Everyone wants their thing to be HTML
blog.mozilla.com
blog.mozilla.com
* One browser vendor implements something
* Some other browser vendor realises its a good idea, copies it
* Every other vendor scrambles to implement some form of it
* Behaviour is formalised into a standard
* Browsers slowly but surely end up on the standardised version
This often tends to create saner and more useful 'standards' than the real ones: InnerHTML, the Quirks Mode box model, Ajax, Canvas, font-face, JavaScript. The downside is slow and 'incorrect' implementations can be slow to fix bugs (although that happens with true standards too). Standards are rarely truly "industry leading" and I'm truly not sure if the standards leading is actually much better in the real world than the present ad hoc development method.
Minor browser engines such as Webkit, Gecko and Opera can't change the market much on their own. They need to be interoperable to some degree for people to start using their additions to the standards on the wider web. HTML5 (nee WHAT-WG) was initially those browser vendors coming together to work on exactly that.
You give Canvas as an implementation before standardization example, yet it was far more iterative than that with Apple's original implementation (for use in the walled garden of their widgets) being presented for standardization with the explicit idea that it was good for Apple if there were interoperable implementations out there.
It's a pretty good example of Co-opetition. (http://en.wikipedia.org/wiki/Coopetition)
The img tag isn't hugely interoperable, but it'd be a sad web without images. <audio> is a huge step up from the current madness of <object> <bgsound> <embed> or <flash>.
Are people out there really using XHTML still? It'd be interesting to do a poll to see how many are using XHTML vs HTML.
So to take your img example it would be analogous to a specification for img where Firefox was supporting PNG (but not GIF) and Safari supporting GIF (but not PNG) and IE supporting neither but Microsoft is at the same time pushing a closed source browser plugin that can show images in a Microsoft proprietary and patented format that sits outside of HTML. That would be a "sad web" too, even if it technically had images.
He's not (imho) against the video tag in any way, just pointing it out as an example of the (inherently) political squabbling and consequent craziness that surround standardization in a multi-billion pound market.