November 30, 2007

Interesting analysis of the fragility of SemWeb meta-data. Calls for explicit human viewable and "auditable" meta-data.

Of course, it's hilarious that the SemWeb people haven't yet managed to solve the problem of how to "address" things with URIs. :-) After all, people have been addressing things successfully with URLs for years.

What is apparently difficult is being clear exactly *what* is being addressed. Because SemWeb promises we can be unambiguous about that, but then seems to have problem distinguishing whether we're talking about the "sense" or the "reference".

In contrast URLs have no problems refering because we're not too fussy about what they refer to.

Could there be some kind of "uncertainty principle" here? One which says you can know "what" a thing is or "where" it is, but not both at the same time?

1 comment:

Scribe said...

Sounds like the more you try to define abstracted concepts digitally, the more trouble you get in. What we have so far (generally speaking) is definitely a hack, AFAICS: everything boils down to 0 OR 1, so you're trying to squeeze an infinity of existence and comprehension into a (false?) dichotomy. Is everything built on top of that (technically speaking) then inherently a hack too?

The net works so far because it's fuzzy. Humans are fuzzy. Language is fuzzy. What I mean isn't necessarily how you interpret it, but it does the job, because we're not the same person.

There's probably some similarity with the idea of measuring happiness here...