1. A commonly-offered definition ("data about data") is vacuous in
actual practice, if not conditioned by further information, since it
indicates no operational or essential distinction between metadata and any
other kind of data. That is, all data is data about something, and
consequently there are no structural, operational or other requirements that
obtain for data-about-data absolutely that do not also obtain for other
kinds of data.
2. I actually think that there is a valid use of the term "metadata,"
but I notice that the same word is used in different contexts to mean very
different things, so I need to know what it means _here_.