In chrome, you switch between "history" of last "interacted" documents, making impossible to our brain to predict which one is coming. We can't quickly perform tasks with open documents because it's not like typing in a keyword (without watching the keys). Instead, it's attention-consuming like typing in an iPad (head-aches) because we have to "track-aim-correct" to hit keys. Not to mention that the behavior is different if you "didn't" interact with it (or an endless loop would make you switch between the 2 foremost docs over and over)
It should be like Time machine layers, even if the graphic motion is not cloned. Not needed. Just the calculation of which window will end-up in front. (please disconnect in your brain the idea of z-index from switching-order)
5 open docs, copy something from each. i.e. 1-copy, 2-copy, 3-copy, 4-copy, 5-copy
Ideal: ⌘`copy, ⌘`copy, ⌘`copy, ⌘`copy, ⌘`copy,
Needed to accomplish: ⌘`copy, ⌘`⌘`copy, ⌘`⌘`⌘`copy, ⌘`⌘`⌘`⌘`copy, ⌘`⌘`⌘`⌘`⌘`copy,
5 open docs, switch windows to get doc #5 (doc 1 in front, doc 5 possibly 3 layers behind, not even necessarily in the last )
Result: It DEPENDS how many times you cycled (and interacted) with others.
Since it's a "history", working with docs 1,2 and 3, could make your ⌘`ing NEVER reach #5.
You could get: 1,2,3,2,3,2,1,2,1,3,2,3,2,3,1,3,2,1,4,5.
20 key-strokes (⌘`) to get Doc #5.
Could it be Chrome's own behavior?
I'd like to make a Feature Request but I'd like to know for sure whether this is a OS or APP behavior, before to submit it.