Daniel Harlow on Quantum Gravity, Black Hole Information, and the Holographic Principle

Daniel Harlow, a quantum-gravity theorist at MIT, traces the line from Hawking’s fifty-year-old black hole information paradox through the modern claim that spacetime itself is emergent — and out the other side to a startling conclusion: applied to the whole universe, the same tools say a closed cosmos has exactly one quantum state, which may force a rewrite of quantum mechanics itself.

Key ideas

1. Gravity’s universality is what makes quantum gravity tractable. Unlike the other forces, everything feels gravity the same way — the equivalence principle — so lessons drawn from unrealistic toy models (wrong number of dimensions, wrong sign of the cosmological constant) plausibly carry over to the real theory. Black holes exist only because nothing can be neutral to gravity: a particle that ignored gravity could simply climb out.

2. Hawking’s paradox is a forced choice among three goods. A black hole behaves like an ordinary thermal system with a finite number of states (set by its horizon area), yet the standard approximation of quantum fields on curved spacetime says it can hold infinitely much. Hawking showed you cannot keep all three of: finite entropy, information preservation (unitarity), and locality (no instantaneous action at a distance). Pick two.

3. The modern resolution sacrifices locality: spacetime is emergent. Over the past fifteen years the field converged on keeping finite entropy and unitarity while giving up strict locality — but only by an amount so tiny it shows up only in operations exponentially complicated in the black hole’s entropy. Location and time are approximate notions, not fundamental ones.

4. The gravitational path integral ‘knows’ things it should not. Feynman’s sum-over-histories, applied to gravity, returns the correct count of a black hole’s states without ever counting them, and (since 2019) reproduces the unitary evaporation Hawking’s own method denied. Harlow treats it like the Oracle at Delphi — it gives right answers you must be careful to interpret correctly.

5. Turn the same crank on a closed universe and you get one state. The holographic principle puts the fundamental degrees of freedom on a region’s spatial boundary; a closed universe has no boundary, so the count is zero — a one-dimensional Hilbert space, a single possible state. Harlow reads this as a signal that ordinary quantum mechanics, built for an observer standing outside the system, simply does not apply to a cosmos that contains its own observers.

Summary

Why quantum gravity is hard — and why progress is still possible

Carroll opens on the frustration of fundamental physics: the Standard Model was finished in the 1970s, yet gravity still resists being folded into the quantum framework. Harlow’s reason for optimism is gravity’s universality. The Standard Model is a ‘smorgasbord’ — nineteen dimensionless numbers fitted to data, different fields feeling different forces in different ways. Gravity has none of that mess: by the equivalence principle everything responds to it identically, which is exactly why a black hole can trap anything at all. That universality means arguments about gravity generalise. Harlow works in deliberately unrealistic settings — one spatial dimension instead of three, the wrong sign of the cosmological constant, extra supersymmetry — because the gravitational part of such models looks alike even when the rest differs, so what he learns there should survive into the true theory.

The black hole information paradox

Fifty years ago Hawking took general relativity interacting weakly with quantum fields — the seemingly obvious way to combine the two — and found a contradiction. A black hole acts like a finite thermal system: it has an entropy and an energy, it obeys thermodynamics, and it radiates. Yet the same approximation says you can stuff an arbitrarily large amount of matter inside, so the internal states are really infinite. As the hole evaporates and shrinks to nothing, where did the information about what fell in go? Harlow frames Hawking’s result as a trilemma — finite entropy, unitarity (information is preserved, in the technical sense that quantum evolution is reversible), and locality (what I do here cannot instantly affect Baltimore) — of which you may keep only two. The relativists’ traditional escape, a ‘baby universe’ pinched off to carry the information away, keeps locality but abandons the finite state count.

Emergent spacetime and the path integral

The consensus Harlow describes instead keeps finite entropy and unitarity and surrenders locality — but only faintly. Locality fails only under operations exponentially complicated in the black hole’s entropy, something no experiment has ever come close to probing. The slogan is that spacetime is emergent: the notion of where and when things are holds only approximately, in certain regimes. The mathematics of that emergence is the genuinely new work of the last decade or so. A second, less fundamental route runs through the gravitational path integral — Feynman’s recipe of summing over all possible histories with a weight fixed by the classical physics. In ordinary quantum mechanics the path-integral and the matrix (‘canonical’) formulations are equivalent; in gravity the path integral is mysteriously stronger. Gibbons and Hawking showed in the late 1970s that it delivers a black hole’s state count directly, and in 2019 it was found to encode the unitary evaporation. Harlow trusts it the way the Athenians consulted the Oracle at Delphi — the answers are right, but misreading them (wooden walls: a hedge, or the ships?) can prove fatal, so he prefers a more fundamental account from which the oracle’s pronouncements can be derived.

The holographic principle and the one-state universe

Holography, inspired by the fact that a black hole’s entropy scales with its surface area rather than its volume, says the fundamental degrees of freedom of any region live on its boundary. Apply that to cosmology — specifically a closed universe, one with no spatial edge, such as a three-sphere — and the count of fundamental degrees of freedom is zero. Not approximately: the Hilbert space is one-dimensional, meaning there is exactly one possible state of the universe, with no fact of the matter about ‘which’ state it is in. A bar-room version of the argument is decades old, but Harlow’s point is that the modern black hole calculations now back it quantitatively. Classically a closed universe is rich — ‘lots of things can happen’ — so a single quantum state is shocking, and, worse, it would let us deduce the global shape of the cosmos without looking out of the window.

What quantum gravity teaches about quantum mechanics

Harlow’s reading is that the one-state result is not a mistake but a diagnosis: quantum mechanics was always a theory for an observer outside the system — the arbitrarily large, slow, careful apparatus of Bohr and Landau that measures probabilities to unlimited precision. In cosmology there is no such outside; the observer is part of the closed system. His proposed fix keeps us inside the system but treats the observer specially, hitting their state with a decohering channel (a mathematical step that averages over their microscopic detail to render them effectively classical) as a law of nature rather than a consequence of interacting with an environment. From this, ordinary effective physics re-emerges — but only up to errors of order e^(−S), where S is the observer’s own entropy. Science, on this picture, has a built-in precision floor set by who is doing it. The claim survives even when gravity is switched off, which is why Harlow thinks it is really about the foundations of quantum mechanics. Carroll, a committed Everettian who holds that observers live inside the quantum state and that many-worlds already handles cosmology, pushes back hard throughout the closing exchange — the two agree only that the question is open, and that changing quantum mechanics to make gravity work was not what either had bargained for.

Speakers

• Daniel Harlow — theoretical physicist at MIT; works on quantum gravity, black hole information, holography, and, latterly, quantum cosmology.
• Sean Carroll — host; theoretical physicist and author, Mindscape.

See also

• Quantum Measurement Problem — the concept page: what happens on measurement, and whether an external observer is needed; Harlow’s proposal is a fresh entry in this debate
• Many-Worlds Interpretation — Carroll’s position in the closing argument; Harlow resists it for cosmology
• Sean Carroll on Quantum Mechanics, Many Worlds, and the Problem of Structure — the host’s own account of the interpretation he defends here
• Janna Levin on Black Holes, Wormholes, and Quantum Mysteries — a companion Mindscape-adjacent treatment of black holes and quantum puzzles
• Michael Nielsen on Collaboration, Quantum Computing, and Civilization's Fragility — quantum information theory, the toolkit behind emergent spacetime
• Sean Carroll — host