- From: Scott Luebking <phoenixl@netcom.com>
- Date: Mon, 20 Dec 1999 11:02:39 -0800 (PST)
- To: charles@w3.org, phoenixl@netcom.com
- Cc: w3c-wai-gl@w3.org
Hi, Charles I tend to be very user-focused. I believe that the goal should be sending a web page which is easily used by blind people who have an average technical background and/or average motivation to learn technology. I think the demo I've come up with on what a page optimized for blind users could look like is very much along the line of this goal. Do you agree with this goal? Do you think that the demo web pages are consistent with this goal? If you agree that the structure of the demo web pages is desirable for blind users, the next question is how to create web pages for blind users which have this type of format? Here's my technical analysis of web page transformation working from a computational complexity point of view. First, we're talking about a transformation T such that T maps page P into page P' with certain desirable attributes. As anyone who has been in this analytical position knows, you got to see if T is going to be possibly an NP-complete type or an unsolvable transformation? Looking at the type of web page which is desirable for blind users, it is unlikely that the transformation can be easily decomposed into smaller steps. For example, how would the transformation easily categorize the links on the page or blocks of text in terms of relevance. The context around the links or blocks of text would need to be analyzed. So, the transformation complexity is probably at least the level of NP-complete. As soon as I figure out something is going to be NP-complete I sure as hell see how I can get out of that situation rather than spinning my wheels finding solutions to a problem of that level of complexity. One of the usual approaches is to decrease the constraints on P'. However, to do that, the ease of use would need to be compromised. The second approach is to increase the constraints on the initial page P. However, often the constraints need to be so severe that P is almost like P' any way. The third approach is to abandon the transformation technique if another technique is available, e.g. in this case generating web pages dynamically in various formats. (Now you might ask why CSS which is a transformation of P into P' is attainable. The issue is that the transformation can be broken down easily into a set of transformation rules. Because of that characteristic, CSS is not NP-complete and therefore readily achievable.) The dynamic generation of web pages in different formats is actually consistent with the goal of XML. The XML provides for the information to be stored in an abstract form which then allows the information to be expressed in different formats. If a search engine retrieves information stored in XML, then expressing it in various HTML web page formats is very consistent with the goals of XML. Retrieving well-structured data, even if not in XML format, still can provide the basis for multiple page formats. There is a difference between blind people and sighted people getting quick access to the most important information. Sighted users can skip past the underbrush because they see where the main information is. The underbrush can be logos, headlines, visual layouts, etc. The visual design for appearance is at odds with what blind users need. A simple design is often not as visually interesting, even though it is easier for blind users. In terms of the groups you suggested, the needs of the groups can be in conflict. For example, some people with cognitive disabilities may prefer more graphics and less text which is the opposite of the needs of blind users. I think that asking web page designers to do both is putting more of a demand on them then doing separate formats. The number of formats probably depends on the application. I think that it is pretty clear that the preferred format for a sighted user is often different than the optimal format for a blind user. There is great variety in terms of needs for low vision people. I'm not sure a single format is achievable, but additional research might help to identify if there are certain key factors. If a web page has no information in auditory representation, like most web pages, then a separate format is probably not needed unless the issue of English familarity is being take in consideration. As a physically disabled person, I don't see that much need for special accomodation in web pages except perhaps links with larger text, larger form elements or fewer image maps. (These are pretty simple and rarely require an overall change in web page structure.) Web page design to accomodate cognitive limitations probably would require more research too. So, I think for many applications I see two main types of web pages, one for sighted and one for blind. (The sighted one could be enhanced for features to help physically disabled and research is needed for low vision and cognitively impaired.) I think supporting two formats is not that much work especially if the software is designed to express the stored information in multiple formats. Again, I'm not saying that text-only is the answer. The organization of the page is very important. I do not see that what I am proposing is not in the direction of a web anyone can use. However, the issue might be the interpretation of that phrase. Is it that anyone can use any page or is it that anyone can get the information and functionality provided by the web? I don't believe that it is achievable for everyone to have the same experience of the web unless the experience is limited for some users in order that other users have the same experience. Do you believe that many web users would support having their web experience limited? Scott > Scott, > > I agree with all your premises. But I think in addition to those itis true > that non-blind users benefit from having the most important information > first. Have a look at the work of Jakob Nielsen - http://www.useit.com/ - or > of UIE - http://www.uie.com/ - among others. > > I think you should consider deaf users, users who have low vision, users with > cogitive disabilities, users who have motor disabilities, and the various > combinations of these. Which leads me to guess at something between six and > twelve presentations. > > The basic point is that the work spent on producing these differing > presentations, which will have to be updated each time the structure of the > "primary" content is updated, would be better spent in most cases on > improving the accessibilty of the original. A case in point is a website > under development that Dave Poehlman recently posted to WAI-IG about. It has > a double version simply because it uses images to provide link text in a > fixed format. This is a straight-out mistake in my view - a little more > thought would have shown that they could do away with the second version and > provide a single version which meant there was no second design to think > about if they decide to change or update. > > I think reducing the amount of work required is a sensible goal, since the > easier it is the easier it is to convince people to do it. (Generally - if it > was always true the world would be a different place.) However I think your > approach to doing so is less likely to lead to the desired results, and at > the same time is (like "text-only versions") likely to become accepted as > sufficient, effectively leaving us well short of the real goal, which is a > web which can be used by everyone. > > Charles McCN
Received on Monday, 20 December 1999 14:02:42 UTC