Ø Frozen Predictions
The pre-registration of every forward prediction, its document, its verification script, and its current standing. Also the conditions under which each one dies.
English only. This is a precision document; a mistranslated digit is worse than an absent page. It is deliberately not translated into the twelve language editions.
A pre-registration is a claim written down and time-stamped before the measurement that will test it — so that no one, including its author, can quietly revise it afterwards. Here the commit hash is the timestamp: these files were committed to the public repository and tagged prereg-2026-08-02. Every number below is regenerated from CODATA 2022 inputs by the scripts in this directory; nothing is quoted from memory.
This is a scoreboard, not a showcase. A registry that lists only the surviving predictions would misrepresent the corpus and destroy the credibility the freezes were written to earn. Two results are recorded here as corpses — the bare neutron–proton row, marked FIRED — 7.24σ on 2026-08-02, and the Hubble prediction 74.3 ± 1.2, marked FIRED — 2026-09-03 on the corpus’s own derived rate — alongside the live ones, including the four frozen at the wave of 4 September 2026: the realised G, the realised neutron row, the age as one cycle, and the expansion rate as a closure. A fired switch is shown, never repaired, never deleted; a new structural row may freeze beside the corpse, never in its place.
The same freezes are read independently, and in plain language, by the verifier who anchors them: Frozen Predictions — 420Code, in plain words. He runs his own scripts against the same CODATA inputs and keeps his own scoreboard. Reading both is the point; a registry checked only by its author is a registry checked by nobody.
Ø On this page
Seven parts — start anywhere.
Ø The scoreboard
Standing as of 21 September 2026. Thirteen rows below: eleven live, two fired. They are carried by eleven registrations, listed under the register — a registration can own more than one row, because one prediction can be read against more than one measurement. AP28’s first reading of G, 6.7206 × 10⁻¹¹, is withdrawn — superseded by the realised value on the author’s ruling of 21 September 2026 — and kept in the register by reference (the dated note).
| Claim | Predicted | Measured | σ | Error | Paper |
|---|---|---|---|---|---|
| Proton mass ratio — the series closed | 1836.152673445 | 1836.152673426(32) | +0.59σ | 10.3 ppt | AP49 |
| Gravitational constant G, realised — against CODATA | 6.6719 × 10⁻¹¹ | 6.67430(15) × 10⁻¹¹ | −16σ | −0.036% | AP44 |
| Gravitational constant G, realised — against atom interferometry | 6.6719 × 10⁻¹¹ | 6.67191(99) × 10⁻¹¹ (LENS 2014) | +0.01σ | +1 ppm | AP44 |
| Neutron-proton mass diff, realised | 2.53098857035 mₑ | 2.53098857(74) mₑ | −0.005σ | −1.4 ppb | AP47 |
| Neutron-proton mass diff, bare row | 2.53099393 mₑ | 2.53098857(74) mₑ | +7.24σ · FIRED | +2.1 ppm | AP30 |
| Age of the universe, one cycle | 13.830 Gyr | 13.787 ± 0.020 Gyr | +2.1σ | +0.31% · window 13.17–14.49 | AP46 |
| Expansion rate H₀, the closure | 67.45 km/s/Mpc | 67.4 ± 0.5 (Planck) | +0.1σ | +0.07% · window 64.4–70.8 | AP48 |
| Hubble constant from the floor inverted | 74.3 ± 1.2 | 67.45 (the corpus’s own rate) | +5.7σ · FIRED | +10.2% | AP18 |
| MOND acceleration a₀ (at H₀ = 67.45) | 1.089 × 10⁻¹⁰ | 1.20 ± 0.24 × 10⁻¹⁰ (syst.) | −0.46σ | −9.2% | AP18 |
| Dark energy | 68.85% | 68.89 ± 0.56% | −0.07σ | −0.06% | AP42 |
| Dark matter | 26.39% | 26.07 ± 0.38% | +0.8σ | +1.2% | AP42 |
| α(0) does not vary — across place, epoch and density | Δα/α = 0, structurally | consistent with zero (Oklo, quasars, clocks) | — | a null · no number of its own | AP51 |
| Visible matter | 4.76% | ≈ 4.885 ± 0.05% | −2.46σ | −2.5% | AP41 |
How to read the σ column. Every measurement comes with a bar: one standard deviation, σ, the experimenters’ own statement of how far their number might sit from the truth. The σ column divides the gap between prediction and measurement by that bar. The sign says which side: + above the measurement, − below. The Error column gives the same gap in the number’s own units. The odds below are the chance of landing at least that far from the truth by chance alone, counting both directions.
