Monday, December 16, 2013

Software Patents Are Patently Absurd

Patents are a set of exclusive rights granted to an inventor over the given invention for a limited amount of time. The primary premise behind patents are to incentivize people to innovate because without that exclusivity, the inventor would not receive any "due benefit" from the invention because of the free-rider problem. Since the 1980s, the United States Supreme Court has ruled in Diamond v. Diehr (1981) and State Street Bank and Trust Co. v. Signature Financial Group (1998) that allowed software to be patentable. For over thirty years, America's legal system has recognized, to one extent or another, that software is covered under §101 of the U.S. Patent Act, as well as Article 1, Section 8, Clause 8 of the Constitution. This past Friday, the certiorari for the case of Alice Corporation Pty. Ltd. v. CLS Bank International was granted by the Supreme Court. If the defendants are successful, it very well could mean the beginning of the end of software patents in this country. Rather than look at this case through the lens of constitutional law, I would like to analyze the implications of the policy that has de facto been in place for over three decades for software. Are software patents an obstruction of innovation or do they open a new pathway to innovation? Would a decision by the Supreme Court to curtail software patents do more harm than good?

Patents in general are implemented with the intention of spurring innovation. However, reality tells a different story, particularly when it comes to software patents. Software is too abstract to be patentable because a software algorithm is "an abstract description of a general way to solve a problem," not to mention that coming up with a legal definition of "software" is not simple. If software were not considered to be too abstract (look at the Church-Turing thesis), then, as the Economist points out, "all manner of harmful monopolies would spring up based on common ideas found in everyday life, such as boiling water to make tea, that could feasibly be used to prevent others from doing the same, or at least require them to pay a license fee."

In theory, software patents are supposed to protect people's intellectual property rights, but in practice, software patents are another example of the government imposing a law that sticks it to smaller companies, particularly those that are trying to enter the market. In order to comply with a patent, a third party needs to know what they are in danger of infringing. Software can easily have thousands of lines of code. Because of the broadness of patent law (GAO, 2012), a company can cry "infringement" on the usage of a few lines of code simply because "it's an innovation." With about 40,000 software patents granted each year, which translates into millions of lines of code, i.e., millions of "ideas", how can you expect a small company to sift through all that data and not infringe on a patent?

There are not only costs of compliance, but also with the patent application itself. Time and labor are rare resources for service-based companies. Having them focus on patent law detracts them from actual innovation. Furthermore, since software development has low initial costs (e.g., a single computer, an Internet connection, relatively low level of training), software development is a low-risk industry, which is yet another reason the government does not need to intervene with its protectionism. Look at Microsoft, Apple, or Google. They managed to topple software companies significantly larger than them. By creating a perceived obligation of acquiring a patent, the government has inadvertently created a significantly larger barrier to entry.

The moment that a company infringes upon a patent is when things get ugly. Patent trolls, which are non-practicing entities (NPEs) that buy abstruse patents and pounce on any business who allegedly violates the patent, create a litigative nightmare, mainly due to the fact that patent litigation is exceptionally expensive. According to the Government Accountability Office, 46 percent of patent lawsuits have to do with software (GAO, 2012). Although NPEs win less than ten percent of their cases (Allison et al, 2010), NPEs brought forth 61 percent of software patent cases. Patent trolls have cost nearly half a trillion dollars in losses from 1990 to 2010. In 2011 alone, the cost that these trolls on society was $29B. These trolls are not necessarily looking to win the case, but pay a settlement to avoid the even more expensive cost of a trial.

Software technological development moves at a much faster pace than the patent system. According to Moore's Law, computer powering doubles about every two years. The patent process takes an average of three years, and that does not include the time it takes to enforce the patent in court. Also, patents shouldn't last for 20 years, but something like the World Trade Organization's TRIPS agreement (Article 33) gets in the way of having the market adapt to technological advancement.

Software patent reform is necessary, especially if the patent system stays intact. One option would be to make the patents indexable (Mulligan and Lee, 2012). There is the possibility of functional claiming, which is the idea that patent is only on the the method of achieving the function, not the function itself (Lemley, 2012). As mentioned above, the patent period should not be twenty years. Shortening the patent period would clear up some of the inefficiencies. Even so, when the patent application process is so long to begin with, it is difficult to implement that change. There is the possibility of being more lax on the "obviousness clause" of the Patent Act (Abramowicz and Duffy, 2012). A bit of tort reform might help here. If the patent is invalid or there is no infringement, the trolls should have to pay the legal fees. Given that they represent a disproportionate amount of the trials, fee-shifting would create a disincentive to file frivolous lawsuits.

There is the idea of an innovation defense, which is problematic because a) it is a damning indictment of software patents because it illustrates the burdensome costs, and b) with all those software patents, how is one to perform cost-benefit analyses at that pace? There is the possibility of creating Defensive Patent Licensing (DPL), which would use the Prisoner's Dilemma to solve patent infringement issues (Schultz and Urban, 2012). However, there are issues with getting people to opt in, as well as enforcement issues. One can implement a maintenance fee to reflect the true social cost of patents. I find the maintenance fee to be amusing because it deals with a negative externality that is ironically caused by the government.

If the America Invests Act has illustrated anything, it is that patent reform is excruciating.
Even the Left-leaning website Mother Jones realizes that patents have "evolved into little more than virtual armaments that big companies use to fight virtual wars with each other." It's hard for proponents of software patents to advocate on the notion of "intellectual property" when the property system is so convoluted where it becomes nigh impossible to define the property. Software patents are much less effective than other patents because information processing is about as decentralized of a process as one can have.

Removing software patents would not only not impede software growth, but would likely accelerate it because companies don't have to deal with all the costs that come with software patents. Innovation took place during the advent of the software industry at an astounding rate from the 1950s to the 1980s, and did so without software patents. Microsoft DOS, the World Wide Web, Linux, and email were all created without patents. History shows us that software innovation can take place without patents in place. As assertive as proponents are about software patents, there is no empirical evidence showing positive effects of software patents. If anything, the data show that these patents ironically either create a negative or neutral net effect (Boldrin and Levine, 2012Torrance and Tomlinson, 2009; Bessen and Meurer, 2008; Pollock, 2006; Bessen and Hunt, 2004; Jaffe, 2004).

Software patents emulate and amplify the worst aspects of patent lawNew Zealand recently did away with its software patents, and I think America should follow suitCopyright law is sufficient because it is simpler than patent law, not to mention less expansive. As long as the programmer creates his code from scratch, the odds of copyright infringement are minute. Tinkering with the software patent system will not adequately eliminate these economic barriers and transaction costs, but making software patents nonexistent would do the trick quite nicely.

No comments:

Post a Comment