It seems that many 'new' interfaces try to do their own thing, and try to go for looks rather than functionality. If you have no consistency and no known functionality, the UI is basically worthless.
There are some ways to get a 'better' UI without breaking so much, like the macOS menu bar that is context-dependant. You save some space in each window, yet because you can only click one menu at a time anyway you don't lose any functionality or consistency. The downside is the extra step of first getting focus on the application you wanted to use (but in most window managers you had to focus the window first before using the window-embedded menu bar).
It's not that a UI can't be made better, it's just that it's hard to sell to people with no knowledge or affinity with basic HIG principles. Such disconnect between reality and 'design' could be found a while ago where feature phones and some laptops came with either round trackpads or round screens. That's a neat artistic thing, but completely useless in a reality where everything is rectangular, including the things you interact with in the world as well as digital content.
Same with the touch-centric UI components getting ported to desktop operating systems that assume lots of swiping and touching where there is none. While possible to execute it reasonably well (I think there is a 'touch mode' in recent windows versions vs. touch-integrated so you can just switch between them instead of having a mess of both worlds in one). Same with Apple's idea where your input device they ship with the computer has the input capabilities matched with the OS and the first-use explanation of the UI when you power up for the first time. And if you don't use it, you're not losing anything, because the traditional buttons and menus are still there, visible, and not in some hidden swipeable area.