- Under 1σ — inside the bar. Agreement: about two honest measurements in three land this close, and one in three lands further out by chance alone.
- 2σ — tension. One in twenty-two. Worth watching, not yet a verdict.
- 3σ — evidence of a real gap. One in 370.
- 5σ — the discovery standard of particle physics. One in 1.7 million (one in 3.5 million if only one direction counts, as the discovery convention usually reads it). Each switch names its kill threshold in advance. Past it, the row is dead by the work’s own rule.
- Above 5σ — dead. The two corpses on this page sit at 7.24σ and 5.7σ. They are shown, never repaired.
Three caveats, because σ is measured against the experiment’s bar and the bar is not the same for every number. G: the same realised value reads −16σ against CODATA and +0.01σ against atom interferometry, and both rows are in the table so the reader can see the issue. CODATA’s bar on G is 22 parts per million, so a 360-ppm gap reads 16σ; the laboratory measurements of G disagree with one another by up to 540 ppm, twenty-five such bars, and the atom-interferometry value (LENS 2014, 6.67191 ± 0.00099 × 10⁻¹¹) sits 1 ppm from the realised value. A bar that tight on a constant the experiments cannot agree on is a statement about the adjustment, not yet about nature. On the scale above the CODATA row reads dead; the switch on G, KS-CCC.3, was registered to test direction — whether the consensus migrates down toward 6.672 as systematics close, within AP28’s one-per-cent bar — not magnitude against CODATA, and it stays live on those terms. There is one value of G. AP28’s first reading, +0.69%, was wrong: it omitted the arena’s holding of the break, which AP44 priced. It is withdrawn, superseded by the realised value on the author’s ruling of 21 September 2026, and kept on the record, dated (the dated note). The neutron’s fired row exposed the omission, and its consequences reached three derivations, revised at the wave (AP44, AP47, AP48). The age: +2.1σ on Planck’s bar, but the claim was registered one lane-time wide, ±0.66 Gyr, because an aware observer measures about one cycle. The switches fire on their registered conditions, not on 5σ by default: KS-45.1 registered ±1.2 and fired at 5.7σ on that width; on the honest ±14.9 it would have read 0.46σ. The registration was the error, and a fired switch is not repaired.
Ø The register
Each registration with its standing, the condition under which it dies, its frozen document and script, and the commit that timestamps it. The numbers are in the table above.
| Prediction | Status | Kill condition — the crux | Document | Script | Commit | Last checked |
|---|---|---|---|---|---|---|
| Proton–electron mass ratio | LIVE | Dies at 3σ once measurement uncertainty ≤ 3.4 ppt with the central value unchanged (now 17.4 ppt; ~3–5 yr); 5σ at 2.1 ppt | mp-me-closed (raw) | verify_prereg.py | 6772e24 | 2026-09-06 |
| the series closed (AP49) · KS-HOLD.1 | ||||||
| Proton–electron mass ratio | SUPERSEDED | Superseded by the closed form, never replaced by it. | mp-me-alpha3 (raw) | verify_prereg.py | prereg-2026-08-02 | 2026-09-06 |
| at order α² (AP30) — superseded by reference, kept — D = 1836.152673444331 at +0.57σ, frozen 2 August 2026 (digest 6ea78dac…). | ||||||
| The base of the chain | LIVE | Fires if the measurement settles on the leaked value, or on neither. | mp-me-closed (raw) | verify_prereg.py | 6772e24 | 2026-09-06 |
| KS-HOLD.3 · AP49 — Bare 21α or the leaked repair: the two bases predict 1836.152673445 and 1836.152673408, 3.7 × 10⁻⁸ apart. CODATA sits between them (+0.59σ / −0.56σ) and decides nothing yet — decided at about ten times today’s precision (12σ at 1.7 ppt, ~7σ at 3 ppt). | ||||||
| The reading of the sixteen | LIVE | Fires if any second-order count in the family needs its object squared, its object through three dimensions, or any count not of this form. No measurement decides it. | mp-me-closed (raw) | — | 6772e24 | 2026-09-06 |
| KS-HOLD.2 · AP49 — structural — A second-order support count is its object through the four dimensional expressions: 21 × 4, 2 × 4, 4 × 4. | ||||||
| Neutron–proton mass difference | FIRED — 2026-08-02 | Empirical limb fired (offset 2.12 ppm vs a 0.288 ppm bar). | ERRATA §E6a (raw) | — (two-route check, see errata) | prereg-2026-08-02 | 2026-09-04 |
| KS-NPP.1 (AP30) — No repair offered: the required correction is negative, every term is positive. Structural limbs KS-NPP.2/.3 open. | ||||||
