What this covers

Daniel Harlow and Sean Carroll discuss what quantum gravity reveals about the foundations of quantum mechanics itself. Harlow, drawing on decades of work in black hole information physics, traces a surprising chain of reasoning: techniques developed to resolve the information paradox—especially holography and the gravitational path integral—when applied to a closed cosmological universe yield a Hilbert space of dimension one. This means only a single state is possible, not that the universe happens to occupy one state among many. The implication is radical: standard quantum mechanics depends on an external classical observer who does not exist in cosmology, so any complete theory must include observers as part of the system and subject them to a fundamental decohering operation. The result is that semiclassical physics and ordinary science emerge only up to errors of order e^(-S_observer).

The conversation ranges across the conceptual fault lines in quantum gravity. Harlow explains how Hawking showed that locality, unitarity, and finite entropy cannot coexist in black holes—forcing physicists to abandon locality and adopt Emergent Spacetime. He argues that the one-state problem persists even when gravity vanishes, suggesting it is foundational rather than gravitational. The Observer Decoherence Postulate is presented as a way forward: treat observers as part of the closed system but decohere them by the laws of nature itself before extracting predictions. Carroll probes tensions with Everettian accounts of time and the role of the Born rule. Throughout, Harlow defends Idealized Thought Experiments as essential to theory-building, and positions the gravitational path integral as an oracle whose answers work but whose physical meaning remains to be decoded.

Sharpest takeaway

Harlow argues that applying the lessons of black hole information physics to a closed cosmological universe yields a Hilbert space of dimension one (a single state), which forces a revision of quantum mechanics itself: the observer must be treated specially, decohered by the laws of nature, so that semiclassical physics emerges only up to errors of order e^(-S_observer).

  • Black hole information is preserved by giving up locality (spacetime is emergent), establishing tools that count degrees of freedom via the gravitational path integral and holography.
  • Applying holography to a closed universe with no spatial boundary implies zero fundamental degrees of freedom — only one possible state.
  • Quantum mechanics relies on an external classical observer that doesn't exist in cosmology, so the observer must be made part of the system and treated with a fundamental decohering channel.

The argument · threads7 threads · 30 claims
0.85

A closed universe yields one state, creating a foundational problem in quantum mechanics itself.

5 pointscentrality 5/5
  • A one-dimensional Hilbert space for a closed universe means there is only one possible state of the universe — not that it happens to be in one state, but that no other state is possible and there is no fact about which state it is in — which appears to conflict with the richness of human experience and with time evolution.

    “There's only one possible state of the universe. It could not be in any other state.”

  • The one-state problem is foundational to quantum mechanics rather than merely gravitational, because as you take Newton's constant to zero the Hilbert space dimension stays one until gravity is exactly zero, when it discontinuously jumps to something large — meaning the problem exists even when gravity is absent and is really about the foundations of quantum mechanics.

    “let's say we have this theory with one state and now we're gonna take the limit G Newton going to zero... Well, the Hilbert space dimension is still one... And then somehow when it gets exactly to zero, then you go from one to something big.”

  • Applying black hole methods (path integral counting and holography) to count the degrees of freedom of a whole closed universe gives the answer zero — equivalently, a one-dimensional Hilbert space — because holography says fundamental degrees of freedom live on the spatial boundary, and a closed universe has no spatial boundary; multiple independent calculations developed for black holes quantitatively confirm this.

    “let's use it to count the number of degrees of freedom for the whole universe as well... and the answer is always the same. And the answer is zero.”

  • Hawking showed that combining general relativity with quantum field theory at weak gravitational coupling leads to a paradox: a black hole behaves as if it has a finite number of degrees of freedom (entropy proportional to horizon area divided by Newton's constant) and radiates thermally, yet you can fit arbitrarily much stuff inside it, so when it evaporates the information about how it was created appears to vanish.

    “Hawking argued that that seemingly obvious way... of combining gravity and quantum mechanics, leads to paradoxes when you try to apply it to black hole physics.”

  • The apparent conflict between a single static state and observed change over time is less mysterious than the one-state problem itself, because time is not a God-given quantity ticked by a divine clock but the time ticked by your watch, which exhibits evolution even in a closed universe with zero energy.

    “what do you really mean by time? You don't actually mean some God-given time ticked by some divine clock. You mean the time that's ticked by your watch. And that has evolution even in a closed universe.”

0.79

Quantum mechanics requires an external observer and is emergent in that limit, not fundamental.

4 pointscentrality 5/5
  • Quantum mechanics does not make sense without an external classical observer or apparatus; it is an emergent theory in the limit where there is an arbitrarily big, slow, careful observer outside the system who can test the probabilistic predictions to arbitrary accuracy — without that observer it is mathematics, not science.

