Accessibility as a property of the dialogue, not the device

A formal-and-political reframing from a 2006 paper, complementary to the Functional / Intrinsic distinction. Three structural moves in one definition; the cleanest correction of one of the field’s most stubborn category errors.

The definition

Goods and services are accessible if the provider of the goods or services offers a means by which a broad range of users may enter into a dialogue with the provider to gain access to them at any given place and time.

From an ATNAC 2006 submission. The paper was not published — venue mismatch with ATNAC’s networks-and-telecoms-engineering focus — so what follows is a dated written artefact, not a citation. The structural moves carried through into the later doctoral work and stand independent of whether the original paper found its venue.

Move one: the unit of accessibility is the dialogue

A device alone is not accessible or inaccessible — it either supports a particular user’s dialogue with a particular provider, or it does not.

If a device only provides a means for some users to hold a dialogue with a provider, then it is not meaningful to describe the device as ‘accessible’, nor can one describe one device as more accessible than another; it either works for a particular user, or it doesn’t.

This is the cleanest correction of one of the field’s most stubborn category errors: talking about an accessible iPhone or an accessible website as if the artefact carried the property in itself. It does not. Accessibility is a property of the relation between a specific user (with a specific capacity profile in a specific context) and a specific provider (with a specific catalogue of devices and channels through which to engage). Move the user, the context, or the catalogue, and the accessibility status moves with them.

The closely related observation in the doctoral framework is that functional accessibility is also relational rather than intrinsic to the device — achievable through at least one successful negotiation between a user and a provider, in a specific context, via at least one available medium-and-protocol path. The dialogue reframing is more general still: it generalises from one specific user / one specific interface to a population of users / a catalogue of provider channels.

Move two: measurement is coverage across an offered range

Measurement of accessibility then becomes a question of coverage: taken together, does the offered range of devices provide a means for all users, disabled and non-disabled, to access the goods and services?

A coverage-set formulation. A provider is accessible if their range of devices and channels collectively supports every user in the target population. The property attaches to the catalogue, not to any single artefact within it.

This is distinct from intrinsic accessibility as formalised in the Measure of Accessibility collection. Intrinsic accessibility names how many pseudo-users a single underlying interface admits; coverage names how many real-population users a provider’s collection of interfaces admits. The two measures complement each other: an intrinsically accessible single channel is better than a functionally accessible specialist channel; a coverage-set of intrinsically accessible channels is better than either.

Move three: the commercial-and-legal version

The same conclusion arrived at from a different angle, useful for audiences who do not engage with the equality framing.

In supplying phones, manufacturers are in competition, and one element, to support legal accessibility requirements, will increasingly be breadth of user-base. In such an environment there is a commercial advantage in demonstrably maximizing the breadth of user-base; ‘demonstrably’, because it may be necessary to convince courts and lawyers that best efforts have been made to provide access to goods and services.

For commercial buyers framing accessibility as risk management rather than as rights, the conclusion still holds: maximise breadth of user-base across your channel catalogue, demonstrably, because the alternative is litigable. Same destination as the equality argument; different vehicle. The destination is the accessibility status of the provider’s offering, not the accessibility status of any single artefact.

The strategic observation embedded in the same paper

Multiple impairments are a common feature of ageing.

The 2006 paper states explicitly that the typical accessibility case is not a single textbook disability but combinations that compound and that change over time — and that the demographic engine for this is ageing, not disability-from-birth. Today’s accessibility-and-ageing pivot is a 2006 claim, written up twenty years ago, that no venue picked up at the time. The paper’s rejection means it is not a citation; the dated artefact in the archive remains.

The same observation surfaces formally in the doctoral framework via the spiky profile construct — the Capability / Capacity / Preference / Requirement four-model architecture handles multi-axis variability trivially because the templates compose along multiple ontologies (visual, sonic, haptic, cognitive, colour-vision, language). The bucket model the field defaults to — blind / deaf / motor-impaired / cognitive-impaired— handles it by failing. See Intrinsic Accessibility for the formal treatment of the four-model architecture, and the worked Mike Smith / David Furness personas in the thesis Personas appendix for the operational granularity.

Reading on