Brian Hogan has a post that I, at least, find provocative. His thesis is that Markdown—and by extension, Org-mode—are holding you back because they’re not semantic. That’s true, in a way, I guess, but both Markdown and Org-mode are Markup languages. That means they’re concerned with producing publishable content: they consider what things look like on the printed page or the screen.
They are, in their way, semantic languages in that they specify what grammatical object is being described. Markdown, and especially Org, can use that information to display the item in a manner appropriate for the output target. Of course, Hogan wants a stronger sense of semantic where the markup describes higher order meanings such as “step in a process” instead of simply “a bullet item”.
The problem is that that always ends in tears. It ends, in short, in abominations like XML. XML is, in fact, what Hogan really wants but even he knows it’s a horrible solution beloved by exactly no one. Still, he says, it’s necessary because otherwise it’s too hard for AI agents and the like to discern the meaning of the markup.
Forgive me for saying that that’s not my problem. My problem is producing a nice looking document that human beings can read, understand, and enjoy. If AI companies want to get rich by training their AI agents on my content, it’s certainly not incumbent on me to make their job easier.
And by the way, why doesn’t Hogan mention Org-mode. It does, after all, meet a lot of his requirements, have a single definition, and allows the easy—for some definitions of easy—specification of backends that can be semantically aware.
So, as Betteridge’s Law of Headlines insists, no, Markdown is not holding us back. I understand Hogan’s views to the contrary and I certainly don’t accuse him of bad faith but I fail to see how Markdown or Org-mode is holding us back in any meaningful way.