Re: [topicmaps-comment] Genetic PSIs [Re: Topic Map domain,paradigmatic PSIs ...]

["Steven R. Newcomb" <srn@coolheads.com>]

This letter speaks to the question of whether to
provide the foundational subjects (the ontology) of
Topic Maps in a single, easily-administered canonical
place on the Web, or whether to provide them in
multiple places, the invocation of any one of which
will be considered an utterly valid and canonical
invocation.

I agree with:

  * What Bernard said.  Enthusiastically.  Replication
    is the strongest -- and only -- way to guarantee
    the survival of knowledge.  The existence of
    multiple addresses for exactly the same subjects
    will protect investments in topic map assets.  And
    it's not expensive to do it!

  * What Lars Marius said, at least partly.  Until Web
    servers become capable of reporting whether two
    address expressions address the same location, we
    must fall back on the heuristic of comparing
    addressing expressions with one another in order to
    decide whether they reference the same binding
    point.  However, the necessity of using this
    heuristic does not prohibit us from having, and
    uniformly advertising, multiple PSIs for the same
    subjects.  Indeed, I think that if we advertise
    multiple addresses, any one of which must be
    regarded as being perfectly canonical, we will be
    advertising an otherwise-little-understood aspect
    of the robustness of the semantic integration
    capabilities of the Topic Maps paradigm.

  * What Eric Miller said.  We must enable people to
    publish their own ontologies, and to do it in
    radically useful ways, without unnecessary
    constraints, and without depending on authorities
    or having to ask permission.

Decentralization and redundancy are the keys to
permanence and stability.  This lesson is to be learned
everywhere we look, not only in biology, but also in
human affairs, and in systems engineering (viz. the
internet).  If we're really trying to get the most out
of our knowledge assets, we must avoid a situation in
which there is any single thing on which they all
depend.  Even if that single thing is ISO, or W3C, or
OASIS, a single point of failure is a single point of
failure, no matter how much we may believe in its
sanctity or indestructibility.  All empires eventually
fall, but knowledge can be preserved for the benefit of
future generations.

We should focus our efforts on protecting the value of
other people's information.  Truly, we are *all* "other
people".  So let's protect our assets from accidental
loss.  In that spirit, let's freely and visibly exploit
the fact that Topic Maps are designed to support
redundant ontological foundations (sets of binding
points for exactly the same concepts).

-- Steve

Steven R. Newcomb, Consultant
srn@coolheads.com

Coolheads Consulting
http://www.coolheads.com

voice: +1 972 359 8160
fax:   +1 972 359 0270

1527 Northaven Drive
Allen, Texas 75002-1648 USA