Part 2: The ratchet

Recovered fragment // Speculative Accessibility // Week 2

The consultancy and charity model isn’t a solution to inaccessibility. It’s a symptom of it.

Rungwell represents the high-water mark of what a Type 0 civilisation can achieve within its own logic. For twenty-five years, the industry’s best answer to Tyrer’s Law has been remediation: sighted people build visual things, and then specialists are paid to translate them for everyone else. The accessible version is always downstream of the visual original. It is, by definition, a lossy copy — information stripped, context approximated, meaning partially reconstructed. Every screen reader workaround, every alt-text retrofit, every accessibility audit is evidence of a system working against its own grain.

The Tyrer Framework names what Rungwell cannot: this isn’t a pipeline problem. It’s a ratchet. Developers don’t forget accessibility through inattention, to be corrected by better training and workshops. They are subject to Tyrer’s Law — they think in visual metaphors, they prototype visually, they reach for dashboards and icon sets and layered interfaces because that’s the cognitive toolkit their discipline has built. Each iteration adds visual complexity that outpaces the accessibility layer. The gap doesn’t stay constant. It widens by design.

The Newtonian Web — fixed objects at fixed coordinates, described to a screen reader after the fact — is what this ratchet produces at civilizational scale. The screen reader is not an accessibility tool. It is a translation service for a visual medium that was never built to be translated.

What the Structural Break makes possible is something categorically different: interfaces generated at the moment of interaction, from shared source data, simultaneously and natively for every sensory modality. Not a visual interface with an accessible version bolted on. Two sibling interfaces — or many — none of which is the original and none of which is the adaptation. This is Type I behaviour. The blind user’s interface and the sighted user’s interface don’t share a parent-child relationship. They share a data structure.

The danger is that without this shift in first principles, AI becomes the most powerful engine of visual drift ever produced. Generative interfaces, procedural graphics, dynamic dashboards, AI-composed layouts — all of it accelerating faster than any remediation workflow can follow. The consultancy model breaks not because Rungwell fails at what it does, but because what it does becomes structurally insufficient at the scale and velocity of what’s coming.

Treating accessibility as a specialised subdiscipline — something you bring in expertise to fix — encodes the Type 0 assumption into the economics of the field. It makes the retrofit permanent. The Tyrer Doctrine demands a different premise: that no critical information may be locked behind a single sensory channel, not as an accessibility feature, but as an engineering invariant. The same class of requirement as uptime, or data integrity, or security. Not something you add. Something a system fails without.

The question the Structural Break forces is whether the field will use the generative moment to finally build from that premise — or whether it will use AI to automate the retrofit and call it progress.