Re: Proposal for Guideline 2 as well as a proposal to trim WCAG 2.0 to 3 guidelines (won't william be glad?)

My response to Wendy's proposal is somewhat complex and cannot be stated
within the terms of the provided questionnaire, which, with regret, I must
therefore decline to answer directly. Before proceeding to give a
substantive discussion of the issues, I should emphasize, from the
beginning, that Wendy's analysis is revealing, thought-provoking, and most
welcome. My primary concern is that the categorizations, notably in
guidelines 1 and 2, are somewhat arbitrary and that the checkpoints do not
fit neatly under the guidelines (except those of guideline 3, which are
somewhat better in this respect).

Of course, there is substantial scope for legitimate disagreement as to
what division of the subject into basic, general propositions best
captures the essential constituents of universal or accessible design, and
can be most effectively employed as a conceptual aid to the understanding
and application of the more detailed requirements which are subsumed under
these broad statements. As has been remarked on enumerable occasions, we
need to arrive at a balance between generality in the formulation of the
high-level guidelines, and concreteness in specifying, albeit still in
quite broad terms, what, conceptually, are the elements of accessible
design, of which the document as a whole is, fundamentally, an
elaboration. With these qualifications in mind, I shall venture to advance
an opinion, for what it is worth, regarding the specifics of this
proposal.

Guideline 1: The concept of "graceful transformation" is here extended
well beyond its conventional meaning. The principal argument for retaining
the term "graceful transformation" was that it had a well understood
technical meaning, as evinced by WCAG 1.0, according to which it referred
to the capacity of content to remain usable across a diverse range of
older and newer software (particularly user agent) implementations. In the
current proposal it is broadened to such a degree as to become a
"catch-all" concept that encompasses a multiplicity of disparate
requirements, the logical connections among which are far from obvious.
These include modality-independence (text equivalents and their
concomitant synchronization requirements), compatibility with user agents,
including older technologies; and the use of style languages to control
layout and presentation. The second of these covers what has traditionally
been meant by "graceful transformation", and I find it hard to discern any
clear reason for connecting it, under a single heading, with either text
equivalents or style languages. One could of course extend the term
"graceful transformation" to include the other two concepts, but if one
were to do so, then why should one not go further and declare that
device-independence (which is what style sheets are, in part, intended to
achieve) is really part of the concept of graceful transformation, and
that everything subsumed under the proposed guideline 2 should actually be
brought within guideline 1? Stated differently, one can either limit the
concept of "graceful transofrmation" to its conventionally understood
meaning, or broaden it, in which case I do not see any non-arbitrary
reason why it should refer only to interoperability and compatibility,
text equivalents/modality-independence, and style languages (separation of
content and structure from presentation).

Guideline 2: In effect, the same remarks apply here as well: the concept
of "device independence" is entirely capable of embracing most of the
checkpoints under guideline 1, as well as those which have been proposed
as guideline 2. Thus, within the class of possible output devices, there
are those which happen to be auditory, others which are visual, and still
others which are tactile. Within each of these categories there are
further subdivisions (small and large screens, high and low-resolution
displays, printers, braille embossers, dynamic braille displays, and
various kinds of speech synthesizer). It seems somewhat arbitrary to
separate out modality-independence, placing it under the heading of
"graceful transformation", along with the use of style languages, whereas
the proper representation of structure and semantics, and a range of
further requirements connected with navigation, distracting or
time-sensitive content, etc., are collected under the rubric of
device-independence. In fact, as Lisa has pointed out, the issue of
time-sensitive and distracting content is at least as much of a cognitive
matter as it is a question of device-independence, and thus it fits
uneasily at best into the latter category. More importantly, as I have
already intimated, one can regard modality-independence and style
languages as matters of device-independence, thereby collapsing most of
guideline 1 into guideline 2, with user agent compatibility and
interoperability being the only requirements under guideline 1 that
arguably wouldn't fit (unless one decides somewhat arbitrarily that each
user agent should be treated as a separate "device", in which case the
whole of guideline 1 becomes part of guideline 2).

Thus, the upshot of my analysis is, first, that the general concepts of
"graceful transformation" and "device independence", as used here,
threaten to collapse into one another. If the concept of "graceful
transformation" is extended, as proposed, then the distinction between it
and device-independence becomes arbitrary. One could just as easily use
either concept as a guideline under which to place many of the checkpoints
proposed for guidelines 1 and 2. As I have also argued, some of the
requirements, especially the avoidance of distracting content, movement
etc., are a matter of cognition rather than graceful transformation or
device-independence. One could extent the proposed guideline 3 to include
these phenomena (ensure that content is easy to comprehend, and that it
does not distract or interfere with the reader's ability to navigate, for
example). Next, one could broaden the notion of device-independence, or of
graceful transformation, to cover most of the other checkpoints. The
result would be highly general but, in my opinion at least, hardly
insightful, and certainly it would not be easy to remember as a basic set
of underlying requirements. Nor is Wendy's proposal satisfactory, for the
reasons which I have indicated, and I would not wish it to form the basis
of the next draft. What is needed, rather, is a scheme in which the
checkpoints fit logically under the guidelines, in which there is a clear
connection between the basic principles and the more specific requirements
subsumed thereunder.

Of course, it might be maintained that any division into guidelines and
checkpoints will inevitably be, to some extent, arbitrary. This is
probably true, but some possible divisions are obviously better than
others. I would have liked to present here an alternative scheme, but I
need to give the matter further thought. I think the existing set of five
guidelines (from the latest draft) is quite good: (1)
modality-independence; (2) separation of content and structure from
presentation, together with proper exposure of semantics; (3) facilitation
of comprehension; (4) facilitation of browsing and navigation; and (5)
device-independent handling of user input. One could perhaps argue for:
(1) device-independence, covering text equivalents, separation of content
and structure from presentation, and exposure of structure and semantics;
(2) browsing and navigation, covering site maps, linking, summaries,
search facilities etc.; (3) comprehension and cognition, encompassing
Wendy's proposed guideline 3 as well as the avoidance of distracting
content, blinking/flickering content etc., given their perceptually-based
rationale; and (4) compatibility and interoperability, covering
standard-compliance and compatibility with older technologies.

Received on Saturday, 6 January 2001 23:02:17 UTC