The early UI developers recognized how effective depth cues were, and used it to great effect so that users knew exactly what can be interacted with, what not, where one thing ends and another starts etc. You see this in Windows 3.1.
Sonar uses depth cues a lot, and with control density as high as Sonar's, it would become extremely difficult and frustrating if all depth cues were removed. You can see how the fugly (flat & ugly) theme in Windows 10 affects control density by requiring large gaps between all controls. All we are looking at now is a collection of flat monochrome lines, circles and squares. The only way to see where one ends and another starts is to use enough space to separate them. Yes part of this is because MS thinks touch screens are the future (yea, just like 3D TVs are - I still have some of those cool giant 3D glasses somewhere, haven't used tem in ~3 years). So controls need to be spaced out far enough to allow fat finger control.
There is a good balance between using no depth cues and using too much of it (approaching skeuomorphism). MS grudgingly added back drop shadows in W8, first only to the window that has focus, then later to all windows. Why? Because they realized how important that was, even though it goes 100% against their "nothing that can in any way be construed as skeuomorphism whatsoever" goal. It helps a bit, but that gets us to about 5% of where Windows 7 is. Making title bars colored gets us another 3%, so now we are at 8%. Still need to go 92% of the way.
Add to this that things like the sizing cursor's hit area is now
outside the window (!!) due to the thin border, making it hard to resize the correct window when there are many windows in close proximity.
Honestly, they should not have let the interns design the W8/W10 UI. It shows that whoever did it had no clue and though it would be OK to put a non-windowed phone UI on a highly windowed environment.
Please let's just bring back the real UI designers, they seemed to understand this kind of stuff.