Black-Hole Complementarity

Black-hole complementarity proposes that exterior recovery and smooth infall are compatible because no single observer can operationally combine the two semiclassical descriptions into a cloning experiment. Exterior information is associated with a stretched horizon and late radiation; the infalling description uses interior fields across an approximately smooth horizon. The proposal succeeds only if these are complementary representations of one quantum system rather than independent copies in a common factorization.

Required background. Quantum Information supplies no-cloning, logical encodings, observer algebras, and recovery used to formulate the proposal.

Helpful background. Semiclassical Gravity supplies stretched-horizon and evaporation scales; Quantum Fields in Curved Spacetime supplies near-horizon vacuum observables; Causality and Global Structure supplies the causal separation of exterior and infalling worldlines.

The complementary observables

A traditional statement combines four claims:

• exterior evolution is unitary and information eventually appears in Hawking radiation;
• semiclassical effective field theory is reliable outside a stretched horizon;
• an infalling observer sees low-energy vacuum correlations at a large horizon;
• no observer can verify two independent copies before causal or energetic limits intervene.

The proposed complementary objects are the stretched-horizon or radiation representation of a black-hole degree of freedom and the interior field representation used by an infaller. They are not simply two coordinate descriptions of one local operator. The crucial assertion is that they should not be represented as two freely commuting full matrix algebras that an observer can manipulate independently.

In algebraic language, let $\mathcal A_{\mathrm{ext}}$ and $\mathcal A_{\mathrm{in}}$ be the exterior and infalling observable algebras. Complementarity needs maps from a common logical algebra into each description, together with a rule for their domains. Where the domains overlap, all shared predictions must agree. Where they do not, the theory must say why no state or operation requires both assignments. This is stronger than saying that one observer lacks time to compare two already independent copies.

Susskind, Thorlacius, and Uglum formulate the stretched horizon, information retention, and observer comparison as one proposal Susskind, Thorlacius, and Uglum 1993, §§ 2–4. Its physical motivation is clear; a complete state-independent operator construction is an additional requirement.

Causal protection of no-cloning

The canonical duplication experiment asks an exterior observer to collect radiation containing a diary, decode it, then enter the black hole and receive the infalling diary’s signal. Waiting outside boosts the required interior signal into an exponentially small proper-time window. The uncertainty principle then drives its energy upward, eventually beyond the regime of a low-backreaction message.

This causal argument protects operational consistency in the parameter regime where the waiting and decoding times are long. It does not prove that the complete Hilbert space never contained two factors. No-cloning is a statement about the physical map, whereas causal inaccessibility is a statement about permitted comparisons. A satisfactory complementarity principle should meet both.

The distinction becomes sharper for an old black hole. If early radiation already purifies a late exterior mode while smoothness requires that mode to be entangled with an independent interior partner, an entropy contradiction can be stated without one observer completing the experiment. Harlow reviews why the original causal complementarity argument and the old-hole entanglement problem are not identical Harlow 2016, §§ 5.5–5.7 and 7.1.

A code-subspace model

Consider the two-dimensional code

$$\mathcal H_{\mathrm{code}} =\operatorname{span}\{\lvert000\rangle,\lvert111\rangle\}.$$

On this code, $Z_1$, $Z_2$, and $Z_3$ have the same action and each represents the logical $Z_L$. This redundancy is not three copies of an arbitrary logical qubit: the conjugate logical bit flip is $X_L=X_1X_2X_3$, and acting with $X_1$ alone leaves the code. Measuring $X_L$ tests phase coherence between the logical basis states. The state $\alpha\lvert000\rangle+\beta\lvert111\rangle$ is one encoded state, not $\lvert\psi\rangle^{\otimes3}$.

This finite model shows how multiple operator representations can avoid no-cloning on a restricted subspace. It does not establish black-hole complementarity. A gravitational construction must identify the code subspace, control errors, reproduce near-horizon correlators, and explain consistency when code subspaces overlap or when superpositions leave the semiclassical domain.

Changing the factorization is the adverse test. If the three physical qubits are instead treated as independent outputs each carrying the complete unknown logical state, the desired map becomes universal cloning and is impossible. The distinction lies in the operator algebra, not in the observer’s vocabulary.

Results and open questions as of 2026-08-19

It remains unresolved whether black-hole complementarity, by itself, provides a globally consistent interior for arbitrary old-hole states. Competing positions differ over factorization, code-subspace dependence, horizon state, and whether operational consistency is sufficient.

Evidence or position	Result	Scope and limitation
Stretched-horizon complementarity Susskind, Thorlacius, and Uglum 1993, §§ 2–4	Assigns exterior thermalization and smooth infall to complementary descriptions	Causal protection is semiclassical; a global microscopic operator map is not constructed there
No-cloning theorem Wootters and Zurek 1982, pp. 802–803	Independent faithful copies of arbitrary nonorthogonal states are impossible	Allows redundant logical representations and orthogonal classical records
Old-hole entanglement argument Almheiri et al. 2013, § 2	Late purification conflicts with an independent smooth-horizon partner	Depends on a common $EBA$ state and factorization; it does not select a unique replacement
Code-subspace complementarity	Multiple representations of one logical operator can be consistent on a restricted state class	Linearity, overlap consistency, and error control outside that class remain model-dependent

The conservative position accepts complementarity as a constraint on observer-accessible semiclassical predictions but does not treat it as a complete microscopic solution. The encoding position makes the two descriptions redundant logical representations. The firewall position rejects smooth near-horizon entanglement for sufficiently old states. Another position changes factorization or interior operator domains. Their conclusions depend on which Hilbert space and algebra are asserted.

The conclusions could change after a microscopic model constructs both observer algebras on overlapping state spaces, or after a causal decoding protocol survives complexity and backreaction bounds, or after a theorem shows that complementarity preserves linearity and Born-rule predictions for arbitrary allowed superpositions.

Consistency requirements

A viable proposal should specify the exterior and interior operator domains; reproduce semiclassical low-point correlators where claimed; preserve no-cloning and linearity; quantify approximate commutators and reconstruction error; and explain how causal, complexity, and energy limits apply to an actual observer. Passing only the last test is not enough.

Generic code theory belongs to Quantum Information, local horizon correlators to Quantum Fields in Curved Spacetime, and evaporation scales to Semiclassical Gravity. Explicit string or holographic realizations are treated separately.

Exercise

In the repetition-code example, an observer measures $Z_1$ while another measures $Z_2$. Have they measured two cloned qubits, and what operation would test the missing conjugate information?

Solution

They have measured two representations of the same logical $Z_L$ on the code, not two independent arbitrary qubits. Their outcomes agree because of the code constraint. To test the conjugate logical information one needs an operation sensitive to $X_L=X_1X_2X_3$, which is nonlocal across all three physical qubits. Treating $X_1$ and $X_2$ as independent logical conjugates would leave the code and change the formulation.

References

• Almheiri, Ahmed, Donald Marolf, Joseph Polchinski, and James Sully. 2013. “Black Holes: Complementarity or Firewalls?” Journal of High Energy Physics 2013 (02): 062. Open text.
• Harlow, Daniel. 2016. “Jerusalem Lectures on Black Holes and Quantum Information.” Reviews of Modern Physics 88: 015002. Open text.
• Susskind, Leonard, Lárus Thorlacius, and John Uglum. 1993. “The Stretched Horizon and Black Hole Complementarity.” Physical Review D 48: 3743–3761. Open text.
• Wootters, William K., and Wojciech H. Zurek. 1982. “A Single Quantum Cannot Be Cloned.” Nature 299: 802–803. DOI.