Minutes 03 May 2001 Ready

Same address: http://www.w3.org/WAI/GL/2001/05/03-minutes.html

In attendance:
Loretta Guarino Reid
Jason White
Katie Haritos-Shea
Matt May
William Loughborough
Anushka Perkins
Greg VanderHeiden
Donovan Hipke
Gregory Rosmaita
Charles McCathieNeville

Action Items:
CMN: Document the use of metadata for disclosing cognitive levels. 
CMN, WA: Pull together the techniques for cognitive disabilities and see 
if there are patterns or common threads.



GV: Wendy, Jason, and I have been talking in preparation for the
meeting. We seem to be wrestling with a compound issue: accessibility
vs usability, and cognitive disability gets in there, too. We want to
get people's thoughts. In summary: a comment was made there are
accessibility issues relating to sensory disabilities, and everything
else is a usability issue. It seems very hard to draw the line between
them. What if you can get your hands on the information but you can't
comprehend it? For instance, what if a site is coded in ascii text in
Swahili. Most of us couldn't use it. Is that really an accessiblity
issue? Usability addresses cases where you can get the information and
use it, but you can't use it very well. This threshhold can be
different for every person. What one person finds hard, another will
find impossible. Also the old bugabo: you can make pages that are
usable across the whole range of sight limitations or hearing
limitations or physical limitations, but not the whole range of
cognitive limitations. Of course, there are always limits in these
first cass. We can't really make a symphony accessible to someone who
is deaf.  Cognitive is always this special case. We can talk about
things that will facilitate understanding, but it doesn't go that
far. If we require sites to be accessible cognitively, you'd have to
rewrite a site 6 or 7 times and you'll lose significant amounts of
information each time.
KHS: It doesn't make sense to go that far.
GV: Then we would make things accessible to the blind, but not to the
cognitively disabled. Because that problem is qualitatively a
different animal.
KHS: SHould we remove the requirement for cognitive disabilities? 
GV: That's what the Access Board did.
KHS: But the Access Board is not us; what do we want to do?
GV: If we are going to provide limited support, how far will we go?
JW: Let the author go as far as he wants, and make a claim about it.
GV: We'll make no requirements except that author discloses the cognitive 
level?
JW: That is what we do for other disabilities.
GV: No it isn't. We have requirements that you have to meet in order 
to claim compliance.
JW: With the new RDF you can make claims for each checkpoint individually. 
You don't have to satisfy any set of them.
GV: With RDF, you still have to satisfy a set to put a sticker on for Level
A.
WA: You can still make assertions in RDF.
JW: e.g. you can make an assertion that you satisfy one checkpoint.
GV: We don't want authors doing that.
JW: There is a rationale that there should be a certain set satisfied
to make a claim, but let's save this for another time. Beyond the
minimum requirements for level A compliance, it is all a matter of
disclosing what more you have achieved.
GV: And each of those is binary, that is, either you satisfy it or you
don't. And there would be no such requirements for cognitive
disabilities? I'm not suggesting there be none, just saying that if
all the others are binary, for cognitive disabilities you have to
qualify your claim: "This works for cognitive level up to
such-and-such". Otherwise we all have a zero, because we all fail for
someone with some level of cognitive disability. On cognitive ground,
we have to say we give up.
JW: This is asserting some kind of cognitive rating. Can this be done in 
a meaningful way?
GV: But we wouldn't require any particular level of support. This would 
result in no cognitive in requirements, only disclosures.
JW: I'm just suggesting this as a mechanism
GR: if you are claiming compliance for using language that is appropriate, 
you must document "appropriate"
JW: Is it possible to require anything of that kind for websites in general?

