I think this discussion is drifting out of scope... please... no more
book reviews. The only part of it that is relevant to the WG is the
basic tenent that should be driving all of the efforts of this group:
To promote a standard for HTML that will support the consistency
that Peter Sheerin asked for, so that content providers will be able to
provide data that will be rendered in a predictable, consistent manner
across platforms and browsers, without having to proof their content in
each possible combination.
I really don't think this has anything to do with user guides (which
are, of course, needed, as well). It has to do with tightening the
spec such that there is a core tag set and rendering model that works
in every compliant browser. This in no way precludes extensions, but
rather guarantees that some basic set of capabilities will work in a
predictable, stable way in every browser.
As we near the officialdom of HTML 2.0, I wonder if Dan Connoly's
notion of a compliance data set should be resurrected?