LinLog

I've been thinking a little about HTML 5 today, thanks to a blog entry by Roger Johansson on taking the semantics out of HTML. Apparently, the HTML 5 WG mailing list as gone off into loony-land and Roger is a bit concerned that the accessible, semantic markup he and other have been promoting for years is going to fall by the wayside in order to make things easier on browser vendors. And so far, his fears are not without some justification.

I can't claim to be an expert on HTML in general or the HTML 5 development process, but from what I've read, things do seem to be getting a little...weird. For instance, the requirement that all pages served as text/html will be treated as HTML 5. Apparently the idea is that HTML 5 will be backward-compatible with all existing versions of HTML and, I would guess, XHTML 1.0. Or rather, it will have to be if they intend to keep that requirement. Otherwise they really will "break the web."

There's also this item from the HTML 5 FAQ responding to the "HTML 5 legitimates tag soup" claim. It describes the difference between requirements for conforming documents and requirements for user agents. For example, user agents are required to support the MARQUEE tag, but conforming documents cannot contain it.

My question is: Is this a good thing? At this point, I really don't know. It seems like the W3C is trying to codify the the tag soup that browsers have been forced by circumstances to support into a single, coherent standard. On the one hand, this is undoubtedly a good thing, as it will (if done properly) solve the problem of different user agents rendering things in inconsistent ways.

On the other hand, is this really solving the long-term problem? In the marquee example, I understand that the document and user agent requirements are orthogonal. But if user agents are required to support a tag, won't people feel entitled to use it? And if people aren't supposed to use it, wouldn't it be better to discourage that by deprecating it or writing it out of the spec altogether? Am I just being naive here?

What I'm getting at is that I'm not sure how (or even if) this approach is supposed to fix the generally miserable quality of markup on the web. People relying on the browser to implement things a certain way is what got us into trouble in the first place. By stating that browsers will now be required to implement MARQUEE in perpetuity, but you shouldn't use it, isn't that effectively giving people a license to use it? If the people writing HTML cared what the specs recommended, we wouldn't have MARQUEE in the first place. I can easily imagine people taking the browser rendering requirements as gospel and simply disregarding the document conformance requirements.

While I do have some sympathy for the forgiveness by default philosophy, I think we need to focus a little more on discipline. The laissez-faire attitude of the 1990's left us with a web that was held together with duct tape and bailing wire. Surely we want to do better than that. I'm not necessarily advocating the strict XHTML "validate or die" approach, but surely there must be a happy medium between that and semi-legitimizing every ill-advised vendor extension to HTML out there. Surely.