In software patents, the implementation is described in such general terms that a lot of the time, any implementation can be said to use the method described in the patent. This is my biggest problem with software patents: once you abstract your solution enough, it isn't of much use to anybody, hence the value of the patent to the general public is null while it still allows you to block out competitors that want to find a solution to the same/similar problem although your patent is of no help to them actually solving the problem.
It is this skewed relationship that drives me crazy. Licensing a software patent only allows you to come up with your own solution a lot of the time, it does not save you the development cost.
If you license a patent for some industrial process it saves you huge amounts of research cost so licensing the patent is actually attractive.
Software should only be patentable if the patent provides actual value in solving the problem it claims to solve. Since useable software implementation are so specific, this may rule out patenting software at all.