Embracing Markdown as Hypertext

By R. S. Doiel, 2026-02-19

There has been a flurry of discussion around Markdown after CloudFlare's post about automatically converting cached HTML pages to Markdown for requests that have an accept header of "text/markdown". It fits along side their robots.txt file expression of content signals. Why I think the energy being placed behind the hype around large language models is over done I like acknowledging that Markdown is a hypertext. I've been working on blog posts about that over for the last few months. I fear that the AI hype and eventual crash will distract us from the significance of that realization.

We can have a radically simpler Web by embracing Markdown as a first class representation of hypertext. Now that the cat is out of the bag what if Web browsers acknowledged that too? When I read content on the web I spend much of that time in my browser's reader mode already. It just improves the browsing experience. The reader mode would be far simpler to implement if it could just grab the Markdown and display that simply.

What else can we do with Markdown aside from having content that is easier to read for both humans and machines? Markdown has a beautifully simple way of representing links and lists. We have a really good syndication format that just doesn't give up call RSS. It is trivial to generate an RSS feed for a simple list of links in Markdown. It is equally possible to take that Markdown list of links and turn it into OPML. Markdown like RSS doesn't replace HTML's role in the web but extends the Web's collection of technology allowing for both simpler composition and reading experience. It does so without breaking the behavior we currently experience from the AI bot swarms.

Embracing Markdown in this way makes sense. Markdown already in the dialect of rMarkdown enabled Jupyter notebooks. It seem appropriate for it to be embraced further both for machine processing and for the browser.