Independent Research
Research, published in full with its sources and its data. Essays, written at the length their subjects deserve. Thought, set down while it is still taking shape.
Latest Work
- The Announcement Carries the Cue: Markup, Boundaries, and the Notation of Pre-Training Corpora
Notation as an unrecorded training variable, measured on three fronts — what corpora carry, what readers use, what writers impose — and the pure frame those measurements imply. - The Missing Format
Structural markup is a perfectly reliable boundary cue: a census of ten corpora, a converter study, and a probe on what its reliability removed from the curriculum. - Glaring Fireball
A proposal to measure what pre-training pipelines discard: three studies on structure imposed on prose that was never composed with it. - Hierarchical Symmetry Selects Log-Poisson Cascades: Classification, Uniqueness, and Stability
A formal result classifying log-Poisson cascades under hierarchical symmetry, with uniqueness and stability results. - The Last Fingerprint: How Markdown Training Shapes LLM Prose
The em dash as markdown leaking into prose, tested by suppression across twelve models from five providers.