Prediction Locks and Falsification Registry
Prediction Locks
This page records DT/FDS prediction locks, pre-registered statements, refined descendants, and their failure conditions.
A prediction lock is not a claim of certainty. It is a public commitment to a model identity, its dependencies, and its demotion criteria. Predictions may be supported, weakened, demoted, quarantined, or abandoned according to the stated outcome tables.
FDS-G1 / G1DE-M3/4 — Projection-Locked Background–Weyl Residual
Current Leading Prediction-Locked Branch
Status: Prediction-locked refined branch; exact-pilot evidence-selected; production validation pending.
FDS-G1 is the refined descendant of the X1 horizon-ledger direction. The current leading empirical branch is the projection-locked G1DE-M3/4 model.
Locked late-time residual pattern:
s < 3, μ(a,k) = 1, Σ(a,k) − 1 = −¾(3 − s) ̅sH(a), ̅sH(1) = 1.
Here ̅sH(a) is the normalized horizon-response output shape. It is not a uniquely identified microscopic source–kernel factorization. The model is a sparse background–Weyl residual, not a flexible dark-energy or generic modified-growth fit.
Current exact-pilot matched evidence hierarchy:
M3/4 > Mκ > const-Σ > G1DE-2 > CPL > ΛCDM.
This result should be read as a preliminary exact-pilot evidence selection, not final cosmological confirmation. Production nested-evidence refinement, expanded weak-lensing / EG / 3×2pt likelihoods, full wide-prior baseline sensitivity, and independent replication remain pending.
Forward prediction lock
- Background: future BAO/SN data should continue favoring s < 3 over the s = 3 ΛCDM-like limit.
- Weyl/lensing response: expanded lensing and EG/3×2pt likelihoods should support a suppressed Weyl response with shared ̅sH(a) shape.
- Growth response: μ(a,k) should remain near 1; large |μ−1| comparable to |Σ−1| would weaken the Ward-suppressed Ricci-leakage branch.
- Projection lock: free κ should not decisively beat κ = 3/4.
- Model identity: no independently sampled free A(a,k) function should be required.
Demotion / failure conditions
- If free κ decisively beats M3/4, the exact 3/4 isotropic projection lock is demoted.
- If constant-Σ decisively beats the output-response-shape model, ̅sH(a) is demoted while the Weyl-amplitude branch may survive.
- If data require |μ−1| comparable to |Σ−1|, the Ward-suppressed Ricci-leakage branch fails.
- If a free A(a,k) response function is required, the model becomes a generic dark-stress source rather than G1DE.
- If production-level evidence returns to CPL or ΛCDM dominance, the current G1DE observational branch is demoted.
Full prediction registry: FDS_G1/prediction_registry.md
FDS-X1 v1.2 — Horizon-Maintenance Dark Energy
Historical Precursor Bridge Note
Status: Released / frontier physical consequence (P3). Historical precursor prediction.
Core claim: If cosmological horizons are finite distinguishability boundaries, then horizon-maintenance cost naturally has scale ρHM ~ H2MPl2, with physical non-phantom tendency and late-time approach toward w = −1.
Not claimed: This does not derive general relativity, quantum gravity, the cosmological constant, or a complete cosmological model. The note is a frontier physical consequence (P3), not a completed theory.
Failure condition: Robust high-precision confirmation of constant w = −1 with no density evolution demotes the dynamical version. Robust unavoidable physical phantom behavior falsifies the strict non-phantom version.
Relation to G1: X1 is retained as the earlier horizon-ledger bridge prediction. G1DE-M3/4 is the refined G1 branch: it adds Weyl-normalized projection, μ ≈ 1 growth suppression, the 3/4 optical projection lock, output-response normalization, matched evidence controls, and explicit demotion ladder.
Full prediction registry: prediction_registry.md
Note. Prediction locks are not claims of certainty. They are public commitments to model identities, dependencies, and demotion criteria. Pre-registered predictions are recorded before specific outcomes are known. Prediction-locked refined branches carry forward-looking test channels.