1. The Question

What the question actually asks

Accessibility is not a technical question with a technical answer. It is not legal, although the law has things to say about it. It is not ergonomic, although ergonomics is part of the picture. It is political — in the sense that the answer depends on what kind of society you take yourself to be building, and on what relationship that society has with the people who use it.

Most technical writing on accessibility quietly elides the political character of the question. WCAG is a checklist of testable criteria; it is extremely good at being a checklist. But a checklist does not tell you what counts as accessible — it tells you what counts as conformant to a particular set of decisions made by a particular working group about which technical properties can be tested reproducibly. Conformance and accessibility are not the same thing. The conflation is convenient and widespread; it is also the source of most of the category errors the field makes.

The political spectrum

The answer to what is accessibility sits somewhere on a spectrum. The two ends are starkly different positions, and most of the contested ground in the field is the territory between them.

At one end, accessibility is the capacity to insert a disabled user into existing society. This is the position implicit in most assistive technology built since the 1980s. The society is taken as given; the inaccessible interfaces are taken as given; the work of accessibility is to graft a translation layer onto the user so they can interact with what already exists. Screen readers translate the visual web into speech. Switch-access scanners translate complex interaction into single-button affordances. Captioning translates audio into synchronous text. Each is, on its own terms, a substantial engineering achievement; each is also an admission that the underlying interface was not designed with the user in mind, and the cost of that omission is now being paid by the user’s assistive layer.

At the other end, accessibility is the property of a society that ensures universal access to its goods and services. On this view, the work of accessibility is the work of designing the underlying systems — interfaces, policies, infrastructure — so that they admit many user-platforms from the start. The translation layer is not absent; it is built in. The user does not arrive at an interface designed for someone else and try to make it work; the interface already adapts to who they are.

These are not technical positions. They are positions about what a society owes its members. The first treats accessibility as a charitable accommodation extended toward people who would otherwise be excluded. The second treats it as a baseline precondition for membership. Neither is unreasonable in isolation; the gap between them is where most of the argument lives.

Why most assistive tech sits at the insertion end

If you only have a hammer, then everything looks like a nail.

You can write a program to translate text into speech. You can write a program to scan through menu items at a configurable rate. You can write a program to add captions to a video. What you cannot write is a program to change the assumption that the video should be primarily visual, or that menus should be primarily clicked, or that web pages should be primarily read with eyes. The first three are tractable. The fourth is political.

The hammer is software. Software is the assistive-technology field’s strongest tool, and as a consequence the field has largely framed accessibility as a problem software can solve. That framing carries a hidden constraint: it pushes the answer toward the insertion end of the spectrum, because that is the end of the spectrum where software-only interventions are sufficient. The universal-access end requires changes to interfaces, standards, organisations, and procurement practices — categories of change that no individual programmer can ship. The temptation, then, is to treat accessibility as a software problem, because software problems are the ones we know how to attack. The cost of that temptation is two decades of bolt-on assistive technology that takes the underlying interface as a given.

Why current vocabularies underdetermine the question

Three vocabularies are usually offered when the question of what is accessibility is raised. None of them resolve it.

The legalvocabulary names accessibility as a matter of rights and obligations. Different jurisdictions name those rights differently — the ADA in the United States, AODA in Ontario, the European Accessibility Act, Canada’s Accessible Canada Act. Each of these reads accessibility as something a service owes to a class of users by virtue of disability status. The vocabulary is precise about who must comply and what the penalties are for failing to do so. It is silent about which interfaces are accessible and which are not, except by reference to technical standards that the law itself does not author. Legal vocabulary tells you whether the consequences for failing have teeth; it does not tell you what success looks like.

The ergonomicvocabulary names accessibility as a property of a fit between user and tool — the right size of grip for the hand, the right contrast for the eye, the right latency for the motor pathway. Ergonomics is good at what it does. It is also scoped to physical properties of the interaction surface; it has very little to say about what the interface should be for, who it should be for, or what should happen if the user’s body or context falls outside the ergonomic envelope. Ergonomic vocabulary tells you whether a particular knob is graspable; it does not tell you whether the knob should exist.

The usabilityvocabulary names accessibility as a special case of usability for users with disabilities. This is a friendly framing and often a productive one — ergonomic and cognitive accommodations that make an interface usable by a disabled user often make it more usable for everyone — but it conceals a category error. Usability is a commercial property:does this product satisfy its target users well enough to compete? Accessibility is a political property: does this society make its goods and services available to its members regardless of physical capability? They are not the same question, and the answers do not coincide.It is possible to have a wonderfully elegant and usable interface that will score low in terms of intrinsic accessibility, and an accessible interface that is almost unusable. The usability vocabulary cannot capture that distinction.

The categorial difference between usability and accessibility

The distinction is worth making explicit because so much of the field operates as if the two collapse into each other.

Usabilitymeasures how well an interface serves a defined target population. The target population is decided by the business — who is buying, who is being marketed to, who the strategic plan says is in scope. Once the target is fixed, usability is measurable: task completion rate, time-on-task, error rate, satisfaction. The answers can be high or low; the question they answer is well-posed only relative to the target.

Accessibilitymeasures how well an interface serves the population that exists independent of the business’s targeting decisions. The relevant population is not who the business chose to serve but who the society includes as members. A reading-only screen reader user, navigating a website built for sighted users, is not outside the population the website should serve; they are outside the population the website was built for. Those are different judgements.

Both questions are legitimate. Both are answerable. They are not the same question, and an interface can do well on one while doing badly on the other. The clearest illustration is in the field itself: specialist assistive devices — the PacMate, for example, with its replaced keyboard and forced text-to-speech — are functionally accessible to their target users at high cost in usability. They are accessible in the sense that the relevant user-population can complete tasks; they are unusable in the sense that the experience is profoundly different from what a typical user encounters. Conversely, a beautifully designed web app with no regard for screen readers can score in the high nineties on every usability metric and still fail to serve a substantial fraction of the people who are entitled to it. Both situations exist. Both are common. A vocabulary that cannot tell them apart is a vocabulary that cannot say what it is for an interface to be accessible.

Where this collection goes

The remaining five pages of The Measure of Accessibility develop a position that takes the political character of the question seriously and answers it formally rather than rhetorically. Briefly:

• Functional Accessibility (page 2) gives a definition with notation: the property of successful negotiation between a specific user and a specific provider, achievable via at least one medium-and-protocol path among those available.
• Intrinsic Accessibility (page 3) generalises: the property of an underlying interface that admits successful negotiations across many user-and-context profiles, formalised through the pseudo-user set independent of any specific provider.
• Equivalent Experience (page 4) argues against utilitarianism in accessibility, and treats the timing dimension of equivalent experience as an equality question.
• The Shlaer-Mellor lens (page 5) is the methodological substrate that lets the formal definitions on the previous pages be built rather than just stated.
• Communities of Practice (page 6) reframes inaccessibility as community dysfunction and opens onto the multi-agent territory that the 2029 framework resumes.

The order is not arbitrary. Each page rests on the one before it. A reader who wants only the political framing can stop after page 1; a reader who wants the formal definition can stop after page 3; a reader who wants the whole position should read all six.

Continue

Next: 2. Functional Accessibility →