176 weeks agoWeb is too fast (XML was lucky)

I was working on my own scheme for semantic web and one thing that I got right, I believe, is that it’s core has to be as simple as possible. The reason for this is that the web got such a momentum, that simplicity has a number of benefits over completeness — if it’s simple and useful then many people will adopt the standard and come up with their own implementations and such. If something can be understood in half an hour it has a thousand more chances for success on the web compared to something that takes a day to digest.

XML was lucky not to have more simplistic competitors. As a developer I (and many others) like to have an overhead view of everything we use in our software, but that is not the case with XML — there is just too much about it (namespaces, schemas, DTD, quirks and such) so one has to settle with knowing only a surface of it. It was a good choice to make RDF simpler than XML.

I think of XML as a serialization medium instead of markup language. The name XML stands for EXtended Markup Language, but the only place it is used as markup is in XHTML, as far as I know. I really don’t understand why a markup language has to be so hierarhical. Text markup is mostly a ‘flat’ thing. Strict tree structure of XML is good for arbitrary data representation, but there are simpler ways to accomplish that (while still having human-editable representation).

OK, I’m not that stupid to question that XML rules the world of serialization now, so I’ll just get over it. I just have to make sure I use as little of XML features as possible, just to keep things under control. (I don’t understand people who try to shoot themselves in the foot by mixing namespaces of RSS, DublinCore, RDF and looking for more). Keep it simple, stupid.

Nowadays web is different than is was in ‘98, and it favours cleaner solutions much more. Newly introduced technology has a chance of becoming part of the infrastructure only if it is lightweight enough. XML-RPC is not the most all-encompassing solution, but it adds so few new concepts that it can be picked up and used almost immediately, which is reason for it’s success. (Compare to WSDL, a solution from W3C, that is aimed at the enterprises rather than humans). So I place my bid on “tiny technologies”.

Web is smart, not intelligent.

Questions?