The 2026-08-04 news round updates the scorecard against the current literature with no new analysis of our own — radical honesty in both directions. CONFIRMING SIDE: JUNO's Neutrino-2026 update (207 days) reports sin²θ₁₂ = 0.3036 ± 0.0064, the sharpest single measurement of the solar angle ever — +0.49σ from the frozen prediction 0.306747, with the central value moving toward it from the first 59.1-day run (0.3092); ICARUS's first standalone result (no ν_μ disappearance, new 3+1 exclusions at 90% CL) adds another null leg exactly as N_fam = 3 requires; the sequentially-valid DESI reanalysis localises the evolving-dark-energy signal almost entirely in one redshift bin (LRG2 — without it the evidence collapses, mildly pro-ΛCDM), continuing the dissolution trend the w = −1 watchdog predicts; two new independent birefringence analyses (Planck scale-dependence β = 0.30° ± 0.05° with the constant model Bayes-preferred; PR4 field-level 0.32° ± 0.12°) stay consistent with the 0.2424° seed; and the CMB-SPA H₀ = 67.19 ± 0.38 lands +0.11σ from the parameter-free budget value 67.15 (correlated legs, never a new independent hit). WEAKENING SIDE, equally prominent: the CMB-SPA combination (SPT-3G D1 + ACT + Planck) gives n_s = 0.9679 ± 0.0033 — the central value sits ABOVE the TFPT band edge 0.967 for the first time (edge at −0.27σ, preferred branch 0.9611 at −2.06σ); this is not yet the dated kill ('robustly ≥ 0.967') but is typed an honest mild weakening and watch-flagged; the DESI DR2 effective-mass posterior keeps the Σm_ν = 0.0588 eV floor under +3.1σ model-dependent pressure (unchanged, on record). AND ONE BRIDGE DIES: the muon g−2 seam-vertex bridge's dated kill condition (2026-07-02) has fired — WP25 (Phys. Rep. 2025) adopts the lattice HVP as the SM baseline and the BMW hybrid evaluation (Nature 2026) confirms it, experiment − SM = 0.5σ, so the converged Δa_μ lies outside 2.879×10⁻⁹ ± 0.5×10⁻⁹: the vertex identification is excluded in its present form, the bridge row is typed kill_channel, and the exact compiler number δ₂/(2π) plus the compiler core are untouched — the watchdog delivered exactly what it was built for. Operationally the round also froze the Vela GO-programme preregistration (p9_go_programmes_v1.yaml) ahead of the expected 2026/27 giant-glitch window and recorded the CHIME/FRB Catalog-2 + RepCat4 data availability in the FRB experiment folders.

The frozen log-comb frequency is confronted with the two largest untouched ranges in nature — (i) the ultra-high-energy cosmic-ray spectrum (Auger Open Data, 76k events, 0.1–144 EeV = 2.99 comb periods, the largest ln-E range anywhere; smooth model with knots frozen at the published spectral features): the kernel comb is absent (MC p = 0.49, off-kernel rank p = 0.77; injection power 97% at 5% amplitude, 23% at the predicted 1.7% — the full-statistics Auger/TA spectrum is the dated decider); and (ii) the primordial power spectrum — the one natural bed where the log-clock is MOTIVATED (inflation e-folds): the frozen ω = 2.583 sits inside the published Planck log-oscillation search band with no detection, and the predicted 1.7% amplitude lies a factor 1.7 below today's 95% bound (~2.9%) — a zero-parameter dated decision point for CMB-S4-class data (a future bound below 1.7% at the frozen frequency with no detection kills the primordial reading).

Four probes implement the strategy shift from visible numbers to state mixing — the burst-observable covariance IS strongly block-diagonal (p = 0.006 against a spectrum-preserving null) but the found blocks are exactly the standard Faraday-vs-emission sectors (default astro); no forbidden region exists in the polarization state space beyond the marginals; the delay-embedded energy dynamics have effective rank 0 (memoryless, no three-mode manifold); and within-session energy sequences are TIME-REVERSIBLE at catalog statistics (reversal-null p = 1.0) — an arrow-free cascade that bounds every directed-recovery reading of burst trains and matches the threshold-amplifier contract's prediction.

