W3C Web Accessibility Initiative

Authoring Tool Accessibility Guidelines working group meeting

6-7 October 2000 Bristol, UK

Registration | Agenda | Minutes | Required Readings | Logistics | Participants

The main goals are to:

  1. address issues with incorporating WCAG 2.0 when it comes out.
  2. determine how to incorporate ideas from the Device Independent Authoring Workshop,
  3. develop techniques and testing procedures.


Registration is essential and can be done via the online registration page. Please Note: If you are already registered for the Web Content Accessibility Guidelines meeting, please register for Saturday only - you will already be registered for the Friday afternoon. Registration closes 29 September 2000.


Registration is open to members of the Authoring Tool Accessibility Guidelines Working Group, and developers of Authoring Tools interested in the work of the group.


Friday 6 October - Afternoon
Joint session with WCAG working group meeting. Discussion of preliminary work for WCAG 2.0 and how to integrate it into ATAG.
Saturday 7 October - 9:30 - 5:00 (with breaks)

Required readings

Participants should be familiar with discussion on the group mailing list (consult the archives if necessary) and the following material:


Meeting place

The Friday afternoon session is hosted by HP Labs Bristol.

Hewlett Packard
Bristol, England, UK
Map and directions

The Saturday session is hosted by the University of Bristol.

More detailed information will be provided by the beginning of September


Block reservations for any nights between 2-7 October have been made at the following hotels until the deadlines shown. Participants to quote "W3C/HP" when registering.


Bristol is the largest city in the South-West of England, situated 120 miles west of London.

HP Labs is situated 5 miles north of Bristol city centre. Coach transport will be provided each day between the Labs and the recommended hotels.

To reach Bristol by car from London Heathrow Airport, takes about 1 hr 30 mins via the M4 motorway.

Trains from London Paddington station to Bristol Parkway (5 mins taxi from HP and Posthouse Forte Hotel) take 1 hr 20 mins and run every hour. Trains also run to Bristol Temple Meads (city centre and Bristol Marriott Royal Hotel) from London Paddington. The Heathrow Express links Paddington to Heathrow Airport and takes 20 mins.

There are frequent buses from Victoria Bus station to Bristol.

Bristol International Airport is 6 miles south of the city and provides daily direct flights to Paris and Amsterdam.

(information copied from the Device Independent Authoring Workshop).


Friday afternoon - see WCAG Meeting minutes




ATAG 2.0

WL Think of authoring tools as authoring tools rather than enhancing accessibility.

CMN Outside of our charter. Do we want to ask W3C if we can do this?

WL Yes.

GR We've at least laid the ground work.

IJ There are things with HTTP that I would put in a UAAG, but it's not accessibility but something that should be done. HTTP spec does not discuss fragment anchors.

WL We'll think of others like that as we move forward (in ATAG).

DB Part of the doc or something separate? How does that affect conformance?

CMN Do we want to take on that work? If not us, should someone else? Should someone do it at all?

MK I can imagine those things for UA but not for AU.

Action WL: send to the mailing list a list of items that are not purely accessibility that could be included.

DB Things that are not about accessibility muddies the waters for conformance, particularly for policy-makers. Can we e-mail these ideas to someone else.

JR We have enough work to do with techniques.

CMN Additional requirements will help us make a better case to lawmakers.

WL We're talking about ATAG 2.0, IJ is concerned about ATAG 1.0 techniques. Requirements is not a necessary word. I'm getting a different look at what disability means. There is a point where you realize there is a continuum. There is a gradual diminution (?) Retinal conceit is an issue in our business. We need to look at disability more generally.

DB Worry about where we draw the line. I can look at any products and have suggestions. No resistance to say that those things that people have difficulty with as they age we should include. Bad navigation fits in there.

GR I'm worried about false dichotomy between usability and accessibility. One is a subset of the other. Need for general activity within the W3C.

CMN We could take on as a work item that we publish the non-accessibility bits as a Note. Do we want to take that on?

yes - WL, GR

DB Suppose that in a tool the arrangement of the menus is not correct, is that what we're talking about?

WL yes.

GR In UA we've had arguments about "accessibility" vs. "usability." A tool that is usable by everyone is accessible. We need to reinforce that accessibility should be considered from the onset. We need an integration of guidelines.

CMN We collect any requirements that we reject from our guidelines and publish those in a Note. "These are things that are good ideas."

Resolved: publish a "things that are good ideas" as a Note.

CMN There is a redundancy between checkpoints that should be resolved in ATAG 2.

/* discussion about Microsoft menus */

GR Accessibility/Usability is in eye/ear of beholder.

DB I agree, they are not different, but we need boundaries.

IJ One reason it might be useful to maintain a distinction is it reflects the real world.

WC Overall roadmap would help. Heard lots of requests for. Usability activity.

CMN Propose to comm that bump up priority on arch document?

WC Yes.

CMN How urgent is ATAG2?

WC How much implementation experience? Need for one is driving WCAG 2.0

GR With device independence concern with tool increasing due to complexity.

