> There is some level in every implementation of anything where you have to say, "variations at this level aren't part of the invention." I'm just saying there is no bright line telling you how high that line can be before the invention is not patent-worthy, so maybe whether or not the thing is implemented with software is below it.
Whether or not the thing is implemented with software is irrelevant. The problem with software patents and the like is the nature of the thing itself: an algorithm or user-interface element, as opposed to a device or manufacturing process.
To put it simply, the proper domain of patents is *technology*: how one can take advantage of the laws of physics in a novel and non-obvious way to accomplish some useful end. They shouldn't cover something you could, in principle, do in your head (e.g. algorithms), even if in practice you make use of existing technology to speed up the process or supplement your memory. A patent on the look of a device or an abstract interaction concept is even further afield.