- From: Charles McCathieNevile <charles@w3.org>
- Date: Mon, 30 Jul 2001 08:37:46 -0400 (EDT)
- To: Matt May <mcmay@bestkungfu.com>
- cc: <w3c-wai-gl@w3.org>
I have responded to this in alternating paragraphs. It is about philosophy, which is important in the long run but may or may not be a distraction in the short term. Anyway, I think it is important enough to keep talking about it - taking a step back from the details is painful and slow, but ultimately useful for me nearly always. Anyway here it is, in the form of alternating paragraphs from Matt and my responses: On Sun, 29 Jul 2001, Matt May wrote: I need to get a little philosophical here to describe (I hope) precisely the kind of rathole this constant debate is dragging us into. The issue I have with 3.4 which precedes all others is that we are not justified in telling authors we know more about their content than they do. This is an ill-conceived assumption which will backfire if it makes its way out to authors. CMN I am going to try and show that we are not justified in telling authors we know anything about their content, but that we do not need to in order to tell them how to express it in a way more useful to more people. MM The reason the W3C has authority to produce documents such as WCAG (and have an audience) is as a result of technical leadership. It produces the protocols and formats, and for that it has the right to state authoritatively how they are used. CMN I don't agree. I think W3C can produce WCAG is that it gathers a number of experts in different areas of accessibility, and those experts work hard to figure out how to useWeb technologies in order to make content accessible. MM What the W3C does _not_ have the authority to declare is what people are to say and how they should say it. CMN W3C does not attempt to say anything about what people should or shouldn't say. It does, through WCAG, attempt to provide some guidance about how to ensure that whatever it is being said can be understood by as many people as possible, regardless of disability. MM I don't object to the presence of Guideline 3. (In fact, I think it might be better placed above Guideline 2, since most interaction is done after comprehension...) It is, however, the least technical and the least normative of the four, and that's no accident... CMN I think the reason it is no accident is in fact that it is the area we have done the least actual work on trying to set technical requirements for, and the one where it has been hardest (this is my personal opinion of my experience) to actually get any of the work that we have done incorporated into any of our documents. MM ...The full cycle of comprehension is dependent on: - The domain knowledge of the author (that is, the originator/communicator of the message); - The skills and tools of the author to craft the message; - The medium of communication; and - The skills and tools of the recipient to interpret the message as accurately as possible. CMN Agreed. MM What we need to do is to rely on the author's knowledge of the chosen subject matter and ability to communicate it as effectively as s/he is able. To that end, we can suggest best practices outside of technology to augment that knowledge (such as "write clearly and simply" and "emphasize structure"). We can even suggest to them that images or other media can be beneficial for important concepts. CMN In other words, we can suggest methods such as taking advantage of technology (for example a screenreader), doing things that are more or less technology independent (such as writing styles), or things that blend the two (using a tool to help analyse an image and produce some other representation to be cleaned up by the author). MM It is completely nonsensical, however, to create blanket requirements respecting the clarity of a message relative to the number of images present, FOG index, or any other ratio... CMN Agreed MM ...Comprehension is neither reliable nor quantifiable. If it were, there would be no need for tests in school, and anyway, if one were held, everyone would have perfect scores. CMN If comprehension could not be reliably quantified (not perfectly, just reliably) then there would have been no point to testing in schools, but it seems that most of the schools I went to and most people I meet believe that isn't the case. MM ...It is, alas, not that way. No words-to-pictures (or, for that matter, words-to-anything) algorithm makes all content quantifiably clearer, or more universally understandable. CMN This is a big claim. A really big claim. Perhaps it is correct, but I think it is also irrelevant to the real world. Any algorithm that can make some kind of content almost universally more accessible is probably a win for us, and I think there are some algorithms that can be defined well enough to be useful in the guidelines. (I have proposed 3 now). MM 3.4 should acknowledge, at least implicitly, that its effect is limited to the ability of the author to communicate using non-textual means. It should be there to draw an author's attention to another means of making a point to users. CMN Agreed MM ...What we need not to do is to turn this into anything more than a "where appropriate". CMN I don't think so. I think that we need to look for ways that we can actually lift the technical level of this to something that can be tested, and provide some more useful guidance in terms of specifics. I also believe that this is possible. I don't think we are yet at the stage, even with all the suggestions I have seen, where we can replace this general advice checkpoint, but I think we can certainly augment it usefully. MM ...It is an untenable position to say we're sure that for all content, any image you produce, irrespective of quality or subjective relevance, is better than none.... CMN Obviously MM ...We _must_ respect the role of the author if we expect to influence authors. CMN Agreed. We must consider the role of tools, and we must remember that there may be user needs that not every author is good enough to meet on their own. (for example, some people have great difficulty writing in a way that people can understand. They don't have a right to have people understand everything they write, any more than a person who cannot sing has a right to have everyone appreciate their lack of melody). MM I feel I've already made all the points I need to make regarding content in the web design process, as well as the complexity of communicating messages visually. Frankly, the messages I've seen regarding, say, the insufficiency of verbal communication with abstract art fail to sway me. In fact, they only make me more steadfast in my position: the author's grasp of her or his own subject matter is and should be the determining factor in offering illustrations. The author doesn't care what the W3C says about content, nor should s/he. CMN The author's grasp of their content, but further their ability to create or find content in different forms, will be major factors in whether or not they manage to make their message accessible to everyone. The advice of those who are working with people who have difficulty reading seems to be fairly consistent that graphic representation will help them. It is clear that bad graphics are not helpful to many people, and that graphics without text is not helpful to lots of people. We need to remember those as we go forwards with this. But we don't need to stop working here yet either. cheers Charles
Received on Monday, 30 July 2001 08:37:47 UTC