Time & Gravity · T.10 · Where a clock can be seen to lose information, and the counting that says it should not · Step 9 of 9 · 14 min read
Page's Counting Argument
A black hole's horizon only grows, its entropy is its area, and it radiates. Hawking's calculation says the radiation carries no information and its entropy rises forever. Page's counting argument says that if the whole process is quantum mechanics, the radiation's entropy must turn over halfway and fall back to zero. This page builds the counting argument on ten qubits, checks it against Page's formula, and says plainly what in this story is measured, derived, and still open.
Fixed boundaryNo theory of everything here: six evidence labels, one open question per page.
Nothing on this page is a theory of everything, and this site proposes none. Every claim carries one of six labels, each tied to a ledger row: MEASURED, DERIVED, BOUND, PROPOSED, OPEN, or SINGLE RESULT. Preprints never rise above PROPOSED. Each page ends with one open question that the next page takes up.
Plate T.10-V · Hawking's straight line rising beside Page's curve, which rises with it, turns over at half the qubits and falls back to zero; dots from the ten-qubit toy sit on the curve. Drawn by timegravity/make_plates.py from tools/timegravity_kernel.py.About this illustrationThis plate is a deterministic drawing, not an AI-generated image and not a photograph. The curve is the kernel's own trajectory; the shading exaggerates a clock-rate difference of about one part in 10^16 per metre so that it can be seen at all.
Instrument · Hawking's line against Page's curve
The plot shows the entropy of the radiation against the number of qubits that have left the black hole, out of N in all. The straight line is Hawking's: every emitted qubit is thermal, entropy rises by ln 2 each time, forever. The curve is Page's closed form for a random pure state of N qubits: it rises almost along Hawking's line, turns over at N/2, and falls back to zero when the last qubit leaves. The dots are a toy: random states on 10 qubits, each cut's entropy computed from the eigenvalues of the reduced state with the same pure-Python eigensolver the verifier uses. The N slider redraws the line and the curve. The knockout removes unitarity: the radiation is thermal all the way and the entropy never turns over, which is also the verifier's planted defect. The second button drops Page's correction term −(m−1)/2n to show how far the naive count misses. Keyboard: Tab and arrows. With reduced motion on, nothing animates.
Predict before you revealBefore you move the slider: for a black hole made of N qubits that evaporates as ordinary quantum mechanics, when does the radiation's entropy peak?
Peak entropy, Page's formula at N/22.966 nats
Toy at N = 10: max |toy − Page|6.3e-04 nats
Entropy at the last cut0.692 nats
Turns over?yes, at N/2
Counting, not gravity: a random pure state's smaller part is nearly maximally entangled with the larger, so the radiation's entropy follows the smaller side and comes back down.
I · What has been observed
Pairs of black holes merge, and the signals, including the remnant's ringdown, are consistent with general relativity: the first gravitational-wave signal, GW150914, was observed in 2015 by the LIGO and Virgo collaborations, and the same public data were reanalysed independently in 2021 to test the area theorem. Those two rows are what every DERIVED statement on this page is about: the black holes we have actually seen.
The 2021 reanalysis of GW150914 found the remnant's horizon area larger than the sum of the two initial areas, the first test of Hawking's area theorem on a real signal.In 2025 the LIGO, Virgo and KAGRA collaborations reported GW250114, a binary black hole signal with a network signal-to-noise ratio of 80, and stated that the remnant area is larger than the sum of the initial areas to high credibility, with the ringdown mode frequencies constrained to within 30 percent of the Kerr spectrum. The figure is quoted as the paper's own claim; this page does not recompute it. Plugging published median masses into the area formula would look like a test and would not be one, because the result comes from posterior sampling, not from medians.
In Plain Words · Extra-Easy Explanation
Burn a book and the smoke looks random, but in principle every letter is still in the smoke, scrambled. Hawking's 1975 calculation said a black hole is different: its smoke is truly random, and the book is gone. Page's argument says: if a black hole is a quantum system like any other, then after half of it has turned to smoke, the smoke must start carrying the book, and by the end the smoke holds everything. Draw the amount of randomness in the smoke over time and you get a hill, not a slope.
The hill is not a theory of black holes. It is counting: what a random quantum state of N parts must look like from any one part. The open question is how gravity manages to make the real smoke follow the hill, and whether it does.
Everyday Analogy & Controls Primer
Everyday analogy: a shuffled deck is dealt one card at a time. Early on, knowing the dealt cards tells you almost nothing about the rest. Past the halfway point, every new card is mostly predictable from what is already on the table. The uncertainty in the dealt pile rises, peaks at the middle, and falls to nothing when the last card lands. Page's curve is that, for quantum cards.
