Re: comments re draft version 2.0

David Carlisle also wrote:

> In my own case I hope to first use tex4ht to get to presentation
> MathML, and then use XSL and some knowledge about the subject matter
> of the document to convert the presentation MathML to Content
> markup. But this last stage is hard to do in general, It's easier if
> you have individual knowledge about the subject area of the document.


Such individual knowledge goes with author and reader, not those who
write formatting or translating code.

Beyond that subject areas evolve, and we do not want tomorrow's markup
to be stale the day after tomorrow.

For these reasons I think that content markup should be based on type
declarations for symbols using *relative*, rather than absolute,
semantics.  The public system of types needs to be rich enough that
expressions formed using the usual overloaded operators that we
commonly find in, for example, arXiv source can be imported into
computer algebra systems with no more than minor anchoring of the
relative semantics by the computer algebra system user.

I suppose that it's OK to have default absolute semantics for K-12*
publishers, but even that ground can move over lifetimes.  Just as the
U.S. K-12 curriculum undertook "sets" in the mid 1900's, one should
not rule out the K-12 appearance of "categories" by, say, 2050.

A system based on absolute semantics will always leave out more than
it includes.

                                     -- Bill

* U.S. colloquial usage for education commonly associated in the U.S.
with ages 5-18.