Black-Hole Interior Reconstruction

An operator “behind the horizon” has no unique boundary representation until one specifies the state family, boundary region, gravitational dressing, code subspace, time band, perturbative order, and desired error. Two-sided entanglement wedges, small-code recovery, state-dependent mirror operators, and radiation-island constructions answer different versions of the question. None currently establishes one simple, state-independent interior algebra on the full boundary Hilbert space. Evidence is assessed through 2026-08-20.

Required background. Quantum Information supplies operator algebras, recovery, and monogamy, while Gauge Theory and String-Theory Preparation supplies the holographic encoding and large-N regime. Helpful background. Quantum Gravity supplies relational observables and semiclassical limitations, and Black-Hole Information supplies the general interior and information-theoretic debate.

What must be reconstructed

Consider first the eternal two-sided AdS black hole, dual to the thermofield-double state

$$|\mathrm{TFD}\rangle =\frac{1}{\sqrt{Z(\beta)}} \sum_n e^{-\beta E_n/2}|E_n\rangle_L|E_n\rangle_R .$$

The union of both complete boundaries has a semiclassical wedge containing the two-sided interior. A relationally dressed bulk operator in that wedge can therefore have a representative in the joint left-right algebra within a low-energy code. This geometrical statement is much weaker than representing the same operator on the entire right CFT alone for every microstate. The two-sided geometry and its state preparation are developed by Maldacena 2003, §§ 2–3, pp. 3–12.

For a one-sided or evaporating black hole, four inputs become decisive:

1. State domain: a small neighborhood of one equilibrium state, a microcanonical band, or an exponentially large family.
2. Algebra and dressing: simple effective fields, a wedge algebra with center, or a fully boundary-anchored relational operator.
3. Region and time: one boundary time band, both asymptotic boundaries, an early/late radiation region, or their union.
4. Accuracy and complexity: fixed-order correlators, a uniform channel norm, or a decoder with bounded complexity.

A valid reconstruction claim states all four. Agreement of a few low-point correlators does not imply a linear operator on the full Hilbert space, and information-theoretic existence does not imply efficient decoding.

State of the evidence through 2026-08-20

The comparison table separates controlled constructions from adverse cases. Dates refer to the cited primary result, while the assessment cutoff is 2026-08-20.

assumptions and setting	evidence or position	adverse limitation or discriminator	primary source and date
Fixed semiclassical code subspace, fixed QES wedge, regulated operator algebras, perturbative relative-entropy equality	Wedge operators have boundary representatives on a region whose entanglement wedge contains them; overlapping regions can give equivalent code-level representatives	Enlarging the code until wedge assignments change destroys the uniform equality; existence is perturbative and need not be computationally efficient	Dong, Harlow, and Wall 2016, complete Letter, pp. 1–5
Equilibrium large-N state, small algebra of simple exterior operators, limited excitations and time band	State-dependent mirror operators can be defined by modular-like relations and reproduce desired effective interior correlators on that small domain	Extending the same rule across large overlapping state families raises linearity and ambiguity objections; the construction is not one operator on the full Hilbert space	Papadodimas and Raju 2013, §§ 3–6
Old evaporating black hole, factorized early radiation and near-horizon modes, ordinary quantum mechanics and effective-field-theory smoothness	Monogamy gives an adverse obstruction to simultaneously assigning a late mode the entanglement needed for both exterior unitarity and a smooth infalling vacuum	Changing factorization, code-subspace, or non-isometric assumptions changes the inference, but any alternative must identify exactly which premise fails	Almheiri, Marolf, Polchinski, and Sully 2013, §§ 2–3
Semiclassical evaporating models admitting replica wormholes and an island QES, with controlled saddle continuation	After the Page transition, the island can lie in the radiation entanglement wedge, supporting reconstruction of its effective algebra from radiation in the code description	Replica continuation, ensemble interpretation, nonperturbative factorization, and microscopic decoder remain additional assumptions; a QES saddle is not an explicit universal decoding circuit	Penington et al. 2022, §§ 2–4

The first and fourth rows are controlled semiclassical evidence for code-subspace reconstruction in their specified models. The second is a constructive but disputed response to the one-sided interior problem. The third is a negative result conditional on its subsystem and effective-field-theory premises. Treating these rows as contradictory without aligning their state class, region, and algebra would erase the main scientific distinction.

Competing positions

Code-subspace position. The conservative view identifies an interior operator only after a QES wedge and code algebra are fixed. Relative-entropy equality then guarantees a representative to the retained order. This is established perturbatively, but it need not continue across an exponentially large state set or through a wedge transition.

State-dependent position. For an equilibrium state $|\Psi\rangle$ and a small algebra ${\cal A}$, a mirror mode may be defined on ${\cal A}|\Psi\rangle$ through a relation schematically like

$$\widetilde O_\omega A|\Psi\rangle =A e^{-\beta\omega/2}O_\omega^\dagger|\Psi\rangle, \qquad A\in{\cal A}.$$

This can encode the entanglement required by effective interior correlators without demanding a simple global operator. The competing objection is that different equilibrium states or overlapping small algebras can prescribe incompatible actions on the same vector. The proposal is therefore meaningful only with a sharply bounded domain and consistency checks; Papadodimas and Raju 2013, §§ 4–6 do not establish a state-independent full-space map.

