[topicmapmail] PSI repository

[Murray Altheim]

Alexander Johannesen wrote:
> Murray Altheim <m.altheim@open.ac.uk> wrote:
> 
>> Because this has little value if it's "Murray's PSI Repository" or 
>> "Alexander's PSI Repository", or hosted by a vendor or even a 
>> university, the question of hosting venue is really paramount.
> 
> You don't mention why this is. In my view, if there was a
> repository, I'd use it. Maybe I'm too pragmatic to get it,
> but any repository is a good start. I don't care who hosts
> it, as long as there is an open license to use it and a
> generosity in the service. Is there some hidden integrity
> needed that I've missed?

If you're a hobbiest, you might not care about longevity, security, or
trust, or about the specific domain being used. But if you are going to
build upon a set of PSIs and are running a business or work in government,
you likely have a very different set of criteria about what consitutes
acceptable practice. For example, I've worked at NASA and at Sun, and
neither would use PSIs that started with "http://altheim.com/...". Sun
would either use Sun's domain for its own stuff (like Java), or would
use an industry consortium or standards body. NASA would likewise
probably use a NASA domain or a standards body (but probably not an
industry consortium, though I'm not aware of any rules against this).
There's a US Federal XML Working Group whose task it is to work out
such policies and best practices.

> Say I put up Alexander's repository. Why would that be
> a bad place for all of us to start sharing the PSI's?

There's no reason for you to keep from putting up the repository, but
I don't think you'd find too many takers from among certain sectors.
Now, you could put up a registry (rather than a repository) and you
might have interest. A registry differs from a repository in that it
simply registers the location of PSI sets, pointing users to where
they exist, but doesn't physically maintain them. If you got lucky,
you might end up being the Google of PSIs. But you'd have to figure
out what motivation people would have for using your particular
registry, as people (and the net) tends to follow network effects,
i.e., gravitate towards popular sites.

>> This is likely to
>> be handled within various domains by organizations within those
>> domains, such as say, the United Nations, the American Medical 
>> Association, OCLC, US Library of Congress, Reuters, Spanish Knitting 
>> Association, etc.
>
> So why these, but not a university or even a company?

Because universities generally don't have it within their charter to do
this kind of thing, so it's usually hosted by some faculty or staff
member out of interest. This kind of thing costs a lot of money to
manage (in terms of people time, but if popular it might even impact
computing/network costs).

As for companies, they have proprietary interests. And even if they
don't, there's no reason for other companies (particularly competitors)
to trust that they don't. And they are notoriously unstable, even the
biggest ones, both in terms of how they manage projects, and their own
existence. They die and are bought out by other companies, change their
names, etc. If the petroleum industry had used a bunch of PSIs that
started with "http://enron.com/" they'd have a problem right now. And
even if Enron was still around, why would Exxon or BP trust them not
to alter or remove the PSIs?

The reason I mention some of these organizations is that they already
publish things similar to PSI sets. Now, the Spanish Knitting Association
is too small (I'm guessing, but you never know about those Spanish
knitters) to warrant inclusion of their PSIs at the United Nations, nor
would the UN likely consider an application, so they'd probably just
publish their PSIs on their own web site. That's what I'm doing with my
own work.

Note that even the W3C has decided that they don't want to get into the
business of registry/repositories, last I heard. It's not really in
their remit.

>> If you're talking about a general publication venue, the most obvious 
>> one for this group is probably OASIS.
> 
> Ok, so is anybody interested in starting such a body?

You sound interested. You just need several OASIS members.

One other venue that might work would be to create a sourceforge.net
project (or under some similar venue) whose purpose is to publish PSIs
and try that. But I think your best bet (in terms of getting a wide
variety of interest) is to do this through a trusted consortium that
has this as an existing task, such as OASIS.

Put it this way, the reason you don't see PSI repositories popping up
is because they're expensive to set up and maintain, require a trusted
host, and because individual publishers often have venues within their
specific domains that are already suitable in communicating with the
communities of practice that matter to them (e.g., Spanish knitters).

I don't mean to be a wet blanket, but until you could demonstrate
legitimacy, longevity, etc. to someone for your site, they won't use
it. This is of course true of many things in life, not just PSI
repositories. Because PSIs are by their nature meant to be stable
within their intended lifespan, a repository site must be just as
stable.

Murray

...........................................................................
Murray Altheim                         http://kmi.open.ac.uk/people/murray/
Knowledge Media Institute
The Open University, Milton Keynes, Bucks, MK7 6AA, UK                    .

   Rather than join the 81 signatories to the Interational Criminal
   Court, the Bush Administration has blackmailed 39 countries into
   signing bilateral agreements granting Americans immunity from war
   crimes. What do they fear?
     http://www.guardian.co.uk/yugo/article/0,2763,975476,00.html
   USA for the International Criminal Court: http://www.usaforicc.org/