The information-loss paradox and the status of proposals

For a finite-dimensional pure state of radiation R and remaining system B, their reduced von Neumann entropies agree and are bounded by the logarithm of the smaller dimension. A Page estimate further assumes a state chosen typically with the Haar measure; it is not the entropy of every state or the output of any arbitrary evaporation dynamics. As the available radiation dimension grows and the remaining dimension shrinks, the typical entropy rises and eventually falls. This illustrates what unitary statistical behavior can look like.

The exact fine-grained entropy is distinct from a coarse entropy obtained by treating emitted modes as independent thermal quanta. Correlations invisible to a few simple measurements can change the former substantially. Likewise, an entropy curve does not supply individual amplitudes, a practical decoding procedure, or a complete evaporation S-matrix. These differences matter when interpreting a gravitational calculation that predicts a Page-shaped entropy but does not explicitly construct all the underlying microstates.

Semiclassical gravitational entropy prescriptions associate candidate regions, including a possible interior island, with a generalized entropy. One first finds extrema with respect to the candidate boundary and then selects the least entropy among the admissible extrema, including the empty-island candidate. This ordering is physical: merely fitting two smooth functions and taking their lower envelope does not derive a gravitational entropy formula. A calculation must specify its action, state, boundary or radiation algebra, saddles, and approximation regime.

Controlled examples in holographic settings and evaporating dilaton-gravity models produce Page behavior through a change of dominant extremum or replica saddle. In two-dimensional gravity, the gravitational entropy uses the appropriate dilaton expression rather than a literal four-dimensional area. The transition sharpness belongs to the leading approximation; corrections can modify a sharp crossing. Even successful entropy calculations retain questions about analytic continuation, factorization or ensemble interpretations in particular models, reconstruction, and the evaporation endpoint.

The scope of island proposals extends beyond models with a nongravitating external bath. Antonini, Chen, Maxfield, and Penington's 2025 primary analysis argues for their compatibility with massless gravity and asymptotically flat radiation, while emphasizing the chosen observable algebra. A radiation algebra restricted to suitable outgoing radiative observables is different from one enlarged to include the ADM energy, which can generate time translations and alter the question being asked. An entropy independent of a retarded-time cut for the enlarged algebra need not refute a Page curve for the restricted one.

These are arguments with specified gravitational dressing, observables, and reconstruction assumptions, rather than a universal experimental demonstration of information recovery. Many-copy tomography or highly complex operations can probe information unavailable to simple semiclassical observables, but that does not provide an efficient decoder for a real astrophysical hole. A responsible status statement identifies which model gives a controlled entropy result, which algebra defines radiation, and which microscopic claims remain unsupported. A Page curve is strong evidence within its stated framework, while the detailed final state and endpoint still demand more physics.