The US Patent and Trademark office is starting to recognize that it has a software patent problem and is soliciting suggestions for how to improve software patent quality. A number of parties such as Google and EFF have filed comments.
I am on record against the idea patenting software at all. I think it is too difficult for programmers, as they are writing code, to constantly check to see if they are violating existing software patents, which are not, after all, easy to identify. Furthermore, any complex piece of software is likely to violate hundreds of patents owned by competitors, which makes license negotiation costly and not straightforward.
However, given that the abolition of software patents seems unlikely in the medium term, there are some good suggestions in the Google and EFF briefs. They both note that the software patents granted to date have been overbroad, equivalent to patenting headache medicine in general rather than patenting a particular molecule for use as a headache drug.
This argument highlights one significant problem with patent systems generally, that they depend on extremely high-quality review of patent applications to function effectively. If we’re going to have patents for software, or anything else, we need to take the review process seriously. Consequently, I would favor whatever increase in patent application fees is necessary to ensure that the quality of review is rock solid. Give USPTO the resources it needs to comply with existing patent law, which seems to preclude such overbroad patents. Simply applying patent law consistently would reduce some of the problems with software patents.
Higher fees would also function as a Pigovian tax on patenting, disincentivizing patent protection for minor innovations. This is desirable because the licensing cost of these minor innovations is likely to exceed the social benefits the patents generate, if any.
While it remains preferable to undertake major patent reform, many of the steps proposed by Google and EFF are good marginal policy improvements. I hope the USPTO considers these proposals carefully.