Finite lifetime
Conventional dimension-5/dimension-6 unified-theory channels remain archived as historical comparison calculations.
Archive record · August 16, 2026
Prediction Update
This page records what changed, what was added, and what did not change in the UTFANSWF prediction surface as the controlled revision produced a v23 pre-release add-on checkpoint.
Historical record: This page preserves the dated state of the project. For current claims and status, use Results and the Full Prediction Index.
Record date: . This is a dated website research record, not a third-party cryptographic timestamp. The v22 PDF remains the current live public release. These controlled results are carried in a separate v23 pre-release add-on ZIP ahead of the formal September 1 launch. The record date remains unchanged.
Prediction changed
Proton decay becomes proton stability under the exact Z6 candidate.
The earlier v22 conventional unified-theory branch produced finite proton-decay lifetime estimates of roughly 1034–1036 years. Row 28 tests a different structural result: the decay operators themselves are forbidden while the exact residual Z6 is preserved.
Γp = 0
Isolated ΔB = 1 proton decay is forbidden while exact Z6 is preserved.
ΔB = 3k
ΔB = 1 and ΔB = 2 are forbidden. The first allowed baryon-number violation is ΔB = 3.
Supersession rule: the finite v22 proton-lifetime numbers are no longer the active UTFANSWF proton prediction under the Row 28 exact-Z6 candidate. They remain visible as the historical/conventional branch rather than being erased.
New predictions and closures
Two new testing surfaces are added.
First allowed violation: ΔB = 3
The new falsifiable signature is not ordinary single-proton decay. The candidate instead permits baryon-number violation only in multiples of three. No trinucleon lifetime is yet claimed because the ΔB = 3 Wilson coefficient/scale is not fixed.
MNisoQCD = 941.223 ± 35.184 MeV
A target-isolated Nf=2+1 continuum/chiral reconstruction closes the isospin-symmetric nucleon mass scale using 24 production CLS ensembles across five lattice spacings.
Fine structure · Row 27
A new blind numerical checkpoint, not yet a final prediction.
The blind electromagnetic chain reached
α⁻¹(0) = 137.03735243
before unblinding. The verified post-mixed rung is 137.00222764, and the withheld comparison value was 137.035999177.
Boundary
Source-complete handoff remains open.
The 137.03735243 value is retained as a legitimate blind checkpoint. It is not promoted here to the final source-complete UTFANSWF fine-structure prediction.
Fine-structure recordCosmology / H₀
The public UTFANSWF distributed-H₀ candidate remains unchanged.
Historical publication context: the published v22 paper reported a distributed result near 72.7 km/s/Mpc. The later controlled post-release distributed result, 73.134202 km/s/Mpc, became the public website candidate. Both are preserved in the record rather than silently overwriting the earlier publication.
The controlled post-release distributed output is preserved as
H₀ = 73.134202 km/s/Mpc
The v23 hardening work separately tests the conventional/background cosmology fit. That technical distinction remains in the reproducibility record and will be stated explicitly in the paper; it does not replace the public distributed-H₀ candidate on the website.
Status
Distributed candidate preserved.
The hardened v23 cosmology machinery remains reproducible and its separate background-fit diagnostics are retained for technical review. The public prediction displayed by UTFANSWF remains the distributed value above.
Current cosmology recordNo current numerical change
Recent additions do not presently force changes everywhere else.
The August 15–16 work does not currently require a numerical revision of the existing axion, black-hole ringdown, or Neutral-Zone testing surfaces. Those remain active under their existing gates until a controlled propagation audit demonstrates otherwise.
Failed routes do not become predictions. The attempted Flavor-Origin / primitive-texture branch failed its full texture test and is not incorporated into the prediction set. It remains part of the research provenance only.
Executable provenance
The dated prediction record now points to the sealed v23 add-on checkpoint.
python -m harness.run_all_gates
18 / 18 executed gates: PASS
Row 20: 14.205046 / 13 = 1.092696
Fresh-extraction benchmark: 14.59 s
Superseded pre-release executable SHA-256:
d42ce9908d46356f17d9dd30409e12f78cfb5b7e7c1525e233487e54803764bd
The formal v22 main.tex and main.pdf remain protected and unchanged. “100% PASS” refers to the executable canonical harness and does not close separately documented open research boundaries.
Portability hotfix: Row 0.4 now carries
mpmath 1.3.0 inside the v23 executable itself. The canonical runner also preflights NumPy, SciPy, pandas, and PyYAML before Row 0 and bootstraps any missing external package. The advertised command remains python -m harness.run_all_gates.