Computational substrates
When the material does part of the thinking.
Voltage, movement, yarn, chemistry or collective behavior is not incidental hardware. Its physics becomes an operation, a memory, a constraint or a language.
Independent research laboratory
Lathilda explores other ways computers can be built, programmed and experienced.
Foundation established. First flagship not yet selected.
Most software assumes the same arrangement: silicon executes instructions, state moves forward, and the machine stays politely invisible.
We treat that arrangement as one possibility among many. What happens when voltage, yarn, mechanical motion or a collective of small robots becomes part of the computational idea? What changes when programs reduce as graphs, run backwards, handle effects explicitly or identify themselves by content?
Lathilda builds concrete answers: languages, runtimes, machines and public instruments. The result should not only be unusual. It should reveal a capability, constraint or experience that conventional computation hides.
02 / Two axes
The axes intersect, but neither is a list of favored tools. A project earns its place by asking a real question about the medium or the way execution works.
Computational substrates
Voltage, movement, yarn, chemistry or collective behavior is not incidental hardware. Its physics becomes an operation, a memory, a constraint or a language.
Alternative execution models
Evaluation can be lazy, parallel, reversible, effect-aware or content-addressed. Different rules change what can be expressed, optimized, inspected and composed.
03 / A recurring practice
A public computational instrument is not a third research axis. It is how a real substrate or execution model becomes something people can touch, hear, steer or understand.
The instrument may be a browser experience, a game, a physical machine or an interactive proof. Its public appeal never substitutes for technical substance.
Touch a node. Watch the graph find another shape.
04 / Candidate register
No flagship has been selected. These cards are candidates awaiting a tournament on mechanism, contribution, feasibility and public experience.
Step through computation in both directions and expose which operations genuinely lose information.
Execution modelPatch equations into a live, visible circuit and feel numerical behavior as a continuous physical process.
Substrate + instrumentMake interaction-net reduction inspectable enough to compare what parallelism gives and what it complicates.
Execution modelBuild a textile whose topology, touch and deformation participate in the computation rather than merely housing it.
SubstrateSelection tournament
Familiarity with Haskell, WebGPU or an existing project gives no automatic advantage.
05 / Preserved systems
The new identity does not rewrite the past. These systems keep their own histories, questions and review states.
A reusable GPU compute-and-render system. It remains powerful infrastructure, not the laboratory thesis and not a required dependency.
A deterministic cascade study with its own project-local evidence and review state. Preserved, not retrofitted as proof of the new charter.
06 / Open laboratory
We are building the foundation now. Follow the work, inspect the systems, or return when the first candidate survives selection.
Explore Lathilda on GitHub