    “I don't really think quantum mechanics makes sense without that external observer. I don't think it's really science. It's mathematics, but it's not science. For me, quantum mechanics is an emergent theory in the limit where you have this external observer and apparatus which is sort of arbitrarily big and slow and careful.”

  • The one-state answer arises because quantum mechanics is being applied to the whole universe, which is only valid if there is an observer outside looking in; the one state is telling us there is no external observer, so for a closed universe we must build a theory in which everything (observers, apparatus) is part of the system, and that theory will only approximately reduce to quantum mechanics in some limit.

    “the reason we got this answer of one state is because we're trying to apply quantum mechanics to the whole universe. And that's only correct if there's an observer who's outside of the universe looking in.”

  • It is not hard to argue that if quantum mechanics is to have a probabilistic interpretation at all, then the Born rule is the unique rule it must use; the real open question is why have a probabilistic interpretation in the first place.

    “I think it's not hard to argue that if quantum mechanics is going to have a probabilistic interpretation, then the Born rule is the one it has to be... But that's not the question. The question is why have a probabilistic interpretation at all?”

  • Quantum mechanics requires an additional axiom beyond the Schrödinger equation: a physical inner product on Hilbert space, which is physical only because of the Born rule; without the Born rule you would have merely invertible linear evolution on a vector space with no physical inner product, so Everettians cannot escape the Born rule because deriving it always smuggles in the inner product.

    “a true dyed-in-the-wool Everettian does not think that quantum mechanics happens in Hilbert space. I think they think it happens in a vector space.”

0.79

Observers must be treated as part of the system, always entangled and made classical only by fundamental decoherence rules.

4 pointscentrality 5/5
  • Harlow's framework posits three axioms for observers: (1) an observer semiclassically has a finite number of degrees of freedom and can do physics only up to e^(-that number) error; (2) an observer is classical, having a stable pointer basis stable under environmental interactions; and (3) observers are never pure — they are always entangled with the rest of the universe because they are produced out of the world they live in, and their pointer basis is the basis where their density matrix is diagonal.

    “the first was that the observer semi-classically should have some finite number of degrees of freedom and they can do physics up to E to the minus that number of degrees of freedom”

  • Harlow's proposal keeps the observer as part of the system but, before computing anything, subjects the observer to a decohering channel on their pointer basis — averaging over their microscopic degrees of freedom to make them classical — not because the observer truly is classical but because the concept of an observer is meaningless otherwise; crucially this decoherence is part of the fundamental laws of nature, not merely environmentally induced, since in a closed universe the environment is also part of the system.

    “I keep the observer as part of the system, but in doing calculations, using the one state, I always subject the observer to a decohering channel that essentially averages over their microscopic degrees of freedom to make them classical.”

  • Wave function collapse is subjective: the wave function collapses for you when you learn the result of an experiment, even though in principle an outside observer could recohere you and the apparatus; the result counts as having happened for you because you became irreversibly decohered with it, and any recohering operation would destroy the convenient fiction of free will and coherent memories — so shared scientific reality rests on mutual decoherence.

    “my rule is that the wave function collapses for you when you learn the result of the experiment.”

  • Science as a concept is fundamentally approximate, with a lower bound on its error given by e to the minus the entropy of the observer (S_observer); you cannot write a number to better precision or believe anything more accurately than that, because at that scale memories and ink could fluctuate into different configurations — and in quantum cosmology, unlike ordinary settings, you cannot take S_observer to infinity to remove this limit.

    “I claim that science as a concept is approximate with a lower bound on the error given by e to the minus the entropy of the observer.”

0.59

Black hole information paradox is resolved by observer-dependent physics with exponentially suppressed accuracy bounds.

3 pointscentrality 4/5
  • Harlow's observer rule provides a mathematics for black hole complementarity: an observer who stays outside an evaporated black hole computes the Hawking radiation to be pure (entropy zero, unitary S-matrix), while an observer who fell inside computes it to be mixed (consistent with entanglement across the horizon), and these differing answers are consistent because the laws of physics differ for the two observers, with each limited to e^(-S_observer) accuracy.

    “this is actually a rule that gives you different rules for different observers and you can study that case.”

  • An evaporating black hole leaves behind a closed baby universe with only one state, whose holographic encoding is a rank-one projection; information falling in hits this projection and bounces off through entanglement in the Hawking modes into the radiation, so the purity of Hawking radiation is crucially tied to the baby universe having a one-dimensional Hilbert space — a two- or three-dimensional space would make the radiation mixed.

    “the holographic encoding of that one state is just a rank-one projection. So the information goes into the black hole, it hits that rank-one projection and bounces off, goes through the entanglement in the Hawking modes out to the Hawking radiation, and that's how the state is pure.”