This gets back to problem of highly specialized, complex material.
GR: When it comes down to real life cases, what is appropriate is whatever
is 
legally required. A uUniversity site can claim that it only need to satisfy 
the cognitive level requirements for entrance to the university. It will be
a
content and context related judgment.
JW: This is my reason for suggesting there is a problem in requiring that a
site be cognitively accessible. A general requirement doesn't take context
into account. 
KHS: Do we have a cognitive requirement?
WA: I find "requirement" disturbing. I don't think we require anything.
GV: We always have to think in terms of hard requirements. Countries
are taking our guidelines and adopting them. If we say this is theiR
problem, not ours, we are saying they need to change our
wording. Whatever problems we leave in, they must take out. But W3C
has gone on record against modifications to our specs.
KHS: Hard requirements are also important for machine checkability.
WA: You can verify whether there are alt tags, but that doesn't make 
it a requirement.
KTS: All things we do have to be measurable.
GV: I'm not sure they have to be machine measurable.
KTS: They have to be if we want them to be automatable.
GV: Automtation facilitates a lot of things. But even if the
requirements are not automatable, if they can be objective, it would
be good. (So everyone can agree if they have been satisfied)
JW: We have a range of options regarding cognitive.
 WA: In 1933 I clearly remember something like 20 as the number of
people in the world who understood the General and Special Theory of
Relativity. Making it so a million people could understand it would be
out of the question This is a parallel to what Greg is trying to say.
JW: An interesting question: if something is a matter of accessibility
rather than usability, then the solution must make a difference for
someone who has that cognitive disability. It could only be an
accessibility requirement when you can demonstrate that some group
with an identifiable cognitive disability would be able to understand
it with the requirement met but they wouldn't be able to understand it
otherwise. One needs to make a distinction between cases where it
would help and those where you can make it easier to understand but
the page would still be beyond the reach of that group. Most people
aren't in a good position to judge this. If a disability just affects
reading, like dislexia, does that come within the set of cognitive
disabilities?
GV: People who have learning disabilities say there are cognitive,
language, and learning disabilities, and that they are all
different. Like low vision and blindness are different
disabilities. You'll get a lot of pushback if you group learning
disabilities under cognitive disabilities.
JW: Maybe we do need to develop those distinctions, since the solutions 
for those categories are likely to be different.
WA: Some of those things can demonstrably be helped. That's the point Ann 
has been making. Illustrations can help for a class of disabilities
GV: Those with reading disabilities? 
WA: Her husband has a condiiton where he can read, but without illustations 
he can't get through a site. Picture books vs chapter books
JW: There might be certain types of contents where you can simplify or
add illustrations wthout losing the meaning, whether or not this makes
the content accessible to people with different cognitive
disabilities. You'll only be required when it will make a different
for someone with a documented disability. This is a diffult
distinction for a non-specialist to make. It is a consequence of
applying the accessibility vs usability distincton.
WA: There is heavy insistence that is does help, and it has taken me a 
while to come around to that side.
JW: I'm not yet on that side. It depend so the context.
WA: Does Gregory think that the presence of illustrations could be used to 
overcome problems at priority level 1
JW: This is just one of the techniques for improving the
comprehension. One among a range of techniques. How are we going to
deal with these in the guidelins? As part of a priority scheme? with
conformance requirements? Where? When are they accessibility issues
(vs usability)? Do we want to make that distinction. These are the
questions that have been raised. We solicit opinions and
suggestions. Greg's argument was that any strategy used for cognitive
disabilities (language, consistency, color distinctions,
illustrations, speech) can make a difference for a certain group of
people between accessibility and inaccessibility.
WA: the question has been raised as to whether illustrations qualify
as a priority 1 sort of thing. That is, there are people, regardless
of the name of the disability, who can't deal with pure text. They
must have illustrations.
 GV: Is this a widespread category? Although it goes against our basic
principles, in our guidelines we have also had do deal with
practicality issues. With other disabilities, we talk about techniques
that don't change the appearance of the page. Alt text, labels on
tables, using things properly, etc. The cognitive suggestions are
quite different. They change the appearance of sites significantly to
everyone (not just those who need it). Is AltPict needed?
WA: AltPict is what I am opting for. Supplementary picture
GV: Somethat that would accompany text, but could be refused.
JW: Every image and audio file is refusable anyway
GV: The mass market gets "regular" picture, but people could request 
addition illustrations.
JW: There are many strategies related to cognitive disabilities. Any
one could make a difference to certain groups of people. Greg argued
that that means each becomes a priority 1 under some set of
circumstances.
CMN: People are building interfaces where adapting content to each person 
is fairly standing practice.
MM: The presentation adapts itself, but the content is either the same
or written in an inverted pyramid style so you can chop it off at some
level and still have the remaining content fit together. How
appropriate is the content to the audience? The content is as
