Re: [xml-dev] Are hyperlinks presentation or content?

["W. E. Perry" <wperry@fiduciary.com>]

John Cowan wrote:

> But knowledge of what the source provides and knowledge of what the target
> expects does *not* constitute knowledge of the mapping between them.  If I send
> you a document in Piat, having that message in hand and knowing what you want
> to find out from it leaves you not one whit forrader. Nor will a whole stream
> of documents in Piat help you.  At most you can
> do traffic analysis.

What I can do is answer first the question 'can I do something useful with
this?'. The answer will occasionally be 'no'. Remember that I am a node rendering
a service in which I am specifically expert. By design, I do not require that
users of that service (downstream from me) or sources of data which that service
requires (upstream) share specific knowledge of my expertise, as they must if I
demand particular instantiations of data as a condition of performing processing.

You and I have had this conversation before, and as before we are now addressing
different problems. Your objection makes it clear that you want a guarantee that
submitting data to my processing node will result in an execution of my process
and a useful result. By design there is no such guarantee. A significant part of
the expertise of my node is in knowing what its own data needs are and thereby
being able to determine what input it can or cannot do something useful with. By
contrast, I am seeking to provide a node which can in most cases perform
specifically expert processing for users who only generally, even vaguely
understand what it does, or how. The benefit of my approach is that the user is
relieved of responsibility for a priori agreements with me, just as those
agreements are the most significant cost of your alternative. The cost of my
design is that some fraction of the processes which it would have been desirable
to execute cannot be.

Respectfully,

Walter Perry