As someone writing in a version of SGML back in 1988, I instantly recognised HTML when it appeared in the early 1990s and was able to be quite productive. The essential idea, that of separating content from formatting, was much the same. You wrote text, added structure in the form of headings, lists and tables and then it was up to whatever you sent it to, to render it in appropriate fonts, typefaces and so on. All very neat.
But bit by bit the language became corrupted, with Netscape starting the rot, as I recall. From the earliest 'font' tags to later table madness and then finally HTML being bastardised into a page layout tool by people who had spectacularly missed the point of a 'Mark up Language' in the first place.
The end result is HTML which is nigh on unreadable to the naked eye. And a job for a computer to render reliably, which is partly why web sites look different in different browsers, even now in 2008.
Thankfully, the common sense that is CSS came along to partially save the day, although most web pages are still uncomfortable hybrids of old HTML, new CSS and bodged DIV and table layout. In short, it's a royal mess.
Hopefully, CSS will win out in the end and pages will get purer again? Hopefully. In the meantime, I continue to hand-code my main web page (3-Lib) in raw HTML and the code loads like lightning on most desktop and mobile devices... Not that anyone appreciates it these days...