> One could say you patented the end result of a mouse trap that doesn't close while you're baiting it, or one could say you patented the end result of having a mouse trap with a remote switch, covering every possible implementation of that switch.
Either way, the real "end result" is an unclosed, baited mouse trap. You can't patent that. What you can patent is device which makes achieving that state a bit easier, provided it meets the other criteria, such as non-obviousness.
> A slide-to-unlock switch on a pocket touch-screen device is not qualitatively different.
First, there is no "switch" involved. The hardware (touch screen) is not the subject of the patent. Second, this isn't a software patent; it covers a virtual user-interface pattern, not an algorithm. This isn't so much a case of patenting what a mouse-trap does (which would be bad enough) as patenting what a mouse-trap *looks like*, which is even further outside the domain patents were intended to cover. As I said before, a trademark would be much more appropriate here.
Regarding actual software patents, however, such as a patent on a compression algorithm or a form of encryption, the qualitative difference is obvious: You build a mouse-trap, but you *evaluate* an algorithm. An algorithm is just as complete in your head, or worked out with pencil and paper, as it is running on a computer. The computer is just a tool to speed up the process, just as the pencil and paper serve to supplement your memory. The mouse-trap, on the other hand, is of no direct use to anyone until it's given physical form.
Not that any of this really matters, of course, as regular patents are just as devoid of legitimacy as software patent. They're not the same thing, but that doesn't make them any more just.