Re: Task Approach to conformance.

Interesting ideas - see comments below.
On Sun, Oct 22, 2006 at 06:18:19PM -0500, Gregg Vanderheiden wrote:
> In looking at a web page one person could view the whole page as being used
> to accomplish a task.  Others could view it as being a collection of many
> tasks.  In fact in those cases it probably would come down to being a task
> nested in tasks nested in tasks.

Also, tasks might span "pages". This gives rise to the problem of what is in
the scope of a conformance claim. Suppose I am reviewing a nested list of
categories of books that will eventually lead me to the title and author of
the book I want. Depending on my delivery context (desktop machine, mobile
telephone, voice browser, etc.), how this information is spread across URIs
will, let us suppose, vary.

If we think in terms of the "book identification task", then we could argue
that there is a certain starting point, identified by a URI, from which all
tasks are available through user interface elements defined within the content
itself. Links, user interface controls, etc., are all user interface elements
for this purpose. I think it is important to restrict ourselves to tasks
achievable via user interface elements which are part of the content, since it
is only these over which the content creator has control - so they could serve
as the basis for defining the scope of tasks. Thus, for example, a task
involving a search engine wouldn't qualify since it requires the use of either
user agent functionality or Web content (e.g., the interface of the search
tool) that goes beyond what can be reached via user interface
controlssuppliedas part of, or derived from, the content itself.

Suppose furthermore that once the bibliographical record of the book has been
retrieved, the user is given the option to borrow it from a library. If the
guidelines require that all tasks meet the minimum conformance level, then the
borrowing task should certainly qualify since it can be carried out from any
of a number of URIs that lie within the scope of conformance. Assume for the
sake of the example, however, that the borrowing task is actually not the
responsibility of the author of the book Web site - it simply involves links
from bibliographical entries to a library site that handles the borrowing
operation from that point on.

The maintainer of the book search Web site must be able to exclude the
borrowing operation from the scope of conformance, since she or he has no
control whatsoever over the independent entity that is the library. If this is
right, then we could say that all tasks which can be performed via the
rendering of content, or the operation of user interface controls obtainable
solely by accessing URIs that match a pattern specified in the conformance
claim, must be accessible according to the minimum level defined in the
guidelines.

However, this lets in Gregg's favourite example of the shopping site in which
everything but the order completion page conforms. I can't think of a way to
exclude this without opening up the content developer's responsibility to
everything that is linked from her/his own Web content.

Let's summarize the conclusions from the above.

We could think in terms not of tasks, but of content rendering and user
interface interaction. All content rendering operations, and interactions with
user interface controls created as part of the content accessible directly or
indirectly from a given URI, must conform to the guidelines at the specified
level, provided that such rendering and interactions can be carried out solely
by dereferencing URIs matching a pattern given in the conformance claim. This
gives us the basic scope of conformance - a "starting URI" and an optional
pattern covering other URIs that may be accessed while rendering content and
performing user interface actions within the scope of conformance.

The pattern serves to restrict the scope of the conformance claim (in fact, as
in the current WCAG draft, we could allow for multiple patterns).

But how should we handle Gregg's shopping site without preventing my book Web
site from accessing the borrowing services of a library? Here is one
suggestion. The difference between the two examples is that the legal entity
(the corporation) responsible for the shopping site has control over the order
completion page, whereas the owner of my hypothetical book site doesn't have
any legal relationship with the library - I'm just making use of their
services, for example, via published Web APIs that the library offers.

I propose the following requirement: a URI pattern cannot be so designed as to
include only part of a task which it is clear from the content included in the
scope of the conformance claim was contemplated by the developer of that
content, in circumstances where the said developer would have been able to
exercise control over the design of the excluded portion of the content,
whether directly or through a legal relationship with the creator of the
latter content. Here, "developer" perhaps isn't the right word - I really mean
to refer to the legal entity, whether individual or corporate, that controls
the creation andor maintenance of the content.

Received on Monday, 23 October 2006 08:51:05 UTC