In a comment on a post titled "Content Negotiation Considered Harmful, Again" are two of the main reasons why XHTML isn't successful in the fields it really ought to be (such as publishing scientific articles):
Inline MathML/SVG is not worth a) losing incremental rendering in Firefox and b) coping with the differences between XHTML and HTML.
The "differences" include things like document.createElement vs document.createElementNS (which, though it can be worked around, should really be handled transparently by the browser) and having to make sure all the non-XHTML character entities are converted to numerical entities.
Update: Apparently incremental XHTML rendering will be fixed in Gecko 1.9/Firefox 3. If jQuery etc can deal with the Javascript variations that should be problem solved. Non-numerical entities are still annoying though...