Re: An Open Letter to Oracle on the Topic Of Accessibility

[Jason White <jason jasonjgw net>]

Kenny Hitt <kenny hittsjunk net> wrote:
> Hi.  In my view, Gnome accessibility will never succeed.  It will sometimes
> get close, but will never make it all the way.  I came to this conclusion
> after using Gnome versions from2.2 through 2.28.  I've seen several
> occasions where a change in Gnome broke accessibility.  Combine that with
> the fact apps aren't required to be accessible before being included in
> Gnome, and you have a recipe for failure.

I actually think Gnome accessibility efforts are remarkably successful, given
the limited resources available and the nature of the problems that need to be
solved.

The role of operating system distributions in deciding what to package, and
which applications to offer as defaults, should also be borne in mind. The
decision process doesn't have the same top-down control structure that a
proprietary operating system vendor can exercise. The decentralized community
is part of what makes free software great, notwithstanding the difficulties
that can arise when negotiations are underway in regard to resources,
priorities, what to include in a release, etc.

I think the accessibility situation would be considerably better if user
interfaces were specified in a much more abstract way, and then styled
according to the needs and capabilities of the user, and of the input and
output devices. Obviously, doing this properly would involve designing an
entire UI infrastructure from the ground up. Until that happens, I suspect,
accessibility will remain to some extent an "add-on", and therefore
vulnerable. Another way of saying this is that the more application developers
have to think about "accessibility" as a discrete, separate phenomenon that
needs to be taken into account, the more accessibility is likely to lose,
despite constant "education" efforts and repair strategies to deal with the
deluge of regressions.

To be specific, I think the user interface component of XForms, and the
somewhat similar Universal Remote Console specification, are indicative of
what is possible, and that's only the beginning. Web interfaces might also
help here, depending on the extent to which the necessary semantics are built
into the core infrastructure of the Web; there's a danger that add-ons such as
Aria will become entrenched as long-term solutions rather than as interim
measures to be phased out once the underlying formats become sufficiently
expressive.

As to me, I'm staying in my Emacs and console sessions as much as possible,
but I'm glad Gnome accessibility is here when I need it - and I really do need
it, in some circumstances, especially, at the moment, for the Web. My main
reason for using Gnome only as needed is not so much a lack of accessibility,
but rather that I don't like the WIMP (Windows/Icons/Menus/Pointers) paradigm
and the type of user interface to which it gives rise. This is not to say that
I reject graphical user interfaces - in fact, there are interfaces that make
full use of a graphical display without imposing the limitations of your
typical WIMP-style interface - but I find user interfaces which are based
fundamentally on a point-and-select model rather than a language/command-based
interaction far too restrictive. Text editors are a superb example of this;
compare Emacs (or Vi, if you prefer) with a typical GUI/WIMP-based editor, for
instance. (I'm well aware that both Emacs and Vi can run as full X
applications, which shows why it would be wrong to characterize my point of
view as critical of graphical interfaces as such.)

As a final thought regarding where this is all headed, there remain open
questions as to the future role of Web applications. "Firefox wants to be
Emacs", as pointed out in this post: http://tjic.com/?p=8916
and I think chromium is headed even more strongly in that direction. If user
interfaces, including the free software/open-source community, are headed
toward an era in which a descendant of what we today call a "Web browser"
becomes the desktop, then what would make most sense is a Javascript-based
"assistive technology" (for want of a better term, that can process
user interfaces built on Web technologies, and which is fully integrated into
the extensible browsing environment.

So that's my perspective, for what it's worth.