| Hubble constant H₀ from the floor inverted | FIRED — 2026-09-03 | Fired by AP48 on the corpus’s own derived rate, as AP46 §9 pre-registered. | H0-KS45.1 (raw) · erratum (raw) | verify_family.py | prereg-2026-08-02 | 2026-09-04 |
| KS-45.1 (AP18) — The registered ±1.2 omitted the ±0.24 systematic on a₀; on the honest width, 74.3 ± 14.9, the floor’s relation survives at 0.46σ (KS-45 live). Shown, never repaired. The corpus’s H₀ entry is now KS-ASM.1. | ||||||
| Expansion rate H₀ — the closure | LIVE | Dies if the expansion rate, as the data settle it, lands outside 64.4–70.8 in either direction | H0-KS-ASM.1 (raw) | verify_family.py | wave-2026-09-04 | 2026-09-04 |
| KS-ASM.1 (AP48) — — including if the Cepheid ladder’s 73 is confirmed as the expansion rate; on that edge flat ΛCDM at the sky’s partition dies with it, and KS-STRETCH.3 fires too (the age would be 12.78 Gyr) | ||||||
| Age of the universe — one cycle | LIVE | Dies if the age, as the cosmological data settle it, differs from one cycle by more than one lane-time, 0.66 Gyr, either way | age-KS-STRETCH.3 (raw) | verify_family.py | wave-2026-09-04 | 2026-09-04 |
| KS-STRETCH.3 (AP46) — (an early-dark-energy age near 12.9 fires it) | ||||||
| Neutron–proton mass difference — realised | LIVE | Dies if an improved measurement departs from the expression Δ_bare/(1 + α²/8π) beyond the bar, 7.4 × 10⁻⁷ mₑ (0.29 ppm). | neutron-KS-FLIP.1 (raw) | verify_family.py | wave-2026-09-04 | 2026-09-04 |
| KS-FLIP.1 (AP47) — Frozen beside the corpse above, never in its place. The absent α³ term (KS-FLIP.3) waits on ~20× better B_d metrology. | ||||||
| Gravitational constant G — realised | LIVE | As apparatus systematics resolve, the adjusted G migrates ~300 ppm downward toward 6.672 × 10⁻¹¹. Convergence upward toward 6.7206 kills it, with no fallback: the fork is withdrawn (dated note, 21 September 2026). | G-KS-CCC.3 (raw) | verify_family.py | wave-2026-09-04 | 2026-09-04 |
| KS-CCC.3 (AP44) — Tests existence and direction, not magnitude. AP28’s first reading, 6.7206 × 10⁻¹¹ at +0.69%, frozen with this registration as the structural value — withdrawn 21 September 2026, superseded by reference, kept. | ||||||
| α(0) does not vary across the record history | LIVE | Dies on a variation of α(0) at 5σ or better, by two independent methods, standing after their systematics are reviewed | alpha-no-variation (raw) | — | c9c02cd | 2026-09-11 |
| P-GLASS.1 · KS-GLASS.3 (AP51) — a null; the anchoring is outstanding | ||||||
| Visible fraction | LIVE — nearest to firing | Dies if Ω_b/Ω_total central value exceeds 4.81% at sub-0.5% precision (~5 yr) | visible-fraction-KS41.1 (raw) | verify_cosmology.py | prereg-2026-08-02 | 2026-09-04 |
| KS-41.1 (AP41) | ||||||
| Dark-sector clock | LIVE — conditional on D48 | Dies if Ω_DM/Ω_b at z ≳ 4 matches the z = 0 value, or a dark-matter particle is detected (KS-42.1) | dark-clock-KS42.6 (raw) | verify_cosmology.py | prereg-2026-08-02 | 2026-08-03 |
| KS-42.6 (AP42) | ||||||
Commit = the commit tagged prereg-2026-08-02 (the freezes of 2 August) or wave-2026-09-04 (the wave) in the repository; that hash is the timestamp of record. The papers of the wave carry their own freezes, anchored by the independent verifier and published beside each proof at /proofs/instruments/. The verifier keeps his own account of the same freezes, in plain words and with his own scripts, at lucid.rodeo/prereg — and the wider companion to the body of work at lucid.rodeo/the420code. Where his reading and this page differ, the difference is the useful part; both are public. Last checked = the date the standing was last re-computed against measurement. Every row the root verify.py carries is re-computed on every push, and last was at the wave; the dark-sector clock is not one of them — it awaits a redshift-evolution measurement, and its August date is the honest one. Two further documents are not predictions and carry no row: RIGIDITY.md (raw) (why the integer 21 cannot move — summarised above under “Why 21 is forced”, with its dated note (raw)) and PROTOCOL-lattice-qcd-mapping.md (raw), a sealed-envelope procedure for the one blind test still available (its machinery — the template, the readiness check that refuses a freeze with a blank field or a lattice number in it, and the digest-and-commit step — is seal_lattice_freeze.py; the derivation itself is reserved to the author, alone, by the protocol’s own §1) — explicitly not a freeze.
