we have multiple different paradigms today, but due to how big the market is and how big the fraction of people who're not interested in exploring the space is, the only places the innovation happens is in labs. there aren't any ux user studies on an os-level anymore.
except the "desktop" and "buttons on a grid" modes of operation, the biggest one is "scrolling app", followed waaaaaay down by things like tiling interfaces and whatever the hell gnome is doing. there are multiple interesting ideas down there, like the "infinite whiteboard" of oberon or "zui" of eaglemode, or even the "composable widget toolbox" of arcan. it's just that they will never seem use outside a few hundred users at most because they are so different. nobody is interested in checking if users prefer them because the users are also the developers and there's no budget behind them.