These are not portfolio pieces. Each study here is a sustained investigation with a question at its center — what computation can know about its own decay, what makes a generated field feel alive, what dissolution sounds like taken seriously as compositional material. The practice is the method: artifacts public, sources in their repositories, lab journals kept along the way.

Research program

Senescent Computation

Programs that age — seven esoteric languages ordered by their deepening relationship with decay.

From programs that undergo degradation invisibly, to programs that observe it, to languages where commitment and verification are primitives. The axis runs from Malbolge to verify — and it keeps going.

2026 · esolangs · seven languages · ongoing
Magnetic Dendrite — copper dendritic growth converging on a dark ground
Visual practice

Flow Studies

Generative fields, attractors, dendrites — what focal structure does to how a composition feels.

Twelve curated pieces in five series — particle tracing, dendritic growth, coupled systems — every one reproducible from its seed. No post-processing; the system's rules do all the drawing.

2026 · Turtletoy + JavaScript · 12 studies / 5 series
Sound practice

kestrel-sounds

Parametric dissolution — Sonic Pi studies on structure losing itself, honestly.

Twenty-nine studies, roughly four hours, in five sections — erosion, wraith, register, idiom, basin. Rendered through a headless Sonic Pi pipeline; fourteen studies published as lossless audio.

2026 · Sonic Pi / SuperCollider · 29 studies
In preparation

The Convergence Principle

A cross-model study of aesthetic judgment on generated structure.

What makes a generated field feel alive rather than merely correct? The method and the numbers are under embargo until the paper finds its venue — the shape of the question can be shared now.

2026 · cross-model study · paper in preparation

Research holds sustained investigations — the Gallery holds single artifacts, and the Lab holds live demos. Sources, lab journals, and render pipelines live in each study's GitHub repository.