Expect Delays signI’ve blogged before about the problems created by a lack of ADA website regulations, including the difficulty courts have deciding just what  “accessible website” means.* The Circuit Court most likely to shed light on this issue is the 11th Circuit, for the pending Gil v Winn-Dixie appeal presents the question directly. There is, however, a more fundamental problem. It may not be possible to create an objective standard for accessibility. I’ll explain why by looking at the most commonly referred standard, WCAG 2.x and showing that it is impossible to determine objectively whether any website actually conforms to WCAG 2.x at any Success Level.

Objectively verifiable compliance with the law was a principle incorporated into the original and existing ADA regulations. To make compliance objectively verifiable the law and regulations draw a distinction between places and the services or goods they offer. Regulations concerning access to places precise and objective criteria for determining whether an owner has provided non-discriminatory access. For the sake of objectivity the regulations do not guarantee equal access. Some allowable ramps are too steep for individuals with the most severe disabilities and it is common to permit accessible routes that are longer and less convenient than the routes available to those without disabilities. The 2010 and 1991 standards incorporate a host of compromises that sacrifice full accessibility in order to create an objective standard.

The regulations also include an explicit rejection of the idea that all of the goods and services offered by a public accommodation must be accessible. The most obvious example concerns books. A bookstore must be physically accessible to a blind person but is not required to carry braille books. Where entertainment is offered the regulations require equality of opportunity in terms of seeing or hearing the entertainment, but only in the case of movie theaters do the regulations require that non-verbal content be translated into verbal form. Even then a public accommodation is only required to make audio description and closed captioning available when digital movies are distributed with such features – the public accommodation itself is not responsible for the adequacy of either the captioning or the audio description. (See, 28 CFR §36.303).

The distinction between place and product becomes blurred in the world of the internet. Websites deal in information, and in most cases that information is seamlessly integrated into the metaphorical space represented by the website itself. The website is both the place and the content. For websites “accessible” has a double meaning. In a physical place the goods or services are accessible if you can get to them. In a website accessible means both getting to the content and being able to use it.  As a result the Web Content Accessibility Guidelines include success criteria that cover both how a user gets around the space represented by a website** and whether the user can use the informational the content of a website.†

Another critical difference between physical spaces and websites is the bias in web development toward the visual. Human interactions with computers have been mediated through screens almost from the beginning and for most developers and users the human facing part of a website is first and foremost pictures and words on a screen.

When the information provided by websites was primarily verbal‡ this was not a problem. Whether a disabled person saw the words or heard them through a screen reader the information was the same. Now, however, websites include and even feature non-verbal information. When “the medium is the message” equality of use is impossible. No verbal description can convey all the information in a picture. No closed captioning can convey all the information in a song or even the spoken word. Making non-verbal content accessible always involves compromise in how much information is provided in a form that can be used by a person with a disability.

Equally important, the translation of non-verbal to verbal information involves a guess as to the intent of the creator by a technician or a software robot of some kind. What the image is may be obvious. Why the image was chosen involves speculation.

I can make this problem more concrete by looking at a specific bit of content we are all familiar with – click bait. As I read online news I frequently find a picture of a woman in a bikini with the headline like “Ten celebrity clothing fails that everyone is talking about.” WCAG Success Criterion 1.1.1, which is at the most basic “A” level of success, has this requirement for that picture:

All non-text content that is presented to the user has a text alternative that serves the equivalent purpose. . . .

For this particular image the purpose is to make me click on it so I’ll go to another website where I will be presented with advertisements. A picture of a woman in a bikini fills that purpose because many users will not be able to resist the promise implied by the caption that clicking will lead to more pictures of scantily clad women. The obvious text alternative for this picture –  “picture of a woman in a bikini” –  isn’t likely to fulfill this purpose for a person who is blind. Such a user is probably not tempted by the prospect of reading a nine more equally unalluring text alternatives. A different text alternative – “how to make a million dollars from home” – might serve the equivalent purpose, but of course it would have nothing to do with the picture itself. And an explicit statement of purpose – “clickbait photo” – would be the worst at accomplishing the purpose.