Ø The span — the field each prediction is made in
The same axiom and the same single measured input, the fine-structure constant α, are read at every scale the corpus reaches. This is the same set of predictions as the scoreboard, sorted by the field of physics each one is made in and the size of the thing it describes.
| Field | Scale | What is predicted | Paper | σ |
|---|---|---|---|---|
| Particle physics | 10⁻¹⁵ m — a proton | The proton-to-electron mass ratio | AP49 | +0.59σ |
| Nuclear physics | 10⁻¹⁵ m — a nucleus | The neutron–proton mass difference, realised | AP47 | −0.005σ |
| Gravitation, metrology | 10⁻¹ m — a torsion balance | The gravitational constant G, realised | AP44 | +0.01σ vs atom interferometry −16σ vs the CODATA adjustment |
| Galactic dynamics | 10²¹ m — a galaxy’s outskirts | The MOND acceleration floor a₀ | AP18 | −0.46σ |
| Cosmology — the dark sector | 10²⁶ m — the observable universe | The dark-energy, dark-matter and visible fractions, from one timescale | AP42 · AP41 | −0.07σ · +0.8σ · −2.46σ |
| Cosmology — the chronology | 10²⁶ m — one cycle | The age of the universe | AP46 | +2.1σ |
| Cosmology — the expansion | 10²⁶ m — the closure | The expansion rate H₀ | AP48 | +0.1σ |
What the span shows. These are not one field’s predictions. They run from the weight of a proton to the rate at which the universe expands — from the quantum to the cosmic — more than forty powers of ten in length scale, from the femtometre of a nucleus to the Hubble radius — and each is derived from the same axiom and the same one measured number. Seven of the ten live rows that carry a σ land inside one standard deviation of the measurement they face: the proton’s mass ratio, the neutron’s realised difference, the realised G against atom interferometry, the acceleration floor, the expansion rate, and two of the three dark-sector fractions.
What the rows share. One input, α — and that is the point: one input, no dials. The rows are not independent tests, and they are not offered as though they were. The acceleration floor is derived from the expansion rate and the expansion rate from the cycle and the partition together, so those sit on one chain read at three points; the three dark-sector fractions come from a single timescale, not three derivations. Counted as families rather than rows, the live set is four at most — the mass scaffold, the gravitational coupling, the cosmological chain, and the dark partition — and even those are coupled, since the rate is a closure of the partition and the cycle. One of the seven, the realised G, reads +0.01σ against atom interferometry and −16σ against the CODATA adjustment; both readings are in the table above, and the reader is entitled to weigh them.
What it does show. Where a structural omission was found, paying it moved three separate derivations toward measurement rather than away. The bare neutron–proton row fired at 7.24σ on 2 August 2026; pricing the held distinction that row had treated as free brought the difference to −0.005σ (AP47). The same omission, priced at the arena, moved the gravitational constant from its first reading, +0.69%, now withdrawn, to −0.036% (AP44). The same mechanism at the cosmological scale replaced the floor’s inverted 74.3 — which fired — with a rate closed out of the partition and the cycle, 67.45, at 0.1σ against the relic glow (AP46, AP48). Three limbs, one correction, all three improving. That is what the span supports. It is evidence that the framework behaves as a structure: an omission found in one place, once accounted for, propagates into other fields and improves them there too. It is also evidence of overlap with nature: seven rows inside one standard deviation, from the quantum to the cosmic, from one measured input. Evidence is not yet final proof. Final proof is settled at the switches — the conditions listed above under which each of these rows dies — and every one of them is live.
