The problem is that everyone (browsers + developers) need to agree on a replacement, which is notoriously hard to do. Chrome got laughed out of the room with Dart/Flutter.
A nice starting place would be if one of the major browsers added an easy to use extension point for swapping out languages. (at least I'm not aware of this existing, nor that I've looked)
What would that actually mean in practice that would be different from how Dartium did it?
Such that one could throw together, I dunno, news.ycombinator.com/index.z and see if they can do a better job than HTML/CSS.
As far as the actual solution to try for, table stakes is a native implementation of something that looks like React (but presumably reimagined from scratch).
What would be the use of that? It would help people who want to make prototypes of a different web with no hope of it ever getting adopted, but that's not a huge market.
> table stakes is a native implementation of something that looks like React (but presumably reimagined from scratch)
Like a component layer? Browsers have tried with e.g. XUL or WebComponents, how would you avoid the problems of those?
So people can experiment with it. Creating a good place to experiment is the best way to make lots of good ideas happen. If 10000s of people can play with ideas instead of the 10 groups with enough time and energy and expertise, you get a much richer exploration of the space. Heck if it was easy I'd be doing it right now.
> Like a component layer?
I'm not aware of anyone creating a component layer that is even passable for abstraction compared to React's programming paradigm. Usually because they stay imperative-first when all signs seem to point to declarative-first being the more scalable model.
> how would you avoid the problems of those?
Umm... they're terrible, so I would just not copy them? Look at XUL for seconds and try not to vomit. WebComponents adds abstraction of HTML in Javascript, rather than in HTML / declaratively, which is obviously never going to work as a foundational technology. It is far too imperative-first. I would start by copying React... but then doing everything differently because most of React's design is forced by being implemented in JS and having to work around CSS being quirky. The greenfield version would do a much better job.
(This is not any better solution than GIF or aPNG though for the problem in the article.)