Five preregistered axes test what FRBs could BE in TFPT, one level above the number tests — (i) the first FRB axis with a NAMED transduction (bursts as probe pulses through the relaxing magneto-ionic medium; DM/RM are known linear line integrals) finds the nightly medium state of FRB 20240114A does not require a second relaxation rate (AIC prefers single memories), and the intra-session follow-up (two sources, FRB 20240114A + FRB 20201124A) finds the strong within-session RM variance carries NO temporal memory at all against an injection-validated gate — the burst-sampled scatter is per-burst/magnetospheric, not a relaxing medium state, so the two-rate ratio test at ln3/ln(3/2) = 2.7095 has nothing to act on at burst cadence (data-limited; the reading survives only in the untested hours-to-days band or via a tracked rotation-measure injection event); (ii) the first FRB comb test with a NAMED clock (τ_mod = cumulative |ΔRM|, the S14 medium-state clock) is null at detectable amplitude (3 gate-passing sessions, Fisher p = 0.81); (iii) the parity-without-rate diagnostic holds all three predictions — a nominal m=4 PA excess is typed as distribution misfit by the dated specificity battery (odd-harmonic placebos fire equally, a 3-component smooth null absorbs it at p = 0.64), class switches are rate-free (tooth p = 1.0), and the dominant PA class persists across 45/45 sessions — consistency, standard magnetospheric physics predicts the same; (iv) the episode-level transfer ladder (quiescence gaps + episode-integrated energies on the teeth, the last untested aggregation level) is null across 8 sources (gap Bonferroni p = 1.0); (v) the Z₂ leaf-class reading of the repeater/one-off dichotomy fails its sharpest morphological test on CHIME Cat1 — morphology does predict repetition (CV-AUC 0.833, replicating Pleunis+2021, default astro/selection), but BIC prefers three morphological modes, not two. A transduction-invisibility theory contract (5/5 machine checks) shows a coherent threshold amplifier with uniform character coupling produces burst trains exactly and statistically blind to the transfer spectrum — the 20+ FRB intensity/timing nulls are what that ontology predicts, and an unequal-weight coupling makes the same functional discriminating.

Six structurally untested FRB axes preregistered/addended and run on committed catalogs — the FORCED 2D ladder coupling (a kernel rung multiplies time by (3/2)^k AND energy by (2/3)^k; all prior tests were marginals) is cleanly null in both sources under an exact energy-shuffle null; the circular-polarization HANDEDNESS alternation (the only Z₂-deck-visible polarization observable — PA is mod π and |V|/I is even, so all earlier polarization tests were deck-blind by construction) is null over 3454 signed-V burst pairs; and the size-space DSI on the burst-energy distribution is null in both sources after a lognormal/KDE/GMM population-null battery (the famous energy bimodality absorbed, the pulsar PG.01 result replicated in the FRB energy domain; a first-pass λ=8 artefact from a re-seeding GMM sampler was caught by split-half and free-λ probes and dissolved — on record). The μ₄ phase–time helix (coupled polarization–time phase at the frozen ω) is also null once the drift-robust circular-shift null is applied — the naive permutation 'signal' is slow position-angle drift, present at any ω. The new non-tooth operator proxies are also null: KC.05 finds no C4-specific μ₄ block protection in PA-class transitions, and KC.06 finds the multivariate polarization lag spectrum is not unusually close to {1,64/729,1/729}. And the shared-seed block now carries a dated FORWARD band: Cabibbo + θ13 + Ω_b predict β = 0.2413 ± 0.0018° (β never entered the fit; ACT DR6 sits at z = −0.36) — LiteBIRD / Simons Observatory will test this band blind.

The seam clock is now probed as an OPERATOR SYMMETRY, not only as a number — (i) the QGEO S_off leakage statistic (μ₄ character block-diagonality of the boundary energy form, the operator form of the seam premise) is certified on a reconstruction pipeline with noisy eigenpairs (theory contract: exact leakage < 1e-30, μ₄ clock-angle zero set read off the operator, instrument requirement ~1e-3 relative spectral precision) and then run on a REAL measured boundary operator: the KIT4 open EIT archive (38 tank measurements, 16 electrodes) — the homogeneous tank realises a μ₄-block-diagonal ND map at the instrument floor (S_off = 7.6e-5), all 28 anisotropic targets leak with the generic isotropic signature, and the decisive μ₄-positive 4-inclusion configuration is preregistered as a future measurement any KIT4-class lab can run; (ii) the shared-seed reading is upgraded to a DECODER test: one latent u fits β, Ω_b, reactor-θ13 and the Cabibbo angle simultaneously (χ² = 4.10, dof 3), AIC prefers the 1-parameter shared decoder over the saturated model, 0/14 single-swap neighbour link-decoders beat it and it sits at the 0.0th percentile of 2000 equal-complexity random placebo decoders — the cross-channel ratios are architecture-specific (consistency, not proof); (iii) a per-source comb-PHASE coherence test (RC.04) bounds the persistent boundary-clock reading on FRB repeater sessions (null at ε ≳ 0.15–0.3).

