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

8 November 2001 WCAG WG Minutes

From: Loretta Guarino Reid <lguarino@adobe.com>
Date: Thu, 08 Nov 2001 17:01:54 -0800
Message-Id: <200111090101.RAA01596@patagonia>
To: Web Content Guidelines <w3c-wai-gl@w3.org>


In attendance:

JW: Jason White
JM: Jo Miller
ASW: Andi Snow-Weaver
GSW: Gian Sampson-Wild
LGR: Loretta Guarino Reid
AP: Annuska Perkins
KB: Kynn Bartlett
KHS: Katie Haritos-Shea
PB: Paul Bohman

JW: Strategy for next week's Face-2-Face meeting:
Move forward on conformance, maybe technology specifics and success criteria

Conformance: - 2 options proposed for how to proceed
1. consider kinds of criteria to use for minimal subset
2. work through guidelines, comparing against proposed criteria

Conformance is used by other WAI documents, has broad implications.

JM: does anyone have thoughts about how to structure discussion at F2F so 
there is progress from the last F2F? 

JW: Because of W3C process, need to have conformance scheme. Argument is 
there needs to be a division between minimum set and the rest. All or 
nothing will lead many organizations to take "nothing". Basic level of 
conformance that can be built upon seems a better strategy. Suggestion 
from Kynn: checkpoint-by-checkpoint claim beyond the minimum. 

Jo: Tp give people incentive for going beyond minumum - speak in terms of 
phases?

PB: phases could be defined in many ways - standard way of defining phases?

JM: phasing would describe implementation (retrofitting or building in 
accessibility)

PB: is phase 1 equivalent to all priority 1 checkpoints? 

KHS: building off phase wording, "Web Content Guidelines and Implementation"

PB: but are phases the same as conformance definitions?

JM: parallel; just different terminology.  There is controversy because 
the perception is that people will stop as soon as they reach level 1, so 
if your favorite guideline isn't in level 1, it won't be implemented.

KHS: Some things that aren't easy to do might affect a larger population

JW: Certain requirements are more important than others, so that if 
content doesn't meet those, it will exclude certain groups of people 
entirely. Need to implement those first because benefit of remaining items 
won't be achieved. This is still the implicit idea when we discuss minimum 
sets.

PB: Any sort of identification as minimal compliance will lead to that 
being where people stop. In practice, it is much more common for people 
to choose as small a set as possible. What's a manageable number? 
between 10 and 20. Implications of minimal set are worrisome  - it defines a
set of things you don't have to worry about

JM: Agree. we see this happening. The converse problem is people won't 
try at all. Idea - levels on the way to full conformance? 

JW: "Limited conformance"?  Requiring too much leads people to adopt 
their own subsets. 

KB: Do we have any evidence this really happens?

KHS: Governments considering adopting guidelines may drop the idea if 
current technology doesn't support it. Based on my experience as a 508 
coordinator, if all of WCAG1 were required for 508, it wouldn't have happened.

PB: I understand Kynn's concern, but section 508 is an example of this 
happening. It is based on WCAG1 priority 1 guidelines, but reworded based 
on what was considered important or doable. My biggest concern is that 
someone else will set the standard 

KB: Standards will be set somewhere else no matter what we do. 508 is an
example of someone who needed WCAG as a toolkit for assembling a policy, 
but since it wasn't structured that way, they had to do more tweaking and 
rewording. On the topic of claiming compliance: consider the HTML spec. Is 
it ok to say I'm only compliant with 50% of spec and claim minimal compliance?
But there is strict vs transitional compliance. This means there are 
effectively 2 levels of compliance. Might be a good model for us to look at. 

JM: Yes! "transitional" was what I was going towards with "phase"

KB: Consider the modularization of xhtml 1.1 as a model, and tools to put 
them together. This may be a WCAG3 item. Make WCAG something from which 
you can build a policy, rather than building in a policy, which is what 
compliance labels may end up doing. Use a more value-neutral label. 
Also publish a policy built using this framework.  Don't think we can ever 
define a true minimum, since someone will always be excluded.

