Corpus synthesis · 31 Aug 2026
The black-hole-universe literature, audited
An adversarial, paper-by-paper audit of the published “is our universe inside a black hole?” literature. 58 entries — 51 papers plus 7 imported measurement instruments — each tiered under one preregistered rule and double-checked by two independent reviewers. 55 read in full; every figure below traces to a primary source or a two-seat verdict.
1What the base layer actually contains
Peer-reviewed journal articles only. Parsed from the record, not recited.
| Tier | Count | What it means |
|---|---|---|
| Consistency-only | 32 | a construction shown compatible with observation — predicts nothing measurable |
| Qualitative-directional | 7 | a direction (sign, inequality) but no calibrated window |
| Calibrated-falsifier | 4 | a number and a threshold a measurement can cross — entries 7, 31, 44, 51 |
| Theoretical-obstruction | 3 | a no-go theorem — entries 5, 22, 48 |
| Prospect | 3 | a “detectable” claim whose amplitude is not yet computed |
| Support (no tier) | 7 | imported measurement instruments, not BHU-claim papers |
| Gated (unread) | 2 | entries 42, 47 — paywalled, left gated |
The field is dominated by prose: 32 of 58 entries predict nothing measurable. They show the idea is not ruled out — which is not the same as evidence for it. Only fourentries put a number against a threshold. A fifth, Gaztañaga’s Λ = 3/rS², displays a number that is fixed from the measured Λ rather than predicting it.
Five ways a “prediction” dissolved under audit, each with an exemplar:
- Fitted-not-predicted (25): a number read off the datum it “explains” — the number and the reason it cannot fail are the same clause.
- Instrument-fired, not the claim (7): the falsifier crosses an instrument’s threshold (a mass ceiling), while the source only says a heavy star “would put in serious doubt or simply falsify” the cosmology.
- Correction sized to the measurement (44): when the base model died at 8σ, the fix offered is an uncomputed ~4% correction whose size is read off the datum it must reproduce.
- Unreproduced from the stated inputs (51): a printed threshold no route from the paper’s own inputs reaches; the connecting step omitted.
- Attributed threshold (31): the falsifying number is credited to others, not derived by the claimant.
2The live-falsifier ledger
Two FIRED, two LIVE, plus the directional curvature watch. Each threshold has a primary receipt.
| Entry | Status | Threshold | Current data |
|---|---|---|---|
| 7 · Brown–Lee–Rho (CNS) | FIRED | neutron star M ≳ 2 M☉ falsifies the kaon-condensation chain (PRL 101, 091101) | fires the Mmax≈1.5 M☉ instrument, not CNS — the source gives CNS only “serious doubt” |
| 44 · Pourhasan et al. (white hole) | FIRED | exact scale invariance, ns = 1 (JCAP 04(2014)005 §4) | Planck 2018 VI eq (19): ns = 0.9649 ± 0.0042, 8σ from 1. Successor is an uncomputed ~4% correction. |
| 31 · Smolin (CNS) | LIVE, 1.36σ short | a neutron star above 2.5 M☉ refutes CNS (Smolin §4) | heaviest well-measured NS 2.35 ± 0.11 M☉; moving away from firing as the error tightens |
| 51 · Popławski (torsion) | LIVE, unfired | a minimum black-hole mass floor ~10¹⁶ kg (PLB 690, 73) | unreproduced from the stated inputs — six routes tested, none reaches the floor; connecting step omitted |
| 54 · Gaztañaga (curvature) | LIVE, not fired | predicts closed curvature, Ωk < 0; refutes on a confirmed Ωk > 0 (PRD 111, 103537 Eq. 27) | DESI DR2+CMB Ωk = +0.0023 ± 0.0011 (~2.1σ open) — adverse but not a detection. Watched by a standing tripwire. |
Read honestly: exactly two calibrated falsifiers remain LIVE (31, 51), and neither is close — one is drifting away from its bar, the other’s number cannot be reproduced from its own inputs. The two FIRED ones each killed a sub-model, not the cosmological framework. Nothing in the base layer is currently refuted at the framework level, and nothing is on the edge of firing.
3The three paywalled papers, ranked by what a read would unlock
All pre-2010, ~a few thousand won each by interlibrary copy.
- Entry 47 — Sato, Kodama, Sasaki & Maeda (1982). Highest value: the earliest false-vacuum multi-universe mechanism, and the missing member of the branch the Farhi–Guth no-go (entry 48) is about. Reading it answers one real question — does the 1982 mechanism fall inside that no-go, or predate and sidestep it?
- Entry 42 — González-Díaz (1991). Medium: a baby-universe metric-equivalence claim; a read decides whether it adds a result the branch lacks or is another consistency construction.
- Entry 2 — Good (1972). Lowest: a one-page note on nested universes. Provenance only — no mechanism, no observable.
If any single paper is worth an ILL, it is 47 — it converts an assumed obstruction relationship into a checked one.
4Falsifiers the family’s own texts imply but never state
Testable edges the primary papers gesture at without committing to a number. None is yet a stated falsifier.
- The Popławski interior transfer function. The only published multi-paper mechanism with explicit field equations — yet the transfer function from parent-hole parameters through the bounce to any interior observable was never written. Deriving it would state whether any finite-amplitude signature survives.
- The Roupas amplitude. A published “detectable” claim with a named band (µHz–Hz, LISA-class) but no computed amplitude. If a number exists, it becomes a fifth calibrated falsifier; if not, it is a prospect without one.
- The Gaztañaga cutoff. A power-spectrum cutoff tied to the CMB low quadrupole — a falsifier only if the cutoff scale is fixed independently of the anomalies it explains.
- The preferred-axis / spin-parity edge. Longo’s handedness dipole is the instrument for the rotating-parent family’s preferred-axis prediction — the amplitude a DESI spin-parity test would confront. This is the one live cross-check.
- Easson’s cross-programme no-go map. A 2026 obstruction theorem that, mapped across the other published interiors, could retire several rows at once — a theory-internal falsifier rather than a data one.