The problem here is the word “purpose.” For verbal content the presumed purpose of the author is to communicate the ideas represented by the words and sentences, so changing from one verbal form (text) to another (spoken word) doesn’t matter. For non-verbal content that presumption is clearly wrong. If the information that could be conveyed in verbal form the creator would have written a sentence instead of creating a melody or picture. At the very least a picture or melody is intended to present information in a form that is more efficient than text. “A picture is worth a thousand words.” Unfortunately you can’t put a thousand words in an alt tag without making it harder rather than easier for a person with a screen reader to understand.

This creates an impossible problem for human or machine testers looking for objective evidence that a web page is accessible to someone with a disability. It is easy enough for a software tool to determine whether every image has an alt-tag, but the presence of an alt-tag doesn’t guarantee the purpose of the image has been fulfilled. A human tester can do better by guessing at the purpose of an image or sound, but it still only a guess and may reflect the individual tester’s bias. My blog always includes an image whose purpose is to capture the reader’s attention, be a little humorous, and perhaps provide a non-verbal counterpoint to the text. Sometimes and for some readers there is no doubt the image fails. For one human tester the image might be best described as “funny picture of an ostrich” while for another it would be described as “another stupid waste of pixels.” Who is to say which description meets Success Criteria 1.1.1 when neither my purpose nor the reaction of a disabled user can be known.

The problem with “purpose” and alt-tags is only one aspect of the problem. WCAG Principle One is that content be “preceptible.” An object is perceptible if I can see it even if I cannot feel or smell it. It is perceptible if I can smell but not see or touch it. It is perceptible if I can hear but not touch, smell or see it. But is it perceptible if I cannot see, hear, smell, taste or touch it and must instead rely on a verbal description? Have I perceived a rose if I am told there is before me a fragrant red flower with tender petals? Some poets would say yes, but even in poetry verbal description is intended to recall the emotional or intellectual response from an earlier encounter with the real thing. It is not a substitute for that encounter. If I have never seen a rose the word “rose” cannot convey the same information that even a picture of a rose would convey.

At a deeper level rules about making content “perceptible” cannot capture all the information present in  an integrated web page. Because web page design is largely visual the spatial relationship of words and pictures is important, if not critical, to the function of the page itself. Those spatial relationships are only partly conveyed by methods of organizing the text equivalent of a web page, meaning that for even the most “accessible” web page the disabled user will be missing something.

These flaws do not mean the effort at creating regulatory standards is not worthwhile. The effort to comply with WCAG 2.x at some success level is bound to increase the accessibility of the internet to those with disabilities. It does mean though that the solution to the current website litigation crisis will not be found in standards like WCAG 2.x, whether or not they are incorporated into regulations. The way modern websites blend verbal and non-verbal content and the impossibilty of any clear distinction between a website as a “place” and a website as a repository of information or an experience makes website accessibility an inherently subjective concept. Even if WCAG 2.x at success level AA were established as the regulatory standard any plaintiff could claim the website failed to meet those standards because there was some kind of information that was not fully conveyed by whatever words were available to a screen reader or included in a caption.º  The current litigation industry exists because of expense and uncertainty, neither of which would be eliminated by the adoption of WCAG 2.x as a regulatory standard.

Is there a regulatory solution? Yes, but it will require a complete re-thinking of what kind of equality is appropriate for disabled users of the internet. As with the standards for physical access it will require compromise – a person with a disability will never have full access to a website that includes non-verbal information because the full informational content of a picture or sound cannot be put into words. Describing just what the ADA permits and requires for translating non-verbal to verbal information in an objective way is going to be a challenge, but it must be undertaken if we are ever going to see a slowdown in lawsuits over web accessibility.

* See, What is an ADA accessible website?

** See, e.g. WCAG 2.1 success criteria 2.4.x

† See, e.g. WCAG 2.1 success criteria 1.2.x

‡ I’m using verbal in the strict sense of consisting of words rather than its sometimes use as a synonym for oral.

º As noted above this problem is avoided for movies by pushing responsibility to the creators of the content who are not themselves ADA public accommodations. If the closed captions or audio description of a movie are not adequate the movie theatre owner cannot be held responsible. This same excuse may apply to third party content on a web site, but for the website itself the excuse is not available.