Accessible shopping

Back in the day, we covered, to the exent possible, the findings of a Royal National Institute for the Blind study on accessible E-commerce. To the surprise of no one working in the Web-accessibility demimonde, sexy E-commerce Web sites were found to be slightly to very inaccessible.

The problem? You had to pay an inconveniently low fee to read the report (£5), which the esteemed Institute refused to post online. However, we now have a kind of summary available. You won’t find the full report online because, says an RNIB functionary, “the policy of RNIB’s Web team is not to HTML large documents” even though the file is a measly 68 K in length. This of course says nothing of the use of to HTML as a verb.

RNIB also won’t provide a PDF version “because of accessibility issues,” ignoring the facts that simple PDFs can be easily handled by screen readers using Acrobat 5 and that specifically-accessible PDF authoring is now quite possible by direct export from Word 2000. The only format available is indeed Microsoft Word, which is itself questionably accessible.

In any event, the report concludes:

RNIB believes there should be a legal requirement for Web services to be accessible to disabled people within the spirit of the Disability Discrimination Act 1995 (DDA).

Martin Sloan has demonstrated that the Disability Discrimination Act does in fact apply to Web sites.

As many Web sites are set up in the USA, another important development in the legal case is provided by Section 508 of the Americans with Disabilities Act (ADA). This Act mandates that all federal electronic and information purchases made after 7 August 2000 must be made accessible to people with disabilities.

Section 508 has nothing to do with the Americans with Disabilities Act, and 7 August 2000 was not the initiation date (that was 21 June 2001).

Findings, you ask? Findings?

The results were extremely disappointing. All companies in the study failed to meet the needs of blind and partially-sighted people. No company achieved a 100% pass rate against the five criteria. Some Web sites even failed every aspect of the test.

The criteria were:

  1. Text legibility
  2. Presence of alt texts
  3. Presence of noframes content
  4. “Are navigation links easy to use?” (a highly subjective issue without pass/fail criteria)
  5. Passing Bobby

The report tells us, on the one hand –

RNIB recommends use of the Web Accessibility Initiative’s Web Content Accessibility Guidelines, a set of standards which ensures Web sites are accessible for all customers.... While this report does not explore all of the technicalities of Web design in any depth, designers will find sufficient guidance and references to set themselves on the right path to accessible design.

– but fails, on the other hand, to actually apply the WCAG as testing criteria. The worst failing is reliance on Bobby, a testing program that’s as dumb as a mule and only half as reliable. Bobby will frequently complain that even highly accessible sites actually are not – spitting out a worrywart, bleeding-heart litany of errors even though it cannot tell whether or not you have actually made those errors.

Bobby is too primitive and unreliable and has a tendency to give false-negative results, flunking absolutely everything but the simplest text-only sites. (In their heart of hearts, oldschool accessibility advocates truly believe the only just, proper, and noble sites are simple and text-only. They rue the day graphics munged it all up.)

Even writing a complex sentence triggers a warning from Bobby. The NUblog (e.g., a typical recent article) is perfectly accessible but elicits warnings on that count:

The following 4 item(s) are not triggered by any specific feature on your page, but are still important for accessibility and are required for Bobby Approved status....

  1. Use the simplest and most straightforward language that is possible [Is “Shove it, Bobby!” simple enough? – Ed.].
  2. If ASCII art is present [It ain’t – Ed.], consider substituting it with an accessible image.
  3. Identify any changes in the document’s language [There are none – Ed.].
  4. If you can’t make a page accessible [We can and did – Ed.], construct an alternate accessible version.

We do not think it is excessive to state that the RNIB’s survey results must be called into doubt because they derive from an organization afraid to publish Web documents for specious reasons and one, moreover, that uses questionable or indeed irrelevant criteria to judge Web accessibility.

It almost certainly remains true that the businesses surveyed by RNIB do indeed have slightly or completely inaccessible Web sites. It is all well and good if this report, despite its inaccuracies and its ill-conceived research basis, prods E-commerce sites to fix accessibility defects. It remains true that assessing Web accessibility is highly contingent, relies on a great deal of impressionistic human evaluation, and cannot be reduced to a form of automated sloganeering. RNIB’s survey seems to have been devised to elicit bad press and to shame E-commerce sites into cleaning up their act.

The conclusions are probably right even though the reasons probably are not. Ends hardly justify means here.

Posted on 2001-10-15