Web Standards
The W3C's Reinvention of HTML
The WHATWG revolt, the XHTML dead end, and how HTML became a living standard — the plain-English version.
The WHATWG revolt, the XHTML dead end, and how HTML became a living standard — the plain-English version.
Somewhere around the mid-2000s, the future of HTML looked bleak, rigid, and slightly absurd. The official plan was to march everyone toward XHTML — HTML reformulated as strict XML — while HTML itself was declared essentially finished. Then a group of browser makers quietly staged a revolt, and the language we write today is the result. It's a genuinely interesting story about how standards actually get made, told without the drama.
The W3C's bet was that the web's messiness could be disciplined by XML's strictness. XHTML documents had to be well-formed XML: every tag closed, every attribute quoted, and — the fatal clause — served with the right MIME type, at which point a single syntax error would show the user a parse error instead of the page.
That's a fine trade for a machine-to-machine data format. For a web authored by millions of people of wildly varying skill, it was a non-starter. Almost nobody served real XHTML; they wrote XHTML-flavored markup and served it as HTML, getting the syntax tax without the benefit. Meanwhile the actual needs of web applications — richer forms, media, offline, drawing — went unaddressed.
In 2004, people from Apple, Mozilla, and Opera — the folks who actually shipped rendering engines — formed the Web Hypertext Application Technology Working Group. Their premise was almost heretical: describe the web as it really is, specify how browsers really behave, and evolve HTML pragmatically rather than replacing it.
Instead of asking authors to conform to XML, ask the spec to conform to the web.
They began work on what became HTML5. For a few years the W3C pursued XHTML 2.0 in parallel, an even more idealized and backward-incompatible design. It never shipped. Eventually the W3C conceded reality, adopted the WHATWG's work, and XHTML 2.0 was formally abandoned.
The deepest change wasn't a feature — it was the model of the spec itself. The old approach produced numbered, frozen versions: HTML 4.01, done, sealed. The WHATWG replaced that with a living standard: a continuously updated document reflecting what browsers agree to implement, with no more grand version numbers.
For years two organizations published overlapping HTML specs — the W3C's snapshots and the WHATWG's living document — which was confusing for everyone. In 2019 they signed an agreement: the WHATWG living standard is the single source of truth, and the W3C stopped publishing competing versions.
The moral is worth keeping. HTML wasn't reinvented by decree from above; it was rescued from a tidy but unusable plan by the people closest to how the web actually works. "Describe reality, then improve it" beat "impose an ideal," and the language got healthier for it.