I'm Matt. I live in Seattle with my partner Rachel and two cats who run the place. Professionally I build machine learning systems. Unprofessionally I fall into rabbit holes and build things at the bottom.
The papers are real research. The blog posts are long and probably have too many footnotes. The projects are what happens when "I wonder if..." turns into three weeks of lost sleep.
Latest Post
All →Latest Paper
All →Calibrated Falsification of Prior Dorabella Decryptions: A Negative Result on Substitution-Consistency and Language Identification
A calibration-first falsification pipeline applied to Edward Elgar's 1897 Dorabella Cipher. The Packwood (2020) reading fails strict substitution-consistency (one symbol encoding eight different letters), and matched-budget simulated annealing cannot demonstrate Italian as the plaintext language (z = −1.23). Proposes substitution-consistency as the minimum standard for length-matched strict-substitution claims.