Ø Why 21 is forced — a count, not a fit
Every step from one record exists to the twenty-one is forced, and physics has not entered the chain when the count is reached — the whole chain, step by step, is at The Method. The charge answered here is fair as a principle: with a handful of small integers — 3, 4, 6, 21 — and π, and a free choice of where to put them, you can hit any single target you like. So the question is never how close one number lands. It is how much freedom there was — and that is computable.
21 is not chosen. It is built: 21 = 6 × 3 + 3 — six faces of the break (AP24) across three spatial dimensions (AP10), plus three actualisation couplings. Both inputs are derived upstream and both were fixed before any mass ratio, gravitational constant or dark-sector split was computed; the ordering is checkable against the dependency graph, since AP10 and AP28 precede AP30, AP41 and AP42. So the perturbations available to a fitter are not “any integer near 21” — they are: change the face count, or change the dimension count. Here is what each does, substituted consistently everywhere the integer appears:
| N | Where it comes from | mp/me error | G error | Visible fraction error |
|---|---|---|---|---|
| 14 | 2 dimensions | −54.5% | 9.1 × 10¹⁴ × too big | +46.2% |
| 18 | 5 faces | −26.0% | 2.6 × 10⁶ × too big | +13.7% |
| 20 | ad hoc | −9.10% | 138 × too big | +2.35% |
| 21 | 6 faces, 3 dimensions | 1 × 10⁻⁹% | +0.69% | −2.52% |
| 22 | ad hoc | +9.53% | −99.3% | −6.95% |
| 24 | 7 faces | +29.9% | −100% | −14.7% |
| 28 | 4 dimensions | +75.9% | −100% | −26.9% |
Read the G column. N sits in an exponent, αN. One step changes G by a factor of 137 — not 137 per cent, a factor of 137. The mass ratio alone would tolerate 20 or 22 to within ten per cent; the exponential closes that window to a single integer. The same integer gives the proton’s 1836 by a polynomial, and the α terms it carries take the ratio to eleven decimal places; it gives G to within a factor of 1.0004 by an exponential, and the visible share to within a few per cent by a reciprocal — three functional forms, three unrelated observables, and no value but 21 survives all three. (The G column shows AP28’s first reading, the form RIGIDITY.md was frozen with and the exponent test is written against; the realised value, −0.036%, carries the same exponent, and the verdict does not move. The first reading is withdrawn: dated note of 21 September 2026.)
What the next measurements decide. Rigidity shows the integer cannot move within the work’s own family; whether a different family of primitives could carry three observables at once is KS-30.4’s between-family limb — narrowed by the search below, and live. Most of these constants were measured long before this work — as Mercury’s orbit was measured long before Einstein. The logic did not start from them; it started from one record exists, and counted. When a measurement was taken says nothing about how a number was reached; how it was reached is the chain, and every step of it is on the page to inspect. Where a structural choice could have been made by comparison with a measured value, the work fences it: KS-CCC.1 binds the commitment behind G and fires on any refinement whose selection depended on comparison with measured G and whose history is undisclosed. It is live and has not fired. The frozen forward predictions and the sealed lattice protocol add what inspection cannot: dates no reader has to take on trust. The α² coefficient 16 was owed under KS-30.3 until AP49 The Hold paid it at the structural register on 6 September 2026: the sixteen is four conditions of a record held in three dimensions over time, four times four. The switch’s own test stands — the closed form dies at three sigma once the measurement reaches 3.4 ppt.
What a fitted framework cannot produce. Fitting has a signature: everything lands, nothing is ever wrong, every error sits comfortably inside whatever tolerance was quoted. This corpus does not have that signature. The first reading of G landed at +0.69% — 309σ on CODATA’s bar, and roughly thirteen times the spread between G experiments — and stood on the wall at that miss until the logic found what it had omitted; a fitter would have done better. It is withdrawn now, and kept. The neutron–proton row failed at 7.24σ, found and published by the corpus itself. The uniqueness claim was found wrong and corrected the same day. The visible fraction was restated from 2.0% to 2.46σ once the observed value was taken from the middle of its range rather than the bottom. The muon is unreachable and the corpus says so. Five negative results, four of them found and published before anyone outside asked. Fitting has no mechanism for producing its own counter-evidence.
The full argument, with the arithmetic and the sensitivity script: RIGIDITY.md (raw) · verify_rigidity.py · dated note of 4 September 2026 (raw) (the frozen document’s “31σ” for the structural G is corrected to 309σ there, and its H₀ threshold superseded). Its sentence that the α² coefficient 16 has no structural decomposition is superseded too: AP49 The Hold paid KS-30.3 on 6 September 2026. And §4(b)’s “written against a measurement that already existed” is corrected by the dated note of 19 September 2026 (raw) — written against is a claim about method, and it is withdrawn. The frozen document is not edited.