How the controls work in plain words: the N slider is the size of the deck. The straight line is a dealer who burns each card after showing it, so the pile's uncertainty only grows; that is the knockout, and nothing on the table ever lets you predict a card. The second button uses a slightly wrong counting rule to show that the exact hill needs Page's small correction.
Plate T.01-D · Visual Intuition Bridge: the everyday analogy (left) sets up one quantity with two settings (centre), and the verifier's three demands (right) are what the knockout and the sign-flip buttons in the instrument above test.
II · Area, entropy, radiation, and the counting
In classical general relativity the total horizon area of black holes never decreases: Hawking's 1971 area theorem, which is the statement the two gravitational-wave tests above examine.Bekenstein argued in 1973 that a black hole carries an entropy proportional to its horizon area, and Hawking showed in 1975 that a black hole radiates thermally at a temperature fixed by its mass, so that it loses mass and, in that calculation, returns no information about what fell in. That is the straight line in the instrument: thermal radiation whose entropy rises by one bit per emitted qubit without end.
S(m, n) = Σ_{k=n+1}^{mn} 1/k − (m − 1)/(2n), m ≤ n (Page 1993)
Page showed in 1993 that for a random pure state on a Hilbert space of dimension mn the average entanglement entropy of the m-dimensional part is S(m, n) = Σ_{k=n+1}^{mn} 1/k − (m−1)/(2n), only slightly below the maximum ln m, and applied it to an evaporating black hole: if the hole and its radiation together are in a pure state, the radiation's entropy rises almost thermally, turns over when half the degrees of freedom have been emitted, and returns to zero. The verifier draws eight random pure states on ten qubits, computes the entropy of every cut from the eigenvalues of the reduced state with a pure-Python Jacobi solver, and finds the average within 0.002 nats of Page's formula at every cut, peaking at five qubits with 2.97 nats and ending at 0.69 nats. The check is against the formula; the random numbers are only the sample.
In 2019 and 2020 two groups showed that within semiclassical gravity the entropy of the radiation, computed with the quantum extremal surface prescription, follows Page's curve: after the halfway point a region inside the black hole, an island, belongs to the radiation's entanglement wedge. That is a derivation inside a framework, not a mechanism anyone has observed, and the toy on this page contains no island; it contains only the counting.
§ II.1 · The derivation the verifier runs, in four stages
A pure state on N qubits, split into a emitted qubits (dimension m = 2ᵃ) and N − a remaining (n = 2^{N−a}). For a Haar-random state the average entropy of the smaller part is Page's S(m, n); entropy is the same seen from either side.
Stage 2 · The toy
Eight random complex vectors on 2¹⁰ dimensions (Gaussian components, normalised; fixed seed so the artefact regenerates byte for byte). For each cut the reduced density matrix of the smaller side (at most 32×32) is embedded as a real symmetric 64×64 matrix and diagonalised with cyclic Jacobi; S = −Σ λ ln λ.
Stage 3 · The assertion
The sample mean at every cut must lie within 0.03 nats of Page's closed form (it lies within 0.002), the curve must be symmetric about N/2 and must fall to less than half its peak at the last cut. The assertion is against the formula, never against the random numbers themselves.
Stage 4 · The defect
The planted defect replaces the toy's entropies by Hawking's line a·ln 2, a non-unitary model that never turns over; the 'falls to less than half its peak' check fails, as it must, and the page does not build.
III · What this page does not say
Two 2025 and 2026 preprints take opposite sides on what the island calculation means: one argues for islands as the mechanism behind the curve, the other argues that islands only redistribute information when a detector has a blind spot and that the asymptotic algebra is already complete. Both are preprints and neither can rise above PROPOSED here. Unitarity is near consensus; the mechanism is not.
Where the information goes in our universe, with its positive cosmological constant and no asymptotic boundary to hold the radiation, is not settled: the curve is a counting result, and the derivations that reproduce it live in frameworks whose application to our spacetime is itself open. This is the last page of the track, and it ends where the hub began: with a question. The slope in the rate of time is measured to a part in 10¹⁵; what happens to a clock's record when the slope becomes a horizon is counted, derived in pieces, and not yet seen.
Grounding ledger rows cited on this page
Every row below resolves in the site's offline Crossref cache. The lint derives what each set of rows can carry; a person decided the label on each claim above. Rows pass a review date and the build fails until someone rereads the source.
