> EuAccessibilityQualityMark

A European certification scheme for a Quality Mark for Web Content Accessibility

There is to be a CEN/ISSS Workshop on Specifications for a European certification scheme concerning the delivery of a Quality Mark for Web Content Accessibility. The new activity is known as “Support-EAM” and has arisen in Europe to provide an important opportunity for Europeans to attend to the accessibility of Web resources and services. The outcome of the Workshop will be significant in Europe and influential in the rest of the world. This briefing paper presents a perspective that explains the optimism and support for this activity by a number of organizations concerned with promoting an approach that will increase the accessibility of the Web.

We endorse, and wish to see developed, a common language for describing a common set of accessibility-related qualities of resources and services. We would like to see a mark that indicates that a resource or service has been evaluated for accessibility and provides the location of a machine-readable statement, in standard format, that can be used for matching resources and services to users' needs and preferences.

All work on Web accessibility today relates in some way to the work of the W3C/WAI which has developed a set of guidelines that aim to define, in a general form, specifications for "universally accessible" digital materials. The Web Content Accessibility Guidelines are, without doubt, the best in the world and are recognised worldwide as the de facto ‘standard'. The guidelines do not try to relate the qualities of resources to particular technologies or particular disabilities.

Reporting on the accessibility of a resource by using the WAI Conformance Levels does not communicate which types of accessibility features are offered in a resource. Even reporting at the level of a checkpoint (individual technical requirement) may not make clear which kinds of accessibility problems are or are not present, since some checkpoints are meant to correct a number of inaccessible Web programming practices. In addition, people with disabilities are not homogenous, and disabilities are not always compatible. The accessibility solutions for different disabilities can conflict.

For these and other reasons, despite supporting as energetically as possible the development and use of the W3C Guidelines as specification goals for developers, many in the accessibility community have now decided that mandating "universal accessibility" of all resources and services prior to the point of delivery would be an unattainable and unnecessary standard. Instead, they have taken advantage of the developments in technology and now aim to deliver to any individual user, materials that are accessible to that person at the time of delivery. This approach adds significant flexibility with the advantage that:

The new approach implies both that information about the accessibility qualities of a resource or service should be available for potential consumers of the resource or service and that there is no logical way to determine the absolute accessibility of a resource or service in advance of the delivery of that resource or service. In fact, if it is the user's satisfaction that determines this, it is important that the suitability of resources and services should not be prejudiced by a quality mark but enabled by availability of information about the qualities of the resource or service.

Quality marks for materials have generally been applied in two ways.

First, as in the US, a conformance mark (or claim) might signify that a resource publisher has, or has not, satisfied legal requirements. As the test of conformance in the case of accessibility is partly subjective, it is common for such claims to be inaccurate. In addition, immediately after they are made, the resource can be changed and the claim's veracity may not survive the changes. As such claims may also attract legal sanctions or financial implications, so they raise issues of trustworthiness. So there are two points of concern with respect to the reliability of such claims: the quality of the evaluation (or expertise being used), and the time and date of the evaluation. Finally, as stated, even a resource that is universally accessible according to the guidelines may not be accessible for an individual. The use of such a mark is thus somewhat questionable.

Secondly, a quality mark could be used to describe the qualities of a resource. This is the case when a garment carries a ‘pure new wool' mark to indicate that it is made of Australian wool. Those who have found the W3C Guidelines inadequate as a guarantee of accessibility, and turn instead to the question of whether or not the resource as delivered satisfies the immediate user's needs, depend upon information about the resource and its components to ensure its accessibility. That is, they want to know the qualities of the components of the resource so that, if necessary, inaccessible components can be replaced by accessible ones. A mark that points to a trust-worthy, machine-operable description of the accessibility characteristics of resources would enable the matching of such a resource to a user's stated needs. It would also mean that while a single resource might have one or many accessibility flaws, the entire information delivery system may still cooperate to achieve accessibility. Finally, a description of the qualities of a single resource would not attract legal or financial implications as it would not indicate ‘accessibility' or otherwise. It could be expected to be more reliable than a single, quantitative conformance mark.

W3C has developed a suitable form of representation for evaluation results called Evaluation and Report Language (EARL) and a number of tools are available for its creation. The EARL representation form includes information about who made the representation and when.

A quality mark that carries descriptions of accessibility qualities of resources and services would enable the matching of resources to the individual accessibility needs and preferences of users. Such a mark would vastly increase the potential for cumulative, distributed solutions to accessibility and thus be of incredible benefit to those who have significant accessibility needs.

Background Comments

For many years, accessibility experts have wished for a simple solution to the problem – it would be wonderful if a simple, single-quality mark on resources and services could increase their accessibility.

The WCAG Working Group has spent many days debating the possibility of using such a mark and found there are problems that cannot be resolved. One significant issue has always been the nature of an absolute mark – failure and success are too close when there are more than 60 questions to be addressed and failure of any one, often judged subjectively, can mean failing the conformance test. Strategies such as cumulative scoring of points for accessibility features have been proposed but always, when their use to increase accessibility for users is considered, it becomes clear that not every user will benefit from such a grading, and it may work against increasing accessibility generally. There is an open, international forum for discussion of such a mark offered by the WCAG Working Group.

