Lemley recounts the true stories of these and other inventors in Part I, which is a much more entertaining read than the typical law review article. Lemley is not the first legal scholar to discuss the true stories of these "lone inventors"—for example, Lea Shaver examined Edison's contribution to the light bulb, and Merges and Nelson looked at the light bulb, automobiles, airplanes, and radio in their classic 1990 article—but I don't know of anyone who has gone through all these stories (including the "exceptions" like penicillin and the photocopier) so systematically. And Lemley goes beyond anecdotal evidence by discussing a number of studies indicating that incremental progress and simultaneous invention is the norm, not the exception.
In Part II, Lemley examines the problems that these inventor stories cause for the dominant theories of patent law:
- Incentive theory (that patents are needed to incentivize innovation): "The overwhelming prevalence of both independent invention and incremental contribution calls this basic incentive story into serious question." (Or does this just mean that patent incentives are too strong?)
- Commercialization and prospect theories (that patents are needed for development ex post): For "structural reasons monopolists are actually poor managers of an invention" and in the exceptional "singleton invention" cases of penicillin and the photocopier, there was "substantial delay between invention and commercialization." The inventor stories also show that inventors often hinder the development of new uses.
- Disclosure theory (that patents are needed to disclose technical information): This theory cannot justify the patent system because "inventors don't learn their science from patents." Based on my survey of nanotechnology researchers, I have argued that scientists actually learn more from patents than most legal scholars think, but that disclosure still isn't a justification for the patent system. So I agree with this section's overall conclusion, but it is not clear that the inventor stories from Part I can teach us anything about disclosure.
Part III contains the most novel aspect of Lemley's argument: not only might races have some benefits (which has been suggested by others like John Duffy, Michael Abramowicz, and Suzanne Scotchmer), but racing might be the dominant normative justification for the patent system. "In some . . . of the examples . . . in Part I [like the telephone and cotton gin], the inventors were acutely aware of the possibility of patent rights and of the risk that others might obtain the core patents," and even where there's no evidence of a race, "[i]t is possible that these inventors knew they were racing against identifiable [or unknown] others working on the same thing." And these races are not necessarily wasteful: they can result in faster invention, different solutions, or higher quality ideas under pressure. Under a racing theory, "the 'incentive' offered by the patent system is not the promise of a payoff, but the threat of being taxed or even excluded from the market entirely if they lose the race"—a stick, not a carrot.
As Lemley acknowledges, this paper is far from a defense of the patent system on racing theory grounds: "patent races may have specific social costs," and we need to determine "whether the innovation benefits of granting patent rights exceed their costs." (For a concise listing of the costs "in terms of static inefficiency and in lost opportunities for future improvement," see p. 36 of the current draft.) Rather than concluding that racing theory justifies patents, this paper is a call for future research, and Part III concludes with a useful bulleted list of research directions from which to investigate racing. I would love to see this conclusion loop back to the inventor stories from Part I: with all the discussion of patents hindering future innovation in those cases, is it plausible that the gains from patent races outweighed the efficiency losses for those inventions? Of course, this can't be answered definitively (and even guessing at the answer is difficult), but teeing up this question might be helpful for thinking about this theory.