Scite column: [observed: Scite ML citation classification, accessed 2026-11-09] via Scite's public tallies endpoint. The counts are a classifier's output over indexed citation statements, an evidence-synthesis habit borrowed from medicine; they never replace reading the source, and a row with more contrasting than supporting statements is flagged for a human re-read in the label checklist.
AI contribution to the cited results: none. No AI system contributed to any observation, theorem or derivation cited on this page. The page text, instrument and verifier were drafted with AI assistance and checked by the site maintainer against the primary sources listed below.
Rebuild this instrument locally and run the verifier as the acceptance gate: python skills/timegravity/t10/make_t10.py then python skills/timegravity/t10/verify_t10_page_curve.py (and --planted-defect, which must fail). Then python tools/check_timegravity_labels.py.
Show the SKILL.md source
---
name: timegravity-t10-page-curve
description: Rebuild and verify the Time & Gravity T.10 toy - Page's counting argument on ten qubits: Haar-random pure states whose radiation entropy matches Page's closed form S(m,n) = sum_{k=n+1}^{mn} 1/k - (m-1)/(2n) at every cut, turns over at N/2 and falls back, with eigenvalues from a pure-Python Jacobi solver - whose planted defect is a non-unitary toy that follows Hawking's line and never turns over.
license: MIT
compatibility: Claude Code, GitHub Copilot, Cursor, Google Antigravity, OpenAI Codex, Ollama (Qwen 3.8 / Gemma 4)
---
# Page's Counting Argument (T.10 · Time & Gravity)
> **Fixed Boundary (`EN`):** Nothing on this page is a theory of everything, and this site proposes none. Every claim carries one of six labels, each tied to a ledger row: MEASURED, DERIVED, BOUND, PROPOSED, OPEN, or SINGLE RESULT. Preprints never rise above PROPOSED. Each page ends with one open question that the next page takes up.
>
> **Frontera Fija (`ES-MX`):** Nada en esta página es una teoría del todo, y este sitio no propone ninguna. Cada afirmación lleva una de seis etiquetas, cada una ligada a una fila del libro de registro: MEDIDO, DERIVADO, COTA, PROPUESTO, ABIERTO o RESULTADO ÚNICO. Las prepublicaciones nunca suben de PROPUESTO. Cada página termina con una pregunta abierta que la siguiente página retoma.
## 1. Page, canonical URL and governing equations
- **Track page**: `T.10` (Time & Gravity, reading order 9 of 9, dated `2026-11-09`). Page labels **DERIVED** and **OPEN**.
- **Canonical URL**: https://informatico.ai/timegravity/t10/ (Spanish: https://informatico.ai/es-mx/timegravity/t10/)
- **Page's formula**: `S(m, n) = Σ_{k=n+1}^{mn} 1/k − (m−1)/(2n)` for `m ≤ n` (nats). Hawking's line: `S = a · ln 2`.
- **Central numbers**: 8 Haar-random pure states on 10 qubits; mean entropy at every cut within `0.03` nats of Page's formula (achieved ≈ 0.002); peak at `a = 5` with `2.97` nats; final `0.69` nats; symmetric about `N/2`.
- **Generator**: `python skills/timegravity/t10/make_t10.py`
- **Verifier**: `python skills/timegravity/t10/verify_t10_page_curve.py`
- **Knockout check**: `python skills/timegravity/t10/verify_t10_page_curve.py --planted-defect` (must exit non-zero: entropies replaced by Hawking's line, never turning over)
- **Label lint**: `python tools/check_timegravity_labels.py` (and `--planted-defect`, `--mutants`)
## 2. Ledger rows this page may cite (timegravity/ledger.json)
| Row | Label | Source | DOI |
| --- | --- | --- | --- |
| `T10-ABBOTT2016` | MEASURED | GW150914, first binary black hole merger | https://doi.org/10.1103/PhysRevLett.116.061102 |
| `T10-ISI2021` | MEASURED (first area-law test; SINGLE RESULT as a test) | Isi et al. 2021 | https://doi.org/10.1103/PhysRevLett.127.011103 |
| `T10-LVK2025` | SINGLE RESULT | GW250114 area law and Kerr test (young) | https://doi.org/10.1103/kw5g-d732 |
| `T10-HAWKING1971` | DERIVED | area theorem | https://doi.org/10.1103/PhysRevLett.26.1344 |
| `T10-BEKENSTEIN1973` | DERIVED | black-hole entropy | https://doi.org/10.1103/PhysRevD.7.2333 |
| `T10-HAWKING1975` | DERIVED | particle creation by black holes | https://doi.org/10.1007/BF02345020 |
| `T10-PAGE1993A` / `T10-PAGE1993B` | DERIVED | average entropy of a subsystem; information in black hole radiation | https://doi.org/10.1103/PhysRevLett.71.1291 · https://doi.org/10.1103/PhysRevLett.71.3743 |
| `T10-PENINGTON2020` / `T10-ALMHEIRI2019` | DERIVED | quantum extremal surfaces, islands | https://doi.org/10.1007/JHEP09(2020)002 · https://doi.org/10.1007/JHEP12(2019)063 |
| `T10-APOLOGIA2025` / `T10-BLINDERS2026` | PROPOSED (preprints) | the island dispute | https://doi.org/10.48550/arXiv.2506.04311 · https://doi.org/10.48550/arXiv.2602.06543 |
## 3. Tool-neutral agent packet and security boundary (`security-runtime`)
1. **Rebuild locally**: `python skills/timegravity/t10/make_t10.py`; standard library only; fixed seed; writes only under `src/research/timegravity/t10/files/`.
