I'm Brad Gessler. I've been building on the web for a long time, and lately I've been building a lot of it with agents. Somewhere in there I noticed two things at once.
The first is that the browser is astonishingly capable. It validates forms, traps focus in dialogs, transitions between pages, fills in login codes from your text messages, and reads a document out loud to someone who can't see it. None of that costs you anything. It's already there.
The second is that we keep paying to replace it. Every year a new layer of abstraction goes between the person writing the page and the thing that renders it, and every layer makes it a little easier to forget what the medium can do on its own. So we end up with a six-box login code that breaks paste, a modal inside a modal, and a twenty-minute deploy to fix a typo. Not because anyone decided to. Because nobody checked.
Keeping it simple turns out to be the better deal on almost every axis that matters. It's cheaper to build and cheaper to run. It's faster for the person on the other end. It's accessible by default instead of by heroics. And, more and more, it's what an agent can actually understand: plain hypermedia is legible to a model in a way a pile of client-side state never will be.
So I made this as a fun way for people, and for agents, to learn how hypermedia works and why it's worth using directly. Each article takes one thing we habitually over-build, shows it breaking, and shows the smaller thing that already worked. Short enough to check yourself. Loud enough to remember.
The browser already does that.
It's open. Send a pull request. One file, markdown or HTML, no JavaScript for the site's own UI. I'm the dictator here and only sometimes benevolent: I may not read it, may not merge it, and if I do, it may come out rewritten. Best effort, no promises. There's also a llms.txt, written for the agents.