Non-isometric or gravitational-code position. An isometry preserves inner products; a non-isometric encoding can instead map distinct effective bulk vectors to the same microscopic vector. Explicit model constructions preserve inner products to high accuracy on computationally restricted state sets while admitting a kernel on the larger effective space. They distinguish state-specific reconstruction from one reconstruction valid on a whole subspace Akers, Engelhardt, Harlow, Penington, and Vardhan 2022, §§ 3 and 5, especially § 5.4. These controlled model statements do not derive the exact encoding of a specified holographic CFT black hole; that extension still needs a physical inner product, state domain, and error bound.

Exterior-relational position. The most conservative account assigns invariant meaning only to exterior or relational observables whose boundary dressing is explicit, regarding a state-independent local interior field as an emergent patchwise object. This avoids overclaiming but leaves open how an infalling observer’s effective algebra and cross-horizon correlations arise with controlled errors.

A decisive finite-code test

Let ${\cal H}_{\rm code}$ contain states $|0\rangle$ and $|1\rangle$, and suppose a candidate boundary region $A$ has identical reductions,

$$\rho_A^{(0)}=\rho_A^{(1)},$$

while a proposed interior observable satisfies $\langle0|O_{\rm int}|0\rangle=1$ and $\langle1|O_{\rm int}|1\rangle=-1$. No operator $O_A$ can reproduce both values because identical density matrices give identical expectation values for every $A$-supported operator. This is a concrete counterexample to reconstruction on the enlarged code. A state-dependent prescription can reproduce either state’s correlators on separate patches, but it must explain its action on their superpositions and on any overlap.

The same discriminator applies holographically: enlarge the state set or time-evolve beyond the declared band, then compute whether relative-entropy equality, commutators, and recovery error remain uniform. A changing QES, an order-one center fluctuation, or two wedge-distinct states with indistinguishable boundary reductions marks failure of the original claim rather than a paradox in quantum mechanics.

What remains open

No consensus construction currently supplies all of the following simultaneously: a nonperturbative interior algebra, a state domain large enough to represent generic black-hole microstates, linearity on superpositions, explicit gravitational dressing, uniform small error, and an operational decoder. It remains open whether these demands are mutually compatible, whether complexity or non-isometry supplies the correct limitation, and how finite-N effects modify the semiclassical wedge story.

Island calculations sharpen the region assignment but do not alone settle microscopic factorization or the experience of an infalling observer. Conversely, monogamy obstructions identify incompatible assumptions but do not choose which gravitational assumption fails. The evidence does not establish either a universal firewall or a universal smooth-horizon operator on the full Hilbert space.

What could change these conclusions

The conclusions could change when a reconstruction theorem controls a substantially larger state or algebra class, when an explicit counterexample demonstrates inconsistent linear action on an overlap, or when an independent calculation gives a nonperturbative error bound. A reproducible operational decoding protocol, a proof fixing the role of complexity, or a controlled change in boundary condition, field content, or factorization assumption would also alter the assessment. New correlator matches within one unchanged small code would refine evidence but would not by themselves resolve the domain question.

Exercises

Determine the obstruction in the two-state test. Suppose the trace distance between the two reductions on $A$ is at most $0.01$, while an interior observable with operator norm one has expectation values differing by $1.2$. Can one $A$-supported norm-one operator represent it to error $0.05$ on both states?

Solution

For any norm-one boundary operator, the expectation-value difference is at most twice the trace distance under the convention $T(\rho,\sigma)=\|\rho-\sigma\|_1/2$, hence at most $0.02$. If the representation had error $0.05$ on each state, it would still need to reproduce a difference of at least $1.2-0.10=1.10$. That exceeds the boundary bound by a large margin. Therefore no such state-independent $A$ operator exists on this two-state code; the state domain, region, or error claim must be changed.

References

• Akers, Chris, Netta Engelhardt, Daniel Harlow, Geoff Penington, and Shreya Vardhan. “The Black Hole Interior from Non-Isometric Codes and Complexity.” arXiv:2207.06536v2 (2022). Open manuscript.
• Almheiri, Ahmed, Donald Marolf, Joseph Polchinski, and James Sully. “Black Holes: Complementarity or Firewalls?” Journal of High Energy Physics 02 (2013): 062. DOI. Open PDF.
• Dong, Xi, Daniel Harlow, and Aron C. Wall. “Reconstruction of Bulk Operators within the Entanglement Wedge in Gauge-Gravity Duality.” Physical Review Letters 117 (2016): 021601. DOI. Open PDF.
• Maldacena, Juan. “Eternal Black Holes in Anti-de Sitter.” Journal of High Energy Physics 04 (2003): 021. DOI. Open PDF.
• Papadodimas, Kyriakos, and Suvrat Raju. “An Infalling Observer in AdS/CFT.” Journal of High Energy Physics 10 (2013): 212. DOI. Open PDF.
• Penington, Geoff, Stephen H. Shenker, Douglas Stanford, and Zhenbin Yang. “Replica Wormholes and the Black Hole Interior.” Journal of High Energy Physics 03 (2022): 205. DOI. Open PDF.