WL Recommendation rather than requirement. This takes edge off of manufacturers. Recommendation doesn't have same threat.

IJ There are requirements in this document.

GR If we have generalized document we could have an appendix for conformance on accessibility issues (for policy makers).

CMN We resolved we will collect all ideas, at the moment we don't know what we'll do with it at the end. When WCAG 2.0 goes to recommendation, we should point to it ASAP. I propose that we begin looking at producing another Rec when WCAG 2.0 goes to last call. Include links to WCAG 2.0 and any other things we've resolved on.

DB Tools are different than sites. I would urge a more interim release. Integrate fixes and release.

CMN Change list: currently is eliminating one redundant checkpoint and linking to WCAG 2.0.

IJ TBL proposed (for the UAAG and DOM dependency) that when DOM goes to Rec, do you also approve UAAG. Might be possible to piggyback ATAG on to WCAG. You might not have anything to do but sneak in on their call for review.

CMN An appropriate trigger?

IJ Would be ATAG 1.1.

Resolved: When WCAG goes to last call that will trigger us to create ATAG 1.1.

Techniques for Evaluating Authoring Tool Accessibility

CMN Break into groups and discuss. The purpose of this document is a guide for someone who has to evaluate their tool.

/* discussion about the document in two groups */

JR This will be an online process. Wording that people didn't like.


WL work towards single but caveats for special cases. what about W3C wide?

IJ start w/WAI then take that to W3C.


Action: GR write technique evaluations for Guideline 7.

JR What do people do who do not have specialized equipment? Can some of it be done by people w/out?

GR For windows logo testing, what do they do?

DB There are some tools that look at properties.

CMN ISSUE: can we link to tools that will do it automatically?

WL One tool is "machine w/out a monitor"?

CMN ISSUE: Do we check accessibility of tool separately or integrate?

JR I proposed evaluating image maps two ways: 1. can you do this, 2. can you do this with just a keyboard.

CMN ISSUE: what kind of modules do we want? current forms, multimedia, images. things that seem useful: interaction (forms, navigation), structured text.

ISSUE we don't distinguish between WYSIWYG and structured editing. Do we want to differentiate?

GR What about composite tools - those that you can use WYSIWYG or text editor mode of same tool.

CMN ISSUE: suggest that create an RDF graph to keep the results.

ISSUE: consistency in the way the questions are phrased. "yes" always mean that you conform.

MK I disagree. Questions should be clear. Yes can mean fail.

GR Not just "yes or no" but "not applicable."

IJ In UA, we use that.

GR Primary intent of ATAG is for authoring tools. I think this will be used by individuals who will be buying tools. In results, you can see what are N/A for this tool.

DB You need to verify results.

CMN The scheme has "not tested." I favor separating out "N/A" and "not met."

ISSUE: cut up by priorities. tradeoff - more usable, yet means that useful things they don't see.

WL Be sure it will be provided.

GR If you do that, the output would say here are additional things to consider. like bobby.

CMN Give them a teaser.

GR If Tom makes the P1 and P2 map, include that.

CMN ISSUE: checkpoint links need to be clear where they go to. Sometimes relevant to x WCAG and y ATAG. "These are the reasons this is important." This is further reading?

DB You will start storing data?

CMN Goal: build a web interface to let people check boxes. Also will allow test files. Build once and use it with other guidelines.

DB I have a database that you should look at.

CMN IBM has one too.

CMN ISSUE statistically sampling. Do you need to check every image in a clip art gallery?

WL Bettman Archives should be annotated!

JR Choose a few at random or test all.

CMN Word has templates. Do you check all for WCAG conformance?

GR Check them all.

JR really?

GR Rather than go through the image in the gallery, if you had a list of the alt-text for each one then you could go through.

JR It has to be meaningful alt. It means you have to read them all.

GR This gives you some basis that they have gone through it.

JR If you go through a product that doesn't care.

IJ I propose that this be pushed to the claim. "We claim we do this based on the following technique. We sampled 10% and based on that we assume the rest are o.k."

GR Primarily just for evaluators.

JR The industrial would know.

WL They would know what rules they created them under.

JR as a company, we set in place a rule. Do we need to check all?

CMN One reason to have a bunch of RDF is gives ability to say "this is not assessed." Can then add several assessments together. Based on a sample we believe that ... Does that provide a solution?

GR and WL are not sure.

CMN If you have good assessment tool then point to various pieces of stuff that you've assessed. Then can say statistically think it's satisfied it. When you sit and asses, you track which images you've assessed. That info is stored (x-y are checked, z-c are not). Later can check all, or sample some more.

IJ Who is "who" is going to keep track?

JR He's saying we'll combine the images that were checked. finally, have all were checked.

IJ There's a variety of evaluations. They don't combine them until someone says they want to make the combined version.

CMN There are 2 ways to assess conformance :

  1. by contradiction (logic)
  2. find a bunch of things that do.

These are not incompatible.

IJ Who is we and why?


IJ I argue that such a tool should make it possible to combine assertions from different parties. No link to whether W3C keep sdata or not.