The Z2/double-cover READINGS of the recovery comb are now tested — if the comb carries the Z2 sheet parity per period (antiperiodic, periodic only on the double cover), its Fourier power at the kernel ω = 2.583 is exactly zero, so every kernel null was silent about it; the three derived omegas (half-period clock (3/2)³ at ω = 5.17, first antiperiodic harmonic (3/2)⁴ at ω = 3.87, antiperiodic fundamental (3/2)¹² at ω = 1.29) were run through ALL in-hand dynamic channels: the USGS aftershock battery, magnetar, FAST/GBT + CHIME FRB tails, 22 GRB afterglows, the ZTF nuclear transient, the pulsar-glitch legs (2024 Vela giant glitch and the PuMA/IAR daily ToA residuals of three giant glitches), the 9 gate-passing FRB repeater burst-time sessions, crust cooling, and the FRB 20220912A microshot forest — where the half-period reading is the first to pass the per-λ range gate (3.2 periods) and is null. In the GW ringdown-echo channel the antiperiodic reading was already covered natively as the Δφ = π per-bounce phase × (2/3)³ semantics of the Stage-1c signature battery (NO_VARIANT_ECHO on all 10 events). Verdict: clean NULL everywhere testable (well-powered in the quake, GRB, PG.08-J1740 and repeater-session stacks; quake Bonferroni global p = 0.94 over 11 gated ratios). Two nominal excesses are placebo-typed artefacts, on record: the J1740−3015 (3/2)⁴ raw p = 0.0018 is a broad non-specific ω ≈ 3.6–4.0 bump (periodogram peak at 3.73, placebos equally significant), and the repeater-cascade (3/2)¹² Fisher p = 4e-4 is sub-gate (1.5–2.4 < 2.8 periods; placebo λ 60–300 equally extreme). The antiperiodic fundamental remains range-blind in all current data (needs 13.6 e-folds of ln t; the widest curve has 13.1) — an honest data-limited leg. The first real magnetar run is also typed: 6 Swift-XRT outburst curves, stacked kernel ω = 2.58 not special (p = 0.99), a clean surface-firewalled null.

The fixed-point watchdog gains three dissolution axes — axis E fixes a PARAMETER-FREE neutron lifetime from the frozen Cabibbo angle (exact CKM unitarity ⇒ V_ud = 0.97450; with the Czarnecki–Marciano–Sirlin master formula and measured g_A: τ_n = 877.53 ± 0.71 s — on the BOTTLE side of the ~4σ beam–bottle puzzle at −0.38σ from UCNτ, −4.98σ from the proton-counting beam average, with the n→dark-decay exit forbidden by the same no-slot counting as the sterile axis), axis F watches X17 (MEG II null vs PADME's 1.8–2.0σ global excess; Run IV decides), axis G watches the standing R_D(*) excess (HFLAV 3.8σ — the sharpest dissolution target, Belle II/LHCb Run 3 decide). The parameter-free flat budget now also takes a side in the SECOND structure tension: CAMB on the budget geometry gives S8 = 0.824 — CMB/KiDS-Legacy side (+0.49σ), against DES Y6 (+2.95σ). Two more open questions are typed honestly: the cosmological lithium-7 problem under the frozen η (D/H −0.3σ and Yp +0.5σ anchor η₁₀ = 6.04; ⁷Li overpredicted ×3.36 — the UNIVERSAL 20-year problem, not TFPT-specific; TFPT's dated statement: the resolution must be stellar, η cannot move), and the cosmic number-count dipole anomaly (CatWISE ×2.1 kinematic excess, claimed 4.4–5.7σ vs clustering/mask reassessments at 3.3–3.6σ — a dissolution watchdog on the FLRW foundation). A preregistered QPE recurrence search (eRO-QPE2 + GSN 069, the unexplained post-2019 SMBH eruption class) returns a quantified-sensitivity null: the 2/3 kernel tooth lies >17× the observed ratio spread away from the QPE clock. And the recovery-kernel quantum circuit ran on REAL hardware for the first time (ibm_marrakesh, 16384 shots): one-step survivals 2/3 and 1/3 reproduced at −4.2%/+8.9%, while the blind 12-block bend decode is floor/T1-biased (1.96 vs 2.71) — hardware tier data-limited.

A zero-dial flat-budget closure — {Ω_b = φ₀(1−1/4π) [frozen], Ω_c = (2/7)(1−1/4π) [a numerology-flagged post-hoc candidate, an ALTERNATIVE reading to the axion-spine DM branch], ρ_Λ/M̄⁴ = (3/4π²)e^(−2α⁻¹) [frozen], Σm_ν = 0.0588 eV} plus flatness determines the whole flat ΛCDM budget with no fitted cosmological parameter: H0 = 67.15 km/s/Mpc (−0.39σ Planck 2018, −5.67σ SH0ES 2022 — the parameter-free Planck side of the Hubble tension), Ω_m = 0.3133 (−0.27σ), t0 = 13.86 Gyr; honest stresses recorded (ω_b −2.02σ CMB-side while BBN-only sits −0.73σ; ω_c −1.19σ; DESI-DR2 H0 −2.16σ — one correlated low-h direction). It is a composite of already-counted φ₀-seed and α-engine legs plus one flagged candidate, typed pattern-candidate, never an independent hit.

