Pipeline

First protocol to reconstruct the inverse-temperature four-vector β^μ as a unified observable from passive electromagnetic fluctuation correlations — enabling the first direct test of Lorentz covariance of the thermal state.

Foundations of Physics (Springer) · submitted August 2026

Irreversibility from the intersection of quantum mechanics and classical chaos: chaos amplifies irreducible ℏ-scale uncertainty until time-reversed trajectories fall below quantum resolution and become physically inaccessible, despite remaining mathematically valid.

Scientific Reports · minor revision

Reconceives knowledge as well-calibrated credence, dissolving Gettier cases and Hume's problem of induction. The epistemic core of the Bayesian-epistemology program.

Analysis (Oxford University Press) · double-anonymous · submitted July 2026

A theory is scientific iff it can be meaningfully revised by evidence — an updateability criterion that supersedes falsifiability.

Erkenntnis (Springer) · major revision submitted September 2026

Sub-linear O(d^0.67) scaling through 1024+ dimensions, with a 100–1000× speedup over PolyChord/MultiNest. Reframed around Bayesian evidence computation for a statistics readership.

Statistics and Computing (Springer) · submitted July 2026

A GPU-native memetic population optimizer (Convergence–Halt–Invert–Stick–And–Oscillate) that runs a whole sample batch at once, freezing confirmed modes while the rest escape local traps through a convergence–anticonvergence oscillation. 100% mode recovery across all 42 SFU benchmark functions through d=64 — where CPU baselines collapse at d≥8 — at up to 34–39× speedup, and reliable under likelihood noise up to σ=1.0. Open-source via chisao on PyPI.

Swarm and Evolutionary Computation (Elsevier) · submitted June 2026

Coordinate-free derivation of the Hawking–Unruh spectrum from the inaffinity eigenvalue equation alone — no metric, no Kruskal construction. Being recast as a letter, foregrounding its novelty against Israel (1976), now that both companion entropy papers are published.

The Bayesian polarization literature asked the wrong question. Common-evidence convergence theorems fail because partisan media ecosystems violate their premise at scale — responsibility shifts from individuals to institutions.

The implicit Turing test: a system is intelligent iff it can detect that its own interpretive frameworks have failed and revise them under world-pressure. Gödel, Turing's halting problem, and the Löbian obstacle become one diagonal argument — external grounding is the only resolution that doesn't relocate the regress. Journal version being re-aimed; conference variant accepted at MoC7 (below).

Dissolves the Boltzmann brain hypothesis via information propagability and Liouville's theorem. Two horns: ex-nihilo formation is forbidden by symplectic structure; Poincaré-recurrence variants cannot ground the skeptical hypothesis the literature requires.

Goodman's new riddle of induction — why project "green" not "grue" — resolved physically rather than linguistically: projectible predicates are those preserved under the symmetries of physical law. Split off from the demarcation paper, where a referee called the argument "very much original and worthy of consideration."

Talk bundling three results — Loschmidt resolution (Krylov-space proof), 1/4 from symplectic reduction, horizon thermality from inaffinity eigenvalues — into a single geometric story about what any observer can ever know.

SigmaPhi 2026 · Kolymbari, Crete · contributed talk, 9 July 2026

Conference variant of the Reality Wins program — focused on the architectural requirements of grounded artificial cognition.

Models of Consciousness 7 · University of Copenhagen · October 2026 · accepted; not attending