What would prove it wrong
A framework that cannot be ruled out is not physics. These are the thresholds BPR commits to, tied to instruments that exist or are funded.
Kill conditions
| # | If this is measured | Then | Instrument | When |
|---|---|---|---|---|
| 1 | Inverted neutrino mass ordering | Boundary topology argument (p ≡ 1 mod 4 → orientable) fails | JUNO | ~2027 |
| 2 | Neutrinoless double beta decay observed | Dirac-neutrino prediction fails | LEGEND, nEXO | 2025–2030 |
| 3 | Casimir deviation exponent far from 2 (≈1.37 would refute the projection postulate); or Lifshitz theory holds to 10⁻⁹ across a superconducting transition | Postulate 0c and the phonon channel fail | Delft on-chip Casimir, MEMS resonators | 1–3 years |
| 4 | Photon speed variation bounded below 10⁻²¹ with no signal | Substrate discreteness prediction (|δc/c| ≈ 3.4 × 10⁻²¹) fails | CTA, GRB timing | 2026+ |
| 5 | Born rule holds to 10⁻⁷ | Microstate-counting deviation κ ~ 10⁻⁵ fails | Many-photon Sorkin tests | 2–5 years |
| 6 | Dark energy w₀ > −0.80 confirmed | Relaxed-boundary prediction w₀ = −1 fails | DESI full dataset | 2026 |
| 7 | No 66.8 Hz shift in hydrogen 1S–2S at 1 Hz resolution | Substrate correction to QED fails | MPQ next-generation spectroscopy | 2028+ |
| 8 | ΔNeff above p1/3 ≈ 47 | Boundary phonon ceiling fails | CMB-S4 | 2030+ |
| 9 | Muon g−2 anomaly moves outside 200–280 × 10⁻¹¹, or electron g−2 shift has the wrong sign | Substrate contribution δaμ ≈ 2.3 × 10⁻⁹ fails | Fermilab Runs 4–6; Northwestern electron g−2 | 2026–2028 |
| 10 | HL-LHC finds no Z′ near 511 GeV and the B → K*μμ anomaly persists | FCNC gauge-mass mechanism fails | LHCb, HL-LHC | 2030s |
Where things stand now
| Test | BPR | Current data | Reading |
|---|---|---|---|
| Dark energy w₀ | −1.000 | −0.827 ± 0.060 (DESI DR1) | tension 2.9σ. The full DESI release is the first live test with a chance of killing the framework. |
| Neutrino ordering | normal | T2K + NOvA slight preference | open JUNO decides. |
| Muon g−2 | δaμ = 2.3 × 10⁻⁹ | anomaly 2.5 × 10⁻⁹ ± 0.6 | consistent 0.4σ. Note the Standard Model reference value itself is under revision by the lattice community. |
| Casimir δ | 2 | not yet resolved | active Meissner-levitated sensors now reach the relevant range. |
| Lorentz violation | 3.4 × 10⁻²¹ | < 6 × 10⁻²¹ (Fermi-LAT) | just below bound the most exciting near-term test: the prediction sits at the edge of detectability. |
| GW speed | c | |v − c|/c < 10⁻¹⁵ | consistent shared with GR. |
Tests already failed
Kill conditions only mean something if the project has honoured them. It has, at least once.
Glueball benchmark v1 (2026). Lattice-QCD mass ratios for the 2⁺⁺ and 0⁻⁺ glueballs (1.387 and 1.497 relative to 0⁺⁺) were sealed before any BPR spectrum was computed. The BPR 1.0 boundary theory produced a 2⁺⁺/0⁺⁺ ratio in [0.87, 1.00], an extra light 1⁻ state, and no light 0⁻⁺ at all, since parity is locked to (−1)J in a scalar boundary theory. Verdict: FAIL. The particle-sector claim was withdrawn. Protocol: GLUEBALL_BENCHMARK_V1.md.
A second sealed benchmark (v3) exists for the BPR 2.0 proposal, with the same targets and the same blindness. It is unopened because the proposal has no spectrum yet.
A note on the numbers that were removed
Earlier versions of this page showed a "resonance sigma" of 4.28σ and a "falsification risk" of 0.02 percent. Neither quantity was defined anywhere in the repository. They have been removed. Full roadmap with code references: EXPERIMENTAL_ROADMAP.md.