W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > April to June 2001

4/19 meeting notes

From: Matt May <mcmay@bestkungfu.com>
Date: Thu, 19 Apr 2001 14:38:15 -0700
Message-ID: <010101c0c919$1b983860$6601a8c0@sttln1.wa.home.com>
To: <w3c-wai-gl@w3.org>
In attendance:
Wendy Chisholm
Jason White
Annuska Perkins
Gregg Vanderheiden
Loretta Guarino-Reid
Cynthia Shelly
Katie Haritos-Shea
Jeff Isom
Paul Bowman
Andi Snow-Weaver
Donovan Hipke
Matt May

Action Item:
KHS, JW Checkpoint on summaries and linking information (3.n?)

JW Verifiability: relationship between requirements and issues surrounding
testing, verification, clarification. Consensus seemed to be clarification
of troublesome requirements is necessary, rather than excluding or
separating them. We hadn't devoted enough time to defining/clarifying them.
I analyzed the guidelines and their difficulty of verification by checkpoint
on list. It's a first attempt at discerning verifiability and distinguishing
which are not. 1.1 for example we identified the issue of non-uniform
creation of content across formats. This was identified by Len Kasday. Makes
it difficult to determine whether 1.1 has been satisfied.
CS Go over the ones with problems?
JW Obvious ones...
WC What about "write clearly and simply"? any resolution?
CS Defining audiences
GV About defining the content
JW Is what it's already said
CS Anne's site has info on French and Indian war, written for kids, but same
content could be written for college audience as well.
GV "Physics for Poets" as another example. Issue is, we can't make this
measurable because groups could include people who have low enough cognition
as not to know what a web site is.
CS Testing strategies, or document about writing style?
GV Strategies, yes, but testing requires a criterion, and that defines a
cutoff point
JW If there's a scale, the hard part is determining where along the scale
you put it. Testing would have to be relative to that. Almost defeats the
KHS No decision on priorities?
CS Not yet.
KHS Options for alternative language levels. Then it's still not normative,
more foggy, but makes the point that it would be nice to make info available
to a number of audiences, and thus they don't have to dumb down.
CS Techniques for writing for cognitively disabled
KHS And similar documents. But important to say.
CS Going to be hard to draw a line, but knowledge to transfer
GV To go into techniques. We should put requirements on memory, etc. But, if
you're dealing with an audience with low cognition, there are techniques
relevant to that. To make a page that works without words, page needs to be
extremely concrete.
JW Where there's that kind of content, you can't do it
GV So how does one make it meaningful to use? Can't order products if you
can't understand words.
KHS Cognitive techniques document? For users who are writing for that target
GV Use a special document. Not W3C's responsibility, but maybe we should get
someone to do that.
CS Posted as a note or something.
GV Or done by another group, and linked from W3C to encourage targeted
WC How does this affect PWDs?
GV W3C not chartered to determine how to design assistive technologies.
WC How about adaptive content (e.g. Kynn's product, Edapta)
GV That's a general access technology for standard content.
WC I think those definitions are dangerous and broad.
GV If you design something only for PWD, with no value to the mass market,
that's assistive technology. Same product for everyone else, which can be
used by PWDs, that's a mass market technology. Different laws and
regulations to their use, as well as objectives. WAI is charged with making
web sites accessible, but I don't know if we're excluded from doing this.
Within 24 hours, there would be articles in 50 publications talking about
how ridiculous the requirements were. Would be misunderstood really fast.
WC We have a communications department so that doesn't get spread in that
way. Careful scrutiny of everything we publish.
GV If we had guidelines for specialized audiences, it's an attractive
nuisance. Something will happen to have it misused.
JW Different thing from something applicable across sites. What's the main
checkpoint? I don't see a major enhancement from WCAG 1.
GV Designers will say, my sites aren't designed for PWDs. It's a content and
audience issue?
CS It is legitimate concern
MM Could modularize the techniques to say that if you expect your audience
to contain people with this disability, this is what you do with the
GV Should be careful about how to state that. Laws say required for low
vision. 4 years ago, we said, here's everything you need to do. Now
goverments are saying, here's what you must do. We can only put things up
there that are required. Or, we can put up guidelines that are recommended.
Higher-ups are trying not to do that. One of the problems with the US
