Accessibility is an anticipatory duty, and conformance is not the duty

Accessibility reaches a design team as a checklist and as a legal duty, and this tenth of ten units argues they are different instruments. The checklist asks whether a page passes testable criteria. The duty asks what a provider should have done in advance so that a disabled person is not disadvantaged by the way a service works. Power and colleagues found that about half the problems blind participants met fell outside the standard's success criteria (Power and colleagues 2012).

The duty is written on the provider, and it is written in advance

The Equality Act 2010 places the duty on the provider, not on the person disadvantaged. Section 20 sets three requirements: reasonable steps to avoid a substantial disadvantage caused by a provision, criterion or practice, or by a physical feature, and provision of an auxiliary aid. Section 20(6) counts an accessible format among reasonable steps where information is involved (UK Parliament 2010).

The duty asks what a practice does to a disabled person and requires steps before any particular refusal, so a provider cannot wait for a complaint.

Directive (EU) 2019/882, the European Accessibility Act, sets harmonised requirements for a defined list of products and services, including consumer computer hardware, electronic communications, transport, banking, e-commerce and e-books. It limits itself twice: microenterprise services are exempt, and Annex VI sets the criteria for a claim of disproportionate burden (European Parliament and Council 2019). The requirements apply from 28 June 2025. Write the duty as a list and it can be enforced; the list then bounds it.

A conformance standard answers a different question

The standard the industry tests against is WCAG 2.2, a W3C Recommendation, approved as ISO/IEC 40500:2025, built from four principles and thirteen guidelines expressed as testable success criteria (W3C 2023).

The guidelines are said to make content more accessible to a wider range of people and expressly not to address every user need. The working group records significant challenges in defining criteria for cognitive, language and learning disabilities, including difficulty reaching consensus on testability. Conformance is a floor counted in detectable page failures. The duty is a judgement about disadvantage in a person's circumstances. A successor, WCAG 3.0, remains a Working Draft, updated in September 2026 with the conformance model still under review, so 2.2 is the only Recommendation a team can test against (W3C 2026).

A million pages fail the same four checks

The 2026 WebAIM Million analysed the top million home pages and detected 56,114,377 distinct accessibility errors, an average of 56.1 per page and a rise of 10.1 per cent on the 51 found in 2025. Of those pages, 95.9 per cent carried detected WCAG 2 failures, up from 94.8 per cent, on pages holding 14.3 per cent more elements than in 2025 (WebAIM 2026). The failures are narrow and stable: low contrast text on 83.9 per cent of pages, missing alternative text on 53.1 per cent, missing form labels on 51 per cent, and empty links on 46.3 per cent.

Half the problems users meet fall outside the criteria

Power, Freire, Petrie and Swallow ran task-based evaluations with 32 blind users on 16 websites and recorded 1,383 instances of user problems. Only 50.4 per cent were covered by WCAG 2.0 success criteria. Where a problem was covered, 16.7 per cent of websites had applied the recommended technique while the problem remained. There is little indication that implementing the guidelines means people will meet fewer problems (Power and colleagues 2012).

Shinohara and Wobbrock interviewed 20 people. They found that devices sometimes marked their users as having disabilities, that functional access took priority over feeling self-conscious, and that one misperception running through the accounts was that disabled people would be helpless without a device. They propose design for social acceptability (Shinohara and Wobbrock 2011). An audit can report that an alternative text string exists; it cannot report the cost of needing one. Lewis, Martinez, Das and Fogarty (2025) show the boundary from another direction: for 16 people who use screen readers, magnification or braille displays, some access barriers were indistinguishable from deception, so a page can pass every criterion and still mislead the person who reaches it by assistive technology.

The social model puts the barrier in the arrangement

Thirty years after the book that introduced it, Oliver restated the model as the claim that disabled people are disabled by the barriers society raises rather than by their impairments (Oliver 2013). If the barrier is in the arrangement, the obligation falls on whoever arranges it. Mankoff, Hayes and Kasnitz brought the disability studies literature to assistive technology research (Mankoff, Hayes and Kasnitz 2010), and turning that into a method produces ability-based design: Wobbrock and colleagues set out seven principles so that a system answers to ability rather than to a diagnosis (Wobbrock and colleagues 2011).

Contested question: is conformance the duty or a proxy for it?

