08 May 2003 - WCAG WG Teleconference Minutes


Paul Bohman, Doyle Burnett, Ben Caldwell, Kerstin Goldsmith, Loretta Guarino Reid, Katie Haritos-Shea, David Macdonald, Matt May, Jibari Simmons, Cynthia Shelley, Gregg Vanderheiden, Jason White


Avi Arditti, Wendy Chisolm, Michael Cooper, Lee Roberts, Roberto Scano


Proposed definition of Web Site accepted
Proposed definition of Web Content reviewed and discussed; Jason White to incorporate discussion into a new proposal
Preliminary discussion of the proposed reorganization; Gregg will post a link to the current proposal from the GL website so it is easier to find. Everyone is encouraged to read the proposal carefully and respond on the list, since it is hard to make much progress until we know whether we are proceeding from the current base or the proposed reorganization.

Reopened item: whether to require metadata for extended conformance claims.

Action Items

CS: contact chairs after getting feedback on charter changes
KHS: post the rewording of the definition of Content to the list
CS: propose a simpler definition of Content (one sentence?)
JW: propose a definition to distinguish web resource from web content
GV: add a link to GL web page to the Reorganization proposal
BC: incorporate new definitions and wordings in the current draft of WCAG2

Revisions to the Charter

CS: I need to run the IP clause changes by some people before I can approve it.
DB: How long does the charter run?
JW: 13 months, til June 2004
MM: All activities must end by June 2004 because WAI needs to be rechartered then
GV: Cynthia, please contact the chairs directly as soon as you reach resolution

Definition of Content

There were proposals to refine the use of "HTML file" in the discussion of scope, so that it is clear it is only an example.
GV: Media type doesn't seem to be unique to content. Maybe "content element" has a single native media type?
JW: Media type seems wrong. XML data has no media type; it isn't presentational.
GV: Wouldn't it be a text media type?
JW: Do you mean an MIME media type or CSS media type?
CS: I think he means MIME type.
JW: Then text/xml would qualify. But even so, a resource can have multiple network resources associated with it. A single web page can contain multiple media types.
GV: Can we just drop the sentence that talks about having a single media type?
GV: Is it useful to have such a complicated definition for content? I think it is.
JW: In UAAG, they distinguish the meanings of content  in "content model of an element" and in "text content vs non-text content" .
GV: We should define this as web content,and note that while content usually means web content, it can also have these other meanings.
CS: This definition is "the stuff the browser renders" rather than "the stuff between the tags".
CS: This is awfully complex.
DM: What is an example of stuff that is processed to be on the web?
GV: An example is making a purchase
DM: This example of  "processed" would happen on the backend of a web application that wouldn't be subject to accessibility guidelines.
??: An example is stylesheets
DM: Are stylesheets subject to these guidelines?
JW: Undoubtedly
JW: This definition of content will have to be reviewed by other working groups. Are we ready to propose it to other groups?
CS: I'd like a copy-editing pass to try to slim it down
[[Action: CS to produce a simpler definition of content]]
DM: We need this level of granularity to describe what we mean. But for 99% of the consumers of this document, we should be able to say this in no more than one or two sentences.
GV: The definition is actually the first sentence
DM: It defines it, but doesn't clarify it to those who aren't immersed in our discussions.
GV: How about the sentence that is there, then a sentence that says "It includes such things as ...". The discussion of the properties should be  a second paragraph.
KHS: what is the short list?
GV: HTML pages, style sheets, web applications, audio, video, forms, proprietary document formats, script, XML
GV: We don't want to define web content as being only W3C formats; whether they are accessible or not is a different question. They are web content.
[[Action: KHS to posted a revised version of this definition to the list]]

CS: Proposal for simplified definition: Web content is anything with a URI
GV: Any counterexamples?
JS: Some things that can be accessed via a URI are not content.
??: An example is a perl script?
CS: why isn't that content?
JW: Maybe the best way to address things is to rewrite the relevant checkpoint so it doesn't contrast content and presentation.
??: but content and presentation are separate things
CS: In the group title, content means all the stuff the browser renders. In the checkpoint it means the stuff between the tags.
GV: But we are just saying the content needs to be accessible, not the presentation. We aren't saying the presentation must be accessible.
JW: A formatting document object that is enitiely presentation is part of the content.
KHS: In a simple sense, any information that can be accessed through a URI
GV; It has to be broader. It has to include function
KHS: Any information or functionality that can be accessing through a URI by  a user agent
LGR: "user agent" here reintroduces the circularity in the definitions
GV: Drop the "by a user agent"
JW: Last week, we defined a web resource to be whatever can be accessed via a URI. Are these now synonyms? Or how do we explain why content is only a subset of web resources
[[Action: Jason makes a proposal to clarify this distinction]]
JS: If two machines are communcating with each other using the web, this doesn't have to be accessible
GV: web and internet are not interchangeable.
CS: The two machiens could be communcating via http, which is the distinction between web and internet that I've heard most frequently
GV: The web contains the servers
JS: If two servers are communicating, does that have to be accessible? It is the user who introduces the need for accessibility
MM: We have web services based on SOAP, etc that are designed for computer-to-computer communication. The interface to the user isn't part of that definiton. But the XML itself has a responsibility to prevent it from becoming inaccessible. So there is a definition of web content that is anything that has a URI. If we have to divide user content from web services or raw data...
GV: The Web is anything that has a URI? Content can be anything that has a URI which is presented to a user. The User Agent becomes that which presents things to the user.
JW: But my completely arbitrary XML data can be presented to the user just by presenting the raw XML
GV: and that would be web content
JW: I could be monitoring that machine-to-machine communicating via a web application to debug
JS: So it is the responsibility of the web application to present the 1s and 0s in a way that you can see. It is when you interact with that application that accessiblity comes in.
JW: Even if something isn't actually rendered to anybody, but finds its way into another computer and gets rendered indirectly, there is still a responsiblity to carry all the necessary information
GV: content is that which is accessed via a URI and presented to a user. All content is accessed by a machine (PC) and later (through a user agent) presented to a user. If it is never presented to a user, it isn't falling in our area. But this really goes back to the W3C. We shouldn't be defining content. We can suggest the definition as anything accessible via a URI and presented to a user, but they may say we can't limit the definition in that way.
MM: There is a case to be made that if something is presented via URI that is not intended to be rendered in final form directly to the user, but is for the purpsoe of transferring data, in many cases there is no means to make that accessible. If you take the SOAP wrapper itself and try to present it to the user, there is nothing there. I agree that there is an area in which users are not intended to receive that content. It may be damaging to say that WCAG applies to that.
GV: We aren't saying WCAG applies to it. Our definition of content would say "anytning with a URI". And in the guidelines we'd say that they only apply to content presented to the user. Maybe we need to redefine user content?
MM: We aren't defining web content. We are defining the web. The definition of content is still up for grabs.
JW: I am concerned that the definition of content becomes the definition of what is covered by this guideline. We don't need to go down that route. We can define content broadly and say that the guidelines only apply to a subset of content. This is a reasonable approach.
GV: We can enable this by defining content to be user content or say that our guidelines only apply to user content. Jason still has the action item to make a proposal.

