RE: live meaning and dead languages

Thanks Alan

I'll make in line responses, pruning your messages:

> My first reaction to your comment is that it is, at least, overbroad.

I agree.
It was partly intended as stating one extreme of the spectrum, also, it was quite explicitly my thoughts while reading a book by Cupitt who is very radical in his views.

> At worse, the attitude it espouses is a threat to the utility and
> success of the Semantic Web.

We clearly have different views as to how the SW may succeed. I think time will tell; and I also suspect that there may be more than one sort of success in different areas, and different underlying philosophies of meaning and change could undergird different successes.

> 
> Some cases to consider:
> 
> 1) The Semantic Web is for science too. Within that domain, precise
> definitions of many terms are essential if one is to make meaningful
> comparisons.

Scientific language often tries to give a precise snapshot of the state of an on-going historical discussion.
Scientists tend to believe that the current consensus is in some sense 'real' (about a world out there). But history shows that the consensus changes, less rapidly than some other fashions, but still a fair rate of change.
With such change the terms change, and ideas that use to make sense, seem increasingly incoherent (not just wrong, but unintelligible). A different way of thinking about scientific language, a non-realist account, is to look more at how the scientific language works in and with itself (including experimental reports etc. tying us back, to our more basic language usages reporting sense experience). In this way, the classic dispassionate observer reporting reality can be seen as a useful linguistic device, but not more than that. 

I think the use of description logic to critique scientific descriptions is a very useful advance in our ability to make large self-coherent descriptions. I doubt it will change the fact that today's description will be out of fashion tomorrow. & I do not see such applications as a fundamental use case for SW technology. [Although the scale of such tasks do often have global aspects ...]


> 
> 2) Your comments don't square with the experience with programming
> languages, where having terms (functions, operators, data structure
> constructors, etc) not being precisely defined leads to unportability
> and difficulty in maintenance. Why are SW representation languages not
> like programming languages?

I believe that before the Web hypertext systems viewed a reference error as a disaster.
My understanding of the two new ideas that make the Web work are:
- the URL
- 404

404 was revolutionary in that errors are to be expected, and dealt with, without it being a disaster.

I think that comparing semweb to programming languages then we need to have equivalent techniques in which semantic mismatch is as everyday an occurrence as page not found. Not desirable, not good, but not a catastrophe.

Logically this aligns with a paraconsisent position that rejects the notion that single logical inconsistency invalidates the whole model.


> 
> 3) Is the evolving nature of words, while undeniably operant,
> something to celebrate? 

Yes :)
I cannot imagine a world of perfect communication. 
I fear it might be rather dull.

> Perhaps on the level of celebrating creativity
> it is. But the flip side is that the fact that we have so many natural
> language means that large groups of people have no way communicate
> with each other. Even within a single language dialect, idiom, and
> drift over time serve to put in place barriers to communication. 

Along with the joys of trying to understand another's point of view, of getting to know new people, new places, new cultures, new ideas.

> I
> think it would be fair to say that this inability to communicate has
> had dire consequences in our history.
> 
> Why do SW languages need to mimic natural languages? Why not let each
> serve a distinct (and to be celebrated) function?

Natural language too allows for specialized sublanguages that have distinct functions.
These sublanguages (including for example, mathematics, or Java), while allowing humans to communicate, have additional rules to make them more appropriate for distinct functions.

I certainly have nothing against such specialized sublanguages, and in fact, I enjoy them - but I try not to confuse them with the world, or to see them as privileged over the world.


My guess is that the sort of end users my current work is meant to be serving, are mainly ones who are talking (in natural language) about business problems that in many ways are closer to the sort of thing natural language is good at, than the scientific type problems that you mention, which are problems that natural language addressing by spawning specialized, more formal, sublanguages. Thus, at least professionally, I suspect that the sort of amorphous flexibility (and certain lack of rigor) may be a feature rather than a bug.

On a more personal level, I find the idea of social networking type SW apps appealing, where I fear that formal rigor can all too easily be constraining (cf. my comment on gender in the OWL2 primer [1]).

> 
> -Alan
[1]
http://lists.w3.org/Archives/Public/public-owl-comments/2008Nov/0005

Received on Saturday, 7 February 2009 18:16:27 UTC