I think part of the problem with patents for software is precisely the point that has been made again and again -- software is much closer to mathematics than any other type of patent. If you have a large group of people who are smart and tackling the same software problem, then the nature of the problem itself will often funnel them all toward similar solutions. I mean, there is a thing in software engineering called refinement which is the process of reducing a software requirement specification into working code. In some cases, this can even be done mechanically with CASE tools. Since the optimal code for achieving a particular goal is often a logical (or near logical) necessity of the goal itself, it's hard to see what innovation is really being protected by software patents. It seems to me that where the innovation is actually occurring is in the very act of noticing that a problem needs solving in the first place. But that is not what is patented.

EDIT: Just to clarify my last few sentences. Software patents seem to reward the wrong thing. It seems to me that the innovation occurs in the very act of specifying the requirements for the software. In a sense, the software that best implements the requirement is less a creative invention and more of a discovery. Now, I agree that there should be some kind of reward/prize/compensation for companies that choose to pursue unusual goals but I'm not sure that patents are it. Favorable market position may be reward enough.