Linear reading order

Summary: there should be no specific requirement at level 1 for a
logical, linear reading order because (a) it is not clear how this
would apply to all types of content, and (b) it would be redundant due
to the guideline 1.3 requirement that structure be provided in markup
or a data model.

To justify the first of these points, consider a map represented as an
SVG graphic in which there are sufficient descriptions and metadata
provided to enable text-based browsing. There are many possible
approaches to exploring the map, none of which is necessarily more
logical than any other, except by reference to the reader's purpose in
reading the map.

As another example, consider a user interface composed of three basic
structures: a navigational menu, a table of data and a form with input
fields. All six possible reading orders are equally valid, assuming
that the user wants to read all three components, which usually won't
be the case; preference among these depends on the reader's purpose.

Turning now to the second point, if the structure is provided in a
machine readable form then downstream software, such as a user agent,
assistive technology or server-side adaptation mechanism, can expose
the structure to the user, who can then choose an appropriate order in
which to read the content, and, just as importantly, can decide which
portions of the content to read and which should be skipped.
Checkpoint 1.3 already requires that structure be provided.

Thus I conclude that linear reading order should not be a requirement
at level 1, and I have significant doubts as to whether good sense can
be made of the concept in the context of inherently non-linear
structures where reading order depends entirely on the reader's
objectives.

Received on Tuesday, 6 July 2004 21:01:27 UTC