The search, run against the work (4 September 2026). Rigidity shows 21 cannot move inside the work’s own family. The search asks the wider question: could any formula of the same size do what 21 does? The space the charge names was made concrete — the integers 1 to 30, π, the four arithmetic operations, and no more tokens than the work’s own five — and every formula in it was enumerated against the measured ratio. A search can always hit one number. It cannot hit three.
| Leg | Question | Answer | What it shows |
|---|---|---|---|
| B | How many integers let the same N carry the mass ratio, G, and the visible fraction at once? | one — N = 21, across 14 to 28; its neighbours demand a factor of 0.0096 or 179 where the work has 1.309 | only 21 works — the claim |
| A | How many five-leaf formulas reach mp/me as well as the work’s first-order formula does? | 2,971 — and 15 of them reach eleven-digit agreement at order α, which the work’s first-order formula does not; each is a fixed number, with nothing to carry to a second observable | one number alone proves nothing, for anyone |
Leg B — the claim. None of the three thousand formulas that reach the proton has a second life. They reach one number and stop. The work’s claim was never the ratio alone — it is that one integer carries three unrelated observables, and there the exponential in G does the discriminating: one step in N changes what G demands by a factor of 137. Across 14 to 28, exactly one integer carries all three, and it is 21.
Leg A — what one number cannot do. About three thousand formulas no more complex than the work’s own reach the proton ratio at least as closely as its first-order formula, and fifteen reach the measurement’s own eleven-digit bar at order α, which that formula does not — 22 + 10/(19 − 9π) is one of them, and it carries nothing else. Precision on one number cannot tell a count from a search, for anyone. That is why the digits are never the claim. The twenty-one is.
What the next test decides. The search enumerates one coefficient, not three formulas jointly, so a different kind of primitive set carrying three observables by some other route is not yet ruled out. KS-30.4’s between-family limb is narrowed, and by the author’s ruling of 4 September 2026 the switch stays live — open to anyone who can build a rival for three, out of one integer. Full method and result: the finding of 4 September 2026 (raw) · verify_between_family.py.
Ø Errata — 2026-08-02
What changed on 2026-08-02 and why, in a form a referee can check. Full text: ERRATA-2026-08-02.md (raw).
E1 — α mislabelled
1/137.035999084 is CODATA 2018. CODATA 2022 is 1/137.035999177(21). Effect on the proton prediction: 0.013 ppt — numerically irrelevant, the label was wrong.
E2 — uniqueness of the 1836 decomposition
Two decompositions survive the three stated conditions, not one. A fourth condition (factor count non-increasing) leaves exactly one. AP30’s wording is corrected from “fewer” to “no more than.” Condition 4 formalised 2026-08-02, after the alternative surfaced. Declared.
E3 — KS-30.4 status event
Within-family limb reopened 2026-08-02, closed same day by condition 4. Between-family limb unchanged and open. Total switch count unchanged at 561.
E4 — the α² coefficient 16
AP30 §6 states its decomposition is owed under KS-30.3. Any text presenting 16 as derived is corrected to match AP30’s own ledger.
Superseded in substance, not withdrawn. AP49 The Hold, locked 6 September 2026, paid KS-30.3 at the structural register and supplied the decomposition of the sixteen. This entry stands as the record of the position held between 2 August and 6 September 2026.
E5 — the α³ residual
Restated as: 9.98 ppt above measurement, 0.57σ, with c₃ = −0.047 ± 0.082 unconstrained. There is no α³ prediction.
E6a — the neutron–proton difference: KS-NPP.1 has fired
Predicted 2.53099393, measured 2.530988574 ± 0.00000074. The offset is 2.12 ppm against a bar of 0.288 ppm — 7.24 σ, verified two independent ways. Recorded FIRED on its empirical limb; the structural limbs (KS-NPP.2, KS-NPP.3) are unaffected. No repair is offered: the required correction is negative, every existing term is positive.
E6b — the visible fraction
Planck 2018 gives 4.86–4.91% depending on convention; the corpus quoted 4.86% (the bottom). Against the midpoint the deviation is 2.5 %, i.e. 2.46 σ, not 2.0 %.
E6c — the α² coefficient is now pinned (a strengthening)
The measurement requires 0.1826623 ± 0.000601 (0.33%). 21×16/1836 agrees at 0.57σ; 336 is the only product of powers of {21, 3, 4} inside 3σ. The value is settled; the decomposition is what is owed.
