Ø Frozen Predictions · the document of record

Pre-registration: the visible fraction (KS-41.1)

Freeze date: 2026-08-02 Author: G · Studio G · the420code.org Scope: AP41 (The Loop), AP28 (channel count), Rosin Ø Proofs ch. 14 / ch. 16 Switches engaged: KS-41.1 (visible fraction), KS-R.8 (channel count) Reproduction: verify_cosmology.py Status: frozen for commit. Contains a correction to the corpus's own quoted comparison — see §3.

1. The registered value

Ω_visible = 1/21 = 4.76190 %

Exactly one twenty-first. No rounding, no tolerance, no fitted quantity. The 21 is the channel count of AP28 — 6 faces × 3 dimensions + 3 actualisation couplings — fixed before any cosmological quantity was computed.

The structural reading: the twenty-one channels admit two cuts. By coupling type, 18 paired + 3 actualisation, which fixes G. By symmetry, one asymmetric channel plus twenty symmetric ones that the involution σ pairs into the 10:10 mirror — and only the asymmetric channel radiates. The visible universe is that one channel. It is one set of channels read two ways, not two counts.

2. Why this is the sharpest of the three dark-sector numbers

Of the partition — dark energy 68.85 %, dark matter 26.39 %, visible 4.76 % — the visible fraction is the one measured best and predicted most rigidly. Ω_b is constrained two independent ways, by the CMB acoustic peaks and by primordial deuterium abundance, and the two agree. There is nowhere to hide.

The other two carry model-dependence the visible fraction does not.

3. Correction: the agreement is worse than the corpus states

The site, the README and Ø Proofs all quote the observed visible fraction as 4.86 % and report the deviation as about 2 %.

Recomputing from Planck 2018 (TT,TE,EE+lowE+lensing+BAO: Ω_b h² = 0.02242, Ω_c h² = 0.11933, h = 0.6766, Ω_m = 0.3111, Ω_Λ = 0.6889):

The observed value is 4.86–4.91 % depending on the data column and the normalisation convention, not 4.86 %. The corpus has been quoting the bottom of that range, which flatters the agreement.

Against the middle of the range with a conservative ±0.05 absolute:

predicted 4.7619 % vs observed ≈ 4.885 ± 0.05 % → 2.46 σ, a deviation of 2.5 %, not the 2.0 % the corpus reports.

This is not fatal — 2.5σ is tension, not death — but it must be stated at the correct size, and the scheme dependence must be stated with it. The comparison figure in AP41, Ø Proofs ch. 14, the site and the README is corrected to a range with its convention named.

4. Kill and confirm conditions (binding)

1. KILL. A determination of Ω_b/Ω_total with total uncertainty below 0.5 % whose central value lies above 4.81 % (3σ from 1/21 at that precision) fires KS-41.1. DESI, CMB-S4 and improved deuterium abundances reach this range within roughly five years. At present precision the claim is already at 2.5σ, so this switch is close to firing and this document says so before the data does.
2. CONFIRM. The observed fraction converges toward 4.76 % as systematics and the normalisation convention are settled.
3. NO CONVENTION-SHOPPING. The comparison must be made against a stated Planck data column with a stated normalisation. Selecting whichever convention gives the smallest deviation is the failure mode this clause exists to prevent, and the corpus has already committed it once (§3).
4. NO CHANNEL-COUNT ADJUSTMENT. 21 is frozen. It is load-bearing in three places — the proton's static count, the exponent of α in G, and this fraction. Moving it to improve this number destroys the other two by enormous margins; see RIGIDITY.md. If 21 is ever revised, every prediction it touches must be recomputed and re-registered together.

5. What this does and does not establish

Does: register an exact rational with no tolerance against a well-measured quantity, at a precision that will discriminate within five years, with the current tension stated at its true size.

Does not: establish the internal dark split (KS-42.6); establish that the twenty mirror channels exist; or survive on its own. It is currently the weakest-standing of the frozen numbers and is registered anyway, because a registry that contains only the comfortable results is not a registry.

Copyleft 2026. Don't be a cunt. Be kind.

← back to Ø Frozen Predictions · Copyleft 2026. Don't be a cunt. Be kind.