Critiques of software patents is all the rage lately, from bloggers like Daniel Tunkelang to the NPR. The list of problems with them includes that they stifle innovation, that they are tools to beat up small companies and startups, and that they are simply trading cards that big corporations use to protect each other at everyone else’s expense. So why are software patents different from other patents? Why aren’t people arguing about scrapping the patent system entirely?
Last week I had the opportunity to attend a debate-style talk featuring Bob Zeidman (pro) and Prof. Edward A. Lee (con) about software patents hosted by the Computer History Museum, which I found quite helpful in understanding the issues. The motion under consideration was “Software patents encourage innovation.”
First, I should say that I am reasonably familiar with the software patent process, as I am a co-inventor on a number of patents. So while I sympathize with the objections, I also see some value in protecting the intellectual property that happens to be embodied in software.
Second, it seems to me that the motion was poorly conceived. The real question is not whether innovation is encouraged, but whether patents are an effective mechanism for protecting property rights. As Prof. Lee pointed out, patents encourage patenting. Whether true innovation follows is a different issue.
With that preamble, however, I would like to summarize the pros and cons of software patents as a means of protecting the IP embodied in software.
The case for software patents
The essence of the pro argument is that software is just a formal description of process, and process patents have a long and legitimate history. Software patents protect the investment of effort to develop and maintain software, and just as other patents protect other forms of invention. While it is clear that there are some inefficiencies in the way that the USPTO handles patents right now, this was not a fundamental flaw in the notion of a software patent, but rather a shortcoming of the implementation due to lack of resources in the PTO and to the fast-growing field of software engineering.
The real problem with software patents
The essence of the con argument put forth by Prof. Lee rests on the notion that patents are rooted in property rights, and that a key aspect of property rights is the notion of boundaries. If someone climbs over my fence into my back yard, it is clear that that person has crossed a boundary. With certain kinds of software patents, however, this notion of a boundary breaks down. Software patents can be written in sufficiently abstract terms (and are so written intentionally) that they may be applied broadly, but the breadth can be potentially unbounded. One indication that a patent may be too vague is that legitimate domain experts may disagree on the scope of the claims. This kind of disagreement among experts is notably absent in other classes of invention.
To me this seems like a useful characterization of the problem because it can be assessed during the patent review process without having to resort to after-the-fact litigation. Thus in addition to novelty, utility, and non-obviousness, we would need a fourth criterion that defines the boundaries of the patent in a non-controversial manner. If such a definition cannot be articulated to the satisfaction of the inventors (who would want broad claims) and the PTO (who should argue for a more conservative interpretation), that means the patent, as written, is likely to have problems and should not be issued. While I am no legal expert and may well be wrong on this, it seems that the distinction between this approach and what is currently practiced is that claims are currently examined with respect to conflict with prior art, rather than with respect to clear boundaries. While prior art may in fact constitute such a boundary, lack of explicit boundaries is currently not an impediment to patenting.
On the socially- and economically-redeeming qualities of NPEs
Another interesting aspect of this talk was a case for NPEs (“Non-Practicing Entities”, aka “Patent Trolls”). One of the functions that NPEs perform is to add liquidity to the patent market. Bob Zeidman told an interesting personal story in which a large corporation (a former client) built some tools that they had previously been licensing from Bob and ignored his objections that the tools were patented. Rather than going through the expensive process of suing the large corporation directly, Bob sold his patent to a NPE, who then (presumably) went after the alleged infringer. So unlike some of the examples illustrated by the NPR piece, here the NPE served a socially and economically useful function of remunerating the independent inventor who was otherwise without affordable recourse.
In short, software patents pose a challenge to the legal system for two reasons: first, because (often intentionally) their boundaries are poorly defined, and second because the PTO is poorly trained and funded for the task assigned to them. Thus it seems that a reasonable course of action to remedy the situation is to modify patent law to make clearer boundaries a pre-requisite for issuing a patent, and to better fund and train the agency responsible for these tools of economic innovation.