The Markup War Is Over: W3C Folds XHTML 2, HTML5 Wins

For most of this decade, web markup has had two competing futures, developing in parallel and pointed in different directions. XHTML 2 was the W3C’s clean-slate vision. Markup done right this time, rigorous and pure, free of the messy web that exists. HTML5 was the evolutionists’ answer. Extend the language the web actually runs on. Specify how real browsers really behave. Break nothing.

HTML5 began in semi-rebellion by the WHATWG, a browser-vendor working group, and was later adopted into the W3C’s fold. In July the W3C ended the suspense with paperwork. The XHTML 2 working group’s charter will be allowed to expire at year’s end. Resources go to HTML5.

Two months on, with the community’s reaction settled into consensus, it is worth recording what happened and what it means for the people who build and buy websites. The answer to the second question is happily boring.

Why the pragmatists were right

Our view: the right side won. XHTML 2’s purity was real. It was, in several respects, a more elegant language. It was also unshippable, because it refused to stay compatible with the billions of existing pages and the browsers that render them. A markup standard the installed web cannot adopt bit by bit is not a standard. It is a proposal for a different web.

HTML5’s founding philosophy is less romantic and far more useful. Don’t break the web. Specify reality. Evolve in place. Its working method documents how browsers actually parse and handle errors, so implementations converge on shared behavior. That attacks precisely the cross-browser chaos we have chronicled since 2006.

The browser vendors’ de facto lead of the process raised eyebrows, fairly. It also guaranteed the one thing XHTML 2 never had: shipping code. Standards that ship beat standards that gleam.

What’s actually arriving, and where

HTML5 is a large specification arriving in pieces, and pieces are already in the browsers our support matrices track. The ones that matter for institutional sites:

• Native audio and video elements. Media playback as a first-class citizen of markup, no plugin required. Anyone following the Flash story since our iPhone notes knows what that means. Firefox 3.5 ships video support now, Safari and Chrome likewise, with a codec-format split between them that is this transition’s genuine headache. We will return to it when giving practical embedding advice.

• Semantic sectioning elements. Header and nav. Article, section, footer. Names for the structures every page already has, currently written as anonymous divs. Better bones for accessibility, styling, and machine reading. The semantic project we signed onto in 2006, extended into the language itself.

• Richer forms. Input types for dates, email, numbers, search. Built-in validation hooks. The form assistance we have been scripting by hand, migrating into the platform where it belongs.

• Canvas. A drawing surface for charts, visualizations, and interactive graphics in native JavaScript. One more capability shrinking the list of jobs that required a plugin.

Each arrives with the same adoption profile. Usable where supported, harmless where not, if deployed with the progressive-enhancement discipline that has been our spine from the start. Older browsers see a working page. Newer ones see a richer one. No cliff, no fork, no rewrite.

The politics, briefly, because they explain the shape

How the war actually ended is worth two paragraphs of record, since the shape it produced will steer the language for years. The WHATWG formed in 2004 when the browser makers Mozilla, Opera, and Apple concluded the W3C’s XHTML direction was walking away from the web they actually shipped. They began specifying HTML’s evolution themselves, in public. Their working style was everything the committee process was not. One editor. Continuous drafts. Obsessive attention to how existing content actually parses.

The W3C eventually rebuilt its own HTML work around the WHATWG’s document, producing the unusual current arrangement. One spec, two organizations, parallel processes. Awkward on paper, functional in practice, because the browsers implement from the shared source of truth.

July’s decision completes the reunion by retiring the alternative. The lesson for anyone who watches standards: legitimacy followed implementation, not the reverse. The WHATWG had no charter authority at all. It had shipping browsers and a spec that described reality. That combination simply outweighed institutional process.

It is the same lesson the microformats community taught at smaller scale. Standards are agreements that people actually keep. Worth remembering whenever the next grand committee vision and the next scrappy running code diverge. Bet on the running code. The paper catches up.

What this means for sites you own or are planning

The advice, in full: no rewrites required, and none recommended. Existing sites built on clean semantic markup, the only kind we ship, will adopt HTML5 features bit by bit, as browser support and project needs align. The current markup remains valid, functional, and respectable for years to come.

New projects should be planned with HTML5 in mind. The sectioning elements and form improvements are practical soon, and video decisions should weigh the native option. But ‘built in HTML5’ is not yet a meaningful buying requirement. Vendors leading with it are selling the label rather than the practice.

The one genuinely new planning input is certainty. For the first time in a decade, markup has a single agreed future. Money spent on semantic, standards-based structure is no longer a bet on which standard. Just a bet on standards, which we have spent years arguing for.

The longer view

Step back and the July decision closes an era that began years ago. It was the era when the web’s own stewards disagreed about what the web’s language should become. That ambiguity had costs. Hedged tooling. Doctrinal wars. Energy spent on the argument instead of the work.

Its end redirects all of it. The browsers are converging, through Acid tests, JavaScript races, and now a shared spec to converge on. The language is evolving toward the application platform the V8 story predicted. And the sites that kept their markup clean through the uncertainty arrive at the ending owning exactly the right assets.

We will return to HTML5 regularly as its pieces mature into practice. The markup war is over. The building continues, now with one blueprint. And the blueprint, for once, describes the web we actually have.