About
The Interfaces Institute is a building for interaction evidence. It holds 26 collections over 7 open floors, and every row in them came from a document somebody else published. Nothing here is a survey this building ran, an opinion it holds, or a number it estimated.
It is a reading room, not a blog. The unit of work is a collection: a body of evidence that already exists in public — a standard, a registry, a repository — reproduced whole, with the source named on every row and the join declared wherever two sources are put side by side. The Institute supplies the framing and the arrangement. It does not supply facts.
Every room states what it cannot tell you. Not in a footnote — in a box you cannot scroll past. Where a number is a ceiling rather than a count, it says so. Where a column was collected and then removed because it could not be supported, it says that too, and names the column.
The rules
Measure, never ask. Where this building reports how something looks — a bank's brand colour, a design system's body type — the value was taken off the live page by a real browser at a stated viewport on a stated date. No visual claim here was produced by asking a model what it remembered.
An absence is a result. A row where the publisher documents nothing is kept, with its fields empty. Dropping those rows as "incomplete" would invert every finding this building has: what is not written down is usually the thing worth knowing.
Where a fact cannot be verified, no field is offered for it. A schema slot is an invitation to fill it. Several collections here are deliberately thinner than they could be because the fields that would have made them fat were ones nobody could check.
An aggregator is not a source. A URL must point at the page that states that row's fact. A catalogue entry that names a standard, and a summary that repeats a claim, are both refused — a whole 600-record return was rejected on this rule alone and the rejection is published in the Basement rather than quietly deleted.
Corrections are kept. When something published here turns out to be wrong, the fix and the error both stay on the record. The Basement exists for this.
What the building actually holds
Counted from the pages that shipped, by
pipeline/derive_room_traits.py, rather than written down here and left to rot.
26
Each one reproduced from a named published source, with its own limits note.
2,438
Individual records across all rooms on the Collections floor.
14 of 22
Colour measured off the live page, marks fetched from the sites themselves, specimens drawn from each row's own figures. The other 8 are text and tables, and say so on the door.
Who
Mike Oustamanolakis, PhD — Director. The Institute is one of a federation of standards-facing sites, and the same person is behind each of them; the ORCID is what binds those identities together, because it is issued once and never reassigned, while the name is spelled differently in different places.
Using it
Everything here is CC BY 4.0, except where a collection's upstream source imposes its own terms — the W3C documents and the Unicode data each carry theirs, and each collection page names which applies to it. Citation formats are on the same page.
There are no cookies, no accounts and no analytics that follow you; the privacy note is three sentences long because there is nothing else to say. The accessibility page lists what has been audited and, more usefully, what never has.
Corrections are welcome and are the most valuable thing anyone sends. If a row here is wrong, the source it names will show it, which is why the source is always named.