(S.508), and then there were issues with which guidelines to comply to. We
need to think in regulatory terms, unless we want people not to use our
WC What we modularize is our techniques. Write clearly and simply, for
example. We say "6th grade, college, and low cognition" techniques.
JW There are issues with cognitive because some things are just
incomprehensible to some people. You can minimize, and that's as far as you
can go.
GV Do we start layering -- I'll pick high-school as my reading level. That
doesn't help. Too high. If we do start layering, orgs will start saying,
maybe we will aim for lower levels. Cognitive disability advocates have been
targeting below kindergarten.
JW What are the criteria for determining simplicity while expressing the
same idea.
CS Concerned with lowering reading level of the web as a goal.
JW Me too. I wouldn't lower my writing level under any circumstances. I try
to write as clearly as I can, and that's as far as I can go. I write
appropriately for my audience, and that needs a reasonable grasp of
vocabulary to understand.
GV Jason's undergrad study was law and philosophy. Hard to understand the
subject matter anyway.
CS Scientific American article on the semantic web. Still at a college
reading level, but the easiest description so far. Herculean task to get it
down even to that level.
GV You have to think in terms of the web, and if you can't understand those
concepts, you can't understand the semantic web
KHS Have to define levels.
GV Two holes: tools for assessing reading content is "hokey"; we don't say
we should lower the graphic content for people with lower vision, we just
make sure that it can be done alternatively. Same might work for cognitive.
How many different levels are there, and do we make them do one of each?
WC Not even that you have to divide it, but if Tim Berners-Lee writes a
semantic web article, and no one understands it, we can say, here's a
simpler description. It's not that he has to rewrite it, but here's another
way of putting it
KHS One can be soft in terms of level, because even grade level is
JW Good idea, but still one problem. If you say is X is simpler than Y, how
do you determine if X is equivalent to Y, or a good alternative to Y? What
does X have to convey that Y doesn't, how much is in X that has to be in Y?
We have a clear criterion with images, etc., because we make an assumption
to context of material, and ensure that text equivalent is suitable to
assisting understanding. Different versions on cognitive grounds are a
different question.
KHS Can't simplify a photo. Not giving the same thing. Not easy to pin that
down, but we shouldn't give up on it because of that.
JW Not suggesting we don't, suggesting that there's a problem with
KHS But do your best.
ASW How do you regulate that?
WC It's a technique, not something to apply to every page. For those who
want to do level 3.
GV If we keep levels out of it, people can progress naturally. When we add
levels, we add gate points. We say the main content should be lowered in the
guidelines now. Now we're saying we should provide an alternative, which
ostensibly involves adding a second web page. A lot of work. If we require
it, it doubles the size of the web.
JW One should be watching writing style anyway. That might be as much as you
can do in some cases. Otherwise there might be background material or
potential for rewrite/repurposing.
WC You can make something accessible but ridiculously unusable. One site I
know can make their site accessible using video, but that's not how blind
users will learn it. It's partly audience, and partly why you're educating.
E-commerce vs. education sites, for example. Documents vs. applications. Can
we have one set of documents for all?
MM E-commerce sites will require that people are able to read well enough to
use a credit card. Can't require less from sites where something like that
is an issue.
CS Executive summary idea: lots of parallels with cognitive disability. Both
groups are resource-constrained (time vs. cognitive ability)
GV It'd be great if we could tell them to provide a summary at a certain
CS You're supposed to write executive summaries at a low level.
JW Summarize my novel?
WC I don't think it's that ridiculous an idea. Card catalogue has basic
information attached to everything.
MM Same thing with meta description and keywords
JW So, pay attention to grammar, spelling, orthography, using standard
terminology to express meaning, then summarize keywords, provide links or
references to background material for people who don't grasp the content,
applicable to a lot of content out there. Strategies being set out, now
which belong in guidelines, how are they defined regarding requirements and
compliance, and how do they fit in the document structure?
