Ten predictions

The perturbation-robust spectral structure of a system is its causally real, generalising part. This is the master criterion the rest of my work runs on.

The naturals, integers, rationals, reals, and the primality indicator are all generated by exp(x) - log(y). Formalised in Lean 4, with no gaps left open.

A single visibility-of-selector inequality predicts when any system – a model, a journal, a market – will manufacture evidence for what it was already selecting toward.

RH is equivalent to a positivity condition on a single observation operator over an admissible measurement space. The constraint is the statement; the construction is the work.

Distress is set by the phase lag of predictive correction in the auditory hierarchy, and turns on a single coefficient. A phase-targeted intervention will work where loudness-targeted ones fail.

A unified pitch-time representation collapses the two and reaches transformations the separated models cannot. Built; the result is not out yet.

Reportable content corresponds to the slow, perturbation-robust modes of neural dynamics. You should be able to predict what becomes reportable from spectral robustness alone.

Stochastic resonance sets a precise noise optimum for sampling in energy-based and autoregressive models. There is an exact answer to the open question, not just a regime.

Musical and spatial recall is equivariant under transposition: shift the input, the representation shifts with it. A direct neural test will confirm it.

Autoregulated, wave-shaped progression outperforms linear loading on tendon adaptation, because it respects the slowest tissue's recovery rate rather than the fastest's appetite.