Black-Hole Information

A black hole turns ordinary questions about quantum information into a test of what the words system, observable, and evolution mean in gravity. Hawking’s semiclassical calculation predicts thermal radiation while the classical interior ends at a singular boundary; unitary quantum mechanics instead requires the complete state to retain every recoverable correlation. This volume develops the tension precisely. It moves from channels and fine-grained entropy through Page curves, complementarity, quantum extremal surfaces, replica wormholes, reconstruction, complexity, microstates, and operational tests. At every stage it separates a theorem from a model calculation, a controlled approximation from an extrapolation, and an entropy result from a microscopic mechanism of recovery.

Helpful background. Black-Hole Thermodynamics and Semiclassical Gravity supplies horizon thermodynamics, generalized entropy, and backreaction; Quantum Information for Black-Hole Physics supplies states, channels, purification, entropy inequalities, and recovery; Quantum Fields in Curved Spacetime supplies the fixed-background origin of Hawking radiation and the continuum-field qualifications used below.

What is the information problem?

There is no single contradiction called “the information paradox.” A precise version begins by naming five ingredients:

1. the complete input and output systems, or their observable algebras;
2. the initial state and the allowed family of states;
3. the dynamical map, including boundary and endpoint conditions;
4. the observers and operations by which information is diagnosed; and
5. the approximation that supplies the semiclassical spacetime.

For an asymptotically complete process, one may ask whether an in-state $\rho_{\mathrm{in}}$ is mapped to an out-state by a unitary S-matrix,

$$\rho_{\mathrm{out}} =S\rho_{\mathrm{in}}S^\dagger, \qquad S^\dagger S=SS^\dagger=\mathbf 1.$$

But this equation is meaningful only after the asymptotic Hilbert spaces, charges, soft or superselection sectors, and endpoint have been specified. A mixed state of the collected radiation can arise innocently by tracing an environment from a larger pure state. Fundamental information loss is the stronger claim that evolution on the physically complete system loses recoverable information. A formal channel dilation can always introduce an auxiliary environment in the finite-dimensional setting; it does not show that the proposed gravitational theory contains that environment or makes its information available among the complete final observables. Hawking’s original superscattering formulation makes that distinction sharp; the density-operator setup is developed in Hawking 1976, §§II–IV, pp. 2462–2471.

Gravity complicates the subsystem language itself. Gauge constraints and gravitational dressing can obstruct an exact tensor product between “inside” and “outside.” A nongravitating bath can nevertheless provide an ordinary radiation factor, while an asymptotic algebra can provide well-defined observables without postulating a local factor at the horizon. Pages therefore state the algebra, state, regulator, and observer before using an entropy or a recovery map. The perturbative dressing and noncommuting-algebra obstruction are derived in Donnelly and Giddings 2016, §§2–5; their asymptotically anchored, perturbative setting is not itself a theorem about every nonperturbative gravitational algebra.

Entropy is a diagnostic, not a mechanism

Begin with two qubits in the Bell state

$$|\Phi^+\rangle_{BR} =\frac{|0\rangle_B|0\rangle_R+|1\rangle_B|1\rangle_R}{\sqrt2}.$$

The joint state is pure, yet either qubit alone has reduced state $I/2$: an observer measuring one qubit sees a fair random outcome. The missing information is in joint correlations. Thus $S(BR)=0$ while $S(B)=S(R)=\ln2$. This finite example explains why a mixed radiation state does not, by itself, establish information loss. The black-hole question adds the endpoint: after complete evaporation, what physical system, if any, still purifies the radiation?

For a normalized density operator on a declared region or algebra, the fine-grained entropy is

$$S_{\mathrm{vN}}(\rho)=-\operatorname{Tr}(\rho\log\rho).$$

Unitary evolution preserves the entropy of the complete density operator, but the entropy of radiation may rise or fall as correlations move between the remaining hole and its environment. In an idealized finite bipartite system, the random-state benchmark predicts that the smaller factor is nearly maximally entangled. Page conjectured the exact Haar average, and Sen later proved it; see Page 1993, Eqs. (1)–(11) and Sen 1996, complete proof. Its crossover when the effective dimensions exchange order produces the Page curve. Hawking’s leading semiclassical result instead continues the growth of radiation entropy because each newly emitted mode is correlated with degrees of freedom not included in the retained exterior state.

The contrast is powerful but incomplete. Two microscopic channels can have the same entropy curve and radically different density matrices, decoders, or observable effects. A calculation that reproduces a Page curve need not give individual S-matrix elements, identify how an interior mode reaches the radiation, or make the information experimentally accessible. The modern replica and island results, their controlled regimes, and this limitation are surveyed in Almheiri et al. 2021, §§4–11 and Appendix A.

