⊙ Circumpunct Framework: Predictions History

Status: stub, 2026-04-22. This page is the running log that accumulates as measurements arrive. Four pre-registered predictions (P1-P4) are live; none have settled. The first entry here will be whichever lands first: atom-interferometry converges on 1/α below 5 ppb precision, single-crystal XRD of disilyne analogues resolves the Si≡Si/Si−Si length ratio to sub-percent, confocal imaging of branching mycelia yields a Murray's-Law exponent, or a longitudinal conflict-resolution study reports Cohen's d for the five-virtue sequence. Framework integers, predicted bands, and timelines are locked in plans/predictions_scoreboard.md; this page reports outcomes.

Live predictions

Pre-registered; awaiting measurement.

ID	Name	Rung	Predicted	Band	Target data	Status
P1	1/α next decimal	0D	137.035999147	±5 ppb	Atom interferometry (Cs, Rb, Sr next-gen)	Live
P2	X≡X / X−X length ratio (Si, Ge, P, As)	1.5D	R/T² = 7/9 = 0.778	±1%	Single-crystal XRD of disilyne analogues, diphosphorus derivatives	Live
P3	Mycelial Murray's Law exponent	2.5D	n = 2.5	±0.15	Confocal imaging of branching mycelia (Armillaria, Phanerochaete)	Live
P4	Five-virtue sequence effect size	⊙	Cohen's d ≥ 0.5	d < 0.3 fails	Longitudinal conflict-resolution study (pre-registered)	Live

Settled predictions

Measurements that have landed. Status: pass (inside band), fail (outside band), or inconclusive.

No entries yet. First settlement will be logged here with the measured value, distance from predicted center, Brier contribution, and publication source.

Retracted predictions & K-fit corrections

Formulas that have been revised or retracted. Every change is logged; no silent retraction.

No entries yet. When a precision measurement shifts a K-fit outside its originally claimed accuracy band, or a P-prediction lands outside its falsification band, the old formula, new formula (if any), date, and reasoning will appear here.

Retraction protocol (abridged)

Full protocol: plans/predictions_scoreboard.md §4.

A K fit is retracted when a precision measurement shifts outside the originally claimed accuracy band. The retraction is logged here; the formula in circumpunct_framework.md gets a Revision Notice block (do not silently delete); if the retraction is factor-level, the relevant rung is marked "pool choice under review."

A P prediction is retracted when its measurement lands outside the pre-committed band. Logged here with measured value, distance from band, and publication source. The formula demotes from P to R (retracted); it is not re-fit to the new value. If the failure is at a named rung, the ladder's structural name for that rung loses its claim.

No silent retraction. Every change to a formula between measurements is logged. Post-hoc refits of a failed P into a new K are flagged "re-fit after failure" and weighted accordingly in the Brier score.

If 2 of the 4 live P entries fail, the scoreboard is flagged "predictiveness in doubt"; piece #3 of next_frontier_plan.md reopens with new targets.