One position holds that conformance is how the duty becomes operational. The standard is shared, testable and internationally recognised. The European Accessibility Act writes the same idea as harmonised requirements with a proportionality test. WebAIM's own conclusion points the same way. Six error types account for almost all failures, and addressing those few would significantly improve accessibility across the web (W3C 2023; European Parliament and Council 2019; WebAIM 2026).

A regulator has now written conformance into the duty itself. The United States Department of Justice's rule under Title II of the Americans with Disabilities Act, published on 24 April 2024, requires state and local government web content and mobile applications to meet WCAG 2.1 Level AA. The compliance dates are 26 April 2027 for entities serving 50,000 or more people and 26 April 2028 for smaller entities, after an interim rule extended them (Department of Justice 2024). Meeting the standard is what the rule will enforce, not whether a person can finish a task, which is this unit's distinction written into a regulation.

The other position holds that the standard measures a subset of the obligation. Half of the problems blind users met fell outside the criteria (Power and colleagues 2012). The social cost of an adaptation sits outside every criterion (Shinohara and Wobbrock 2011). Over 43 per cent of a decade of accessibility papers concerned blind and low vision people, and the median participant group numbered 13 (Mack and colleagues 2021).

The strongest objection to this unit's thesis is that a reasonableness duty of the kind section 20 creates is weaker than a checkable target. A provider facing an open standard can argue for doing less, while a conformance target is measurable and auditable. Imrie's critique cuts the same way: universal design appeals to technical fixes and advocates the marketisation of access, risking a right to access exercised through a market transaction (Imrie 2011).

What would settle it is a comparison, not another position. A controlled before-and-after on one product would measure what disabled participants can do before and after a documented change in conformance, with disabled people deciding what counts as success. It would also compare specialist against universal solutions on measured outcomes, which Imrie says the literature lacks (Imrie 2011). Both need samples large enough to compare, and disabled participants as decision-makers, which Mack and colleagues show is scarce (Mack and colleagues 2021).

The two sides do not rest on equal evidence. The conformance side has a standard, a regulator's rule and a crawl of a million pages, and all three measure pages. The proxy side has three studies of people, thirty-two blind users in Power and colleagues (2012), twenty device users in Shinohara and Wobbrock (2011), and a corpus of 506 papers in Mack and colleagues (2021). The page evidence is much larger and answers a different question from the one the duty asks. The counter-argument above carries the objection to this unit's thesis. While the dispute is open, a design team should treat a pass as a floor, run a task-based test with disabled participants, and record what they could not finish.

Participation is what this literature asks for, and what it rarely measures

Participation is the prescription, and the literature is critical of the substitutes. Bennett and Rosner analysed the empathy activities that open a human-centred design process. They argued that building empathy can further estrange disabled people from the design process, and that empathy should be guided by lived experience instead (Bennett and Rosner 2019). Spiel and colleagues took stock ten years after that call and identified systemic obstacles that keep disabled scholars at the margins (Spiel and colleagues 2020).

Where the argument is heading

The measurement is moving back towards the person. Das and colleagues interviewed 36 neurodivergent professionals working from home and found access rebuilt by the workers themselves, through workspaces they arranged and communication practices they negotiated (Das and colleagues 2021). The open question is whether the next decade fixes the six error categories that dominate the crawl, or measures what people can do once the page passes.

