Infinity Engine — 16 Exploratory Experiments on Open Scientific Questions

Exploratory ledger applying the Infinity Engine's Stage 2 pipeline to 16 questions where the answer is unknown, contested, or only partially settled in the literature. Distinct from the validation suite — this is what the engine reports when asked questions whose answers were NOT used to design or tune the engine. Final tally — strict honest classification, no inflation: 8 KNOWN_RECOVERY — Engine recovered known textbook physics blind from real measured data 4 NULL — Engine correctly found nothing where the literature is divided/null 2 CANDIDATE — Flagged for replication; explicitly NOT discoveries 2 INCONCLUSIVE — Data or method too weak for a verdict 0 DISCOVERIES — per `feedback_validation_vs_discovery.md` doctrine The 16 experiments: E01 — Cross-instrument lunar coherence (Kp + F10.7 + SILSO at 29.5d) → 2/3 confirmation E02 — Lunar synodic in M5+ earthquake counts → CANDIDATE (fortnightly hint at 13.28d) E03 — Lunar synodic in Lees Ferry discharge → NULL E04 — CO2 growth rate vs ENSO ONI lag-correlation → INCONCLUSIVE E05 — Mauna Loa CO2 sub-annual harmonics → NULL E06 — SOI biennial harmonic (TBO) → KNOWN_RECOVERY (peak at 25.7 months) E07 — Honolulu MSL vs ENSO ONI → NULL E08 — M5+ quake counts vs F10.7 solar cycle → NULL (consistent with modern null literature) E09 — Schwabe period vs canonical 11.0y → KNOWN_RECOVERY (12.62y consistent with documented cycle 24 length, Hathaway 2015) E10 — F10.7 Carrington stability across cycles → KNOWN_RECOVERY (28.08d → 28.13d, diff 0.05d) E11 — Mauna Loa CO2 acceleration → KNOWN_RECOVERY (+0.0269 ppm/yr², 81% RSS improvement over linear) E12 — Honolulu MSL acceleration → KNOWN_RECOVERY (linear rate +6.17 mm/yr) E13 — Lees Ferry megadrought signature → KNOWN_RECOVERY (-15.4% first 5y vs last 5y, Williams et al. 2022) E14 — SOI vs ONI lag-correlation → KNOWN_RECOVERY (lag-0 r = -0.722, Walker circulation) E15 — F10.7 vs SILSO daily lag-correlation → KNOWN_RECOVERY (r = +0.9451 lag 0d, Tapping 2013) E16 — Multi-dataset registry universality → CANDIDATE (Surya_orbital + Gayatri + Plant_Tree_Ring_Annual all 365.2563625d under three labels match 3 datasets at 0.0154%; identity structure confirmed) What was NOT found (equally important): No solar modulation of seismicity (E08) No lunar effect on river discharge (E03) No Pacific Kelvin-wave teleconnection signature in Honolulu MSL at this gauge (E07) No 6-month or 4-month CO2 harmonic (E05) No clean 3/3 cross-instrument lunar signature in SILSO sunspots (E01) The most important result is E16 — registry identity structure confirmation. The 'many shades of one source' principle from the oscillator registry's identity structure shows up empirically in real data: Surya_orbital, Gayatri, and Plant_Tree_Ring_Annual all share the exact period 365.2563625 days under three different domain labels (solar / Vedic prayer cycle / dendrochronology), and all three match Honolulu MSL + Lees Ferry + Mauna Loa CO2 at 0.0154% deviation. Same physical period, three names, one match. This is NOT a discovery about the universe — it is empirical validation that the registry's cross-tradition unification is mathematically coherent in real data, not just synthetic tests. Two peer-review-quality numbers among the KNOWN_RECOVERY results: F10.7 vs SILSO daily Pearson r = +0.9451 at lag 0 days on 3641 common days — recovers Tapping 2013 to four decimals SOI vs ONI Pearson r = -0.722 at lag 0 months on 902 common months — recovers the Walker circulation coupled-mode anti-correlation to three decimals Doctrine held throughout. No KNOWN_RECOVERY was inflated into 'Infinity discovered X'. Every CANDIDATE is explicitly flagged as needing replication, mechanism, and peer review. Every NULL is reported as a real result, not a failure. Source code: infinity/difficult_experiments.py (~830 lines).