The same discipline applies to generalized entropy,

$$S_{\mathrm{gen}}[X] =S_{\mathrm{grav}}[X]+S_{\mathrm{bulk}}[\Sigma_X],$$

where $X$ is a codimension-two candidate, $S_{\mathrm{grav}}$ includes the renormalized area or its higher-derivative replacement, and $S_{\mathrm{bulk}}$ is the quantum entropy of a specified bulk region and state. Extremizing this functional identifies quantum extremal candidates; anchoring, homology, admissibility, and minimization select the leading one. Neither the extremality equation nor the minimum alone supplies a microscopic Lorentzian transfer process.

The thirteen-part argument

The chapters are ordered so that every later claim can expose which earlier assumption it uses.

Chapter	Central question	Result carried forward
Formulations of the Information Problem	Which complete system, algebra, state, and map define the problem?	A distinction among fundamental loss, ordinary reduction, and operational inaccessibility.
Fine-Grained Entropy and Page Curves	What do radiation entropies actually diagnose?	Random-state and semiclassical benchmarks with their assumptions exposed.
Unitarity, No-Cloning, Complementarity, and Observers	Can exterior recovery and smooth infall coexist without an observable duplicate?	Channel, causal-patch, and decoding criteria.
Firewalls, State Dependence, and Interior Consistency	Which assumptions create the AMPS tension, and what can consistently replace one of them?	An explicit interior-consistency test rather than a slogan.
Generalized Entropy and Quantum Extremal Surfaces	How are candidate entropy saddles defined and selected?	Renormalized generalized entropy, quantum extremality, homology, and causal qualifications.
Replica Gravity, Wormholes, and Islands	How do gravitational replicas produce island saddles?	Integer-replica constructions, continuation assumptions, and controlled evaporation examples.
Factorization, Ensembles, and Gravitational Path Integrals	What do connected bulk saddles compute for disconnected boundaries?	A separation of fixed-theory, ensemble, and baby-universe interpretations.
Entanglement-Wedge and Interior Reconstruction	Which operator algebra can be recovered from which exterior region?	Code-subspace recovery statements with explicit error and domain.
Quantum Error Correction, Complexity, and Chaos	When is information spread, recoverable, and feasibly decoded?	Distinct scrambling, decoupling, chaos, and complexity diagnostics.
Microstates, Typicality, and Coarse Graining	Why can restricted observables look thermal in a pure state?	State-counting and typicality tools without assuming a common exact geometry.
Quantum Singularities and the Fate of the Interior	What would it mean for quantum gravity to resolve the interior?	Non-equivalent regularity, evolution, bounce, final-state, and nonlocality criteria.
Operational Observables and Limits of Verification	Which proposed diagnoses can an observer implement?	Resource-, timing-, dressing-, and backreaction-aware protocols.
Comparative Requirements on Information-Recovery Proposals	How can competing frameworks be compared without premature verdicts?	Six paired axes: unitarity/entropy, horizon/EFT, interior/causality, factorization/microscopic completion, dynamics/errors, and operations/tests.

Three models, used for different jobs

The volume reuses three deliberately limited examples. The first is a finite system $B$ coupled to radiation $\mathcal R$. Effective dimensions $d_B(t)$ and $d_{\mathcal R}(t)$ make Page’s random-state theorem, decoupling, no-cloning, and resource estimates concrete. This is a quantum-information benchmark, not a model of local horizon dynamics.

The second is a large semiclassical black hole coupled transparently to a nongravitating bath. It makes a radiation subsystem unambiguous and permits no-island and island candidates to be compared. The setup is particularly useful because it isolates the gravitational entropy calculation from the question of whether a realistic asymptotically flat black hole admits the same control.

The third is Jackiw–Teitelboim gravity and its matrix-integral completion. It provides calculable topology expansions, replica wormholes, connected spectral correlations, and a precise ensemble interpretation. It is not evidence that every individual holographic theory is an ensemble. The distinction between a controlled low-dimensional equality and a proposed universal explanation is maintained throughout.

Notation follows these jobs. $B$ denotes the remaining hole, $\mathcal R$ all collected radiation, $E$ early radiation, $b$ a late exterior mode, and $\widetilde b$ its interior partner. In reconstruction chapters, $A$ instead denotes a boundary region. Entropies use natural logarithms unless qubits are explicitly requested; a value in nats is divided by $\log 2$ to obtain bits. Euclidean replica calculations state their contour and normalization, and Lorentzian conclusions are not inferred from a Euclidean saddle without an additional argument.

