Weblogs: Web Accessibility

SiteMorse gets nasty, accessibility is the victim

Monday, January 24, 2005

The progress towards web accessibility took another step backwards with SiteMorse's publication of their report into the website performance of Accessibility compliance and Service Provision companies, published on the 19th January 2005. The press-release took a swipe at the DRC's public criticisms of SiteMorse testing, and responded by trying to humiliate the DRC's accessibility efforts.

The Report

SiteMorse uses automated testing tools for their tests - checking for HTML compliance, accessibility tests they believe are automatically testable, server response and download times. The aggregation of this data is used to create a league table for each site tested.

In small text on page 3 of the report SiteMorse concede the obvious fact that automated testing does not indicate overall accessibility or eGMS compliance - manual testing is still required.

A web design company called Red Ant tops the league table; and fhios - a respected usability consultancy - is last. While disability organisations - National Library for the Blind, Royal National Institute for the Blind, and the Disability Rights Commission - all form the bottom half of the table. Nomensa proves its mettle by being the other company to pass all automated accessibility tests.

The Press Release

SiteMorse's press release introduction fires its first salvoes at the Disability Rights Commission, and alleges that the DRC's response to this report adds confusion to an already confused market. A number of querulous finger-pointing exercises unfold:

The DRC, RNIB and RNID are the supposed standard bearers for website accessibility, [but] continue to fail even the most basic A/AA requirements.
SiteMorse seem to take exception to Bert Massie's published reply in the Guardian newspaper where he explains why the automated accessibility tests are flawed, by labelling the published letter as "sharp".
SiteMorse reacted by finger-pointing a web page on the DRC website containing an open letter from Bob Niven, the DRC's chief executive, pointing out a missing alternative text on an image. (Notwithstanding that the equivalent content was already displayed in plain html)
SiteMorse seem to take exception to Patrick Edward's (Head of Media of the DRC) comment that usability testing through human interaction over SiteMorse's automated testing procedures, by labelling the reply as "hostile".

The tone of the news release is emotional and personal. Understandably since disability organisations have been critical of automated testing, which is SiteMorse's core offering. As a business SiteMorse is vulnerable, and any valid criticism of their product or service is a crisis of confidence in the business.

Red Ant - SiteMorse's client and supplier

SiteMorse tries to defend its automated process by claiming they received client feedback that shows its clients have made significant improvements to their sites since using [SiteMorse's tools].

One of their clients is Red Ant webdesign. Also Red Ant created the currently inaccessible SiteMorse website. Red Ant offer SiteMorse and Bobby validation as part of their services. And interestingly enough, they top SiteMorse's league table.

Interestingly, Red Ant is the only web design agency on this league table. No mention of Fortune Cookie - who have been awarded more RNIB "See It Right" accreditations than any other web design agencies. No mention of Webcredible - a well established accessibility consultancy. No mention of the Guild of Accessible Web Designers which is a strongly UK-weighted accessibility organisation. How could one even overlook Accessify?. The choice of candidates is suspicious and smacks of opportunism.

Automated testing is flawed

There are three problems with automated testing methods:

They can't test subjective material, this leads to both false-positives and false-negatives
Automated testing is only as good as the written test cases, and the knowledge of the product author - this means there's no compensation for author errors. It also cannot learn - this is where practical usability testing is miles ahead of automated testing
Automated tests can, and are, gamed.

Red Ant gaming the SiteMorse checker

The gaming of automated accessibility testing is a major factor for Red Ant being on top of this months league table. From their home page comes this following howler:


<a href="#top" id="hidemenuxyz"
 onmouseover="javasript:hideAll();"
 onfocus="javascript:hideAll()"
 title="Back to top">
 <img
  src="http://www.redantdesign.com/
       _images/shim.gif"
  alt="this image is link is purely
   for layout"
  width="100%" height="1" />
 <span style="display: none;">
 this text is link is purely
   for layout
 </span>
</a>

SiteMorse's report claims Red Ant's website was error free, so it must consider the above error-free. But consider the following:

The choice of alt text is a barrier to access. Who in their right mind would draw the conclusion that "this image is link is purely for layout" could ever be considered equivalent alternative text to the content offered in an image called shim.gif?
The choice of link text is a barrier to access. Who in any mind would consider the link text of "this text is link is purely for layout" is by any means descriptive of an internal link pointing to a section with the id of top?
A screen reader would read out both pieces of text, giving the user a link of text "this image is link is purely for layout this text is link is purely for layout" - what a confusing mouthful!

This is a serious accessibility problem. One that outweighs the "problems" recorded on the RNIB and DRC websites. Don't even get me started about javascript: in event handlers.

Automated checkers are limited by author knowledge

SiteMorse make an interesting statement about their HTML requirements testing: [HTML standards compliance with the requirements laid down by the W3C and IETF.

Yet the SiteMorse tool also fails to pick up a major technical problem with Red Ant's website. This failure is solely down to the SiteMorse tool developer's lack of technical knowledge over XHTML. XHTML1.1 should not be delivered as text/html - Red Ant makes this mistake, even in the presence of an XHTML1.1 supporting user-agent. This certainly does not qualify a site as being error free.

Technical vs practical accessibility

There is a problem with web accessibility at the moment - its a perception problem. SiteMorse is now public example of this problem. Web accessibility is about giving people access to content, its about removing barriers they themselves can't remove. Looking at SiteMorse and Red Ant's websites and statements there's clearly some technical knowledge about accessibility, but about real people and real problems there is nothing.

Creating accessible websites is about taking the time to think through repercussions of content, thinking about how content on a page will read and sound. SiteMorse and Red Ant evidence no practical knowledge about what is or isn't difficult for users of assistive technologies. They've boiled down accessibility into a series of monkey-see monkey-do checklists - devoid of any real thought. Its almost as if they don't understand accessibility, and this is the nearest they can simulate.

There are too many accessibility experts who insist that meeting technical criteria is absolutely essential for content to be accessible. This is a myth, there is no substance to it at all. Its evident that a number of blind people have been using the World Wide Web from the beginning - even through the Netscape / Microsoft browser wars when there were practically no valid websites. Back when the number of valid sites were less than a tenth of a percent, people using screen readers were using the web almost everyday. It was difficult, but not impossible.

It is possible to meet basic accessibility levels using invalid markup and tables for layout. It is equally possible creating inaccessible swamps with valid HTML and CSS. Web standards is a means of future proofing the accessibility of content, but it currently offers no supportable benefit for pages out in the real world today. An almost valid page is just as accessible to a screen reader as a perfectly valid page. Even more so, a well structured HTML document is far more accessible than a perfectly valid but structureless page full of divs and spans.

Web accessibility is a simple concept. By obscuring it with insistence of technical perfection pushes widespread accessibility out of the hands of the ordinary web page author and into a "skilled proficiency".

Conclusion

SiteMorse have disgraced themselves with this childish outburst. By publicly "promoting" a company they've dealt with as both a client and a supplier, and taking the same opportunity to denigrate detractors creates the impression that only highly-paid technical companies can create accessible websites. And that is a whole load of crock.