Unless we define some method of negotiating character set encoding for
WWW, I think that the solution you outline above could not be
implemented, though given encoding negotiation, it obviously *is* of
great practical value.
Interestly enough, I spoke to 2 technical documentation
writer/translators who were vehemently against Unicode because it does
not support every possible character in the world... they noted that
there are more than 100,000 Kanji, most of which are not used except
in scholarly works.... and they are correct. It does not have codes
for some of the various Indian languages.
Still, as a lowest common denominator, I think it will handle most
(90%) situations, and I think *that* by itself, is a big step forward.