Tuesday, August 11, 2009

Software Patents and the Church-Turing thesis

One of the most interesting arguments against computer science algorithm patents is that the Church-Turing thesis makes them mathematics and therefore not patentable subject matter.

If this was an argument against all patents and intellectual property constructs then there would be a certain consistency to it, but most people that apply it appear to be applying it selectively without understanding it. A simple corollary of the Church-Turing argument reduces all possible material to mathematics.

In short, algorithmic information theory does not allow that theoretically meaningful distinctions among patentable materials exist. There might be utilitarian arguments that do not try to assert fundamental differences, but those arguments are rarely made. This suggests that the arguments both for and against are generally ignorant at best and disingenuous at worst, hardly the basis for sound policy in any case.

I think there are valid benefits from both sides of the argument, but the discourse is rarely intellgent nor does it acknowledge that we are making tradeoffs between shades of gray. I find the gross inconsistencies inherent in most peoples' positions on this topic to be more unreasonable than any particular position in the abstract.