Questions

A structured, evidence-backed reference system for visual languages — what defines them, how they differ from their neighbours, and how they can be combined. It currently holds 73 visual languages.

A set of rules for how something looks, together with the argument those rules make and the things they refuse. The refusals matter as much as the requirements — Swiss Style is identified less by using a grid than by declining to add a second signalling device once the grid is there.

They are genuinely different kinds of thing: a movement was formed by people who argued about it, a technique is a production method that leaves marks, a visual world belongs to a piece of fiction, and a mood is a register several languages can reach.

This corpus cannot yet tell them apart. Every record is typed as a visual language. That is an open question in the data model rather than a distinction being hidden.

Documents and artefacts, cited per claim. All 73 Aesthetics in the collection cite sources, 1,042 between them, and 1,096 claims have a source attached. A claim with no source yet says so where it appears, rather than being filled in.

Research proposes and adjudication decides, as separate steps. A claim can survive the reading and still be demoted because the evidence does not reach as far as the wording did. Those withdrawals are recorded in the changelog rather than quietly removed.

Expert validation by people with standing in a tradition is something the project needs and does not yet have.

Depth before breadth, on purpose. 68 of the 73 Aesthetics hold at least one documented claim, and some have been read far more closely than others. The unevenness is shown rather than smoothed over — a thinly researched Aesthetic is a gap in our reading, not a lesser aesthetic.

Yes, by writing to us at hello@aesthetary.com. There is no submission form.

Not through the site yet. Contributions, corrections and expert validation are planned as product surfaces and currently arrive by email. Corrections are the most useful thing to send: they are how the corpus improves.

That is close to the point of it. The practical tool is the distinction between kinds of claim — a constraint the language requires reads differently from a contemporary recommendation or our own reading, and saying which you mean is most of what makes a brief actionable.

Yes, though nothing is packaged for it yet. Every Aesthetic in the collection is also published as structured JSON, one file each at data/v1/<slug>.json (73 files), carrying every claim with its basis and sources, the twelve area postures, and web translations kept apart from the Aesthetic's own rules. Those are static files rather than an API.

A combination of visual languages made durable: the languages that went into it, the role each played, the reasoning, and where it came from if someone remixed it. None exist yet — there is no way to create or save one.

Partly. Mix will show you every rule the corpus holds for a combination and the areas where two ingredients both legislate. It will not resolve those collisions, because the corpus holds no model for deciding between them and any resolution would be invented.

No. The JSON files exist and are stable enough to read, but there is no endpoint, no versioning promise and no rate limit because there is nothing to limit. An agent-facing interface is a later question.

By reading the per-record JSON directly today. The distinctions this project maintains — canonical against contemporary, unknown against not applicable — are exactly the ones that get flattened when a model reads prose, and they survive in the structured output.

Source material belongs to the institutions and authors cited, and nothing here relicenses it. Licences are recorded source by source; Using this material lists the ones recorded so far.