The WAI and the WCAG need a Revolution

During my accessibility training sessions, I half-jokingly call WAI the Vatican of accessible internet. By this logic, WCAG would be the bible of accessibility. Everyone can know that the Vatican is a little out of date. Unfortunately, this also applies to the WAI. In this article I would like to show why in my opinion this is the case.

For the uninitiated: The WAI = Web Accessibility Initiative is the working group within the World Wide Web Consortium that deals with digital accessibility. I'm not sure if this article will be of interest to people who are only tangentially involved with digital accessibility. Unfortunately, I cannot explain all the details and context that would be important for understanding for those persons.

Article Content

Problems with WCAG 2.x

In my opinion, WCAG 2.x has to contend with numerous serious problems that make it not really future-proof.

So far the WCAG has been quite robust. However, it comes from an age of static websites and is not equipped for dynamic applications. Numerous requirements such as those of the visually impaired, cognitively disabled or neuro-diverse people are inadequately covered. At the same time, it does not seem desirable to further inflate the already very confusing criteria.

The WCAG itself is intended to be technology-agnostic. However, it's not entirely consistent since the term "web" is in the name. The WCAG itself consists of somewhat cryptic success criteria that only become truly understandable when combined with the “Understanding” and “How to meet” documents. I recommend that my training participants always consult the “Understanding” documents first; the WCAG itself is IMO only suitable for experts. The interaction of these documents with each other and with the techniques is not always clear to outsiders.

The conformance model consisting of A, AA and AAA is also not always understandable. The reasons why certain criteria were classified as A, AA or AAA are not transparent.

The principle of full conformance is also outdated. I have described the problem in the WebAIM article in detail. If there is almost no complex offer that is fully compliant, then something seems to be wrong. Either everyone is too stupid or the effort required to conform is not in any healthy proportion to the result. In fact, I consider the principle of full compliance to be WCAG's greatest shortcoming and should be abolished today rather than tomorrow. It is one of the reasons for the success of accessibility overlays. Full compliance is the elephant in the room that none of the experts are talking about.

The problem is not conformance itself, but the different interpretations by experts. All it takes is one expert saying that criterion X is not compliant and the entire website is not accessible. There is no alternative text for an irrelevant image on page X - the website is not compliant. How should you as a service provider recommend this to a customer? It results in countless micro-optimizations with little to no value for disabled people. But it is one of the main sources of income for accessibility experts. And one of the reasons why many developers hate us is because many things cannot be explained in a way that make sense. After all, WCAG 2.2 abolished criterion 4.1.1 parsing, one of the less useful criteria. Unfortunately, this is only incorporated into the national regulations with a delay.

The really big ones like Amazon, Microsoft, Apple and Google formulate their own standards anyway. Amazon Germany, for example, considers itself to be accessible, even though the cookie message has not been implemented accessible for years. Google does not make many of its business-relevant programs such as Analytics or Search Console accessible and thus excludes blind people from the SEO and analytics field. After X years, Microsoft Bing has failed to make its standard cookie message accessible.

The distribution of the success criteria is also not immediately clear. For example, there is an Info and Relationships check step, which checks whether headings and machine-readable labels are present. Another test step, Headings and Labels, checks whether the labels make sense. There is a test step in which the presence of visual labels is checked and another test step at a completely different point that checks whether the machine-readable name is part of the visual label. This makes sense from the internal logic because these success criteria are assigned to different guidelines. In terms of the testing logic, however, this doesn't really make sense. By looking at the label, you can check in one step whether it makes sense, is machine readable, is visually present, is synonymous with the machine readable name and whether an Auto Complete is present if necessary.

In addition to sensible criteria, there is also a lot of nonsense in WCAG, such as the language attribute (Success Criterion 3.1.2 Language of Parts). I also don't think the AutoComplete requirement (1.3.5 Identify Input Purpose) makes much sense: this can be solved better on the client side. My impression is that many criteria are based on outdated assumptions about assistive technologies and the behavior of disabled persons.