JR Not just storing data but integrating.

CMN Those are separate mechanisms.

IJ Some people may want to integrate and compare. Probably won't be W3C. That's a separate issue. changing b/c we have a quality assurance issue. historically we've been vendor-neutral but we may make assertions based on that activity.

WL RNIB should determine it to the extent that it helps blind people.

CMN Where the data is stored and where the tool is is irrelevant. For W3C to ensure that something it provides is reliable we stick it on our server. We are already storing assertions.

KHS It saves people from reinventing the wheel. It's an info center.

IJ 3 topics:

  1. process for doing evaluations and how facilitate
  2. persistence of evaluations (and merging)
  3. verification body. should w3c have this role? are there guidelines for an entity?

Case of persistence is tertiary. Tool is fabulous, want to use it for UA. Who has the final word on valid claims?

CMN I don't hear anything about certification of claims. We're saying we have a mechanism for generating a claim and persisting in order to combine two partial claims to make a better claim. You can use or ignore any partial assessment. When ask for an assessment can choose claims by people that you trust.

KHS In 508 discussing. We're going to organize vendors, advocates, etc. who will create a seal of verification.

IJ Can't we use PICS?

CMN Can use RDF (labels).

IJ This group has not discussed certification. Do you intend to pursue the discussion? Or raised to CG? or QA?

CMN Topic available if anyone wants to raise it. Personally, doesn't matter to us.

IJ By default, any persistent claim on the W3C site has to say "this is the status."

CMN "This is not endorsed by anyone."

Libby Do you want a security mechanism on claims?

DB Can anyone look at at anytime?

WL Yes. The persistence factor is a universal factor. claims have to be dated. "Good housekeeping seal of approval" may change month to month.

CMN Stuff on web is available to everyone at any time. In order to have integrity of data, if someone creates something libel, W3C can reject a claim.

DB With this info being used in court cases, people will have to have passwords to add info and to edit their info. I'm very concerned about this.

JR I agree. Who's to say if it is clearly wrong? Will W3C then do their own assessment of every tool? Also means reading every claim.

GR The monitoring would not be done by W3C. If someone has a problem with a claim they must bring to claim.

WC If just have vendors make the claim you avoid libel. If they can't do self, then send it to them to "clear." If don't get response from that then ... Only think the collection from others is useful for the accessibility of the interface. Concern we are doing the job of the manufacturer.

JR People can make partial claims: developers making ambitious claims, others making derogatory. If W3C is not going to test these and we rely on developers.

CMN You can approach in 2 ways: generalized mechanism for people to make claims. Or we do that and we limit who makes claims on a product. I don't think people will be doing this for malicious products, out of laziness. We have a communication team to address wrong statements, we have QA to monitor as well. If someone says "there are wrong opinions about my product." W3C say, "here are the people who have made this claim. these are experts opinion."

GR Model AFB and other organizations use: whether from org or entity is sent to a contact person at the developer. They have a finite time to respond. Errors of fact are fixed and dissentions are noted. If they don't respond it goes up as is.

WL Exact version of this called, "amazon.com" results sorted by "i am the author" "i am a reviewer." etc. Must be moderator in some way.

IJ Slightly lost on scope of discussion. Piece of discussion that W3C will be making statements about claims, or making claims or processing claims. Is that in this charter? Has it been discussed with comm team? I don't think there exists a "partial claim." UA has a "well-formed claim." This is a tool to get you to a well-formed claim.

WC marketing for 508 if they do the claims and keep up to date. If don't do claim self then not in database and therefore agency X won't find it and buy Y million of them.

CMN Scope of discussion is this tool. What we have brought up here are good issues.

GR Putting onus on conformance claims on developer of tool.

IJ UA says that anyone can claim anything about anything.

/* break for lunch */


/* break into groups to discuss:

ATAG-TECHS relation to AERT


Action WC send specific edits to the list.

Action WC: ask ER WG what they think.


Jan's comments.

SMIL Techniques

CMN These are the languages in development.

JR PDF is not a w3c spec, what is the deal?

CMN Because there ought to be techniques for doing it right.


Guideline 7

IJ The checkpoints do not need to say "in an accessible fashion" - so say once instead of 7 times. The checkpoints should deal w/functionality.

CMN In general, we should point to tools where they are relevant.

WC Framework is there already, we had talked about that.

UAAG review

IJ Need people to review before last call. deadline is 12 October. go to last call on 19 October.

Action CMN: will get commitment from people in the group.

Next meeting

CMN Want one on tuesday? Next one with ER is not this week but next. Therefore if miss this week then the first meeting after this is a joint meeting. What about fortnightly meetings? The goal is to get more people there (and asking for commitments 1/2 as often). The flipside is that we would push for more discussion on the list.

Resolved: Meeting on Tuesday is cancelled. We have lots of action items and get discussion.

Next f2f meeting

CMN we talked about going to the plenary in february (last week in boston). we were one of the first groups to respond. Barring that let's go to melbourne (height of summer).

$Date: 2000/11/08 08:13:13 $ Charles McCathieNevile