I've been hearing about the idea for nearly a decade now and yet it's still not taken off. Something must be missing

Hell yes. I love the idea. It's just that frameworks beat them to the punch of mass adoption.

IMO, 80% of the neat things on the web could easily be made into components, closed or not (preferably closed DOM, fuck you if you need to fiddle with everything outside of what you deliver).

For an easy example: Embedding PeerTube videos just as easily as YouTube videos in random web pages. Even if that is presently possible (and it is), I haven't seen it out there in the wild.

I have built web components and used them. They're actually great. Especially for enhancing and updating legacy software.

although most my focus goes to Hypermedia systems architecture and HTMX. So most things are rendered on the server, and for complex client side stuff, we miggt use webcomponents.

I wrote a web component framework based on vue and sat on the w3c web components working group for a while because I worked at a startup that really needed the shadow dom to solve some significant business problems. The startup decided to just accept the business problems (and no longer exists), and we could never actually get the browser vendors to agree to anything to move the spec forward. It was one of the most futile and demoralizing things I've ever done in my software development career, and now with LLMs it seems unlikely we'll ever get new frameworks and ideas adopted, because it's impossible to get enough adoption to generate enough documentation for LLMs to train on. I'm glad Nolan feels like it was all worth it, because I mostly just feel like hitting everything that uses electricity with a hammer and moving to the middle of nowhere to grow hot peppers and make artisan hot sauces.

(Touch) scrolling seems broken on that website. Jitters/stutters.