Typicality and the Entropy-Coverage Problem

Known smooth microstate families provide unusually detailed bulk representatives of protected states. That achievement does not by itself show that these representatives occupy a measure-large part of a macroscopic black hole’s Hilbert space. The relevant comparison must specify an ensemble, quantize rather than merely parameterize the family, and test observables as well as state counts. As of the evidence cutoff 2026-08-20, no controlled construction establishes entropy-dominant coverage by smooth, weakly curved geometries.

Required background. Gauge Theory and String-Theory Preparation supplies brane charges and protected sectors; Quantum Information supplies density matrices, ensembles, and concentration of measure.

Helpful background. Quantum Gravity provides the semiclassical-limit caveats used below; Black-Hole Information supplies the coarse-graining and information-loss questions that motivate typicality tests.

Typicality is relative to an ensemble and an observable class

Fix a charge-and-energy window with Hilbert-space dimension $D=e^{S}$. For a Haar-random unit vector $\lvert\psi\rangle$ in that subspace and a Hermitian operator $A$ acting within it,

$$\mathbb{E}_{\psi}\langle A\rangle_{\psi}=\frac{\operatorname{Tr}A}{D}, \qquad \operatorname{Var}_{\psi}(\langle A\rangle_{\psi}) =\frac{\operatorname{Tr}(A^2)-(\operatorname{Tr}A)^2/D}{D(D+1)}.$$

Thus, for an operator whose ensemble variance stays of order unity, root-mean-square fluctuations between typical states scale as $D^{-1/2}=e^{-S/2}$. Raju and Shrivastava 2018, §2.1, pp. 3–6 use this concentration result to argue that macroscopically distinguishable classical geometries cannot be generic Haar-typical states. Balasubramanian et al. 2008, §§2.4–2.5, pp. 16–18 instead emphasize how typical-state or ensemble coarse graining can yield a common effective geometry.

This theorem is exact, but its interpretation is narrower than it first appears. Haar typicality is not the eigenstate thermalization hypothesis, does not select the physically accessible observables, and does not prove that an individual state has a smooth bulk dual. A coherent-state basis may span a Hilbert space even though almost every superposition lacks a distinguishable classical geometry. Conversely, finding atypical coherent states is useful without proving that they are typical.

Counting geometries requires quantizing their phase space

A continuous family with many functions or moduli is not already an exponentially large set of orthogonal states. If the gauge-reduced solution space $\mathcal P$ is a compact $2n$-dimensional symplectic manifold with form $\omega$, its semiclassical Hilbert-space size is estimated by

$$N_{\mathrm{geom}} \sim \frac{1}{(2\pi\hbar)^n} \int_{\mathcal P}\frac{\omega^n}{n!},$$

with quantization conditions and lower-order corrections understood. Before using this estimate one must impose constraints, divide gauge orbits, fix fluxes and charges, identify normalizable directions, and determine whether singular boundaries or string-scale regions belong to the phase space. Quantizing a BPS subspace can be protected for counting, but it need not control real-time motion into transverse non-BPS directions; these distinctions are developed in Balasubramanian et al. 2008, §2.1, pp. 12–13.

Two ratios answer different questions. Define

$$f_{\log}=\frac{\log N_{\mathrm{geom}}}{S_{\mathrm{BH}}}, \qquad f_{\mathrm{states}}=\frac{N_{\mathrm{geom}}}{e^{S_{\mathrm{BH}}}} =\exp\!\left(\log N_{\mathrm{geom}}-S_{\mathrm{BH}}\right).$$

Matching the leading entropy requires $f_{\log}\to1$ in a specified large-charge limit, but that alone does not establish a nonvanishing fraction of states. For example, $\log N_{\rm geom}=S_{\rm BH}-\sqrt{S_{\rm BH}}$ has $f_{\log}\to1$ and yet $f_{\rm states}=e^{-\sqrt{S_{\rm BH}}}\to0$. A limit $f_{\log}\to c\lt1$ is an even stronger deficit, exponential in $S_{\rm BH}$. Moreover, $e^{S_{\rm BH}}$ is only the leading approximation to the full state count, so subleading entropy corrections matter when comparing fractions. A deeper throat or more continuous Fourier coefficients cannot substitute for quantizing the family.

A controlled entropy mismatch

For the D1–D5–P system, let $N=N_1N_5$, add momentum $N_p$, and use angular momentum $J$ in the normalization of the cited counting calculation. The macroscopic entropy is

$$S_{\mathrm{BH}}=2\pi\sqrt{N N_p-J^2}.$$

Shigemori counts the states associated with nonlinear completions of supergravitons around empty $\mathrm{AdS}_3\times S^3$. When $N_p$ and $J$ scale with $N$ while $NN_p-J^2$ remains a positive order-$N^2$ quantity, the result is

$$S_{\mathrm{strata}}=O(N^{3/4}), \qquad S_{\mathrm{BH}}=O(N),$$

so $N_{\mathrm{strata}}/e^{S_{\mathrm{BH}}}$ is exponentially small. In the Cardy regime the momentum dependence is also too weak: $S_{\mathrm{strata}}\sim N^{1/2}N_p^{1/4}$ instead of $S_{\mathrm{BH}}\sim N^{1/2}N_p^{1/2}$. These are controlled statements for the counted supergraviton sector, not a no-go theorem for all string states or all possible backgrounds. Higher and fractional modes are precisely among the omitted candidates Shigemori 2026, §4.6, pp. 25–26, and §5, pp. 27–29.

