RE: A certain difficulty

:Jeff Sussna wrote [in response to Mark Birbeck]:
:> I agree that the point of a spec is rigor and completeness, 
:> and that it shouldn't be expected to be a tutorial. However,
:> I DO NOT agree that the invention being specified shouldn't
:> be expected to be accessible.
:
:I really don't see what the point is in making such value 
:judgements. If
:an inventor is unable to convey their idea clearly then so be it. If
:others can explain it well, then great. It is really irrelevant for how
:good the invention is.

If you define the 'good' of a specification in absence of others ability to
manipulate or comprehend it, then so be it. It seems to make more sense to
define the 'good' of a specification in terms of others ability to use it,
certaimly in network domain. That is assuming a priori one wants the
specification to be used and implemented. There is a valuable difference
between dumbing something down and expressing something clearly. Naturally,
either of these positions are value calls.

:> Generally speaking, a complicated design is a bad design.
:
:I'd like to see the mathematical proof for that one! Once again you are
:introducing value judgements.

We don't require mathematical proof to know that unneccesary complexity in
design reduces its potential. Induction from past exprience is sufficient to
give us an inkling of how to go about handling complexity in many kinds of
design. 


:> I believe the frustration with RDF comes primarily from the
:> casting of the model into XML syntax(es), not from the writing
:> of the spec.
:
:I would suggest that the biggest problem is that it is very 
:difficult to
:implement many of the truly radical aspects of RDF/S, and so 
:people find
:it hard to picture how it would work. It's also a bit odd because the
:applications of RDF are not really 'advertised' in the RDF spec.
:
:> Furthermore, inventions are only useful to the extent to which
:> they are used.
:
:A tautology Jeff ;-)



:> If an invention is brilliant but incomprehensible, no one will
:> use it.
:
:Not so sure, myself. What category would you put Bayes in? I don't
:understand it all, but plenty of people are producing some amazing
:stuff. Perhaps we're getting philosophical here ...

Bayesian probility isn't incomprehensible, far from it. I find it hard to
credit that one can understand predicate logic, XML, XML namespaces and
graphs, and not understand basic Baysian probability. Then again even the
simplest ideas can be made incomprehensible.


:> I worry sometimes that RDF will fall prey to a similar history as
:> Lisp and Smalltalk.
:
:Very different. Anyway, does it matter if RDF goes that way? All I'm
:saying is that RDF poses one way in which the 'semantic web' can be
:implemented. In a year or two's time it may well be done a different
:way, but as long as the semantic web is 'built', who cares which
:standard is used to do it? Perhaps I'm too Darwinian in my attitude to
:technology for 21st century sensibilities!

It matters because people don't want to invest in stillborn technology. I
assume it will matter to you because if you do, as you say, want the
semantic web 'built', you are not going to get it with arbitrary
specifications and hoping one of them takes off. Producing specifications
without regard to their ease of implementation is not a sensible way to get
things done.  

Perhaps we can call this strategy "specifying the semantic web by
coincidence", the absence of design being an essential thrust of Darwinism.
Such a strategy reduces the probability of the building of the semantic web
in the near term considerably. Since we are dealing in the realm of network
systems, it should be clear that specs that are not widely adopted have less
value. 

Received on Friday, 25 February 2000 04:18:15 UTC