Writings
What we are building and thinking about — features, releases, and longer-form research — written in markdown. The first things going in are the research papers on the ideas behind Fundamental. These ideas also connect to an external framework, Foundations of Computational Interaction, which names Fundamental as a substrate — see how it relates.
Research
Longer-form papers on the ideas behind Fundamental.
- research
1. Fundamental: A Field Translation Runtime for Relational DOM Interfaces
The paradigm: UI as a shared, inspectable field of meaning.
- research
2. Reading Field: Attention, Memory, and Relationship Awareness in Long-Form Web Documents
A field model for reading attention and document memory that preserves semantic HTML.
- research
3. Evidence Fields: Visualizing Support, Contradiction, Confidence, and Provenance in AI Interfaces
A field-based interaction model for evidence, confidence, contradiction, and provenance.
- research
4. Motion Is Not Meaning: Reduced-Motion Equivalence in Field-Based Interface Systems
A conformance model for translating motion-heavy behavior into static semantic equivalents.
- research
5. A Host-Driven Field Runtime for Portable Interface Behavior
A runtime architecture that targets DOM, Canvas, SVG, WebGL, native, or headless.
- research
6. Patterns as Portable Field Programs for Interface Behavior
A structured authoring model that composes behavior without corrupting runtime vocabulary.
- research
7. Data as Field Participants: Binding Records, Relationships, and Metrics into Interface Fields
A data-binding model for relational interface fields.
- research
8. Explainable Interface Behavior Through Field Diagnostics
A diagnostic framework for explainable interaction behavior.
- research
9. Field Interfaces as Semantic Substrates: A General Model for Relational UI State
Paper 09 generalizes the eight-paper Fundamental family into one portable, falsifiable model of field interfaces as semantic substrates, with a four-term core model, design implications, and a multi-study evaluation plan.
- research
10. Conserved Attention in Interface Systems
Paper 10 of the Fundamental expansion family generalizes the engine's budgeted-attention behavior into a portable, falsifiable interface principle: attention is a conserved resource that systems should expose rather than multiply.
- research
11. Semantic Degradation Paths for Accessible Interactive Systems
Generalizes Fundamental's reduced-motion equivalence into a portable channel-equivalence principle — that the meaning carried by any expressive channel must remain available through an accessible alternate representation when that channel is removed — and sketches a conformance and evaluation plan.
- research
12. Evidence Graphs as User Interfaces for AI Trust Calibration
Paper 12 of the Fundamental expansion family generalizes the Evidence Fields claim into a codebase-independent principle — evidence belongs in a claim-level graph of support, contradiction, uncertainty, and provenance — and offers a falsifiable model plus a four-study evaluation plan.
- research
13. Runtime-Inspectable Interfaces: Diagnostics as a First-Class UX Primitive
Paper 13 lifts the diagnostics-and-inspectability claim out of the original Fundamental family and reframes it as a falsifiable, implementation-agnostic interface principle with an evaluation and conformance plan.
- research
14. Patterns as Behavioral Design Tokens
Generalizes the Fundamental pattern model into a portable interface principle — behavioral design tokens: named, versioned, validated, inspectable behavior contracts with diagnostics and reduced-mode equivalents — framed as a falsifiable research model plus a four-study evaluation plan rather than a measured result.
- research
15. Data Binding Beyond Views: From Reactive Rendering to Relational Participation
Argues that data binding should preserve the relationships, uncertainty, priority, lifecycle, and provenance of records as first-class, inspectable interface state rather than only synchronizing views with records.
- research
16. The Semantic Canvas Problem
Generalizes the Fundamental papers' DOM-canonical thesis into a falsifiable, implementation-agnostic principle — expressive visual surfaces must never become the sole carrier of interface meaning — with a conformance and study plan.
- research
17. Interaction Truth Modes: A Taxonomy for Honest Interface Behavior
Expansion Paper 17 generalizes Fundamental's truth-mode idea into a portable principle that interface behaviors need epistemic labels, defining a falsifiable participant/signal/binding/equivalence model and a four-study evaluation plan.
- research
18. Field-Based Wayfinding for Long-Form Documents
Generalizes the Reading Field claim into a portable, falsifiable wayfinding model for long-form documents and sketches an evaluation plan that does not assume Fundamental is the only implementation.
- research
20. Non-Perturbing Diagnostics for Interactive Systems
Paper 20 isolates one reusable claim from the first Fundamental family — that a diagnostic interface must be read-only, separable, and auditable — and states it as a falsifiable, portable systems principle with a conformance and study plan.
- research
21. Host-Driven UI Runtimes: Portability Through Environmental Injection
Generalizes Fundamental's FieldHost boundary into a portable host-driven runtime pattern, arguing that isolating environment contact behind one explicit host interface is what makes a UI behavior core testable and portable across browser, headless, worker, native, and alternative render targets.
