The case for software patents was always that they would reward inventors and pull capital toward research-heavy fields. Decades of evidence show they have done something close to the opposite. They have created a parallel litigation economy, slowed down the pace of release for small developers, and entrenched the largest incumbents who can afford defensive portfolios.
It’s time to say it cleanly: software patents are bad policy that survived because the people who profit from them are organized.
The empirical record
The Boston University study by Bessen and Meurer estimated that by the late 2000s, U.S. software patent litigation costs alone were running into the tens of billions of dollars annually, exceeding the R&D budgets of much of the industry the patents were supposed to encourage. The Federal Trade Commission, the Government Accountability Office, and a long line of economists have published reports finding that software patents disproportionately produce litigation rather than disclosure, the supposed bargain at the heart of patent law. The famous studies of post-Alice Corp. (2014) decisions, which made some abstract software patents harder to enforce, found no observable decline in software innovation. If the patents were doing their job, you’d see something break when they got weaker. Nothing did.
The patent troll problem
A meaningful fraction of software patent litigation is brought not by inventors but by non-practicing entities, shell companies that exist only to acquire vague patents and threaten suits. The economics are vicious in a specific way: defending even a clearly meritless patent suit through trial costs two to five million dollars, while a settlement to make it go away costs fifty to two hundred thousand. Rational defendants settle. The settlement money funds the next acquisition, the next suit, the next settlement. Small startups, who can’t absorb either the legal fees or the settlement, are crushed first. Reformers have spent fifteen years patching the problem with venue rules, fee-shifting, and inter partes review. The patches have helped at the margins. The underlying disease, which is that software patents almost always describe ideas rather than inventions, has not been treated.
What software actually is
The deeper problem is that software is mathematics expressed in a particular notation. A patent on a one-click checkout, a slide-to-unlock gesture, or a recommendation algorithm is a patent on a procedure of thought, the kind of thing patent law explicitly excluded for centuries before the Federal Circuit quietly opened the door in the 1990s. Copyright already protects specific implementations of code. Trade secret already protects internal know-how. The marginal contribution of a patent regime on top of those is to let companies tax each other’s good ideas, which is exactly what the system has produced.
Bottom line
Eliminating software patents wouldn’t end software innovation any more than the absence of patents on novels has ended publishing. It would shift competitive advantage back toward execution, the thing software companies are supposed to be good at. The patent system survives in software because it’s profitable for lawyers and large incumbents, not because it works.
Leave a Reply