Software developers sometimes protest that software patents rob them of independently-developed code. This is particularly true because so many software patents are obvious and are easily re-invented in the normal course of work. So shouldn't independently-invented code be exempted from patent prosecution?

Indeed, independent invention is permitted under other sorts of intellectual property such as copyright. During the early 80's, in fact, when software was largely protected by copyright, some firms used "clean room" techniques to document that their programmers were unaware of competitors' products.

To some, an independent-invention defense for patent infringement seems counter-productive; one purpose of patents is to encourage dissemination of inventions. It seems bizarre in the Information Age to need to document one's ignorance of leading technology. And perhaps the obviousness of software patents should be directly fixed.

But in this paper, Stephen Maurer and Suzanne Scotchmer make a new point that an independent-invention defense may have another benefit: it may reduce the economic inefficiencies associated with monopoly rights. Under standard economic theory, patent monopolists charge too much, allowing too few people to benefit from their invention. The authors argue that if patent law were changed to permit independent invention, then there would be more competition (through patent licensing), lower prices and more benefit. And as long as the costs of independently inventing are not too low, this policy would not inhibit innovation from occurring.

The setting is one with a patent holder and a large number of potential entrant firms who could independently develop highly similar products at a development cost of K each, which is assumed to be less than the patent holder's initial development cost. With the modified patent policy, these firms can enter the market.

Now if these firms were allowed to freely enter, they would drive the product price down to the point where the initial patent holder would not recoup its initial R&D investment. But if, somehow, only a few firms entered, under an assumption of Cournot ("soft") competition, all firms, including the patent holder, could make positive profits. Maurer and Scotchmer devise a "trick" to insure this latter result. The patent holder can offer a limited number of carefully calculated licenses that permit some firms to enter without having to spend the development cost K. If licenses are calculated so that an unlicensed firm cannot profitably enter, and then a limited number of licensed firms enter, the patent holder extracts all their profits through royalty payments. As long as K is not too small, the patent holder's profits exceed its initial development cost. Prices are lower than with a monopoly and more consumers benefit.

This model depends upon an assumption: the patent holder has to know exactly how much another firm needs to spend in order to develop a similar product (K). Any mistakes might skew the licensing calculation, permitting unlicensed competitors to drive the price down.

Maurer and Scotchmer also point out that a similar result might also be obtained with "narrow" patents instead of an independent invention defense. With narrow patents other firms may "invent around" at a cost K and the patent holder can limit competition by licensing a number of these firms. Again, this argument depends upon an assumption of full information of invent-around costs. It also assumes that the patent holder will not obtain a "thicket" of patents to limit inventing around.