diversity in web UI design

At 08:23 AM 2000-11-01 +0200, Lisa Seeman wrote:
>It is better, because of users with multiple needs that are not all
>addressed any of the user group specific optimized interfaces.
>

AG::

Do we actually know ahead of time whether the unanticipated combination
will be
better served by bending the one solution that has to do for all; or one
out of
several, where one of the several may be optimized for a case that is
closer to
the person's actual situation than is the 'universal' page or site?  

Or are we are all guessing?

The status quo is most often that we get one version optimized for one
interaction situation.

A few people actually take pains to ensure that it transforms gracefully, but
more don't.

Compared to this, a range of selectable options which are optimized for
different situations could possibly be expected to sometimes have one option
among them that transforms more gracefully into your situation than did the
one
choice when you had only one choice.

It is true that in developing multiple targeted versions, the targeted
versions
may be done in narrower, more inflexible ways than would be one common
server-side version.  But that is not a given.  It is William's strong
suspicion, for example; and I have to say there is evidence for such
suspicion.  But it is not logically necessary.  The technical facts allow the
centering of an optimized version to be independent of the breadth of range of
flexibility.  

Part of my problem, in listening to this either/or discussion, is that in my
current working guess it would seem that the optimum site design is a hybrid
that uses some of both techniques.  That it is useful to teach designers both
how to serve the specific UI needs of specific cases and also to teach them
how
to generate flexible encodings of their designs that transform gracefully.  I
genuinely believe that a package where they do some of each is a) more
effective and b) more salable than a purely single-source,
automatic-transformation strategy.  If we try to pick which approach 'they'
will use, the world may indeed go on its merry way paying us some lip service
but little heed.  

On the "make sure there is a maximally robust safety net under everything"
side, I would advocate for a hierarchical table-of-contents style site
index in
text.  Always.  Just do it.  Automate it.  It is going to be the best way for
some people, for both sensory and cognitive reasons, to find stuff.  The
market
for this particular service doesn't go away.  This lets the user do a text
search on the titles with the 'find' function in their browser that they are
facile with already, not some peculiar search interface that works differently
from other sites.

On the "different strokes for different folks" side, I would cite databases
like transit schedules.  Here tabulated data (timetables) are effective for
visual access but a more interactive query-driven application would appear to
be more effective for people operating in audio.  The 'common source' for this
is a database that can't be expressed in web media today.  The best current
solution is a server-side application with different web-encoded views into
it.  Yes, in PF we are campaigning for XML Schemas so that more of the basic
model of the application can be encoded and shared in a Web-standard way.  But
that is not here and now.  People designing Web-delivered services have to
design something now with what is at hand now.

The evidence I have for the following is preliminary (anecdotal) but the
evidence does agree with my prejudices, so I tend to believe it:  "With the
best accessible site design and the best assistive technology, you can make
information retrieval eyes-free almost as usable as a voice portal designed
for
use in audio from the ground up."  Check out TellMe and BeVocal.

Another data point is that there is a graded sense of distance between
different interaction situations.  It is easy to make a reasonable WAP site
out
of a voice-enabled site.  It is not so easy to make a reasonable WAP site out
of an arbitrary Web design.

This is why it is plausible to me that "starting your transformation from
something closer" is actually a factor in having a better result.  Not the
only
factor.  But something that should not be ignored or dismissed out of hand.

[Oh dear -- yet more ideas]

You have to understand that even if the encoding is going to be in terms of a
single deep underlying content model and transformations that express the
common information in different views, I don't believe you can explain the
model to authors without showing them multiple views.  To get the
content-sourcing community to understand the model, and to be able to map
their
message into this framework, you pretty much have to confront them with the
divegent demands of different interface situations, and challenge them to do
the compare-and-contrast across what they would present as customer interfaces
in each.  And in the practical result, you are going to come up with a mix of
common source which can be transformed (as with text content today) and
separate data forming equivalent alternatives (as with captions vs. audio
tracks). 

