What my experiments with various content creation and markup systems over the past 30 years have shown me is that it's far less details of <i>presentation</i> which are crucial, but in <i>enforcing document structure itself</i>.<p>Presentation changes with technology -- I've seen <i>and used</i> systems with toggle-and-light outputs, true ttys (paper), glass ttys, various terminal and console outputs, the "standard" 24x80 terminal, desktop GUIs, and now handheld and mobile GUI devices with sizes from wristwatch to ledger. Other displays may be as large as a city block.<p>And that's just <i>viewable</i> output. TTS (text to speech) and voice recognition are also increasingly present.<p>The same <i>presentation</i> systems don't work across these. But <i>well-encoded semantic content</i> is amazingly robust. I remember learning of the '-man -Tps' (I think) nroff/groff argument -- that is, apply manpage macros, and format for postscript output. The same manual page markup that is readable in a console suddenly becomes pretty-printed (and created a brief market in the late 1990s / early 2000s for "Linux Bible" manpage dump books). Groff has more tricks up its sleeve, and as the Debian dwww package shows, manpages can be converted directly to viewable HTML.<p>But groff is grotty. I knew it (or precursors) once, well enough to turn in several Uni essays prepared via it. But those neural pathways have long since eroded.<p>I'd replaced it for quite some time with HTML, a reasonably versatile structured markup system, particularly given that most of what I was writing was intended for online Web publication at some point.<p>In the past few years I finally cracked the Lion book and started using LaTeX. I'd realised what the blocks were to my earlier attempts (ironically, "easy-to-use" tools such as Lyx actually had gotten in the way), and discovered that it was, as an <i>authoring</i> tool, often far <i>lighter</i> than HTML. Double carriage returns as paragraph breaks replaces seven discrete keystrokes per paragraph. Other constructs are a bit less lightweight, but remain clear.<p>But the real win is in how LaTeX is both a <i>structured</i> and <i>validated</i> document format. Screw up your HTML somehow, or follow some vendor's proprietary extensions, and a browser will say "eh, close enough". Omit a closing brace or backstroke or dollar sign, and your LaTeX compiler will scream at you until it's blue in the face. Unsettling the first few times it happens, but you begin to realise it's right.<p>The Web is an error condition: <a href="http://deirdre.net/programming-sucks-why-i-quit/" rel="nofollow">http://deirdre.net/programming-sucks-why-i-quit/</a><p>But deeper than that, LaTeX offers <i>structure</i>.<p>Documents have titles, and authors, and publication dates. They quite frequently have <i>references</i> and <i>bibliographic citations</i>.<p><i>Text notes -- foot, side, end, whate'er -- aren't some fucking foreign bolted-on concept.</i><p><a href="http://codepen.io/dredmorbius/details/OVmBaZ" rel="nofollow">http://codepen.io/dredmorbius/details/OVmBaZ</a><p>There are other bits and pieces that are missing from both, and LaTeX, not <i>principally</i> oriented to online publication, is short on stuff as well. But then, HTML doesn't have a native concept of reputation-ranked, hierarchical, collapsable comment streams either. Despite Usenet's prior art staring it gloweringly in the face for 25 years.<p>And yes, LaTeX as a direct authoring environment has its downsides. I've actually taken to preferring Markdown for my intial pass through writing documents, and it's a langauge which can readily be front-ended by the GUI-friendly tools most writers will want.<p>And HTML5's semantic structures and the valient attempts by some (see Readability's Developers section and the hNews microformats specs) <i>are</i> actually pretty cool. But until and unless someone steps in to <i>require</i> validated content before they'll pass it (and "someone" tends to be spelled "major search engine", which is spelled "Google" -- who <i>have</i> actually stepped in to police some standards of online behavior and presentation), we're stuck with the fact that crap HTML is still blindly accepted.<p>But if anything, front-ended by Markdown or other simple markup languages, it's HTML that should be replaced by TeX.