Peter Korn's Weblog

Chris Hofstadter, formerly of Freedom Scientific (and of course Henter-Joyce before that) has started blogging. In his recent entry titled "When Capitalism Fails" Chris observes that the fruits of competition have not come as much as one would like in the AT space, especially with things like hardware in Europe paid for by gov't. He further notes that when it comes to getting AT to work with mainstream software, AT vendors are tiny when compared to many of those larger mainstream companies (who would like to at least go somewhat beyond a 508 assessment and into real end-user productivity when used with Windows AT), and so really the right thing is for mainstream companies to fund the AT adaptation work. He says this was a hoped-for outcome from an Assistive Technology Industry Association "AT/IT" committee that failed to come to pass:

The Assistive Technology Industry Association formed a committee to bring the AT businesses together with the mainstream companies so as to form some ground rules for making products compliant with the spirit of 508. One of the first items to which all companies agreed was that the expense of development should be paid by the billion dollar mainstream companies and not by the smaller and less flexible AT businesses.

Unfortunately, this plan never came to fruition. The big businesses would approach the AT companies separately and offer all sorts of intangibles (co-marketing, publicity, etc.) and make claims like “we have more than 20,000,000 installed units of our product and a small company like yours can leverage that…” Of course, as the software was not yet accessible, zero of these millions of users were also customers of an AT company and the likelihood of many blind people getting Federal jobs that require this specific software was low and would represent few screen reader sales. So, in order to maintain their huge Federal contracts by coming into compliance with Section 508, enormous software companies would try to muscle the relatively tiny AT businesses into doing all of the work and receiving little or no benefit.

If the AT vendors stuck together as agreed in the AT/IT committee, the onus would have fallen on the big software manufacturers and the AT companies would have been paid consulting dollars to make changes to the screen readers and the big software developers would have had to modify their products to make them comply with the standards that screen readers rely upon to deliver information to their users.

So, what went wrong?

He goes on to answer his question by saying essential that individual AT vendors broke ranks (especially the smaller ones) and did work not for pay, then failed to do a great job, and in the end you had mainstream companies claiming they worked well and went beyond Section 508 when in fact the user never really got served well.

While I don't doubt that this is how things happened, I disagree with where Chris places the blame. I believe the fundamental problem isn't that AT vendors failed to stick together in demanding money (er, consulting) from mainstream software vendors (some of which, by the way, aren't all that big either). The fundamental problem is that the platform vendor - Microsoft in this case - failed to define an accessibility architecture that did the job needed (as Chris himself notes in a later blog posting titled " Screen Readers and Contextual Information"). Without a clear and pretty complete framework for describing programatically all of the information needed by AT, we are left in a world where any interesting mainstream application needs a lot of customization and specialized adaptation built into every assistive technology in order to work well for users with disabilities.

This is fundamentally untenable. Even with larger companies - like IBM and Adobe and Oracle and AOL (and even Sun), all of which he praises as examples of good companies who do a lot to work with AT vendors - it is simply untenable to work with, track, fund, etc. adaptation and customization with every AT on every platform for use with every software application they create. The necessary investment is large, the overhead (for tracking and testing, if nothing else) unwieldy, and the model fundamentally broken. Chris sung IBM's praises for their work on Firefox accessibility. Rightly so; but that accessibility is visible to Chris because it works with JAWS and WindowEyes (and perhaps soon with ZoomText). But that's 3 Windows AT products out of dozens (and only for visual disabilities, and only on one platform so far). Multiply that by multiple disabilities, and dozens of programs, and add Macintosh and UNIX desktops, and the numbers get untenable. And so instead of the scenario Chris outlined (small Windows AT vendor does some work "pro-bono" for some software application), you have the more likely scenario that the only the bigger AT vendors get consulting dollars for adaptation - which in turn cements their market position and further serves to stifle innovation.

The fundamental problem is the lack of a real accessibility programming contract that would dramatically minimize the amount of custom adaptation needed for interesting applications. This is precisely what we are continuing to build in the GNOME Accessibility Project (and what Apple is continuing to build in Mac OSX, and what Microsoft says will come with UI Automation). While I don't think we'll ever be able to completely do away with all such adaptation work, we should be able to greatly offload the burden on AT vendors. In fact, we're already finding this to be the case with scripting in Orca: the adaptation work for the GNOME instant messenger application was completed in less than a day - and our user expert (and former JAWS script writing department manager) Mike Pedersen said that combined with Orca, GAIM is the best instant messenger interface he's ever used. And in a conversation I had with Mike earlier today he noted that the Orca script for the Evolution e-mail and calendaring application is steadily and rapidly improving, and he is using the combination already for e-mail.

Once we have real richness and fidelity in accessibility frameworks implemented in our desktops and user interface libraries and toolkits, having "more than 20,000,000 installed units" of a product will be more than enough reason for "pro-bono" AT adaptation work - because the AT adaptation work will be a much smaller job, and needed in a lot fewer places. (2006-02-06 23:34:30.0) Permalink


	Peter Korn's Weblog The collected occasional commentary by Peter Korn, Accessibility Architect at Sun Microsystems, Inc.

Some thoughts on Chris' thoughts on capitalism and the AT marketplace