JW: There was strong resistance to the idea that authors should have choice 
for what to include in minimal conformance. The discussion did generate a 
lot of dissension. So this idea is a controversial one. Might be best 
reopened with a specific proposal that explains how this would avoid 
people choosing to drop items that are inconvenient.

PB: Kynn, since you earlier said you don't want to set a minimum,
isn't that what transitional does?

KB: yes, transitional is a compromise for WCAG2. 

JW: One difficulty: if one uses those terms, transitional may be more 
onerous than strict. Transitional conformance maya hve to accommodate 
items that could be implemented by user agents.

KB: It is easier to write strict HMTL than transitional HTML

JM: The difficulty of moving from transitional to strict not our problem.
Success criteria - people are going to pick and choose, no matter what we do.
Regulators tend to pick according to whether there are objective tests or 
measures. Whether we use that as a way of setting levels, other people will.

JW: Reminder - we have a consensus statement that we will only include 
things in the normative document for which we don't have success criteria.

JM: I think this consensus may crumble when someone's favorite guideilne 
is under the knife.

KB: Will we be able to reach consensus about whether we have adequate 
success criteria?

JW: We need to believe that indepedent evaluators would come to the same 
conclusion about whether something satisfies the success criteria. 
Partly hypotheticals, but can be tested for.

GSW: Need to explain guidelines in terms of how they can be tested

KB: Also should consider how we qualify the "indepedent evaluators"? Do 
they need knowledge of content? Can implementors test? Can end user test? 
There will be a difference based on what the user's perspective is and 
the content provider's perspective is, as well as accessibility consultant's. 
Could arrive at valid but different results, based on perspective.

GSW: Implementor needs to test as if they are the user

KB: Not always easy for implementor to do this. Take alt text: it is possible 
for the user and author to reach different valid opinions about whether 
alt text is appropriate. At what point do we say this is within our tolerance?

JW: In the past, this was discussed in terms of 3rd party testing. 

KB: Are you assuming that the 3rd parties can see?

JW: I don't know; just trying to present past discussion. Another issue which
also arises in the Device Independence WG's most recent Principles: the
distinctions between functional presentation and customized presentation. 
Based on whether content achieves the kind of presentation that enables 
users to do what the author expects user to be able to do. Different from 
the user being able to do what the users wants, which may be different 
from what the author had in mind. 

JW: An issue that was raised a few weeks ago is that there are many areas 
where you must provide a normative requirement that is technology specific. 
There are some guidelines where you have definite, normative, 
technology-specific consequences that you can write down as requirements. 
I made a proposal that there should be technology specific *success criteria* 
to supplement general success criteria. The issue is - what is the least 
confusing way of formatting this so the relationship is clear. 

GSW: I like the idea. 

JW: Is a separate technology-specific document needed?

GSW: It is not clear that a separate document is needed.

JW: Quite a few checkpoints don't need technology specific success criteria, 
although we may want to provide examples in different technologies.

GSW: Examples are a good idea; they help clarify checkpoints.

JW: Success criteria are normative; examples aren't.

LGR: Good idea. I need to look at PDF information and makes sure it
all maps to success criteria for some checkpoint. Worried about document 
protection item.

LGR: What about adding new technologies after WCAG2 is done?

JW: This is something that hasn't yet been thought about. Going through the 
W3C process is pretty grueling. What information should go into W3C notes,
which can be changed more easily, and what goes into the W3C recommendation?

LGR: We are still avoiding the conformance question. 

JW: Two tracks are proposed for next week - conformance, and technology 
specific issues. We'll try to keep progressing on technology specific 
documents.Technology-specific success criteria have to be met as well as 
general success criteria. Different views of document for different 
technologies may help manage information.

JM: Those working on the CSS document are very busy, and would welcome 
contributinos from anyone.

JW: Are we missing expertise in any areas where we should be recruiting help?

KHS: Are there issues around appropriateness of including PDF techniques?

JW: We don't feel there is a problem including PDF. There would be a problem 
addressing specific implementations or tools. But PDF is a specification that 
has been implemented by various developers.
Received on Thursday, 8 November 2001 20:02:35 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:47:16 GMT