Re: principles

At 08:40 AM 2000-08-16 -0700, William Loughborough wrote:
>Principle 1: All information must be available entirely through visual,
>auditory, or tactile methods or in any combination required by the user.


Note that the use of the words 'all' and 'entirely' in the above are in
conflict with the principles of reasonable accomodation and effective
communication.  Besides, modifiers detract.  Strike these words and it is a
powerful statement.

The use of "visual, auditory, or tactile" as defining the space of sensory
options is technology-specific.  There is widely available technology in
these areas, but the general principle includes other sensory modes
including haptic, vestibular, and thermal sensations, along with direct
electrical stimulation of neurons.  Targeting all the above three senses is
a strategy, not principle (in principle).

A problem is that I don't know how to state the general principle in the
positive.  It takes a double negative, to the best of my ability to
articulate it.  It is that "Comprehending the information shall not depend
on something the user can't do."  From this we derive the strategy stated a
Principle #1.  And then add a parallel strategy addressing cognitive barriers.
Note that "or in any combination required by the user" is unnecessarily
restrictive and not a valid part of the requirement as stated.  If the
information is completely comprehensible in the user's choice among
single-sense presentations, the requirement has been met.  The inability to
provide perfect equivalents makes it easier in practical terms for the
information source to deliver the information in two out of three senses
than in just one.  So because of this supply-side cost factor, it is
natural that the graceful degradation sequence flows from "all" through
"two" on the way to "one" of the above sensory channels.  Because at each
step you get less 'graceful' and more 'degradation,' in most actual

>Principle 2: Capture structure and implied meaning (semantics) in markup
>or in a data model.
>These are going to be somewhat "academic" in tone/content. I think that
>brevity is permissible so long as subsequent materials elucidate for
>those to whom this sounds "high-falutin'". I recommend leaving out
>"implied" and either "meaning" or "semantics". The "implied meaning"
>seems to apply (mostly?) to the uses of structure/presentation to convey

>semantics. "Content" probably doesn't have markup/datamodel requirements
>but structure/presentation does? Thus the "information" in Principle 1
>and the "semantics" in 2 must both be revealed in accordance with 1. So
>in effect Principle 1 might say that semantics, whether explicit as in
>content or implied through structure/presentation must be... The
>structure is "information" and the "presentation" might well be. The
>general/abstract case is that *all* semantic content must be available.
>So in a way Principle 2 would be redundant if Principle 1 were somewhat


Yes, for the right definition of 'information' the second principle is a
consequence of the first.

However, it doesn't mean that we don't need to break the master principle
down into two facets:  how to meet this requirement as it affects small
chunks of information, and how to meet the requirements as it affects
structure, i.e. the content or semantics that are disclosed through the
association of chunks with other chunks.
Technical note:  The satisfaction of Principle #1 does in practice generate
requirements of a "markup or data model" nature.  The classic example of
this is text alternatives for image data.  Here there is a structure of
equivalence relationship.  The markup or data model has to communicate what
text is equivalent to what image.  Given that raw sound and image data can
convey information, and that some of this information can be captured in
textual 'equivalents,' there is a built-in requirement that there be at
least enough structural model to capture the equivalence relationships.

Note 2:  The peephole effect is the clearest image I know of to explain why
the structure relating different mini- or micro-chunks of information needs
to be articulated well.  [For 'articulated' read "spelled out."]

The data/information chunk that has focus, i.e. it is in display and
occupies the user's foreground attention, is contextualized in the GUI
usage pattern by bits on the screen and in the eyes-free scenario by
concepts remembered by the user.  The two main questions that this context
answers are "where am I?" and "what is this all about?"  The structure
(information atom connections) has to be comprehensible in both of these
alternative views.  In order to support both views with low rates of error
or information loss, it is [very helpful to practically necessary] to make
the patterns of connection explicit or machine-interpretable so that the
morphing can be carried out correctly by automatic processes executed by a

The concrete example to illustrate this is the cell-to-header
relationships in tables.  Effective table adaptation techniques make it
easy for the user to maintain memory of the header context as they read the
cell particulars.
>Principle 3: Design for ease of comprehension, browsing and navigation
>I'm having a lot of trouble with this one, particularly with the
>inclusion of "browsing" which seems to me to be a subset of
>"navigation"? Mainly, I think the directive to "design for ease..." is
>essentially impossible to yield agreed-upon methods/goals since "make it
>simple but not too simple" implies that for some of us "ease of
>comprehension" is hampered by simplisticism.


Principle 3 is also related to Principle 1.  There is an adage "the proof
of the pudding is in the eating."  Principle 3 is the eating that proves
the Principle 1 pudding.  You often cannot define "all the information"
statically to a sufficient degree, so this operational check is a required
piece of the solution.

Browsing differs from navigation in that it includes absorbing some of the
base content, marbled through the motion and vice versa.  Nibble, move,
nibble is the browsing cycle.  Each part informs the other.  You don't know
which link to follow next until you have absorbed some of the actual
content that frames the links in this page.  And you can't read what you
need to read next without exercising the path-control option of following a

As we have learned from "click here" the labels on the hyperlinks are not
sufficient out of context.  This is true not only for this extreme example.

>OTOH I think ease of navigation is indeed quite central to these
>Principle 4: Design user interfaces for device independence
>I believe this will be the saving grace of our goal of Universal Design
>*if we can provide acceptable/practical/usable guidance*. Perhaps
>Bristol will help clarify if we have any chance to continue pursuing
>this goal. If stuff like WAP captures a handhold on handhelds it may be
>difficult to preclude multiple proprietary Webs in lieu of achieving
>accessibility via Principle 4.


Insensitivity to the display horizon (see above talking about explicit
structure and peephole effects) is what the mobile device needs for the
content to be equally appealing as in the large-display context.

On the other hand, if the publishers prepare alternate presentations for
desktop and mobile displays, we at least get to pick the one that works on
a consumer by consumer basis.  The eyes-free consumer subscribes to the
mobile web; the verbally-challenged consumer to the playstation.

Even if the retailing of content goes to multiple forms (c.f. Radio, TV,
Magazines, Newprint) there will still be demand from the information
sources for a common mode at the OEM or B2B interface.  This is why one of
the first XML applications was for newswire stories.  The civil right we
seek is addressed if individuals can opt to tap into the universal source
form as an option to the targeted presentation.

>Principle 5: Compensate for older technologies and missing or
>incompletely implemented features of user agents

>I guess I still think this doesn't rise to the level of a "Principle"
>since in our ivory tower older technologies are superceded technologies,
>failings of user agents are temporary and we don't want to modify the
>Principles just because the telephone operators no longer must insert
>plugs into some spot on a bank of receptacles or turn a hand crank to
>ring one's bell.

This is implied by #4.  The 'device' seen by the page includes the
client-side software.

>This is fun!

Received on Wednesday, 16 August 2000 14:10:56 UTC