BEBEL — Two-Lens Interface Review Checklist
Public research artifact · August 2026

Use automated rules and structured human review as separate lenses. A clean
result from one does not substitute for the other.

1. Prepare the surface
   - Use representative content and states.
   - Include default, interactive, empty, error, loading, and corrected states.
   - Disable transitions before taking contrast or geometry measurements.

2. Run automated checks
   - Validate landmarks, labels, names, roles, focus order, and contrast.
   - Exercise both default and interactive states.
   - Test narrow reflow and keyboard operation.
   - Record the exact tool, version, scope, and failures.

3. Review the semantics
   - What does the interface ask a person to believe or decide?
   - Are evidence, uncertainty, reversibility, and control visible where needed?
   - Does meaning depend only on color, position, typography, or motion?
   - Can an accommodation remove a signal without producing a technical error?

4. Correct the surface
   - Add explicit text when a visual distinction carries consequential meaning.
   - Use a second non-font or non-color channel where the first may disappear.
   - Preserve a record of the assumption that failed and the correction made.

5. Re-run both lenses
   - Confirm the correction under the automated rules.
   - Confirm that the intended meaning survives the changed condition.
   - Check that the repair did not create a new failure elsewhere.

6. State the limits
   - Automated checks cover only the rules they exercise.
   - Structured review is not disabled-user research.
   - Neither lens replaces assistive-technology testing or establishes
     certification, conformance, usability, or universal accessibility.

This checklist is a public explanation of a BEBEL review method. It is not a
certification or legal claim.
https://bebel.io/work/interface-audits/