2. **Acceptance gate**: `python skills/timegravity/t10/verify_t10_page_curve.py` must print `"status": "PASS"` with `turns_over_at_half: PASS`; `--planted-defect` must exit non-zero. Then `python tools/check_timegravity_labels.py`.
3. **Security boundary (`security-runtime`)**: no shell outside the repository, no home-directory access, no outbound network for this task.
## 4. Deterministic verification and falsification rules
1. At every cut `a = 1..N−1` the sample mean entropy lies within `0.03` nats of `S(2ᵃ, 2^{N−a})`; the assertion is against the closed form, never against the random numbers.
2. The curve is symmetric about `N/2` within `0.06` nats and its last value is below half its peak.
3. Planted defect: entropies replaced by `a · ln 2` (non-unitary, thermal all the way); rule 2 fails and `--planted-defect` exits non-zero.
4. `t10_reference.json` matches `manifest.json` and regenerates byte for byte (fixed seed).
## 5. What the instrument shows (reader side)
- Hawking's straight line and Page's curve for a chosen N, with the 10-qubit toy's dots; the N slider redraws both.
- Knockout: non-unitary evaporation, the line only rises. Second button: Page's formula without its `−(m−1)/2n` correction, to show the naive count's miss.
- Predict before you reveal: where the radiation entropy peaks.
## 6. Exact T0–T3 truth boundary
### In the repository today (T0/T1)
- Kernel functions `page_average_entropy`, `haar_random_state`, `reduced_density_matrix`, `von_neumann_entropy`, `page_curve_toy`, `check_page_curve` in `tools/timegravity_kernel_bounds.py`.
### Not on this page (T2/T3)
- The GW250114 area figure is quoted as the paper's own claim and never recomputed from medians.
- Islands are a derivation inside semiclassical gravity, labelled DERIVED; the mechanism dispute is PROPOSED (preprints); where the information goes in our universe is OPEN.
Show the verifier verify_t10_page_curve.py
#!/usr/bin/env python3
"""Standalone stdlib verifier and planted-defect knockout for Time & Gravity · T.10 (Page's Counting Argument).
Canonical URL: https://informatico.ai/timegravity/t10/
Central number: for Haar-random pure states on 10 qubits the mean entanglement entropy of the a emitted qubits matches
Page's closed form S(m,n) = Σ_{k=n+1}^{mn} 1/k − (m−1)/(2n) within 0.03 nats at every cut (achieved ~0.002), is
symmetric about N/2, peaks at 5 qubits (2.97 nats) and falls to 0.69 nats at the last cut; eigenvalues come from a
pure-Python Jacobi solver. Planted defect: a non-unitary toy whose entropy follows Hawking's line a·ln 2 and never
turns over; `--planted-defect` must exit non-zero with FAIL on stderr.
Boundary: Nothing on this page is a theory of everything, and this site proposes none.
