Brion Vibber wrote in part:
Limiting / stripping / segregating HTML now will be a huuuuge advantage in the future, when HTML is replaced by QLZML and nothing supports HTML anymore and we still want people to be able to read and edit the 10 million articles in Wikipedia.
No one even today wants the current QLZML (that is, HTML 4 or XHTML 1). We don't even want all of HTML 2. As QLZML increases the potential vocabulary, we can decide for ourselves what, if anything, we want to include, and how. But we can decide now to accept <tt> and <table> forever -- always parsing them into whatever output markup language is current (*not* leaving them unparsed as we do today).
-- Toby