Well, there is a bit of an issue in figuring out what is obvious. You're probably right about the swipe thing, at least if you start at the edge of the page. But does pinch to zoom really count as obvious?

In terms of real life analogues, I would argue no. Pinching tends to squish things in non-linear ways. The opposite action will stretch things in non-linear ways at best, and do nothing or tear things at worse. The same goes for rotation.

When you look at it in terms of computation though, it is much more obvious. The start and end points of the pinch define a transformation, which may be linear. The same goes for rotation. Of course, you could use other transformations to define the same functions since you are just trying to get a set of numbers from the user. An example are sliders being used for rotation in traditional user interfaces: the slider is left-right, yet the action is a rotation.

So how do you define obviousness in these situations?

At the end of the day I would say that this stuff is more trivial than obvious. Since it does fall on the trivial end, it is probably better to deny patents to encourage uniformity (and avoid user confusion) and competition based on less trivial differentiating factors.