- From: Al Gilman <asgilman@access.digex.net>
- Date: Tue, 13 May 1997 12:49:41 -0400 (EDT)
- To: raman@Adobe.COM (Raman T. V.)
- Cc: jim@arkenstone.org, w3c-wai-wg@w3.org (WAI Working Group)
Yes, T.V., the blind using speech will be beneficiaries of people investing in technology to inform the auto driver. And thanks for the reminder on how we can sample some good stuff. I still want to deal more with another issue Jim raised: "What are our expectations vis-a-vis the allocation of work between client-side and server-side processes?" I am woefully ignorant of the CSS architecture, and I want to admit that my expectation that it is oriented to incremental-refinement tree transforms is just based on general experience and the way people use LaTeX and HTML today. I come at this as a born-again relation-ist. I am worried that people are not going to readily recognize that what will make the Web work best in a speech-output browse is a graph transform on the web of resources that the sighted browse employs, and not simply an alternate presentation of an HTML file viewed as an information tree. My current hotbutton is ensuring the graph integrity of source material so that ability- and preference-smart clients can manage both communication and rendering to construct the virtual-document view that best supports a speech-output dialog. For a concrete example of what I mean by mixing and matching comm and data solutions, check out the "percolation" proposal under http://www.access.digex.net/%7Easgilman/web-access/announce-two.html Am I saying absolutely that adaptation per ability should be in the client? Not exactly. But while I agree with the WAI proposal that we want to increase the understanding of accessibility issues among the information sourcing population, to make this initiative a success I believe we need to _minimize_ the level of understanding required of people involved in the authoring process. It has to be mostly effortless. The "information theory hypothesis" is that we want to ensure the integrity of the information sourced, and not ask for multimode sourcing (although it will be supported in the overall framework). I hear enough screams already from the information sourcing community about how they wish they didn't have to scrub two versions of their web pages. Having the capability to render according to alternative styles is necessary, but not sufficient for a win. Some kernel of information integrity requirements must be satisfied for the diverse renderings to inform. Finding "a few good rules" is one of the critical success factors of this enterprise. -- Al Gilman
Received on Tuesday, 13 May 1997 12:50:15 UTC