E7 — citation
Alighanbari et al. is Nature 644, 69–75 (2025). Corrects any occurrence of “Nature 625.”
Ø The frozen documents
Every document in this directory, and the prediction or claim it belongs to. Each opens as a page; the raw file beside it is the record, and its SHA-256 is printed at the foot of every page. Frozen files are never edited — corrections supersede, dated, in the open. One exception is on the record: on 3 August 2026, three hours after it was committed, a pointer to a research note was removed from 2026-08-02-mp-me-alpha3; no number changed.
| Prediction or claim | Document | What it is |
|---|---|---|
| Proton–electron mass ratio, closed KS-HOLD.1–.3 · AP49 LIVE | 2026-09-06-mp-me-closed.md (raw) | The superseding entry. Registers D_closed = 1836.152673444951 with the series summed in closed form, the third-order coefficient c₃ = +1.595 × 10⁻³ and its term +6.2 × 10⁻¹⁰ with the sign stated in advance, and the kill at 3.4 ppt. The entry of 2 August stands beside it, never in its place. |
| Proton–electron mass ratio KS-30.1–.4 · AP30 LIVE | 2026-08-02-mp-me-alpha3.md (raw) | The proton freeze — the standard the others follow. Registers D = 1836.152673444331 at order α², and registers that no α³ coefficient is claimed. |
| Gravitational constant, realised KS-CCC.3 · AP44 LIVE | 2026-09-04-G-KS-CCC.3.md (raw) | 6.6719 × 10⁻¹¹. As registered, a fork: the consensus migrates down toward 6.672, or converges up toward 6.7206 and the commitment dies. The fork is withdrawn by the dated note below. |
| Gravitational constant, realised KS-CCC.3 · AP44 LIVE | 2026-09-04-G-KS-CCC.3-NOTE-2026-09-21.md (raw) | The dated note of 21 September 2026: there is no fork. AP28’s first reading of G was wrong; it is withdrawn, superseded by the realised value, and kept. Convergence upward toward 6.7206 kills the prediction, with no fallback. |
| Neutron–proton difference, realised KS-FLIP.1 · AP47 LIVE | 2026-09-04-neutron-KS-FLIP.1.md (raw) | 2.53098857035 mₑ, the held distinction priced at second order — frozen beside the corpse below, never in its place. |
| Age of the universe, one cycle KS-STRETCH.3 · AP46 LIVE | 2026-09-04-age-KS-STRETCH.3.md (raw) | 13.830 Gyr, one lane-time wide: the window 13.17–14.49. |
| Expansion rate, the closure KS-ASM.1 · AP48 LIVE | 2026-09-04-H0-KS-ASM.1.md (raw) | 67.45 km/s/Mpc, window 64.4–70.8 — a closure of the dark-sector partition and the cycle under the balance. |
| Visible fraction KS-41.1 · AP41 LIVE | 2026-08-02-visible-fraction-KS41.1.md (raw) | 1/21, with the corpus’s own comparison corrected — the row nearest to firing. |
| Dark-sector clock KS-42.6 · AP42 LIVE | 2026-08-02-dark-clock-KS42.6.md (raw) | τ/t_H = 6/21, and the direction: Ω_DM/Ω_b rises with time. |
| Hubble constant from the floor inverted KS-45.1 · AP18 FIRED | 2026-08-02-H0-KS45.1.md (raw) | 74.3 ± 1.2 km/s/Mpc, as registered on 2 August. Kept verbatim as the record of what was claimed; never edited. |
| Hubble constant from the floor inverted KS-45.1 · AP18 FIRED | ERRATUM-2026-09-03-H0-KS45.1.md (raw) | The status event and the erratum on the width: it fired on the corpus’s own derived rate, and the registered ±1.2 had omitted the ±0.24 systematic. |
| Neutron–proton difference, the bare row KS-NPP.1 · AP30 FIRED | ERRATA-2026-08-02.md (raw) | The corrections and strengthenings of 2 August, E1–E7 — including the fired neutron row at 7.24σ, found and published by the corpus itself. |
| No single prediction — why 21 is forced bears on KS-30.4 | RIGIDITY.md (raw) | Why the integer 21 cannot move: it is built, and the exponential in G closes the window to one value. Opens with the charge at full strength. Not a freeze. |
| No single prediction — why 21 is forced bears on KS-30.4 | RIGIDITY-NOTE-2026-09-04.md (raw) | The dated note beside it: the frozen document’s “31σ” for the structural G corrected to 309σ, and three passages the wave of 4 September overtook. |