Results and limitations

Evidence or result	Controlled statement	Limit, assumption, or adverse case
Reduced BPS phase spaces	A symplectic form can quantize selected supersymmetric solution spaces after constraints and gauge reduction Balasubramanian et al. 2008, §2.1, pp. 12–13.	Quantizing one protected subspace does not show that it contains generic finite-area black-hole states.
Superstrata about empty $\mathrm{AdS}_3\times S^3$	Their counted entropy grows as $N^{3/4}$ rather than $N$ in the specified three-charge scaling limit Shigemori 2026, §4.6, pp. 25–26.	The mismatch does not exclude string-scale states, different backgrounds, or yet-unconstructed fractional-mode families.
Semiclassical coarse graining	In highly supersymmetric examples, ensembles or typical probes can share an effective singular geometry even when special states have smooth representatives Balasubramanian et al. 2008, §§2.4–2.5, pp. 16–18.	This does not prove that every individual microstate is a smooth classical spacetime.
Statistical-mechanics critique	Haar concentration and explicit correlator tests constrain how macroscopically atypical a proposed basis can be Raju and Shrivastava 2018, §§2–4, pp. 3–46.	Applying the critique to all state-dependent, stringy, or non-geometric descriptions requires additional assumptions.
Current programmatic case	Known bubbles and superstrata sample nontrivial black-hole microstructure, while typicality and sufficient state coverage remain open Bena et al. 2022, §§5.1–5.4 and §8.1, pp. 18–22, 29–30.	An advocate review’s research program is not evidence that entropy-dominant coverage has already been achieved.

Competing interpretations and adverse evidence

One position treats smooth geometries as a semiclassical basis or set of coherent representatives whose quantization may ultimately encode far more states than the presently counted families. A stronger version expects horizon-scale structure in typical black-hole microstates. The opposing position stresses concentration of measure: classical geometries with macroscopically different correlators are too atypical to represent most states, and the region where typical differences could appear may already have order-one quantum fluctuations Raju and Shrivastava 2018, §§3–4, pp. 17–46.

The sharp adverse result is the known entropy deficit, not merely the absence of constructions. The empty-$\mathrm{AdS}_3\times S^3$ superstrata counted to date occupy an exponentially small fraction in the stated limit. Deep scaling throats also create a warning: coordinate depth can grow while quantization limits distinguishable states. On the other hand, neither result excludes non-geometric typical states or controlled families with different backgrounds and stringy modes. The unresolved task is to define a measure, quantize a sufficiently broad sector, and compare both entropy growth and a diagnostic set of boundary observables.

Diagnostics

A proposed family has $\log N_{\mathrm{fam}}=0.82S_{\mathrm{BH}}+O(\log S_{\mathrm{BH}})$. Does it cover 82 percent of the states at large entropy? If a bounded observable has order-one ensemble variance, what scale of state-to-state fluctuation is expected for Haar-typical pure states?

Solution

No. The logarithmic coverage tends to $f_{\log}=0.82$, but the actual fraction is $f_{\mathrm{states}}\sim e^{-0.18S_{\mathrm{BH}}}$ up to powers of the entropy, so it vanishes exponentially. The Haar variance formula gives a root-mean-square fluctuation of order $D^{-1/2}=e^{-S_{\mathrm{BH}}/2}$. Therefore this check distinguishes a ratio of logarithms from a fraction of states. Both conclusions depend on the chosen charge window and observable normalization; neither identifies a bulk geometry for a typical state.

What could change these conclusions

The conclusions could change if a geometric quantization or state-counting theorem changes the leading charge scaling of a controlled solution family; if a typical-state boundary observable distinguishes smooth microstates with quantified measure; if string-scale or fractional-mode states acquire a controlled bulk count; or if a new entropy bound or counterexample changes the fraction represented by weakly curved geometries. A valid comparison must keep the ensemble, charges, supersymmetry, measure, and notion of distinguishability fixed.

References

• Balasubramanian, Vijay, Jan de Boer, Sheer El-Showk, and Ilies Messamah. “Black Holes as Effective Geometries.” Classical and Quantum Gravity 25 (2008): 214004. arXiv:0811.0263.
• Bena, Iosif, Emil J. Martinec, Samir D. Mathur, and Nicholas P. Warner. “Fuzzballs and Microstate Geometries: Black-Hole Structure in String Theory.” arXiv:2204.13113 [hep-th] (2022). arXiv.
• Raju, Suvrat, and Pushkal Shrivastava. “A Critique of the Fuzzball Program.” Physical Review D 99 (2019): 066009. DOI. arXiv:1804.10616.
• Shigemori, Masaki. “Counting Superstrata.” arXiv:1907.03878v3 [hep-th] (revised 2026). arXiv.