Xml Of Babel

I'm an old fashioned guy. When I was a boy, Fortran was a pretty neato kind of thing. I mean I'd stay up late wishing I had a computer powerful enough to compile my Fortran. I liked Fortran.

I very soon discovered the error of my ways. I won't recapitulate the wonderful world of languages here. It suffices to say that I don't like Fortran any more. In fact, I don't like Fortran.

In FortranLanguage, all variables with names beginning with the letter "I" are integers. A holdover of that is the well-dreaded HungarianNotation. Which led to all the horrors of M$ naming standards we won't go into here. Suffice to say I particularly don't like this about Fortran now.

The integer conventions come from mathematical notation conventions if I am not mistaking. Fortran was designed with mathematics in mind. It is even how it got its name. And I have to admit that when I see "i" and "j" in programs, I pretty much know it is an array index. I cannot endorse the practice, but it does have enough advantages to make it a low-level gripe (as a programmer convention, not a type declaration system).

But there was something useful about it. If you knew it, and the compiler knew it, and the other programmers knew it, then it was an unquestioned fact of life. You didn't need to describe it to anyone. It didn't appear in your source code in a triple. Over and over again.

Now with the wonderful SemanticWeb it could. I won't try to express it in triples. I don't want to express it in triples. And I don't see what benefit I could ever possibly get out of expressing it in triples. To express it in triples, I'd say, is a solution seeking a problem.

And that's what most XML seems like to me. A million cutesy little solutions seeking problems. How much thought, how many man years - man aeons - man millennia - are going into making all these solutions for problems we don't have?

Well, how many were wasted reinventing ways to fit your old non-XML systems together? How many patches on patches on patches of systems did you write to get information from one cranky old legacy into another cranky old legacy?

The point of XML isn't the cutesy SemanticWeb stuff - that's just the same old ivory tower language wankers doing the same old over-engineered claptrap. It's WebServices. Which is to say, it's making programmatic interfaces to an enterprise's systems publicly available over the Internet, or privately available over the intranet. And that's bloody useful because it means you don't have to re-engineer your cranky old legacies ever again - just use what they do and forget how they do it.

Parallel: http://www.well.com/~doctorow/metacrap.htm

Please note that this paper is rather tongue in cheek and ignores a rather largish portion of the technical aspects of metadata creation and organization. That doesn't make it any less funny.