I think there's an agreement on how animations can be helpful at least in the beginning to learn what's going on, how an application framework behaves.

What I found is that animations are not the only visual help we have... and maybe not even the best:

Animations may be visible only for a short period of time. An animation may indicate that the current screen is part of a certain application or was triggered by a button you pressed - but if for whatever reason you missed this animation, you need something else.

For example I found it unbelievably difficult to follow screencasts of Maemo 5 applications. It all was a mess of widgets thrown on top of each other, I never knew where we were within the application. Had we left a screen and gone back? Had we proceeded to the next screen? Where do these buttons come from?

I slowly realised that this was because I'm trained to understand the meaning of windows that are positioned on top of each other. I'm trained to understand modal dialogs. I see a window in the background and a second one on top and I know: The one on top is probably a sub-window of the one behind it. I need to close it ("cancel") to return there. I'll know that even if I wasn't the one who opened these windows in the first place. This is visual information from a static UI.

I find some of that missing in Maemo 5 because so much is happening full screen (that's "stackable Windows", isn't it?). Plus, at least in the screencasts, the animations aren't very clear and I don't, of course, see/hear when or where the user clicked on an element. This makes it very frustrating to follow these screencasts. I need to watch them 3-4 times before I understand whats going on at all.

So while animations can provide additional information, static graphical information might be even more important and is more persistent, which makes it easier to "read". - When in doubt (or if I'd have to choose), I'd go for static UI, therefore.