The 2026 anomaly round hardens the watchdogs: a sterile-neutrino dissolution watchdog is now a prediction of record (N_fam = 3 exactly — MicroBooNE two-beam, Nature 648 (2025), excludes the single-sterile LSND/MiniBooNE reading at 95% CL; SBND+ICARUS decide this decade), the DESI w(z) preference is dissolving through 2026 exactly as w = −1 requires (DES-Dovekie recalibration cuts 4.2→3.2σ, a calibration error in DES-SN5YR; the Bayesian reanalysis eliminates the DESI+CMB-only preference, ln B = −0.57), the g−2 bridge carries a dated kill condition (a lattice-side HVP consensus kills it — +3.9σ today, WP2025 lattice pull now 3.92σ — while a KLOE-side dispersive consensus restores <1σ), an HFQPO geometric-ladder discriminator is designed (the 3:2 cluster is real but cheap: anchored selection null 18.5%, J1859+226 breaks universality at 9.2σ; the decisive ×1.5 tooth search was never published and is preregistered for the RXTE archive), the first laboratory comb bound now exists on the Planckian σ₁(ω/T) master curve (LSCO x=0.24: ω = 2.583 not special, underpowered at the predicted 1.7% ripple), and the DSI false-positive control bed stands at 0/5 including the Efimov ladder (8.9σ separated from the kernel) and glass/MCT controls (no universal 2.7095 bend).

The previous hardening round added five workstreams: an exact-fixed-point watchdog (the DESI DR2 (w, Σm_ν) pincer puts the joint TFPT point at ~3.7σ — model-dependent, below the 5σ kill — while the α–Λ drift lock excludes 'TFPT + real w(z) dynamics' by ~10^5.7 via optical-clock bounds), a DSI false-positive control (the frozen ω = 2.583 comb detector fired on 0/4 real non-TFPT Omori/DSI cascades — the existing comb nulls are specific, not promiscuous), a repeater burst-time cascade search (first channel past the 2.8-period range wall, 9 sessions up to 4.8 periods: null at detectable amplitude, data-limited at the predicted 1.7%), the PG.08 PuMA/IAR daily ToA-residual comb (first real residual recovery past the gate, ω not special, data-limited by the amplitude wall), calibrated absolute GW echo limits (injection-campaign ε₉₀ ≈ 0.63–1.9 × A220 in the best events, ~7× above the (2/3)⁶ kernel ceiling; the 16 kHz re-run confirms the point-test null is not a lag-resolution artefact), and a quantum-circuit realisation of the recovery kernel (bend 2.7095 exact on the simulator; current FakeBrisbane run stays consistent via per-mode decode from 256 shots, while the strict 3/3-seed combined free-ratio fit is weaker and has no `min_shots_identifiable`).

The GW ringdown-echo channel now covers the 10 loudest ringdowns on real GWOSC strain (incl. GW250114, network SNR 78.6, O4b), all templates redshift-corrected to detector-frame frequencies, with a stacked search over 23 detector streams (p = 0.26, kernel-consistent 0/23), a 12-variant signature battery (energy vs amplitude semantics, μ₄ per-bounce phases as the boundary-birefringence analogue, Planckian lag window, off-source event-gated PSD), a systematic-hardened point test at the theory-fixed C = 3/8 delay (spin scan, skip-first-echo, joint QNM+train fit repairing the short-lag self-absorption), and a first area-quantum Bekenstein–Mukhanov spectral-comb search (ΔA = 4 ln3 line spacing): all null — the (2/3)⁶ upper bound is consistent and robust against every tested signature reading and systematic.

The latest structural round (v231/v233) ties the leptonic CP phase δ_PMNS = 4π/3 = 240° to δ_CKM via the Z2 sheet (δ_PMNS = δ_CKM,lead + π); confronted with the NuFIT 6.0 normal-ordering best fit (212°⁺²⁶₋₄₁) it sits at +1.08σ (consistent, weak power until DUNE/Hyper-K) and shares one independence group with δ_CKM. This is now a pre-registered kill test (v320, sharpened v322): both CP phases are powers of one hexagonal unit ρ = ζ₆, the selected node is the deck order |μ₄| (δ_PMNS = 4·δ_CKM,lead), the prediction band is 240° ± ~9° (the sub-leading bounded by the quark 3λ²), and the nearest wrong hexagonal node sits 60° away — so a δ_PMNS robustly away from 240° at DUNE/Hyper-K/JUNO would falsify the Galois-CP lock. The cyclotomic capstone (v313–v318) makes the whole SM structural sector ℚ(ζ₃₀) with Galois group μ₄ × ℤ₂ (degree 8 = rank E₈), leaving zero dimensionless free parameters — {a, π, v_geo} is the complete input.