He wrote,

The right to decide whether or not I turn on the lights in my house is worth more to me than to my neighbors, so in principle I should be able to buy their permission. The problem is that there are a lot of people living within sight of my house. Buying permission from most of them does no good, since I need permission from all. The result is likely to be a difficult bargaining game, with at least some of my neighbors trying to extort from me a sizable fraction of the value of my land in exchange for their permission to use it.

This suggests that, in deciding how property rights ought to be bundled, there are two important considerations. The first is that, so far as possible, rights should go in the bundle where they are most valuable. The right to control the air a foot over a piece of land is worth more to the owner than to anyone else, so ownership of land usually includes ownership of the space immediately above it. The second is that, since the proper composition of bundles of rights will often be uncertain and may change over time, they should be defined in a way that makes it as easy as possible to trade rights. Property rights should be defined in a way that minimizes the transaction costs of likely transactions.

I wonder if this can be used to look at intellectual property. My instinct is that drug patents make sense, but that software patents do not. (Amazon’s attempt to patent “one-click ordering” was a notorious example of the latter. So was Blackberry’s loss of a patent-infringement case.) My instinct is based on the idea that a drug company has put a lot of effort into research and testing that a copycat company can avoid. However, somebody who comes up with a software concept might have done so in an afternoon, and the copycat company might also have done so.

One way to think of this is that with ordinary property, assigning rights reduces transaction costs. With intellectual property, assigning rights imposes transaction costs between the owner and the infringer. The question should now be whether the benefits of the intellectual property rights exceed the transaction costs that are imposed. With software patents, the transaction costs are very high, and the benefits in terms of stimulating innovation are little, if any.

Note that I have not said that someone should be allowed to copy software. I think that software copyright is legitimate. But patenting an idea for what software might do is objectionable, in my view.

I have just read Friedman’s The Machinery of Freedom for the first time, so expect at least one more post on that book shortly.