As a technical matter, you can implement this with a common database server or
with separate datasets that use cross-links to identify equivalent
alternatives.  The experience delivered to the user can be exactly the same
either way.

Al

>L
>-----Original Message-----
>From: w3c-wai-gl-request@w3.org
[<mailto:w3c-wai-gl-request@w3.org%5DOn>mailto:w3c-wai-gl-request@w3.org]On
>Behalf Of Cynthia Shelly
>Sent: Wednesday, November 01, 2000 3:33 AM
>To: 'Ian Jacobs'; Leonard R. Kasday
>Cc: w3c-wai-gl@w3.org; jacobs@w3.org; kynn@idyllmtn.com;
>asgilman@iamdigex.net
>Subject: RE: General Exception for Essential Purpose
>
>
><quote>
> 2) It is better for designers and users to produce fewer sites that
>meet
>    the needs of more users.
></quote>
>
>Now you're treating designers the way many complain they treat users <grin>
>You're presuming to know their needs better than they do.  Why *restrict*
>designers to single interface?  Why not let them decide for themselves how
>much work they're willing to do to create optimized interfaces?
>
>-----Original Message-----
>From: Ian Jacobs [<mailto:ij@w3.org%5D>mailto:ij@w3.org]
>Sent: Sunday, October 29, 2000 10:56 AM
>To: Leonard R. Kasday
>Cc: w3c-wai-gl@w3.org; jacobs@w3.org; kynn@idyllmtn.com;
>asgilman@iamdigex.net
>Subject: Re: General Exception for Essential Purpose
>
>
>"Leonard R. Kasday" wrote:
>>
>> Ian, Kynn, Al
>>
>> Thanks for all the detail on the 2.0 philosophy but I'm still not sure I
>> understand the essentials.  Would you indulge the following lapse into
>> math-ese.
>>
>> Consider
>>
>> the set of all user groups U1, U2, U3... with different sets of abilities
>> and disabilities.
>>
>> and the guidelines UA and GL for user agents and web content.
>>
>> 1. Is the goal of WAI to produce UA and GL guidelines such that if both
>are
>> followed in their entirety, than each user groups U1, U2, U3... will have
>> available maxium feasible access to all web sites?  Here "maxium feasible"
>> means the maxium that can be obtained with current knowledge and
>technology?
>
>At first glance, yes. Users have needs, we try to write guidelines
>to meet those needs, by assigning responsibilities to meet those
>needs to different parties. It's up to us to choose the scope of those
>guidelines, how many problems in the real world to account for, etc.
>
>> 2. And is it completely acceptable to fulfill this goal by providing each
>> of the user groups U1, U2, U3,.... with different versions of the site S1,
>> S2, S3... ?
>
>Yes, but:
>
> 1) I think that it may not be possible to meet some needs anyway,
>however
>    large the set of sites is.
>
> 2) It is better for designers and users to produce fewer sites that
>meet
>    the needs of more users. Also, this doesn't take into account the
>    issue of providing content that has been tailored to specific needs
>    and therefore may be inaccessible to other users. (The issue of
>    whether accessibility has to be measured on the client side or
>whether
>    it can also be measured on the server side, as long as users have
>    access to full content, etc.)
>
>> If at all possible, please answer with one of the following
>> - "yes"
>> - "no"
>> - "what  _are_  you talking about Lenny?"  <grin/>
>
>So, yes and yes. As I mentioned in an earlier email, this is a new
>model that I'm playing with in my head, and so I expect it to be fragile
>at this stage of its existence.
>
> - Ian
>
>
>> or offer rephrasings of a sentence or two (with or without math-ese) to
>> which you can say "yes".
>>
>> Len
>>
>> p.s.
>> Also, if this is the philosophy, I don't understand where Kynn's
>"minimally
>> accessible" fallback fits in.
>
>--
>Ian Jacobs (jacobs@w3.org)  
<http://www.w3.org/People/Jacobs>http://www.w3.org/People/Jacobs
>Tel:                         +1 831 457-2842
>Cell:                        +1 917 450-8783
>

Received on Wednesday, 1 November 2000 13:56:06 UTC