It's almost like we need some kind of lightweight markup language like HTML that lets us focus more on the content, with less boilerplate... so something like markdown in other words, then?
I don't think the problems Russel mentions are as big of a deal. I've never encountered an issue due to the inconsistent spec. The basic markdown practices are so simple they just work, and when they don't it's clear you need to make a slight adjustment. At worst you are missing something you want to use that's not available in the flavor at hand. In the places I commonly write markdown there is no need for a complex tool-chain to render it as HTML, it's either a real-time part of the apps I use, or I consume it as plain text.
I'd be curious how Mozilla is finding it. I know it's a favorite among bloggers, but I'm not sure about such rich documentation as MDM, the quality of which has never been better in my view.
I think we need to invent new ways to store and retrieve enriched/inter-related/hierarchical data in a way that's convenient, interoperable and ubiquitous. I can't think of any though, can you? Something like a relational database, but lightweight like plain text, but richer, like a website.