"""
from __future__ import annotations
import hashlib
import json
import sys
from pathlib import Path
ROOT = Path(__file__).resolve().parents[3]
sys.path.insert(0, str(ROOT / "tools"))
from timegravity_kernel_bounds import PAGE_TOL_NATS, check_page_curve # noqa: E402
FILES_DIR = ROOT / "src" / "research" / "timegravity" / "t10" / "files"
def verify_reference_files() -> dict:
manifest = json.loads((FILES_DIR / "manifest.json").read_text(encoding="utf-8"))
out = {}
for entry in manifest["files"]:
raw = (FILES_DIR / entry["filename"]).read_bytes()
assert len(raw) == entry["bytes"], f"{entry['filename']}: byte count mismatch"
digest = hashlib.sha256(raw).hexdigest()
assert digest == entry["sha256"], f"{entry['filename']}: sha256 mismatch"
out[entry["filename"]] = digest[:12]
ref = json.loads((FILES_DIR / "t10_reference.json").read_text(encoding="utf-8"))
for a, toy, page, _h in ref["curve_a__toy_mean__page__hawking_line"]:
assert abs(toy - page) < PAGE_TOL_NATS, f"reference curve misses Page's formula at a = {a}"
return out
def main(argv: list[str] | None = None) -> int:
args = argv if argv is not None else sys.argv[1:]
planted = "--planted-defect" in args
try:
res = check_page_curve(planted_defect=planted)
assert res["matches_page"], f"toy entropies miss Page's closed form by {res['max_dev_from_page_nats']:.3f} nats (> {PAGE_TOL_NATS})"
assert res["turns_over"], (
f"the radiation entropy never turns over: final {res['final_nats']:.3f} nats against a peak of {res['peak_nats']:.3f}; "
"a non-unitary toy follows Hawking's line"
)
assert res["symmetric"], "the curve is not symmetric about N/2"
assert res["defect_rejected"], "planted non-unitary toy was accepted"
files = verify_reference_files()
except AssertionError as exc:
print(json.dumps({"status": "FAIL", "folio_id": "T.10", "error": str(exc)}), file=sys.stderr)
return 1
print(json.dumps({
"status": "PASS",
"folio_id": "T.10",
"canonical_url": "https://informatico.ai/timegravity/t10/",
"central_number": {"n_qubits": res["n_qubits"], "samples": res["samples"], "max_dev_from_page_nats": res["max_dev_from_page_nats"],
"peak_nats": res["peak_nats"], "final_nats": res["final_nats"], "tolerance_nats": PAGE_TOL_NATS},
"geometry_checks": {
"matches_page_closed_form": "PASS",
"turns_over_at_half": "PASS",
"symmetric_about_half": "PASS",
"planted_defect_knockout": "REJECTED_AS_EXPECTED",
},
"reference_files_sha256_12": files,
}, indent=2))
return 0
if __name__ == "__main__":
raise SystemExit(main())
Primary sources
Abbott, B. P., et al. (LIGO Scientific Collaboration and Virgo Collaboration) (2016). Observation of gravitational waves from a binary black hole merger. Physical Review Letters, 116(6), 061102. DOI DOI 10.1103/PhysRevLett.116.061102[Scite ↗]
Isi, M., Farr, W. M., Giesler, M., Scheel, M. A., & Teukolsky, S. A. (2021). Testing the black-hole area law with GW150914. Physical Review Letters, 127(1), 011103. DOI DOI 10.1103/PhysRevLett.127.011103[Scite ↗]
Abac, A. G., et al. (LIGO Scientific, Virgo and KAGRA Collaborations) (2025). GW250114: Testing Hawking's area law and the Kerr nature of black holes. Physical Review Letters, 135, 111403. DOI DOI 10.1103/kw5g-d732[Scite ↗]
Hawking, S. W. (1971). Gravitational radiation from colliding black holes. Physical Review Letters, 26(21), 1344–1346. DOI DOI 10.1103/PhysRevLett.26.1344[Scite ↗]
Penington, G. (2020). Entanglement wedge reconstruction and the information paradox. Journal of High Energy Physics, 2020(9), 2. DOI DOI 10.1007/JHEP09(2020)002[Scite ↗]
Almheiri, A., Engelhardt, N., Marolf, D., & Maxfield, H. (2019). The entropy of bulk quantum fields and the entanglement wedge of an evaporating black hole. Journal of High Energy Physics, 2019(12), 63. DOI DOI 10.1007/JHEP12(2019)063[Scite ↗]
Antonini, S., Chen, Y., Maxfield, H., & Penington, G. (2025). An apologia for islands. arXiv preprint (not peer reviewed). DOI DOI 10.48550/arXiv.2506.04311[Scite ↗]
Geng, H., Karch, A., Perez-Pardavila, C., Raju, S., Randall, L., & Riojas, M. (2026). Seeing Page curves and islands with blinders on. arXiv preprint (not peer reviewed). DOI DOI 10.48550/arXiv.2602.06543[Scite ↗]