important as the producer and recipient. There needs to be something
that says for all content X, X + images is better than X withouth
images. Rules like that could end up in the guidelines. People have
issues with creating hard and fast rules if you put something in where
you don't give content producer some way of saying what I've done is
better than what I had before because it satisfies the guidelines, GV;
didn't quite get that
MM: The concern is that the content itself has to be figured in. Rules
have to be contextual. If you are looking at a document to decide
whether to comply with this or not, if you aren't sure you can
actually help with this particular content, it makes it harder to
embrace this goal.
JW: What do we think of this idea: the responsibility should be that
the cognitive demands of the content be exposed in the content.
KHS: I think it should.
GR: It should be .
WA: What is the technique for doing this?  
KHS: tagging in metadata 
WA: With what? third grade reading level? requires college education?  
KHS: It will be left up to what is used in educational circles.  
JW: Assuming it is technically feasable, reliable, and useful, we appear 
to have agreement on disclosure requirement for cognitive demands at some
priority level 
GV: There isn't clearly a way to do this. In talking with folks familiar 
with those levels, they are suspect 
WA: also printed on toys
MM: We can't reliably say that this site is comprehensible by X. We must 
say it is intended for X level. 
GV: A concern I have is people just labeling: this site is intended for
people who can see", etc.  
KHS: They can't do this if they follow the other guidelines.
CMN: Even if they are suspect, they aren't useless. This doesn't seem
like a remarkably difficult thing to do. This won't fix everything for
everyone.  
GV: We are just saying to describe the level, not fix problems for the 
cognitively disabled.
CMN: That would be appalling, but you should at least describe.
GR:The requirement is to use appropriate language; you must also
document what criteria you have used. Yes, it will be subjective, but
people will have to address and provide explanation about criteria
JW: Someone opposing this would argue that a requirement to document
is difficult because 1) it would be the only requirement requiring
more than a rating, since it would require more documentation.  This
is a disincentive. 2) Someone could say it is taking reponsibility
from the Working Group and shifting it to each content provider to
decide the criteria for appropriateness. I don't necessariily agree
with these arguments, but want to put these forward.
AP: The suggestion to document the intended audience, is that in 
metadata or separately documented? 
CMN: I will take an action item to document the use of metadata for
this. The first objection is specious. Providing this is less work
than providing alt text
JW: Greg suggests it goes beyond that
CMN: I don't think it does.
GR: I would love it if people used EARL. metadata is sufficient. We
envision ways of extracting metadata for something useful
CMN: Best practice may go beyond this, but minimal requirements
should be automatable and reasonably trivial to do.  
JW: It could be highly misleading if the measure is inappropriate or 
unhelpful. Second question : if we what an assertion of cognitive level 
included, what else should be a requirement in the cognitive area.
CMN: Maintain the simple language requirement, the requirement for
increasing the use of multi-media. The requirement for consistency
aids this area.
WA: Just the requirement to document the cognitive level will be a huge
step forward 
JW: Should there be anything beyond that?
KHS: Are we including 3.3?  
CMN: We should also require cognitive support, not just document it.
 JW: That is the question. What about the issues Greg has raised about
context-dependent issues, that we can't reach everyone, that lines
need to be drawn, etc.
CMN: These are bold assertions.
GV: We need to come back to not "what would be helpful" but what can
we do? What goes into requirements (ought to be a law)? what goes into
usability (encouraged, but not required)? what goes into techniques
for targeting this audience specifically (could be techniques for item 2; 
but want to regroup by disability). The third item is fuzzy because
W3C is doing standards, not guidelines for specialized sites. If we
start mixing these goals, we risk confusing people and losing ground
on other items. The longer our guidelines, the less effective they
will be. Can we stay lean and mean? But I don't like that cognitive
disabilities is on the edge, so lean and mean pushes them off that
edge.
WA: Most of this is in guideline 3, right?  
KHS: We did some rewording of 3.
CMN: I want to talk about actual checkpoint proposals, rather than
hypothetical guidelines. We should write down text about what to do,
who is helps and hurts.
JW: The problem is that it is not really feasible to put down criteria
for these requirements. We can't say what should be required for all
content. That's the problem we are dealing with. Which leads to the
question of status of requirements. Are they the same, or dealt with
in conformance differently?
CMN: We have been trying a top down approach to deciding whether to
solve cognitive disability problems. I suggest we try a bottom up
approach. Put the techniqes on the table. Outline what can be done,
what is needed. Then we can loook at these and see if we can pull out
patterns, or common threads that can be expressed as requirements.
JW: Who wants this action item 
CMN, WA: take action 
JW: Any more ideas on how to deal with these kinds of requirements?
JW: To summarize, there is a general preference for the disclosure of 
cognitive level. Beyond that we don't have consensus. 
GV: I'm not sure where the discolure goes. There is some feeling that 
something is needed besides just disclosure.

Meet again next week


Katie Haritos-Shea
11809 Waples Mill Road
Oakton, Virginia 
 22124-2113
 USA 
703-620-3551
Mobile: 571-220-7777

ryladog@earthlink.net
kshea@fedworld.gov
kshea@ntis.gov