EuroAccessibility, a group of 25 leading organizations in Europe, formed an association to work collectively towards accessibility including working to provide a way for indicating accessibility of resources and services. EuroAccessibility had a Task Group that based their work on the W3C/WAI's WCAG and aimed to quantify accessibility. They were not the first to try this, and like others, found the task extremely difficult. Many others have now decided the task was not the right one – that a machine-readable comprehensive statement of accessibility qualities is far more useful both in the accessibility authoring and in the process of matching user's needs and preferences than a single quantified score.

In the US, at a strategic moment, a mark was created for commercial purposes– the US Federal Government exercised its muscles by passing the now famous s.508 that acted quickly as a quality mark in the US. It was associated with a huge market player and had a very good effect. Those companies interested in US Federal Government related contracts, and these included many international companies, started to pay attention to accessibility. It worked worldwide. The US Defence Department is economically the largest education market in the world, for example, and so being able to sell into that market became a priority. Add to it all the other related markets and we can see why s.508 was so successful. But s.508 does not even pretend to guarantee accessibility. Other governments may be able to use the strategy to apply economic pressure to increase attention to accessibility needs but s.508 is not suitable as a conformance mark.

Support-EAM can learn from the s.508 experience, however. Where bodies are being asked to declare their failure to satisfy the law, they are unlikely to report honestly, even if they can do so competently, and experience shows very few can do that in the complex accessibility context. A single quality mark that expresses in a binary way that a component or some content conforms to the WCAG/WAI or any other similar set of criteria, can be wrong for too many reasons:

Suggested Approach

We suggest a starting point for the recommendation of the WAC WS as follows.

Establish a conformance mark that describes the accessibility qualities of a resource or service offered on the Web, as described above.

The benefits of this approach are:

In general, qualities descriptions of resources and services in common format can be shared even when those using them are working to different standards, such as s.508, or WCAG Levels A, or AA or partly WCAG and some other criteria. This ensures valuable flexibility and yet maintains interoperability.

We believe the “qualities” approach will support good business practices and accessibility better than any single “all or nothing” quantitative mark could. A likely concern is that it is too hard to produce such a description of all the qualities and so simplicity favours a single mark. So far, all known attempts to develop single quality marks have been based on processes that, in fact, aggregate a list of qualities. For those creating the mark there is, then, no difference in the amount of work required.

A useful basic description of a resource might simply state that the resource has text and an image but does not have sound, so the accessibility of the image can be checked for a blind user but may not be an issue for another user. A complex description would contain detailed information, perhaps to identify a synchronisation file to accompany captions and a text transcription for a film.

There are now many tools available to help with the description process. They produce descriptions that can be read by humans and machines. Humans can use them to predict the accessibility of resources for a given individual and machines can use them to match resources to the user's needs and preferences. Access to a description of the accessibility qualities of a resource is essential to the matching process – in fact, without such a description matching cannot be achieved even if a resource is "universally accessible".

Comments

Please identify yourself before your comment.
Lisa seeman:

I would like the mark to be constructed that it will enable uses to find content that is suitable for their specific scenario. I wish to include describing content in turns of suitability for different physical abilities, cognitive abilities and learning styles. I am not sure that WCAG levels achive this.

The focus of the quality mark should be in describing the accessibility related qualities of the page. The mark should not prejudice the author against accommodation for some disabilities by providing basic accessibility quality mark to content that is inaccessible to some groups of users.

This will involve developing terms useful for different abilities - beyond checkpoint conformance

This may involve extending reporting languages

This will include identifying how the resource accessibility has been provided (relationship to services)

Chaals: Added a correction of history, above. [removed from txt above - was in ref. to statement about previous work - (Chaals: No, they didn't try to quantify accessibility. They started working towards an agreed methodology for testing WCAG checkpoints.)]

In response to Lisa, the reporting language EARL has no difficulty in supporting, as is, descriptions of more detailed information than WCAG conformance. Indeed, it is not tied to WCAG at all, but is a general reporting language that can be used for reporting the results of testing software against various quality checks, or testing people's knowledge of specifications...

Andy Heath: I don't have anything to add to the technical case made in the main body and in reply to Lisa except to say that I personally support the argument for a "Qualities" mark not something binary that organisations could hide behind. I do think any label of this kind needs to support the use of technologies like EARL that can be used today and tomorrow and as techniques improve used again to reveal more information - updatable over time without involving organisations having to do stuff like allocate resources to update an evaluation.

Alan Chuter (ONCE Foundation): The UWEM methodology is not intended prove that something is accessible, but rather that according to the methodology a resource complies with WCAG. There is a phrase in the WCAG introduction makes it clear that it does not guarantee accessibility. But WCAG is here and it is being used, unlike more sophisticated schemes that are still vapourware (or is there now a widespread use of EARL to annotate real-world resources that people can actually use?). Most web site owners and designers still have trouble coping with WCAG half decade after it was published; the possibility of certifying each aspect of each resource in a way that real people will be able to use is a some way off I fear. I agree that annotation is a better idea, and claims for compliance to a subset of checkpoints, but I doubt whether it's practical reality for the near future. We need a solution today. I agree that the certification scheme should provide a mechanism to express results in EARL, but that is independent of the evaluation methodology.