In general, an important topic is the question of comprehensibility: The WCAG and its informative understanding and how-to-meet documents are all sorts of things, but they are not understandable to outsiders. The question remains whether it should be an expert document - which it de facto is today - or a help for people who are not knee-deep in accessibility and want to use it - that is not what it is today. Basically, you always have to know all 60 A and AA criteria in order to be able to use them. But along the way, no one reads into 60 criteria. This means that basically every organization with a digital offering would need an accessibility specialist.

Another example is the complicated translation policy. I was marginally involved in the German translation of WCAG 2.1, which still doesn't exist today. The translation was carried out competently by competent people. To this day it is not clear to me why it was rejected, but that was the end of the effort to translate it into German. There won't be a new translation effort any time soon.

It is true that the WCAG has proven itself. Overall, it should cover most use cases and doesn't need that many updates. That's a good thing, considering that there were 10 years between 2.0 and 2.1 and now almost five years between 2.1 and 2.2, even though these are relatively small changes. Many specific requirements are also more likely to be covered in non-normative documents such as the ARIA Authoring Practices, which appear to be updated more quickly.

The Web Accessibility Initiative needs reform

On the other hand, WAI, like many standardization institutions, appears to be degenerating into a bureaucratic monster. On the one hand, these endless discussions can be explained if you consider the impact of the rules - after all, they are somehow relevant to almost everyone in the world who has to do with technology. On the other hand, you cannot keep up with developments. This can also be seen in the fact that WAI has little to say about electronic documents (except ePub), native apps and desktop software and focuses almost entirely on websites and web applications. It's true that desktop software is becoming less relevant, but that doesn't apply to native apps.

The 10 years from WCAG 1.0 to 2.0 were still explainable. But not the ten years from 2.0 to 2.1 and five years from 2.1 to 2.2. If you consider that authorities need at least another five years to implement these regulations into national law, you simply have to realize that we are hopelessly behind the technical development.

The WCAG should be able to develop as dynamically as HTML5 in order to no longer keep up with new developments. For the lame authorities and everyone else, one could offer a stable version of the guidelines, as is common practice with browsers today, and for the others the WCAG could be dynamically developed further.

WCAG's processes are pseudo-democratic. Yes, anyone can participate - theoretically. Practically, you need to have a lot of time, deep technical knowledge, perfect technical English and a few other qualities. The tone on the mailing lists can also be rough if you're not part of the clique or have a lot of knowledge, so it's not for the faint of heart. I have no doubt that everyone involved is putting in a lot of effort. But everyone can participate and contributing is different than openly inviting people to do so. At the moment the WAI is not a good example of inclusion. It is reminiscent of all venerable institutions like Wikipedia, where you first have to earn your spurs in order to be taken seriously. The committee is not interesting for young people.

In today's times, we need other options for co-determination that, on the one hand, involve more diverse groups and, on the other hand, also enable faster decisions. For example, why isn't there a wiki that would make the discussion lower-threshold than GitHub?

In my opinion, it must also be possible to go beyond the WCAG without being eaten by the customer. It's nice that the WAI gives recommendations for design for cognitively disabled people, but not like this It's nice that recommendations are non-binding.

Is WCAG 3 the answer

The problem of the parallel existence of several sets of rules can be solved. The changes planned so far will not throw all the old rules overboard. This means that what is WCAG 2.x AA compliant will probably also be 3.0 compliant. Transition periods can ensure that old offers gradually meet the few new requirements. New offers can start directly with 3.0 after a certain period of time. WCAG 2.0 and 2.1 already exist in parallel today. 2.2 will probably also exist in parallel with the other two, as some countries still have WCAG 2.0 as their standard.

We have learned from many experiences that one should not set expectations too high. We will be disappointed either way. If WCAG 3 comes at all, which I currently doubt, it will solve a few problems and create a few new ones. However, it is clear to me that WCAG 2.x is fundamentally unsustainable due to the problems mentioned above. However, I think that faster document updates are needed. The WAI's current system for negotiation doesn't seem to really work, so you can't react well to current developments.

At the moment I wouldn't bet that there will be a WCAG 3 at all, I rather believe that George Martin will finish his Song of Ice and Fire by 2030.

More on Accessibility Consulting