The sources this unit rests on

• Bennett, C. L. and Rosner, D. K. (2019). The Promise of Empathy: Design, Disability, and Knowing the "Other". Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems. doi:10.1145/3290605.3300528 A critical analysis of empathy-building activities aimed at understanding disabled people. It argues that such activities can further estrange disabled people from the design process, and would guide empathy by lived experience.
• Das, M., Tang, J., Ringland, K. E. and Piper, A. M. (2021). Towards Accessible Remote Work. Proceedings of the ACM on Human-Computer Interaction. doi:10.1145/3449282 Interviews with 36 neurodivergent professionals working from home during the pandemic. It found access being assembled by the workers themselves, through workspaces and communication practices they arranged.
• European Parliament and Council (2019). Directive (EU) 2019/882 on the accessibility requirements for products and services (European Accessibility Act). https://eur-lex.europa.eu/eli/dir/2019/882/oj The directive giving accessibility an enforcement route beyond a public sector duty. It sets harmonised requirements for a defined list of products and services, and limits itself by a microenterprise exemption for services and a disproportionate-burden test.
• Imrie, R. (2011). Universalism, universal design and equitable access to the built environment. Disability and Rehabilitation. doi:10.3109/09638288.2011.624250 A commentary critiquing the theoretical content of universal design. It argues the concept is reductive and functionalist, that marketising access risks reducing a right to access to a right exercised through a transaction, and that it is ambivalent about specialist solutions.
• Mack, K., McDonnell, E., Jain, D. et al. (2021). What Do We Mean by Accessibility Research?. Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems. doi:10.1145/3411764.3445412 A corpus study of accessibility papers at CHI and ASSETS since 1994. It coded 506 papers from 2010 to 2019 and found over 43 per cent concerned blind and low vision people, with a median of 13 participants for disabled and older groups.
• Mankoff, J., Hayes, G. R. and Kasnitz, D. (2010). Disability studies as a source of critical inquiry for the field of assistive technology. Proceedings of the 12th international ACM SIGACCESS conference on Computers and accessibility. doi:10.1145/1878803.1878807 A review importing the disability studies literature into assistive technology research, with two projects it informed. It shows the literature changing what the authors treated as the problem.
• Oliver, M. (2013). The social model of disability: thirty years on. Disability and Society. doi:10.1080/09687599.2013.818773 A retrospective written thirty years after the book that introduced the social model. Oliver restates it as the claim that disabled people are disabled by the barriers society raises rather than by their impairments.
• Power, C., Freire, A., Petrie, H. and Swallow, D. (2012). Guidelines are only half of the story. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. doi:10.1145/2207676.2207736 An empirical study of the problems met by 32 blind users in task-based evaluations of 16 websites, yielding 1,383 instances. Only 50.4 per cent were covered by WCAG 2.0 success criteria, and 16.7 per cent of websites applied the recommended technique without solving the problem.
• Shinohara, K. and Wobbrock, J. O. (2011). In the shadow of misperception. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. doi:10.1145/1978942.1979044 An interview study of 20 people on how assistive technology use is shaped by social context. It found that devices sometimes marked their users as having disabilities, and proposes design for social acceptability.
• Spiel, K., Gerling, K., Bennett, C. L. et al. (2020). Nothing About Us Without Us. Extended Abstracts of the 2020 CHI Conference on Human Factors in Computing Systems. doi:10.1145/3334480.3375150 A workshop position paper, cited as that and not as empirical evidence. It takes stock ten years after Mankoff, Hayes and Kasnitz called for disabled people to be actively involved in research about them.
• UK Parliament (2010). Equality Act 2010. https://www.legislation.gov.uk/ukpga/2010/15/contents The statute carrying the reasonable-adjustments duty this unit reads. Section 20 sets three requirements, each framed by a substantial disadvantage and answered by such steps as it is reasonable to take.
• W3C (2023). Web Content Accessibility Guidelines (WCAG) 2.2. https://www.w3.org/TR/WCAG22/ The conformance standard the industry tests against, built from four principles and thirteen guidelines and approved as ISO/IEC 40500:2025. It states that it will not address every user need. The successor is W3C (2026), W3C Accessibility Guidelines (WCAG) 3.0, https://www.w3.org/TR/wcag-3.0/, a Working Draft of 10 September 2026 with the conformance model still under review.
• WebAIM (2026). The WebAIM Million: the 2026 report on the accessibility of the top 1,000,000 home pages. https://webaim.org/projects/million/ An automated evaluation of one million home pages in February 2026, the eighth in the series. It detected 56,114,377 errors, an average of 56.1 per page, with 95.9 per cent of pages carrying WCAG 2 failures.
• Wobbrock, J. O., Kane, S. K., Gajos, K. Z. et al. (2011). Ability-Based Design: Concept, Principles and Examples. ACM Transactions on Accessible Computing. doi:10.1145/1952383.1952384 The paper proposing ability-based design as a refinement of accessible computing. It argues the field centralises disability rather than ability, and sets out seven principles.
• Department of Justice (2024). Nondiscrimination on the basis of disability: accessibility of web information and services of state and local government entities (final rule), and the 2026 interim final rule extending the compliance dates. https://www.ada.gov/resources/2024-03-08-web-rule/ Requires state and local government web content and mobile applications to meet WCAG 2.1 Level AA, with the compliance dates extended to 26 April 2027 for entities serving 50,000 or more people and to 26 April 2028 for smaller entities and special districts.