Full comment
I'm prepared to be convinced otherwise! Like I said, I'm speaking only from intuition. I have a piece of paper on my desk. I put my hand on it and slide it. It does pretty much what I ask it to. I would say it's 'well behaved'. If I have some slippery surface and a slippery jell(y|o)-like object, it will wobble. To me, that's a sign that I need to pay more attention to it because it's not 'well behaved'. Modelling not-well-behaved objects for interfaces seems counter-intuitive and (I suggest) takes more mental energy to interact with. This is the point I (hope I) made above. As for the rotation, the origin-normalising in my previous comment is both affine and rotational (yes, I didn't consider that when writing it). But if you look at the examples, they also have rotational momentum. Paper doesn't have that, as it's got low mass and is dampened by the friction. The closest I could get to the Facebok examples is waggling a piece of paper about on the end of a stick, or moving a heavy object on a slippery surface, and that's not how I'd model an interface! I just clicked 'compose' in GMail and the compose 'window' just appears. Likewise in my text editor, and creating a new window in Mac OS Finder and creating a new tab in my browser. Granted, there are lots of subtle (and not so subtle) movements in various operating systems. But it's a world apart from Google's Material Design demo video [0] and the 2d and 3d waggle [1]. Mac OS introduced the bouncing icon in the dock about a