Pulling up a reply by me regarding alternative protocols to the web (i.e. something like Gemini):
You can't possibly "win" against the web if your objective is "better features" because anything is already possible on the web. You have to choose a different playing field, and I see the best chance in a protocol that's more lightweight and less flexible (to hopefully avoid the kind of invasive tracking and anti-patterns we have here), but which focuses on providing features to get rich sites up and running with little effort.
One of my ideas is based on the fact sites all have a hierarchical structure which is usually presented by a menu. Said menu is implemented by the website by rendering items as links and structuring them in submenus.
But what if instead, the protocol included a specification to define the hierarchy which the browser can use to show a menu itself? Desktop browsers can show a menu bar at the top, but on mobile it could be a hamburger menu instead -- all without CSS and viewport queries.
If we think further, what about a set of higher level components that mirror patterns we encounter on the web pretty often? Instead of a dumb image element, we could have an image list component which can be displayed as a dumb list of links by less capable browsers, and as a proper grid with thumbnails and captions by more powerful browsers. Or how about a lightbox component, something which to this day requires JavaScript on the web?
Trade flexibility with more "bang for the buck". That's the only way I can see a protocol that's supposed to be an alternative to the web gaining traction.
❤1
1