Choose a reader route

For a first systematic pass, read the chapters in order through quantum extremal surfaces, then continue through replicas, factorization, and reconstruction. This route shows why an island formula answers a carefully posed entropy question while leaving further microscopic and operational questions.

For a shorter first pass, follow these three linked questions:

1. Why can radiation be mixed while the total state is pure? Learn to trace out a purifier before diagnosing fundamental loss.
2. Why should radiation entropy turn over? Work through the finite-system Page benchmark and its limitations.
3. What changes when an island contributes? Identify the joint entropy region and compare extremized candidates.

For the logical structure of the paradox, continue through complementarity and AMPS, then operational verification and comparative requirements. For the quantum-information route, connect algebraic recovery to scrambling and decoding and typicality. For interior physics, connect state dependence to reconstruction, microstates and quantum singularities.

A useful self-diagnostic is to ask four questions about any claim encountered along the way:

• Which state and complete observable algebra are being discussed?
• Is the result a theorem, a controlled gravitational saddle, or an extrapolation?
• What norm, entropy, correlator, or operational probability measures success?
• Which observer has the time, copies, control, and energy needed to perform the test?

If any answer is missing, the statement is not yet a complete information- recovery claim.

Boundaries with neighboring subjects

This volume does not rederive the Bogoliubov transformation or renormalized stress tensor. Those fixed-background foundations are treated in Quantum Fields in Curved Spacetime. It does not replace the generalized second law, semiclassical backreaction, or stable black-hole thermodynamics treated in Semiclassical Gravity. Instead it asks what their outputs imply for fine-grained purity and recovery.

General gravitational dressing, relational observables, and candidate microscopic programs are developed in Quantum Gravity. AdS/CFT dictionaries, holographic entropy derivations, D-brane state counting, and explicit fuzzball or microstate-geometry constructions belong to String Theory and Holographic Black Holes. Here those subjects enter only where they answer an information-theoretic criterion. Classical horizon definitions, Penrose diagrams, mass inflation, and singularity theorems remain with the geometry, global-structure, and dynamics volumes.

These boundaries prevent a common error: treating success in one layer as success in all layers. A unitary microscopic theory can lack a reconstructed semiclassical interior; a QES calculation can reproduce an entropy without an efficient decoder; a smooth microstate geometry can be unrepresentative of a typical state; and an analog horizon can test kinematic Hawking robustness without testing gravitational unitarity.

Results and open questions

The evidence summary below is current through 2026-08-19. It separates results whose mathematical content is established from the still-unresolved question of what they imply for a single microscopic black-hole theory.

Result or proposal	Evidence status	Scope and limitation
Page’s average-entropy formula, Page 1993, Eqs. (1)–(11) and Sen 1996, complete proof	Proved for a Haar-random state in a finite bipartite Hilbert space	It is a benchmark, not an evaporation Hamiltonian or a locality statement.
Quantum extremal surfaces and islands, Engelhardt and Wall 2015, §§2–4 and Almheiri et al. 2020, §§2–6	Derived semiclassically in controlled holographic and low-dimensional bath models	Replica continuation, contour choice, nonperturbative corrections, and general four-dimensional applicability remain additional questions.
Islands beyond an auxiliary nongravitating bath, Antonini et al. 2025, §§2–4	Recent arguments and examples place radiation islands in massless-gravity, asymptotically flat, and gravitating-radiation settings	The radiation algebra, background symmetries, perturbative compact-operator construction, many-copy access, and nonperturbative reconstruction remain consequential qualifications.
Entanglement-wedge recovery, Dong, Harlow, and Wall 2016, complete Letter and proof	Established for a regulated code under the theorem’s relative-entropy and complementary-algebra hypotheses	Gravitational reconstruction can be approximate, and existence does not imply an efficient physical decoder.
Replica wormholes and factorization, Saad, Shenker, and Stanford 2019, §§2–5 and Marolf and Maxfield 2020, §§2–5	Connected saddles, ensemble correlators, and conditional alpha-sector factorization are explicit in solvable models	Their interpretation in a fixed, single boundary theory remains open and competes with sector, brane, full-spectrum, and other completion proposals.
Reflection-positive QFT reconstruction, McNamara and Wang 2026, §1.1, Theorem 1.1	A finite, multiplicative, reflection-positive closed-manifold partition function determines a unitary compact functorial QFT under the theorem’s source-class axioms	Applying the theorem to gravity requires an exact factorizing object, the complete charged spectrum, and control of the gravitational path integral; it does not itself supply that ultraviolet completion.

