Here's how I think Microsoft can get a quick and sustained longterm win in GUI (/s but only partially) -

1. Drop all its GUI stacks apart from legacy Win32

2. Port KDE Plasma to Windows (with aliased bindings to support traditional explorer.exe calls so as to not break user apps)

That's it. There really aren't many significant apps that use the rest of Microsoft's stack apart from Win32 that won't recover from this, and Microsoft itself will just see a massive drop in its own costs with a massive rise in user satisfaction.