[tm-pubsubj-comment] stability and reliability of PSIs [Re: Moregrumbling]

[Bernard Vatant <bernard.vatant@mondeca.com>]

This thread is for the issue of trust and reliability of PSIs.
Is it in the scope of the TC to recommend anything about it?

> | I'm not sure about what we want for that matter. Do we want
> | *anybody* to be able to publish PSIs?
>
> Yes! Of course! Why on earth would we want to limit who is to be able
> to publish PSIs?

I agree with that "internet spirit freedom" in principle ... But look at what happened for
example for html <meta> tags. It was supposed to help search of relevant resources, and
was let completelely open, free and dreadly simple to publish. The result was quickly so
awful, and with such terrible misuse and abuse, that search engines do not use them any
more as reliable features, and have built alternative complex strategies to figure which
are reliable resources and which are not for a given keyword, in which notion of
"authoritative" pages is important. See for example this recent interesting article.
http://searchenginewatch.com/searchday/02/sd0613-links.html

The same way I'm afraid for example Semantic Web strategy based on RDF metadata is bound
to meet the same pitfalls, I think that if it is as simple to publish a PSI than to set a
new URI and put some <meta> under it, we're bound also to the same kind of disaster. That
is my only concern. Now I don't figure if it is in this TC scope to recommend anything
about it.

> | Is not that somehow in contradiction with the need of stability and
> | trust?
>
> It is not for us to decide what need people have for stability and
> trust. I don't see why we should make decisive judgments about this on
> behalf of all the people who will ever use this.

Of course. But we have to stress the fact that PSIs need stability and some kind of
authoritative label to be reliable and trustable, if they don't want to meet the same
pitfalls as <meta> tags did meet.

> I'm sure some PSIs will be unstable whatever we do, but I don't see
> that that's a great tragedy. Published subjects. like the web, do not
> break just because a URI returns a 404. Things still *work*, only not
> quite as well as before.

Of course. But the less 404 you get, the better you are :)

Bernard