministically split would accept or decline exactly the same bets she would in a different, genuinely chancy universe obeying the von Neumann picture — so the practical and cognitive roles of chance are recovered, and that behavioral match is all 'the universe is chancy' means.
“the kinds of bets that a rational agent would accept or decline would just happen to be exactly the same ones that she would accept or decline in a completely different universe which was chancy.”
The collapse postulate and its substitutes—measurement, consciousness, macroscopicity—all rely on vague terms unsuitable for fundamental physics.
- Eugene Wigner proposed that conscious agency causes the collapse of the wave function, drawing the line between systems that obey standard quantum evolution (inanimate) and those that cause collapse (animate), which he welcomed as evidence that physics requires a non-physical mind; but this animate/inanimate distinction is as hopelessly vague as the term 'measurement' it was meant to replace.
“Wigner had this idea that it was precisely conscious agency that caused departures from the standard quantum mechanical equations of motion that caused the collapse of the wave function. Indeed, he thought there was in this a new definition of the difference between animate and inanimate physical objects.”
- The collapse postulate is fundamentally inadequate because it places the ordinary, imprecise word 'measurement' at the foundational level of physics; replacing it with equally vague substitutes like 'macroscopic,' 'thermodynamically irreversible,' or 'conscious' constitutes no real progress, since all such terms lack the precision required of a fundamental physical law.
“It involved at the fundamental level an English word "measurement," okay? That things behave differently when a measurement is going on and when a measurement is not going on. And it was clear that this English word... didn't have anything approaching the requisite precision to play that kind of a role in a fundamental physical account of the world.”
- The Copenhagen approach posits two distinct fundamental laws of evolution: the linear Schrödinger equation that governs when no measurement occurs, and a stochastic collapse postulate that operates during measurement to select one definite outcome, making measurement an active rather than passive process.
“apparently there are two fundamental laws of the evolution of the physical state of the world. One of which... are these linear equations of motion, and the other, which is in force when measurements are going on is something called the collapse postulate which selects out one or the other of these possibilities for you”
Foundational quantum mechanics was suppressed by institutional forces, geographical isolation, and historical personalities rather than philosophical merit.
- Everett's original paper is unclear at crucial points not through Everett's own fault but because his thesis advisor John Wheeler, afraid of Niels Bohr's reaction, dulled and muddied Everett's point, even writing letters to Bohr trying to assure him Everett was not really criticizing him.
“the fault of his very unresponsible, in this case, thesis advisor, John Wheeler, who muddied and dulled Everett's point all over the place because he was afraid of Bohr's reaction.”
- Despite universal access to communication technology, attitudes toward the foundations of quantum mechanics are strikingly geographically well-defined, with Oxford serving for many years as the isolated epicenter of sympathy to the Everettian view.
“attitudes about the foundations are incredibly geographically well-defined... Oxford was for many years the sort of isolated epicenter of sympathy to the Everettian view.”
- Niels Bohr's enormous influence over the foundations of quantum mechanics derived from extraordinary personal charisma rather than the clarity of his arguments, since brilliant people who spent an hour with him had their lives changed yet then spouted gibberish about foundations for the rest of their careers, and his actual writings (such as his response to EPR) are nearly impenetrable.
“there was just this long string of brilliant people who would spend an hour with Bohr, their entire lives would be changed. And one of the ways in which their lives were changed is that they were spouting gibberish that was completely beneath them about the foundations of quantum mechanics for the rest of their lives.”
- The term 'hidden variables' is a historical misnomer: in a theory like Bohm's, these added variables are not hidden at all but are precisely the things you observe — the actual particle positions — while the wave function does the unobserved guiding.
“These variables for unfortunate historical reasons were called hidden variables. But if you study something like Bohm's theory, they're the opposite of hidden. They are the thing you see”
- The theoretical physics community's neglect of Bohm's 1950s hidden-variable theory was enabled in part by political forces: Bohm was harassed out of the United States by the McCarthy committee soon after publishing, which made it much easier to ignore his work.
“one of the things that allowed people to ignore his work was the work of the McCarthy Committee in the '50s. He was harassed out of the country soon after publishing his theory. That was something that made it much easier to ignore his work.”
- Yakir Aharonov possessed an almost contradictory ability — a quick, sure intuitive grasp of how quantum systems would behave, routinely predicting astonishing outcomes correctly without being able to explain how, such that observers could not learn his method even while watching him.
“I don't know that I've ever met anyone in my life who had as quick and as sure an intuitive grasp... about how quantum mechanical systems are going to behave”
- Worrying about the foundations of quantum mechanics was so profoundly unpopular in physics departments of the late 1970s and early 1980s that Albert faced proceedings to expel him from his PhD program and was forced to complete an assigned, calculation-heavy thesis instead of one he chose.
“There were within fairly short order proceedings instituted to expel me from the PhD program... worrying about issues like that was profoundly unpopular.”
Bell correctly framed the measurement problem as a forcing dichotomy between incomplete wave functions and modified equations.
- John Bell framed the measurement problem as a dilemma: either the Schrödinger equation isn't everything (the wave function is an incomplete description requiring additional variables, as in Bohm) or it isn't right (the equations must be changed to add stochastic collapse, as in GRW); this dichotomy organizes the serious solutions to the problem.
“Bell nicely said in a paper that what the measurement problem shows is that either the Schrödinger equation isn't everything or it isn't right.”
- In contrast to the Wigner, macroscopic, and thermodynamic-irreversibility camps, John Bell is the figure who showed how to think through the measurement problem correctly, establishing a criterion for what could even count as a potential answer and how to evaluate candidates against one another, setting the field's productive agenda ever since.
“Bell is widely regarded as the person who showed how to think through these questions correctly, who established a sort of criterion as to what kind of thing could even be counted as a potential answer to this problem and what couldn't”
- Among objective collapse theories, the GRW (Ghirardi-Rimini-Weber) approach is probably a better strategy than Penrose's gravitationally-induced collapse approach.
“I think there were reasons why GRW's strategy is probably better than Penrose's strategy.”
Classical statistical mechanics shows the principle of indifference cannot justify probability assignments from epistemic symmetry alone.
- In classical statistical mechanics the probability distributions are not a priori truths but empirical statistical laws whose numerical values we believe because of their empirical success; they feel intuitive only because such general statistical regularities have been deeply hardwired into us by natural selection since we were fish.
“they have the status of statistical empirical laws of how the world is arranged, okay? And we believe them precisely because of their empirical success.”
- The a priori principle of indifference — that perfect symmetry in one's epistemic situation between two possibilities requires assigning each 50% probability — is suspect because in classical statistical mechanics it is used to derive substantive empirical facts (like heat flowing from hot to cold bodies) seemingly for free from 'I have no clue,' when an equally symmetry-respecting and more honest response is simply to refuse to assign any probabilities at all.
“I went from, "I have no clue," to, "Of course I should expect that heat is gonna flow from hotter bodies to cooler bodies."... How did I get all of this for free?”