  • Hawking effectively showed you cannot simultaneously have all three of: (1) the black hole having a finite number of degrees of freedom, (2) information-preserving unitary evolution, and (3) locality (what I do here cannot instantaneously affect there); you must give up one of the three.

    “Hawking said there are three things you might want, you get to pick two.”

0.59

Effective field theory and semiclassical physics emerge mathematically from the one-state model with exponential error limits.

3 pointscentrality 4/5
  • In the models with a single state plus the fundamental observer-decoherence rule, effective field theory and semiclassical physics emerge mathematically, but only up to errors suppressed by e to the minus the observer's entropy; this is poetic because it is exactly the limit on the accuracy of science, and although radical, it is the least bad option Harlow currently sees for making sense of the one-state result.

    “what happens mathematically is that effective field theory emerges, but only up to errors that are suppressed by e to the minus S observer.”

  • Unlike ordinary quantum mechanics where the path integral and canonical (matrix) approaches are equivalent, in gravity the path integral is stronger: by summing over geometries (as Gibbons and Hawking showed in the late 1970s) it correctly knows the number of degrees of freedom and entropy of a black hole without ever counting states, something the canonical formalism cannot reproduce.

    “Gibbons and Hawking showed in the late '70s that by using the path integral approach to gravity... it actually knows how many states the black hole has, even though if you tried to actually count those states by going to the canonical formalism, it wouldn't work.”

  • The positive progress of the last 10 years is a mathematical demonstration that finite entropy, unitarity, and weakened locality (three-star) are mutually self-consistent in toy models — but these models are not realistic, and even a consistent theory is not necessarily the correct one until it predicts something testable.

    “we showed that you can have one, two, and three-star.”

0.46

Studying idealized toy models and stark paradoxes yields structural insight even when immediate testability is lacking.

3 pointscentrality 3/5
  • It is very hard to write down a theory of quantum gravity consistent with all existing data, so most ideas — including many Harlow works on — are ruled out almost immediately, which is why physicists resort to studying unrealistic toy models with the wrong number of spacetime dimensions, the wrong sign of the cosmological constant, or too much supersymmetry.

    “it's very hard to write down a theory that's consistent with the data that we already have. So most ideas are ruled out almost immediately. In fact, including many of the ideas that I work on”

  • An undue focus on what can currently be measured is not constructive when trying to construct a new theory; like Einstein's near-light-speed trains or Maxwell's analysis of Saturn's rings (confirmed only 100+ years later), studying stark paradoxes in idealized situations is valuable for finding a theory, with testable terrestrial predictions coming later once the structure is understood.

    “when you're trying to come up with a theory in the first place, I think a sort of undue focus on what can currently be measured is not constructive. That comes later once you understand the structure of the theory better.”

  • A recurring lesson over 50-100 years of quantum gravity research is that black holes are easier to analyze than cosmology, because with a black hole you can sit outside as an external observer and drop things in, whereas in cosmology you are always part of the system that is interacting with you.

    “something that we've learned again and again over the last 50 to 100 years of trying to do quantum gravity is that black holes are easier than cosmology.”

0.38

Gravity's universality under the equivalence principle gives it explanatory power beyond messy standard model forces.

4 pointscentrality 2/5
  • The gravitational path integral should be treated like the Oracle at Delphi: it gives you correct answers when consulted, but you do not always understand what those answers mean, and misinterpreting them can be fatal — so one should seek a more fundamental framework from which the path integral's consequences can be derived rather than treating it as the end of the story.

    “I think of it like the Oracle at Delphi. It's something that you consult and it tells you the answer, but it's not clear that... You don't always understand the answer that you're given.”

  • Gravity is special because everything feels it in the same way (the equivalence principle), unlike the standard model forces where some fields feel a force and others don't; this universality gives gravity an inevitability and generality of argument that the messy 19-parameter standard model lacks.

    “there is something universal about gravity that is not so much the case for the other forces of nature. If you look at the standard model, some of the fields feel electromagnetism, some of them don't”

  • The resolution many in the field have converged on over the last 20 years is to give up locality — captured by the slogan that spacetime is emergent, meaning the notion of where and when things are is only true in certain approximations; you can have finite entropy, unitarity, and a weakened 'three-star' locality where violating locality requires doing something exponentially complicated in the black hole entropy.

    “the view that many of us have converged on gradually over the last 20 years is that you have to give up option three, which is locality.”

  • Black holes can only exist because everything feels gravity the same way; if a particle could be neutral under gravity and not feel it, it could escape from a black hole, so there would be no black hole.

    “the only reason that black holes can exist is because everything feels gravity in the same way. Because if you had a particle that didn't feel gravity, it could escape from a black hole.”