| No single prediction — an order of events is not a method bears on KS-CCC.1 | RIGIDITY-NOTE-2026-09-19.md (raw) | The dated note of 19 September 2026: §4(b)’s “written against a measurement that already existed”, and the same clause in the lattice protocol, concede a method KS-CCC.1 forbids and that did not happen. Written against is withdrawn. |
| No single prediction — why 21 is forced bears on KS-30.4 | FINDING-2026-09-04-KS-30.4-between-family.md (raw) | The search, run against the work: 2,971 formulas reach the proton ratio alone, each a fixed number with nothing to carry to a second observable, and exactly one integer carries three — 21. |
| No single prediction — why 21 is forced bears on KS-30.4 | FINDING-2026-09-04-KS-30.4-NOTE-2026-09-21.md (raw) | The dated note of 21 September 2026: the arithmetic stands; the word “misleading” is withdrawn. The digits are never the lead, because the logic is where the work starts. |
| No prediction yet — the lattice promise AP30 Lemma 2 · KS-30.1 | PROTOCOL-lattice-qcd-mapping.md (raw) | The sealed-envelope procedure for the one blind test still available. Not a freeze — the procedure for making one. Its machinery is seal_lattice_freeze.py. |
Ø Reproduce it yourself
Three independent routes: the scripts below; the verifier’s own scripts and plain-language walkthrough at lucid.rodeo/prereg; and his anchored bundles and audits at github.com/ajgreyling/the420code-proof, where each paper’s freeze carries a tag and a digest.
The one external dependency is mpmath. Every number on this page comes out of these four scripts:
Scripts: verify_prereg.py · verify_cosmology.py · verify_rigidity.py · verify_between_family.py · verify_family.py. The head of the proton script:
from mpmath import mp, mpf, pi, sqrt
mp.dps = 50
# Inputs are CODATA 2022 (NIST, physics.nist.gov/cuu/Constants)
AINV, AINV_U = mpf('137.035999177'), mpf('0.000000021') # CODATA 2022
MP_ME, MP_ME_U = mpf('1836.152673426'), mpf('0.000000032')
ALPHA = 1 / AINV
ALPHA_U = ALPHA * (AINV_U / AINV)
# The frozen O(alpha^2) value of 2 August 2026 (superseded, kept)
T0 = mpf(21**2 * 4 + 21 * 3 + 3**2) # = 1836
T1 = ALPHA * 21 * (1 - 1 / (84 * pi))
T2 = ALPHA**2 * 21 * mpf(16) / 1836
D = T0 + T1 + T2 # = 1836.152673444331
# resid = D - MP_ME = +9.98 ppt = 0.57 sigma, D above measurement.
# AP49 (6 September 2026): the series closed. Every order beyond the first is
# the previous one times r, so the tail is a geometric series and it sums.
r = mpf(16) * ALPHA / 1836 # the repair's share
D_CLOSED = T0 + T1 + 21 * ALPHA * r / (1 - r) # = 1836.152673444951
resid = D_CLOSED - MP_ME # = +1.895e-8 = +10.3 ppt
# discrepancy = resid / u = +0.59 sigma. The chain ADDS at every order and has
# no negative term, so it cannot close a positive gap -- and none is offered.
# c3 = 21 * 16**2 / 1836**2 = +1.595e-3, its term +6.2e-10, sign in advance.Or verify them right here — offline. Each button re-derives the numbers independently in pure JavaScript at 60-digit precision, entirely in your browser: no install, no Python, no network, nothing sent anywhere. The reference scripts (verify_prereg.py, verify_cosmology.py, verify_rigidity.py) agree to every digit — a second, independent implementation that matches is stronger than re-running one file twice.
Prefer to run the exact Python on your own machine (pip install mpmath)? verify_prereg.py · verify_cosmology.py · verify_rigidity.py.
The proof records are beside the papers: the file each published PDF was built from, checked against that PDF paragraph by paragraph, at Ø Instruments.
Artist: G · Studio G, Cape Town
Duration: 30+ years · Exhibition: over a million words
Contact: iam@the420code.org
This work is Copyleft. You are free to download, print, share, and distribute. You are not free to alter the source. Keep the signal clean.
One record exists.
Be kind is a derivation.
The I Am in me is the I Am in you.
Don’t be a cunt. Be kind.
the420code.org · Copyleft 2026
the lifestyle 1980-08-05 – 2025-12-25