WHATWG severs collaboration with W3C on HTML

Fans of HTML will be either thrilled or annoyed by the news that there will soon be two independently maintained standards claiming to be the authoritative definition of HTML. The Web Hypertext Application Technology Working Group (WHATWG), a team comprised of representatives from various browser makers, announced that it is terminating its collaboration with the World Wide Web Consortium (W3C) on the standardization of HTML 5. In WHATWG's account of the split, it is continuing to develop the "living standard" of HTML, while W3C's HTML 5 specification is a frozen "snapshot" of HTML. Based on the public statements and history between the two groups, the underlying issue has more to do with the standardization process than it does with technical differences between their versions of HTML. Nevertheless, competition over who has the right to declare their vision of HTML the official standard will likely cause headaches for web developers.

WHATWG founder and chief public spokesman Ian Hickson posted the news to the WHATWG mailing list on July 19. Hickson (who is a Google employee) had been the primary editor for both the WHATWG and for W3C's HTML Working Group. According to the announcement, Hickson will continue to be the lead editor of WHATWG's HTML work, but leave the W3C editor position. WHATWG will formally be a W3C "community group" (CG), and will continue to use a W3C-hosted issue tracker. On the latter point, however, Hickson and the W3C did go through the existing bugs filed against the previously-unified HTML specification and clone separate copies, one for each of the now distinct specifications.

The upshot is that WHATWG now regards its version of HTML as the definitive standard, which will continue to evolve, without declaring numbered versions. Hickson concluded "My hope is that the net effect of all this will be that work on the HTML Living Standard will accelerate again, resuming the pace it had before we started working with the W3C working group."

Hickson cited two reasons for the split. First, the W3C separated out several parts of the HTML 5 specification into distinct sub-specifications (such as the 2D canvas element, postMessage, and server-sent events). The result, he said, was "an increasing confusion of versions" of the specification, in response to which WHATWG "went back to just having a single spec on the WHATWG side which contains everything I work on." Second is a divergence between the WHATWG and W3C processes. Attempting to explain what that means, Hickson described WHATWG's process as "fixing bugs as we find them, adding new features as they become necessary and viable, and generally tracking implementations." In contrast, he said, the W3C HTML working group is focused on "creating a snapshot developed according to the venerable W3C process."

What, work in a group?

To the untrained ear, those reasons might sound like WHATWG simply does not want to participate in a standards process at all. But there is more pointed criticism of the W3C on the WHATWG site, which calls the term HTML 5 a "buzzword" and enumerates several specific differences between the two standards, linking most of those to W3C HTML working group decisions that WHATWG evidently found disagreeable. The decisions that WHATWG critiques date back to mid-2010, although many of them seem to be connected to the working group's recent attempts to finalize HTML 5 over the course of 2011-2012 — such as disagreements over the actual wording of the specification and its inline advice (as opposed to, say, contradicting definitions of HTML elements or attributes).

WHATWG's disinterest in finalizing the standard was also evident in a blog post Hickson made in January 2011. That post announced that WHATWG's version of HTML would drop version numbers altogether, on the grounds that "the technology is not versioned and instead we just have a living document that defines the technology ." The HTML "Living Standard" terminology persists in the group's current communication. W3C, in contrast, is still moving forward with finalizing the HTML 5 specification and its successors. In an April message to the HTML Working Group list, Maciej Stachowiak said the W3C has begun to extend the HTML Working Group's charter to tackle "HTML.Next" and will proceed to examine proposals, including "the WHATWG HTML specification, which we anticipate will be one such proposal."

Viewed in that sense, the split between the two groups is not so much a forking of HTML into separate proposals as W3C's HTML 5 is a "frozen branch" of WHATWG's trunk. Or it would be, were it not for both groups' claims to represent the official standard for HTML. Hickson made this claim explicitly in his email, calling WHATWG's HTML Living Standard the "canonical description of HTML and related technologies." W3C's claim to ownership is less overt, but the group and its founder Tim Berners-Lee have written and formalized HTML since its inception in the 1990s. The first draft was written in 1993, with subsequent revisions published as IETF RFCs before the W3C process took over.

WHATWG was founded by a group of browser makers who felt that W3C's process was too slow and bureaucratic, so perhaps it is miraculous that the two groups were able to collaborate so successfully as long as they did. In any case, the practical question is what the split means for HTML developers — and, by extension, web users.

The great thing about standards is ...

The principal reason for concern is that if the two specifications drift in different directions, web developers could be forced to add even more workarounds to match both, or else sites could be branded "WHATWG HTML compatible" or "W3C HTML compatible," reminiscent of a return to the dark days of serious browser incompatibilities. In addition, WHATWG's "living standard" approach has its own critics, principally on the grounds that a constantly-in-flux document makes for a poor standard. A commenter named Mike asked on the 2011 blog post "How do you make a test suite and a browser compatibility chart for a “living standard”? It sounds like HTML is becoming a sort of Wikipedia revision style chaotic nightmare." Similarly, Jukka Korpela said:

Indeed, the current presentation of WHATWG's incarnation of HTML does sport a Wikipedia-style "last updated" timestamp at the top (the last update was July 24, 2012 as of press time), plus floating tooltip-style markers that point out certain paragraphs as "ready for first implementations" from the margins — both features that hardly inspire confidence in the specification's stability.

Steve Faulkner replied to Hickson on the WHATWG list and took direct issue with WHATWG's assertion of canonical-ness. "The claim that HTML the living standard is canonical appears to imply that the requirements and advice contained within HTML the living standard is more correct than what is in the HTML5 specification." In particular, Faulkner pointed out that WHATWG's specification, like W3C's, cover only browser implementations and not authoring recommendations (which tackle critical issues like accessibility in addition to stylistic advice). On that front, Faulkner said, the W3C's specification has the "more accurate set of requirements and advice, that takes into account current implementation realities, thus providing [authors] with more practical advice and thus end users with a better experience."

Hickson and WHATWG have not responded publicly to Faulkner's message, but perhaps that is to be expected. It seems clear that WHATWG's primary motivation is continuing to add to HTML and related technologies as quickly and as often as its members wish. Of course, that highlights another thorny issue. As a standards-setting organization, WHATWG is not particularly accountable; membership is by invitation-only and Hickson can only be removed as HTML editor by the members. Other members of the public are welcome to join the mailing lists as "contributors," however. W3C may not be particularly democratic either, but its flavor of bureaucracy is more diffuse, with various working groups, interest groups, coordination groups, and boards.

The H postulated in 2011 that the seeds for the divergence of the two groups could be traced back to WHATWG's dislike of W3C's HTML 5 logo and related branding effort. One would hope that a core web technology like HTML would be above that level of triviality, but the alternative reasons given in public are not much more satisfying. With any luck, though, the web will eventually route around the damage — one way or another.