I wish I had some slides for the typesetting section so that I could 
make a short presentation, but I don't have a thought out position to present.
I do believe the following, and would appreciate you bringing them up 
if you are not planning to talk about them in your talk:

1.  Most computations have some equation *input* and output.
    OpenMath should not ignore this fact.

2.  Equation editors, and more generally, documentation systems
    are very complicated.  All of the rationale for why OpenMath
    is good for use with computation applies to using OpenMath for 
    these systems.

3.  Converters to and from TeX, SGML, etc, should be written.
    These converters are no different *in concept* from converters in
    a computer algebra system.

    Converting from TeX, etc., is much harder because they have no semantic
    information (which is why we should try to influence SGML).
    You can only guess or use comments.

4.  Whereas a computer algebra has a programming language, most 
    equation editors do not have one.  This is a problem for extensibility.
    It argues that formatting should be done elsewhere, either the generating
    program (e.g., a computer algebra system), or by a flexible program
    filter that is included in a standard openmath distribution -- one that
    can be extended at runtime.

5.  I think that the above point argues for the creation of a typesetting 
    language (using the simple OpenMath structure) in much the same way
    as there is planned a "core" language for OpenMath.

    Eg, the language might include typesetting primitives:
        fraction, subscript, superscript, grid, ...
    Note that these do not specify exact positioning information,
    just general semantics similar to the "core" semantics for computation.
    This allows (for example) TeX to use different layout for "inline"
    expressions versus "display" expressions.

    Some provision for incorporating the semantics must be made with
    a primitive to disambiguate ambiguous notation.
    Similarly, some provision should be made to associate typesetting
    with an expression.  I think another optional field similar to the
    atribute field should be added to OpenMath.


Feel free to make a slide out of this.

Good luck with your talk,

        Neil