I've worked on a wide range of projects that use a number of different approaches to this. At the end of the day, you usually want at least some parts of it to some extent. If you don't want every page the user sees to be full of loading spinners at first, whatever is serving the initial HTML needs to know at least something about what the UI components are like and how to lay them out. In fact, the closer the initial load to the final result, the better. On the other hand, you also want interactivity, especially in this day and age. Very few web applications are just static data you look at and don't interact with. So you also need some frontend stuff that adds all of that magic. So all these frameworks are essentially trying to solve this problem, how do you get both of those things without just duplicating and syncing all of that code between the frontend and the backend. I do agree that many of the solutions we see today are immature, but when it's done right, it's wonderful.