http://www.w3.org/TR/microdata/
As someone who has had serious reservations about the so-called semantic web being highjacked by academics who are out of touch with actual web development, I gotta give props to Hixie & co. for the microdata spec.
Granted microdata is almost certainly the least sexy part of HTML5 but it might be worth mentioning.
The microdata spec is uncommonly level-headed. Devs who are more at home slinging jQuery than they would be writing OWL ontologies could get on top of the spec with little effort.
Efforts like schema.org will only help push microdata into the mainstream.
WebGL/WebCL etc. are not part of HTML5 either, doesn't mean that they are not as game changing as the biggest features in HTML5.
But the thing that may keep Flash kicking for years to come may actually be advertising. It's a pretty small matter for Google to reencode their youtube library as WebM (which they are doing) compared to getting all web advertising firms to switch from Flash (for which they know well and have well established tools and workflows) to HTML/HTML5 based advertising.
--
Surely it's also not simply being pedantic to insist that things like CSS3 (!) and SVG don't get lumped into the "HTML5" catch-all. Pity the poor designer or developer just getting to grips with this stuff who has to figure out what someone means when they say "HTML5".
Nevertheless, as Mark Pilgrim said: "HTML5 will continue to be popular, because anything popular will get labeled "HTML5"." (http://diveintomark.org/archives/2011/01/09/dive-into-2010)
--
I personally wish that there was a different term than "HTML5" to refer to the additional work happening around and on the W3C HTML5 foundation. But, unfortunately there is no replacement term that quite captures the full scope of what most people mean when they use the term "HTML5". I share your distaste: it has a version number in it for crying out loud, it can't refer to something vague! I considered trying to use my post to promote "HTML5+" or something similar so I would not contribute to the corruption of "HTML5". But I decided that was the impossible fight the popular usage and chose to instead pick a new term for the narrow reference "W3C HTML5".
Not only that, but it can also bring speech to text to the web[2], potentially add motion gestures for navigation, games, etc. Imo this is the last piece of the puzzle (along with websockets, bytebuffers, etc) to making the browser a full OS from a user's perspective.
[1] I would have said the devices tag, but that has recently been dropped from the html5 standard. Ericson labs had some really great demos of it on a modified webkit engine)http://www.w3.org/TR/html-media-capture/, too bad. [2] see x-webkit-speech attribute of the input tag for chrome's implementation
Now, there's a plainly-described parsing algorithm in the HTML5 spec that handles conformant documents correctly, and non-conformant documents sensibly, and third-party implementations like html5lib[1]. Opening up a corpus the size of the Internet to interested individuals and researches has to be a pretty important achievement, I'd think.
