Critique
Analyze a site, screenshot, or HTML. Do not recreate it. Observation is not interpretation.
Constitution · readable surface
HTML Design Lab is not a portfolio, a landing page, or a template kit. It is a library of visual thinking patterns — ways information should be seen, compared, and understood. HTML is the thinking medium. The question every specimen answers: does this form make the structure of the information clearer?
Knowledge does not start as UI. A reference is observed, then deconstructed, then named as a pattern, then tested as live HTML, then — only if it survives — adapted into a project. Interfaces are never copied. Forms are derived from information structure.
A paper, a book, a site. Specimen, not source file. Observe before you interpret.
A named principle that can travel. Maturity is explicit. Copies are not patterns.
Isolated HTML. One question. Own CSS. The lab shell never bleeds in.
Composition of surviving patterns. Built last. Recorded under Own HTML while it iterates.
The same chain is a graph in Knowledge. Markdown notes remain in DESIGN_KNOWLEDGE.md for git, not for reading in a browser.
The repository is a small operating system for visual thought. Tokens stay global. The lab shell is quiet chrome. Experiments never import lab CSS. Languages and modes are lenses, not pages.
tokens/tokens/languages/.lab/experiments/patterns/references/languages/modes/projects/dna/briefs/Form follows information structure. If the underlying structure is a relationship, the interface should not flatten it into a list. If it is a sequence, it should not be shuffled into cards.
This is the contract every experiment is judged against. Structure on the left, form in the middle, live specimen on the right. Empty cells are honest: the lab has not tested them yet.
| Structure | Form | Specimen |
|---|---|---|
| Relationship | Graph | Interactive Knowledge Graph |
| Sequence | Timeline | Progressive Timeline |
| Process / causality | Flow, traced path | Animated Process Diagram |
| Comparison | Matrix, shared frame | Interactive Comparison |
| Mechanism | Simulation | Negative-Feedback Loop |
| Spatial relationship | Map, field | Spatial Information Map |
| Overview → detail | Zoomable canvas | Infinite Canvas |
| Emergence | Progressive construction | Scroll Narrative |
| Structure-in-image | In-situ annotation | Layered Image Annotation |
| Hierarchy in language | Syntactic entrance | Kinetic Typography |
| Hierarchy as tree | Tree, nested frames | Not yet — the spatial map is a field, not a tree |
| Uncertainty | Opacity, range, branching | Not yet a dedicated specimen |
A specimen is not done when it looks finished. It is valid when it addresses at least one of these questions, and produces a reusable insight rather than a polished page.
Languages are lenses, not themes. The lab shell currently speaks Notebook + Editorial: warm paper, ruled type, quiet chrome. Experiments are allowed to contradict the shell. That contradiction is the point of isolation.
Naming a pattern too early freezes a guess. Nothing is promoted by taste alone.
Analyze a site, screenshot, or HTML. Do not recreate it. Observation is not interpretation.
Take a pattern into another domain. Name what remains and what changes before implementing.
Build the next specimen. Smallest change. Preserve existing behavior. No unrelated rewrites.
Infer Design DNA. Do not restyle the source. Decompose website → principle, never the reverse.
The laboratory opened 15 August 2026. Everything here is a first cut, not a canon. Exploration before standardization.
HTML is the reading surface. Markdown is the note behind it. Start with a live page, then open the source if you need the text.