Certain ideas are so reasonable that it's hard to imagine you ever considered any other alternative. The Earth revolves around the Sun. Tomatoes are edible. A program without goto statements is easier to read. XML is a great way to describe data.
Prior to XML, every time somebody needed a way to exchange data, they spent days making decisions that were ultimately beside the point: instead of working out the characteristics of the particular data in question, they either had to work out a new scheme for exchanging data in general or had to adopt one of over a dozen cumbersome existing approaches.
Prior to XML, every small discrepancy between how two companies represent data was a crisis. Now, if both conform to XML, it is reasonably straightforward to use XSLT (itself an XML application) to convert between formats.
In retrospect, one of the most remarkable things about XML is that it took so long to happen. Goldfarb, Mosher, and Lorie invented Generalized Markup Language (GML) in 1969; GML slowly evolved over the next seventeen years into ISO-standard SGML. Full-blown SGML is powerful enough, but it's cumbersome.
The much lighter-weight (but narrowly focused) HTML provided - both by its general success and by some of its problems - the impetus for a small, simple, web-friendly language to model data. XML models data just as well as SGML, but years of collective maturity have led to the design of a much simpler solution.
This simplicity has already proven key to XML's success in many areas where SGML never caught on. It looks like XML will become the Mother of Standards: some of us expect that the majority of data-related ISO standards adopted in the next five years will be XML applications.
(You can click here to stay on the main line of this article and skip the sidebar below.)
If simplicity is a virtue, XML is saintly.
In contrast to HTML 4.0 and earlier versions, all XML documents (including XHTML documents) must be "well-formed." Many existing HTML documents make sense to one browser while their syntax totally confuses another. This is because HTML is a very "loose" standard, so that when a browser sees sloppy HTML, it must do its best to make sense of it.
In principle, smarter browsers sound like a nice thing. In practice, they're not. Ask anyone who ever wrote a document, had it look great in Internet Explorer, and discovered Netscape couldn't make head or tail of it. Not to mention anyone who ever tried to write an HTML browser.
XHTML doesn't solve all browser differences (you can bet that future browsers will still have proprietary features). Still, it tremendously increases the chance that a document that looks great in one browser will do just fine under another browser.
My e-mail address is [email protected]. Normally, I check this at least every 48 hours, more often during the working week.