This assessment should be reassessed if a controlled single-theory calculation fixes the gravitational integration contour, if a reconstruction theorem gives nonperturbative errors for a parametrically large code subspace, or if an operational protocol overcomes both decoding resources and backreaction. A new Page-curve calculation alone would not trigger a changed microscopic verdict.

Check your understanding

A calculation finds the unitary Page curve for a black hole coupled to a bath. Which conclusions follow immediately, and which require extra evidence?

Solution

The calculation immediately supplies a fine-grained entropy for the declared radiation region in the stated model and approximation. It is evidence against the monotonically rising semiclassical Hawking entropy within that setup. It does not by itself determine individual density-matrix or S-matrix elements, construct a Lorentzian transfer mechanism, prove factorization in one theory, give an efficient decoder, or show that an observer can measure the entropy. Each stronger conclusion needs its own algebra, dynamics, error estimate, and operational assumptions.

What a satisfactory account must deliver

A complete proposal should define a single microscopic state space and dynamics, reproduce the controlled semiclassical exterior in its validity regime, explain the fine-grained entropy and correlations of radiation, state what replaces or reconstructs the interior, respect causal and algebraic consistency, and quantify its approximation error. It should also say whether its recovery map merely exists, can be computed, can be implemented without fatal backreaction, and yields an observation distinguishable from competing accounts.

No one calculation in this volume is allowed to stand in for that entire list. Page’s theorem is a benchmark for random states. AMPS is an incompatibility argument under named assumptions. Islands are controlled generalized-entropy saddles in specified models. Entanglement-wedge reconstruction is a code-subspace statement. Complexity can protect consistency while obstructing verification. The final chapter keeps these achievements and gaps in separate columns, which is the most reliable way to see both how much has been learned and what remains genuinely open.

References

• Almheiri, Ahmed, Thomas Hartman, Juan Maldacena, Edgar Shaghoulian, and Amirhossein Tajdini. 2020. “Replica Wormholes and the Entropy of Hawking Radiation.” Journal of High Energy Physics 05: 013. DOI. Open manuscript.
• Almheiri, Ahmed, Thomas Hartman, Juan Maldacena, Edgar Shaghoulian, and Amirhossein Tajdini. 2021. “The Entropy of Hawking Radiation.” Reviews of Modern Physics 93: 035002. DOI. Open manuscript.
• Antonini, Stefano, Chang-Han Chen, Henry Maxfield, and Geoff Penington. 2025. “An Apologia for Islands.” arXiv:2506.04311. Open manuscript.
• Dong, Xi, Daniel Harlow, and Aron C. Wall. 2016. “Reconstruction of Bulk Operators within the Entanglement Wedge in Gauge-Gravity Duality.” Physical Review Letters 117: 021601. DOI. Open manuscript.
• Donnelly, William, and Steven B. Giddings. 2016. “Observables, Gravitational Dressing, and Obstructions to Locality and Subsystems.” Physical Review D 94: 104038. DOI. Open manuscript.
• Engelhardt, Netta, and Aron C. Wall. 2015. “Quantum Extremal Surfaces: Holographic Entanglement Entropy beyond the Classical Regime.” Journal of High Energy Physics 01: 073. DOI. Open manuscript.
• Hawking, Stephen W. 1976. “Breakdown of Predictability in Gravitational Collapse.” Physical Review D 14: 2460–2473. DOI.
• Marolf, Donald, and Henry Maxfield. 2020. “Transcending the Ensemble: Baby Universes, Spacetime Wormholes, and the Order and Disorder of Black Hole Information.” Journal of High Energy Physics 08: 044. DOI. Open manuscript.
• McNamara, Jacob, and Zhencheng Wang. 2026. “Wormholes as Red Herrings: Reflection Positivity and the Reconstruction of Unitary Quantum Field Theories.” arXiv:2607.01322. Open manuscript.
• Page, Don N. 1993. “Average Entropy of a Subsystem.” Physical Review Letters 71: 1291–1294. DOI. Open manuscript.
• Saad, Phil, Stephen H. Shenker, and Douglas Stanford. 2019. “JT Gravity as a Matrix Integral.” arXiv:1903.11115. Open manuscript.
• Sen, Siddhartha. 1996. “Average Entropy of a Quantum Subsystem.” Physical Review Letters 77: 1–3. DOI. Open manuscript.