12 April 2001 WCAG WG Minutes

Summary of action items and resolutions



JW The role of verifiability in the guidelines and how to determine if a checkpoint has been satisfied.

CMN There is a thing about subjective vs. objective tests.

JW Are there clear criteria to determine if a requirement has been met and can those requirements be applied automatically or not is not the issue. Is it possible to apply the criteria so that they have the same result in many cases.

CMN We could throw these requirements out all together which is stupid. We would lose important descriptive stuff. it's not just "is this well illlustrated." there are things we take for granted. You will get 6 different answers for descriptions by 6 different people.

Or we could push to a lower priority level. I think that is a mistake. It effectively trashes those requirements.

Or separating them out: here are good ideas that we don't know how to deal with. it's a good start. I don't think it is the solution. we have to work out the answers. they are clear problems for access. we are here to solve them. What are the questions I would ask myself? We haven't figured out how to explain what we mean very well. We should aim to improve it.

JW you favor my item 1.

CMN It's only a beginning point. It's not the end point. there is an item 0 in there which is that we can come up with more useful things.

CS Perhaps we can also publish that list of things that we need help with. A sticky example from a recent thread: illustration and writing style. The requirement to either write appropriately or to a certain reading level. If the requirement is write to 5th grade, I would be opposed. To the audience is sticky. Appropriate and audience are both mushy terms.

WC Writing at level causes some to lose audience.

CS USA Today written to 7th grade, New York Times 12th grade.

CMN That depends on where you are. If you are writing Japanese, you have defined levels. It is easier to say "this is 6th grade level or this is not." Basic tests: vocabulary and grammatical structure. In some languages are very automatable. In other, generally possible. If we say "write to level X" whatever we set it at, it will not be correct. So how do we decide if it is as simple as it can be?

CS There is a danger to say, "you must write to level X." People stop worrying about reading above level X which is not good.

JW What we did in 1.0 was to say, "if you are writing a particular kind of content, then there is a certain level of complexity inherent in what you are trying to write that is irreducable." So how do you determined if you've achieved that?

WC The real way to determine is with usability testing.

CS e.g., Slate. A general topic magazine, assumes knowledge. more than USA Today. We can not disallow that.

JW Right, or my thesis. It is written for people familiar with the references. It's what one expects in academic writing.

The issue is that it makes it relative to the content. Is audience something that should be taken into account? In other areas we've been reluctant to take it into account so as not to exclude certain groups of people.

CS When people define a style guide, that's where they start.

CMN I disagree that "there is no way to verify if that has been simplified." I've been arguing that there are ways of doing this - automatically. Run a tool over a document, do thesaurus substitution, "is this the same as what you said." if yes, then we've made this easier to read. The other method is essentially automatic translation but from one language to another, say "write in simple format." these are clearly demonstrated in babel fish and MS Word. If look at the tools, perhaps that will solve the problems.

JW What about people who don't want tools to change their writing style?

CMN We're explaining how to make content more widely accessible, but you don't have to do this.

CS If we are an informative document, people have used WCAG 1.0 as the basis for laws. I would be unhappy if because of work we do, it becomes a law that you have to write to 6th grade reading level. We just need to be cognizant.

CMN I don't think a law like that should be enacted.

JW I think CMN's solution has interesting aspects. Unlike other checkpoints this one is relying on tools to perform testing or substitutions. Go beyond and specficy techniques. When we discussed this 2 years ago, people who had expertise in cognitive disability were very distrustful of reading level tests. We need to find out what the experts feel are the relevant tests. There is a difference between testing and conformance requirements.

CS There are documents written entirely in Calculus. therefore a MathML document needs to be able to make some claim.

KHS One possible solution is to require a simpler language alternative. If you have to have a text alternative for images, then have a simpler alternative.

WC Yes, even if you don't write it but can find a reference to link to.

CS But i've met a physics researcher who could not describe what she does to her freshmen class.

KHS That's life in general. You can't explain to someone who does not have a child what it is like. You can try to explain, but they'll never really know.

JW Simple language versions are not equivalent.

WC Text equivalents don't really provide an equivalent of images.

JW The person who hears it knows what they need to know to use the site. If you can provide the same content with better language then do it in the original.

KHS I think that deals better with targeting a particular audience.

CS Like an executive summary.

JW Are we going to define these requirements in greater detail, if so what is the process for doing so?

CMN I get the impression we're ratholing. I am not convinced that these are different in nature than the requirements we already have. They are simply things we can't explain well enough for us all to agree on. therefore, this should head to the list. It is complex and will take energy and careful thinking. For JW to say that something could not be simplified versus something could not be reasonably simplifying. It takes a lot of work to simplify some things just like it takes work to make some equivalents. Separate "it can't be done" from "it takes a lot of work." We need more exploratory discussion figuring out what we don't know and what we do.

JW I never said something could never be simplified, but if it can be then you should do that to the primary content.

CMN You said it about your thesis.

WC What about the action to divide them into subjective and objective checkpoints?

JW My analysis was that subjective and objective was not useful. I chose not to use it in my analysis. I tried to go on the criteria, "would someone who is making a judgement tend to agree in the majority of cases." it's not clear if that is true or not.

CMN Send this to the list?

JW Waiting for GV and ASW to respond. Went through them all. "This part of it is easy to determine this part is not and here are the reasons why." Most have aspects which you could determine easily and automate without too much trouble, also aspects that require judgement.

WC Will that hold across technologies?

JW Tends to be similar across technologies. e.g. 1.1 the existence of equivalent is usually explicitly given in a markup language.

WC But the example where you have description of image in text, in educational.

JW Right, should be clear when read text. When have 2 versions of same content where no clear correspondence. then becomes difficult. It doesn't have to be explicit relationship in markup, but where there is one addresses existence but not adequacy. Where isn't, it is clear (due to juxtaposition).

WC JW You should publish your analysis. Then, I think our efforts should go towards techniques. Let's let the draft sit for a while and see how they will withstand techniques. We can rat hole on language for checkpoints, but think we should see what light we can shed on them by standing up against techniques and technologies. Have lots of ideas from last 2 face to face meetings, let's finish those up.

CMN Let's find several types of content and see how they withstand WCAG 2.0.

JW I'll publish that to the list. Then there is the need for specific criteria where vagueness is particularly evident.

WC Just so you know: i'm working with a technical editor (Cory Knobel) and hopefully a graphic artist to help clean up the language and make it look pertier. e.g. a couple of his suggestions: instead of "consistent" say "predictable."

/* WC will get together next week with Matt and Cynthia to discuss HTML and server-side techniques */

Face to face

JW any preferences?

CMN June 18-21 in Amsterdam. Along side with HTML and XForms. Also ATAG.

WC And ER would meet then as well.

CMN Another long range is Australia in Mid to late November.

WC Yes.

JW Other things going on then? Not sure about me.

WC Other option is 4-8 in June in conjunction with INET.

CMN Other offers? Hard for me to make that.

KHS Scotland?

WC U of Dundee, have said could be at a meeting, but did not ask to host.

Action CMN find a host in Frankfurt and check out Amsterdam by Tuesday.

CMN Drawback with Amsterdam is hard to get a place to stay.

KHS University?

WC 1st week of June is out - many of us in madison for HFWeb meeting. need to resolve quickly since at 9 weeks and have to give 8 week notice.

JW Distinguish between general checkpoints and specific checkpoints. That's an open question.

CMN I expect people will use "checkpoint" and mix together.

$Date: 2001/04/12 23:37:51 $ Wendy Chisholm