13 Mar 2012

Why It's Mathematically Impossible To Avoid Infringing On Software Patents

from the scalability-problems dept

In 2008 James Bessen and Michael Meurer came out with a truly excellent book, Patent Failure. It's chock full of excellent information and a pretty wide survey of the research showing just how much patents harm innovation. While I don't necessarily agree with the "solutions" proposed, the key thesis of the book makes a tremendous amount of sense: to have a functioning market, you need property with clear borders. If the borders aren't clear at all, the end result is that no one knows when they're trespassing or even what they're buying, and the benefits of a market collapse, and instead you get mired down in legal disputes. That's exactly what we're seeing with patents today. Of course, one of the key reasons for this -- as we've been explaining for years -- is that patents are not property -- and thus the attempt to force property-like rules on something that is naturally abundant is going to make it impossible to creates reasonable boundaries.
Tim Lee wrote about the book, highlighting this very point right here on Techdirt soon after the book came out in 2008. Apparently, it's stuck with him. Lee, along with Christina Mulligan at Yale, have built on that idea to write an excellent research paper that explains how it's effectively impossible to actually avoid infringing on software patents. The key? It's a scalability problem.

Lee and Mulligan have written up a shorter summary of the piece at Ars Technica that makes the point clearly. Because software and software patents don't have "defined boundaries," you really have to go through every single software patent to make sure you don't infringe -- but that's a problem that's insurmountable:

we estimate it would take at least 2,000,000 patent attorneys, working full time, to consider whether all these software-producing firms have infringed any of the software patents issued in a typical year. Even if firms wanted to hire that many attorneys, they couldn't; there are only 40,000 registered patent attorneys and agents in the United States.
This isn't surprising. While some people assume that patent infringement is all about one company "copying" another, in the vast, vast majority of cases it involves independent invention (often of the obvious next step in a process). The infringement couldn't be prevented, because the companies were just building what they needed to build to serve the market, and it's basically impossible to check to see if you actually infringe on another patent. Some patent system defenders pretend it's easy to find these patents, but that displays a lack of understanding about the true size of the problem.

As Lee and Mulligan note, companies infringing on software patents have nothing to do with companies trying to copy others or "get something for free", and everything to do with the fact that it's "mathematically impossible for them to do anything else."