In RSS vs. Atom, Erik writes : "XML may not have been intended to be a manually manageable format, it is used that way everyday." I both agree and disagree on this sentence, because it mixes two camps of the XMLers. For the youngest ones, I must notice the 1999's "XML in 10 points", which says XML is text, but isn't meant to be read. Despite the firm will to prevent XML from being humanly read, it is text. This was particularly true when we only had DTDs, which are human friendly. Do the test, try to navigate with incremental search inside a DTD with entity, element and attribute: quite easy. In the same text, it is sayed that HTML was never meant as well to be human readable, which is a half-truth to say the best. Even if Tim Berners-Lee made many efforts to provide tools to author and browse the new WWW, HTML was designed to be authored easily and quickly without dedicated tools. This goal has been completely lost between HTML 3.2 and CSS2, but now we can make beautiful human-readable HTML code, and that's why so much HTML is still done by hand nowadays.

At some point in the process, some years ago of course, companies like MicroSoft decided that DTD was not enough, so was born XML Schema. XML Schema is the perfect illustration of how a text format can be human unreadable. It has frightened lots of newcomers (well, at least me), because at the time in 2000 when I was reading Dr. Dobbs Journal, there was no compelling reason to use Schema. Strong validation wasn't a requirement when the primary thing to teach was the difference between "well-formed" and "valid". So both camps still exists, the industrials who made us realize that 'S' in "SOAP" doesn't mean "Simple", and the artisans, happy to have an accessible hierarchical text format.