WC Metadata being discussed. Already in WCAG 1, as well as document
collections. Don't know what's in 2.0, but metadata is one thing to explore.
KHS How can one say "this page is optimized for low vision", etc., so that
someone comes up with optimized pages by search?
CS RDF description for conformance.
WC Partly expressed in evaluation & repair tools working group.
GV Part of the problem is that we require the level to be lowered. First
thing is to require that it be lowered to an appropriate level for the page.
Next thing is to require it be lowered even further. The other idea is to
require that an alternative page be used. There are always going to be
people who can't access the content. We could say that pages require
everything to be in pictures and sounds, because people with cognitive
disabilities can't use the site otherwise.
LGR Between a rock and a hard place. Feels qualitatively different.
GV There are some places where this is not different from other
disabilities. We provide alt text that is a lot less info than the image.
There are some types of information where a symphony is captioned. Not a
meaningful translation. We say in guidelines to do x, y, z, but we have no
idea how to do that sometimes, and we should say "all audio that can be
expressed in words, must be." We just assumed people would ignore it when
they didn't know how to do it.
CS When it becomes a law, they can't do it.
GV Right. What was done in S.508 was providing access to keyboard control.
How do you provide keyboard control to a paint program? So they said to
provide keyboard access where the action can be described in words. It seems
to hold up. Eloquent way to put it. Similarly, I can't caption Beethoven's
5th. There are some parallelisms, but we have some work to do in some other
JW Might be able to be expressed in a way where they can't ignore
requirements. What should be added to the existing requirements to address
this? Does ERT group have a way to address this?
GV Measuring language level of text?
JW Yes. Question as to whether language levels were good enough.
LGR Are we going to require a certain level?
JW We wanted to see if content can be labeled, and require that it be
identified. Issue with that is, is that an ER issue, WCAG issue, how do we
identify that level?
LGR Identify that level in the metadata?
CS Yes.
LGR How can we actually evaluate that?
CS Yes. It's a big deal.
WC On list, Anne suggested several ways to evaluate the reading level.
Someone pointed out English is rather well-defined, but Japanese is more
complex to evaluate.
KHS Automated thesaurus as well.
WC Relations can be done in metadata
CS Talking about EARL at face to face, needed to be a way to express
automated and human tests.
WC One reason we began working was to document human-tested content.
JW Needs to be some scheme of classification to be used. Not one standard
that everyone's satisfied with.
CS Does it have to be that complicated? Assert: My content is optimized for
college-level English speakers.
MM Still have the problem of a document that claims a certain level could be
put in front of 100 people, and 50 will choose each as being easier to
comprehend. Doesn't help establish what's "easier".
KHS We can't assert a level?
MM Can assert it, but can't prove it.
KHS Basic things that have to fit in.
CS Content metadata tagging: not 100% accurate, but helpful. Anne had said
that she wanted all content accessible to second-graders, but that she
wanted there to be content for second-graders
LGR More content?
CS No, just more available.
LGR How do we say in order to be accessible, you have to target these people
CS We should say to assert it.
PB Meta tag info has potential. This labels content for an audience or group
of people. Alternative to cognitively disabled is that you can exclude
certain documents, but making alternatives, it doesn't really address that.
Do we want an additional meta tag for executive summary? My concern is that
meta description is usually one sentence for search engines - insufficient
for the purpose.
JW RDF used to classify, not meta. It appears that access-related assertions
will be done in RDF. Could add checkpoint defined a certain way about
summaries and linking to background info. Could be introduced as RDF
assertion. Guidelines don't mandate a level of accessibility, only that if
you make claims, they be done in a particular way. Each checkpoint you claim
to conform to must be satisfied. Claims should be made appropriately. There
is a possible checkpoint coming out of this.
GV Labeling doesn't make them more accessible. All we've done is to tell
people, come in or stay out.
CS We make things more accessible as a whole by helping them find accessible

Received on Thursday, 19 April 2001 17:38:56 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:36 UTC