I'm as curmudgeonly as they get. Other than DRM, I think I'm actually pretty happy with what I see in HTML5, though unfortunately there's still no way to actually force Web developers to use just a minimal markup set.
If I could force every last goddamned Web dev in the universe to stick with <html>, <body>, <header>, <article>, <aside>, and <footer> tags, along with <p>, <em>, <strong>, <a>, <blockquote>, <ul>, (and list items), form and text elements, and <img>, I think I wouldn't shed a tear. And I could then write my own goddamned CSS to render their pages in a sane and readable fashion. Once (as opposed to the 1000x I'm approaching now via Stylebot).
Table elements would be accessible for a fee, waivable for actual data presentation.
Don't forget <main>, <section>, <cite>, <dfn>, <i> and <b>, which, when used properly, can add significant value to the content. (There are a number of reasons why text might be default italic/oblique or bold, and emphasis is only one of them. HTML not only allows us to remove the ambiguity, it also allows us to tell the reader more, like what language that foreign phrase comes from, without interrupting the flow of reading.) I'm also rather a fan of <figure> when that's appropriate, since it provides a sane way to caption images that need non-optional captioning. <dl>, <dt> and <dd> have their uses as well. But yes, the idea that a document should be universally stylable based on the nature of its content should be at the top of the list.
Yeah, there are some additional tags you could toss in. I even thought of adding div and span, which have their occasional uses.
The issues I've seen tend to run the other way though: far too often sites abuse low-level elements (or higher level ones) to accomplish things the wrong way. Example time:
• Google+. They're too good for <i> or <em>. Italicized or bolded text gets its own CSS class. Which you can't reverse engineer. The entire CSS shitpile (thanks to minification) is a mess of utterly nonsemantic elements. Which I'm far, far, far too well aware of having written a very extensive personal stylesheet to fix its many, many, many UI/UX presentation issues. If you're exporting your G+ posts (which I have as I slowly exit the system), use the JSON export option which includes your original marked-up input. That is far, far, far more portable.
• Blogger's "dynamic" templates. There's something to be said about a design for a text-oriented web site that is so utterly broken as to make it impossible to actually read the text. It's so broken (and dynamic) that I can't get it to sit still long enough to actually fix it with Stylebot. I've actually written bloggers asking them to change their templates so I can read their content. Oh, and my usual fallback, Readability, fails miserably as well. Another Google property, whaddyaknow.
• Some Indian news journal site. A stunning cascade of intricately nested divs. Someone was clearly doing the needful ... to gin up consulting revenues. But I have a doubt.
• Some old-school Web 1.0 sites where ... all text is bolded (b, strong { font-style: normal }), or written inside <h4> tags (WTF?), or background images are used (more WTF, plus a plaintive "why?!!").
• Paragraphs separated by <br> elements. No <p> tags. Better: Long streams of <br> tags inserted into RSS feeds. These can be addressed with sibling rules: "br + br + br { display: none; }".
• Paragraph indents accomplished by consecutive entities. That's what CSS rules are for. Sigh.
• Mixing px and pts or px and ems for any character sizing / positioning. IF IT'S TEXT, USE PT OR EMS.
• Fucking with letter spacing.
• Random <span> elements with hardcoded, px-sized, fonts.
• Inline styles.
• Tables used for text layout. Yeah, pg, I'm looking at you.
Don't forget the divitis of responsive web design - one of my utter bugbears. Should people jump ahead of technology so much, or wait for Flexbox support to catch up and then do it? I know what I think...
HTML5 isn't just about the tags. It's an ecosystem of technologies that are breaking down the barriers of the browser to enable more native-like experiences over the web. Take for example how Google Maps in the browser can "access your phone's location". Think about what it's doing - there's a browser API that web developers can call that retrieves data sensed off the user's mobile device. This used to only be possible with native apps.
Also, the HTML5 ecosystem is more exciting because of the renaissance with javascript (nodeJS for server-side javascript, many JS libraries and frameworks) and the entire developer tooling that is making web development workflows much more productive than in the past.
"(nodeJS for server-side javascript, many JS libraries and frameworks)" - the saddest part, as if we don't need language innovation anymore... One to rule them all.
I'm as curmudgeonly as they get. Other than DRM, I think I'm actually pretty happy with what I see in HTML5, though unfortunately there's still no way to actually force Web developers to use just a minimal markup set.
If I could force every last goddamned Web dev in the universe to stick with <html>, <body>, <header>, <article>, <aside>, and <footer> tags, along with <p>, <em>, <strong>, <a>, <blockquote>, <ul>, (and list items), form and text elements, and <img>, I think I wouldn't shed a tear. And I could then write my own goddamned CSS to render their pages in a sane and readable fashion. Once (as opposed to the 1000x I'm approaching now via Stylebot).
Table elements would be accessible for a fee, waivable for actual data presentation.