I am grateful to IP Central for raising the question "what the devil qualifies as a patent on software, anyway?" The blogger adds that it is not an easy question. Although I had not really thought about it in that way I have to agree.
IP Central points to a working paper by Anne Layne-Farrar of The AEI-Brookings Joint Centre for Regulatory Studies entitled "Defining Software Patents: A Research Field Guide." Now although I don't usually agree politically with the the institutions that published this paper, I commend this work.