On Tue, Jan 25, 2022 at 10:27 AM Rick Jelliffe <
rjelliffe@allette.com.au> wrote:
> [...] JSON is clearly more expressive from the POV of how many types of data and
> structures can be represented directly [...] I would say XML is more expressive from
> the POV of the purpose-roles of data that can be represented directly with the available
> delimiters [...] XML is also more expressive from the POV of what constraints can be
> enforced [...]
All of this is so interesting to read. So, there's some things that needs to be made a
bit clearer which is what role semantics or expressiveness has within and outside of
standards. The freedom of JSON is worthless unless there's some knowledge about
how the data is to be interpreted, and the constrictions of XML are worthless in an
open data model. There's a funnel of ideals of data exchange from the least
constrained to the most constrained, both in terms of the structured data as well
as the models they're wrapped in, and I'm not even going to attempt claiming that
there's any superiority to anywhere along the funnel in terms of usefulness; the
"winners" of the standards wars have always been lurking in the middle of the bell
curve of history (I think), making things possible without satisfying everyone. There's
a kind of beauty in these "flawed" models and structures, both in what they currently
enable as well as what future standards they inspire.
I feel the same way about RDF, where epistemology has gotten in the way of
doing things "properly", so we try with what we've got or the models we cooked up
at the time of writing, and then try to evolve things over time. RDF might be a fine
intermediate to the next level, it seems to be holding on, if nothing else from the
lack of good alternatives.
Also, a good standard "language" of data interpretation is not the same as a simple
implementation of said standard. You can probably create a standard really close to
some ideal of data exchange, but that will probably suffer from how hard it is to
implement and use (and vice-versa). Semantics of data is a fun exercise that soon
turns to torture and pain and suffering when you try to get expressive.
Cheers,
Alex