Definition of Web Site

((Much approval of Jason's proposal from the group))
GV: It gets across that a web site is not a server
DM: and it is not everything linked from the page
GV: I recommend using "organizations" rather than "corporations"


MM: You have two groups, and they have minimum and recommended items?
GV: One suggestion is that the first group be called "core" so we'd have core checkpoints and extended checkpoints.
MM: How do you reconcile these 4 categories into a conformance scheme.
GV: Core are required, extended are not required.
JW: If the core are required, is there any reason to divide them into subcategories.
GV: In most standards, you have core and optional items, but if you do the optional items, you have to do them a certain way. So optional items have requirements. If I make a success claim, how do I report what I'm claimin?. If they all have names and numbers, you can do it explicitly. Or there may be a global claim like Core+5 or something like that.
JW As long as there is a clear distinction between suggestions and success criteria, we are ok
??: I like "core" and "extended"
GV: There was a comment on the list that core and extended were being defined as AT compatible and direct access. We aren't doing this. A concern voiced on the list is that writing clearly and simply is no longer core. That is correct; that is no longer a core requirement in this proposal. But the text must be decypherable so that AT could work with it.
CS: What happens to group C in this scheme (specialized for specific access).Are they just part of extended?
GV: Yes. We could break extended into 2 categories. I didn't do this because of lack of time, and because it can be very hard to do.
CS: I'm still not sure which is better, claims for different levels or continuous claims (i.e. core + N).
MM: More than just seeing the number of extensions satisfied, I'd like to see an endorsement for each checkpoint required for a certain disability. It might be a better way to encourage people to address coherent sets of requirements. Then you can claim support for people with hearing problems, etc.
GV: We had a long discussion that we wouldn't create certifications that were disability-specific. It quickly leads to people saying "I made a blind site." We can reopen that decision. But you can claim the specific guidelines you address, so people can search based on the guidelines that are important to them.
JW: Many checkpoiunts are needed by a variety of areas of accessibility. So it is important to be able to list which checkpoints were addressed.
MM: I think that should be a minimum requirement for claiming + anything.
GV: But then we are requiring metadata be added to every site
MM: But can't they use EARL or something? This is just the output of the evaluation tool. You'd have to evaluate to claim any level of conformance. You are just asking people to put that data into a form that can be parsed.
GV: Let's put this down as an open issue.
BC: It is only required if you mark something as a minimum +. So you only need metadata above minimum.
GV: Companies have said that this kind of reporting implies a warranty, and they are unwilling to makes such a warranty. So it could remove the incentive to move beyond the minimum.
MM: You run into the implied warranty with any claim. Institutionally, everyone is just going to say whatever level they attain. Those making + claims are probably individuals who are competing to make the most accessible site.
GV: We need to look at the reorganization proposal and decide whether that is our new base, or whether we are going to stay with the previous organization. But we can't move forward without that decision.
KHS: I don't think this is off-base, and I think it is a fresh approach.
MM: One of the think that makes this difficult to discuss is that it is a wholesale change. We are going to have to ruminate on it.
GV: We aren't going to make a go/no go decision in this conference call. I'm just trying to get a sense from the folks present.
[[Action: BC to make the changes necessary to apply the new terminology definitions to the document]]
??: How do we distinguish techniques?
GV: required vs recommended?
JW: success criteria vs additional items (as in current draft)
GV: recommended practice? there is a difference between extended checkpoints vs extended success criteria. How do you report that?
JW: We could divide the items into success criteria and suggested practice that is non-normative (and hence non-testable, non-recordable)
GV: When I tried to do that, some of those non-normative items are things you'd want to search for in metadata. If they are non-recordable, you'd never  be able to search for them.
JW: One other approach that Wendy suggested is normative inclusions and exclusions. Under some circumstnaces, things that aren't normally required become required.

$Date: 2003/05/09 14:07:59 $ Loretta Guarino Reid