Yesterday I had the pleasure of attending a Show-Me Institute conference on education policy. One theme that was echoed by a number of conference participants is that after decades of study, researchers have been unable to quantify what makes a good teacher or a successful school. We know that certain schools, such as KIPP, work much better than other schools. But replicating those successes at scale has proven maddeningly difficult. When someone tries to take a successful school and use it as a model for producing a large number of equally successful schools, something invariably gets lost in translation. Generally speaking, a successful school can only be replicated through a labor-intensive process of apprenticeship, in which key personnel for the new school spend several years at the existing school learning the details of how it works. Obviously, that makes the process of replicating successful agonizingly slow.
I’ve read (although I can’t find a good source right now) that development economists in the mid-20th century discovered similar problems when they tried to export American technology to third-world countries. They hoped that if they helped poor countries build American-style factories and sent them manuals and technical advisors to explain how to use them, that those third-world countries could start producing manufactured goods and rapidly increase their standard of living. Unfortunately, things didn’t work out that way. Duplicating American infrastructure overseas turns out to be a lot more complex than anyone imagined.
In short, a central problem in both education policy and development economics is that technology is surprisingly sticky. Merely observing someone do something innovative is almost never sufficient to replicate that innovative activity elsewhere. Policy wonks in both fields would love to find a way to mass-produce successes, but that turns out to be maddeningly difficult. Tacit knowledge turns out to be surprisingly important.
During yesterday’s conference it occurred to me that the stated goal of patent policy is precisely the opposite. There, the concern is that duplicating new technologies is too easy: So easy, in fact, that merely observing a final process is sufficient to reproduce the product that produced it. Patents, the argument goes, are needed to artificially inflate the stickiness of innovation so that inventors can recoup a larger share of the value created by their invention.
I think that as in education policy and development economics, many people debating patent policy vastly underestimate the stickiness of new inventions. The most obvious example is business methods. Successfully implementing a given business strategy involves a great deal of subtle business judgments and attention to detail that would be all but impossible to glean from casual observation (or to capture in a patent application, for that matter). The natural stickiness of business processes gives innovative entrepreneur a big head start over copycats, because it can take years of trial and error to successfully duplicate another company’s business strategy. Patent protection is simply gratuitous.
The same is largely true of software. As any programmer will tell you, going from an innovative concept (even one spelled out in great detail) to working code involves a lot of hard work. You inevitably encounter implementation details that weren’t obvious at the outset, and you frequently have to re-engineer portions of your code to work around them. At best, having an existing product to copy gives you a very general roadmap about what the final product should look like. It doesn’t save you from having to figure out the messy implementation details that are the bulk of the work anyway.
And indeed, a mediocre programmer is likely to take far longer to reproduce the work of a really talented programmer—if he can do so at all. The history of the web is strewn with examples of “me too” companies whose programmers produced slow, buggy, poorly-designed software despite the existence of far more innovative sites ripe for imitation. Even Microsoft and Yahoo, two of the most innovative companies on earth with an army of PhDs, took several years to reach parity with Google’s search engine.
I don’t know as much about other industries, so I won’t hazard any guesses about the stickiness of innovation outside of the software industry. But this does seem to be an under-appreciated issue in discussions of patent policy. Most patent proponents blithely assume that duplicating innovative technologies is a straightforward process. If that assumption is false in a given industry, it makes the argument for patent protection in that industry much weaker, because the natural stickiness of new technology automatically gives innovators “protection” against knockoff without any need for government regulations designed to enhance that stickiness.
The Technology Liberation Front is the tech policy blog dedicated to keeping politicians' hands off the 'net and everything else related to technology.