Accessibility

A first use: accessibility

If you caption, translate, simplify, or describe other people's work, most of what you make circulates with your name already stripped. A sealed adaptation names its adapter, binds to its source by hash, and carries its fidelity limits, so the credit and the caveats survive every copy. The format is the other half: one plain-text file a whole class receives and returns alike, native to screen readers and to any tool that parses text or JSON. Accessibility is the first proving ground for KNOBE, with its own ethical stakes.

This page describes KNOBE's accessibility design. For accessibility-related use in courses, see the Education Profile.

The KNOBE instructional approach is being designed for large university courses: high-enrollment settings of 100 to 500 students that operate under strict privacy and accessibility requirements, conditions that make the question of what travels with a knowledge object concrete rather than theoretical. Instructional pilots are being planned with collaborators at UC Davis; none are yet active.

KNOBE records accessibility adaptation lineage: who adapted a work, from what source, and with what fidelity limits. It does not by itself certify that an adaptation meets a standard such as Section 508 or WCAG; that determination remains with the responsible institution. KNOBE is an independent open protocol, not an official UC Davis standard or product.

When materials are captioned, translated, simplified, or remediated, the adaptation is knowledge work. It carries labor, judgment, and responsibility. But adaptations usually circulate severed from their sources: the adapted object travels without the original, without the adapter's contribution recorded, without a clear account of what changed or what fidelity limits remain. The labor disappears, and the link to the source disappears with it.

The knowledge object itself has to carry more of what used to live outside it.

A KNOBE keeps the adaptation bound to its source by hash and credits the adapter in the record. The plain-language adaptation in the Downloads section is a worked example: a simplification of the white paper that records what it was adapted from, who adapted it, and what it does and does not preserve, making legible the gap between parties who do not share the same access to the source, with the labor of that crossing kept visible.

The adaptation lineage in three steps: a sealed source; an adaptation sealed with fidelity limits that chains to that source by hash; credit and lineage that travel with the adaptation wherever it goes.

  1. Sourcea sealed work, named by hash
  2. Adaptationsealed with honest fidelity limits
  3. Wherever it goescredit and lineage travel with it

The adapter's labor stays visible, and the adaptation cannot be mistaken for the full source.

This is now a tool, not just a pattern: the Studio Adapt tab seals a plain-language version, translation, transcript, or description with this lineage built in, guided by two plain questions: what does this adaptation preserve, and what does it omit. A one-page guide for disability-services staff and TAs is in the instructor toolkit.

Try it now
Seal an adaptation

Load a sealed source, answer what the adaptation preserves and what it omits, and download one file that carries the credit and the chain. Runs in the browser; nothing is uploaded.

Open the Adapt tab →

Same KNOBE, many ways in. A KNOBE lets everyone receive and return the same knowledge object without requiring everyone to use the same interface. The whole class works in one format: the body is plain text a screen reader or braille display consumes directly, and the payload is ordinary JSON that any tool can parse, so a student using assistive technology engages the same file as everyone else, not a separate accessible version delivered late. The file can be read as plain text, inspected by a verifier, transformed into audio, adapted into a visual layout, summarized by an AI assistant, or used inside a learning platform, and through all of it the object remains the same, with its context, lineage, and verification record attached. A parseable format does not make the content accessible by itself: alt text, captions, and plain-language versions remain the author's and the adapter's work, which is exactly the work the lineage records and credits.