- research
22. Accessibility Conformance for Expressive Interfaces
Generalizes Fundamental's reduced-motion work into a renderer-agnostic conformance contract — semantic source, redundant carrier, reduced-mode equivalent, preview, lint, and a non-inferiority study protocol — for making accessibility claims about expressive interfaces falsifiable.
- research
23. Relationship Registries as an Interface Primitive
Paper 23 generalizes Fundamental's relationship-registry idea into a portable interface principle — relationships as first-class typed, queryable runtime state — with a four-term model (participant, signal, binding, equivalence), four falsification conditions, and a four-study evaluation plan.
- research
24. Interface Memory: Worn Paths, Dwell, and Returning Context
An expansion position paper arguing that interfaces should treat accumulated meaningful attention (dwell, revisitation, reading trails, review context) as a first-class, inspectable memory substrate rather than an ornament, with a falsifiable model and a four-study evaluation plan.
- research
25. Semantic Motion: When Animation Carries Meaning
Paper 25 of the Fundamental expansion family isolates the claim that semantic motion (carrying causality, continuity, hierarchy, origin, destination, relation, or state change) must be authored, tested, and degraded differently from decorative motion, and frames it as a falsifiable, reproducible interface-design model.
- research
26. Field-Based Evaluation Methods for Interactive Systems
Paper 26 generalizes the Fundamental family's measurement approach into a portable, falsifiable framework for evaluating field-based interfaces beyond task completion and preference.
- research
27. A Taxonomy of Interface Participants
Generalizes the Fundamental paper family into a falsifiable taxonomy of interface participants — defining participant, signal, binding, and equivalence — and a conformance/evaluation plan for treating relational meaning as a first-class substrate rather than an ornament.
- research
28. From Layout to Topology: Interfaces as Graphs of Meaning
Paper 28 abstracts the Fundamental paper family's core insight into a renderer-agnostic interface principle that meaning relationships should be first-class, inspectable interface state rather than visual side effects, and proposes a four-term model (participant, signal, binding, equivalence) with an evaluation plan.
- research
29. AI Interfaces Should Show Support, Not Just Sources
Expansion Paper 29 generalizes the Evidence Fields claim into a portable interface principle: showing that a source supports a specific claim, with a falsifiable participant/signal/binding/equivalence model and a four-study evaluation plan.
- research
30. Platform-Native Expressive Systems Without Replacing HTML
Paper 30 of the Fundamental expansion family generalizes the first eight papers' core claim into a renderer-agnostic, falsifiable model for treating relational meaning as a first-class interface substrate over semantic HTML.
- research
References
The shared bibliography for the paper family.
Releases
What shipped, and what it means.
Features
New capabilities, explained.
- feature
One Engine, Four Runtimes — The Zero-DOM Core
How a renderer-agnostic core lets the same field run on the page, in WebGL, with no document at all, and natively on Apple platforms — and what that discipline costs.
- feature
render: 'none' — The Invisible Field Is the Baseline
Why the engine default is render: 'none', and why a signals-first field is a different thing than a particle background you turn down.
- feature
The Empty Canvas That Costs Every Frame
Why a transparent canvas tanked our framerate, and why the bottleneck was the compositor, not the particles.
- feature
The Field Was Never Particle-Bound — It Was Fill-Rate-Bound
Why a slow field is almost never the particles, and the DPR2 / mix-blend trap that actually costs the frame.
- feature
The Interface is a Field, Not a Screen
The manifesto: why an interface is a shared field of meaning, not a pile of components, and what it costs to make that field real.
- feature
The Silent Contract Gap — Bugs That Don't Throw
A whole bug class on CSS custom properties: faithful writes, no enforced reader, zero exceptions — and the discipline that closes the loop.
- feature
Your Design System Already Has Four Forces
The four Natural Fields aren't new ideas you adopt; they're the relationships your interface already has but never names.
- feature
The invisible fields
Twelve real-data examples where the field never shows a particle — what they are, why the data is honest, and what they taught me.
Notes
Shorter thoughts and updates.
- note
Meaning Should Have Behavior
The synthesis piece: why display-only meaning inflates until it lies, what a conservation law does about it, and where accountable behavior — receipts, not screenshots — goes next.
- note
Fundamental, explained to a developer
The developer-tier companion to 'explained simply' — the mental model (concepts/tokens/metrics/diagnostics/patterns/projections), field() vs apply(), designed-vs-natural, association ≠ coupling, signals-first CSS feedback, and the minimal working mindset.
- note
The Bats Had Momentum: What Fundamental's Field Is Missing
Fundamental flocks like a boid but doesn't move like one — bodies push matter without recoiling, so momentum leaks. A proposal for first-class mass and Newtonian reaction, and the results we'd expect from it.
- note
Fundamental, explained simply
What it is, what it can do, how it borrows from real physics, and why it's different — in plain language, from magnets to the deep end.
- note
Welcome to Fundamental writings
What this section is, and a quick tour of how it renders diagrams, math, and code.