The key point is that patents serve a transactions cost reducing role in pooling human capital for innovation when compared to the alternative of contracts and trade secrets. In the absence of patent rights, it would literally be impossible for a company to contract for exclusive rights to any future inventions developed by an R&D worker with all future employers of that R&D worker. And even if it were possible, the contracts would be practically unenforceable because the R&D worker cannot erase from her head the knowledge and skill relevant to the invention. The patent system is the cheapest way for government to facilitate a free-market in human capital.

Does this theory have any practical implications for proposed patent reform? Yes for several reasons.

First, it reminds us that reform should undertaken with the background or alternative of trade secret and contract law in mind. Anything that increases the cost of obtaining patents (including, for example, post-grant oppositions) will tend to drive some inventors away from obtaining patents and toward maintaining trade secrets. The result could be slower dissemination of innovation, (much) higher wages for employed-inventors who can easily move from company to company, but less compensation for inventors overall since there is no easy way for a market for an invention to develop. For most companies that do R&D, even an imperfect patent system looks better than this alternative. (Note, however, that companies not doing R&D -- or stealing it -- should prefer it!)

Second, it reminds us that, in terms of the administrative costs of governmental systems, patent law saves time and money by clearly delineating facts about who, what, when, where, and how inventions were developed. Patents are useful when and because they give companies the right to stop former R&D employees from taking their work to competitors. Contract and trade secret law will always be available at the margin to handle details. The point is that in a world with clearly disclosed and publicly searchable patents and assignments, company managers can focus more on how new technology should be marketed, manufactured, and distributed, and less on how to prevent employee defections (or in planning raids of other companies' R&D employees). It's partly the bright-line, standardized nature of patent rights that make them useful as an alternative to contract. Some reform proposals (including, to the extent that it does change existing law, the proposal for damages apportionment) could be counterproductive in this regard.