I'm not convinced that patent examiners could make good decisions on software given infinite time. A couple of reasons: Let's suppose you had access to all the

I'm not convinced that patent examiners could make good decisions on software given infinite time. A couple of reasons: Let's suppose you had access to all the source code in the world. Given a description of a patented invention, is there any way to find out if it is already implemented somewhere? The answer is no, there is no decidable method for doing that. The proposition that there is, violates Rice's theorem. By contrast, in other areas, for example drug discovery, a patent covers chemicals of a certain class having certain substructures. Whether another chemical is covered is algorithmically decidable. Let me emphasise that: people are complaining about the tools to find prior are are bad, but we can mathematically prove that perfect tools don't exist. I'd suggest that the burden of proof, that sufficiently good tools can exist, should fall on those advocating patentability of software. But of course, patent examiners don't even have access to all this code. Unlike in drug discovery, where the entire business relies on patents so any discoveries have been filed with the patent office, software companies don't need patents to do business so the vast majority of software ideas aren't filed with the patent office.