Prof. Robert P. Merges, University of California at Berkeley School of Law
February 15, 2014
Under U.S. patent law, it is irrelevant whether an infringement defendant copied from the patentee or independently invented the patented invention. Many commentators have decried this aspect of patent law. The consensus is clear: Patent law needs an “independent invention” defense.1 Until it gets one, it will in the eyes of many be illegitimate. In the meantime, many observers presume that the U.S. patent system will invite a good deal of unnecessary patent litigation – litigation that amounts, in most cases, to rent-seeking, pure and simple.2 As Mark Lemley puts it: “An independent invention defense would eliminate the [patent] troll problem,”3 because very few companies that manufacture and sell innovative products can be shown to have copied product ideas from obscure patents.
The traditional response to this charge depends on incentive theory. A single exclusive right, it is said, reserves a product market for one firm. Under certain conditions, only this level of exclusivity can ensure an adequate return for the cost and risk of developing a complex technology. Absolute liability, in this sort of set-up, is required regardless of whether several inventors might arrive at the same invention simultaneously. Put simply, a monopoly market position is necessary to stimulate the investment required to develop certain technologies. This is so despite the obvious double costs of this arrangement: first due to the wasted resources that go into duplicative effort to win the “race for the patent”; and second as a result of high consumer prices in the monopoly market resulting from the patent.
This is a plausible story theoretically. But this “need for market exclusivity” idea seems to have lost much of its power. The reason is simple enough. The story is based on two assumptions that seldom seem accurate: (1) very high-cost research projects that (2) culminate in a single market-covering patent. Most contemporary inventions outside the pharmaceutical context are incremental, the result of modest discrete investments made in connection with ongoing R&D. And for this reason, in most cases today individual patents cover but one small component, or one aspect of one component, of large, multi-component technologies. So, outside certain special cases of very high-cost research (most notably pharmaceuticals), it is difficult to make a case for exclusivity based on a single market-spanning patent.
This may well be one reason that it has become more common for patent scholars to call for proof of copying in patent law. The factual assumptions underlying the theoretical case for strong market exclusivity have dropped away. But patent law’s disregard for independent invention lives on. Surely, if a single market-spanning patent is not required to call forth inventor effort and investment, then the rationale for an absolute liability regime is gone. Which can only mean that it is time to do away with that regime by instituting a robust independent invention defense.
Two Points for the Defense
In the face of such a critique, is there anything that can be said in favor of the current U.S. rule? I think so, though as I make clear later, my defense is only a partial one. I think there are two points in favor of the current rule: (1) Research in cognitive psychology shows how subtle copying can be, so proof of actual copying may be hard to come by, which means that a rule requiring proof of copying might result in undercompensation for true innovators; and (2) a hidden advantage of the current regime is that in an odd way it actually encourages researchers from different organizations to disseminate and share technical information.
Because patent law dispenses with the need to prove copying, it also makes it unnecessary for research organizations to construct elaborate systems to prevent the inflow of “outside” information. There is no advantage in proving the absence of outside information in a lab or research group, so this information is allowed to flow freely. The result is a culture of open dissemination, which adds a great deal to the rapid diffusion of technical information and therefore presumably innovation. Requiring proof of copying, in other words, would prompt research groups to invest in elaborate systems whose purpose is to help prove that copying did not in fact occur – because no outside information was allowed into the hands of researchers.
On the first point, the literature on technological “diffusion,” together with research by psychologists on “inadvertent plagiarism,” demonstrate that ideas are sometimes copied in obscure and subtle ways, leaving little or no evidence that copying has indeed occurred. When faced with a task requiring creativity, people regularly produce a piece of information they believe is original but in fact clearly comes from someone else.4 Taken together, this research points toward a mechanism by which new technological ideas might plausibly move from one set of researchers to another. The point is that this movement can take place at a level that does not draw much explicit attention. Even the name of the field suggests how subtle communication can be.5 The process is diffuse, extended over space and time, lacking a distinct, identifiable communicative moment. Given this mechanism, proof of copying may be very difficult to come by.
Next, we turn to the literature on optimal standards of care in tort law. I use it to describe what would happen if U.S. law changed to require patentees to show copying. Potential patent infringement defendants – a class that includes virtually all people and companies performing research and development (R&D) – might well impose strict limitations on receipt of technological information, so as to rebut allegations of copying, and therefore reduce liability.6 That would be bad. Technological communities thrive on ubiquitous and unregulated communication. Patent law as it stands encourages this, by making proof of copying irrelevant in patent cases. As a consequence, under the current regime researchers (potential patent infringement defendants) have no reason to restrict their access to technical communications. This accords with the economic literature on tort law, which shows that investments in precautionary measures can sometimes lead to inefficient reductions in activity levels.7
The idea here is that the fear of liability for patent infringement might lead firms to erect thick barriers to the inflow of outside information. This might very well represent a substantial loss in social welfare, because of the many benefits of permitting robust and free-flowing information among firms. But tort rules sometimes lead to this result. As long as a certain choice increases overall private payoffs, that is the choice the firm will make. This is true even if the choice leaves society worse off. Arguably this would be the precise result that would follow from serious restrictions on inflows of outside information. (Another way to say this is that private barriers to inflowing information impose a negative externality on society.) The current rule of absolute liability prevents this result, by making it irrelevant for infringement liability purposes whether or not outside information ever flowed into a firm.
As evidence that the restrictions can be costly, we need look no further than copyright law and common law rules on the theft of ideas – both of which require proof of copying, and both of which have led potential defendants to invest in restrictive measures to guard against access to third-party information. As the torts literature suggests, these measures, if employed by technical researchers, might well lead to a serious reduction in technical communications.
The Place of Prior User Rights
The obvious downside of the current regime is that sometimes, perhaps quite often, an infringement defendant really has invented independently: The research team did not in fact learn or copy, directly or indirectly, from the patentee. In these cases, as the literature shows, absolute liability imposes significant costs. My point earlier was that these costs must be weighed against the benefits of information freely shared under the current absolute liability regime. And it is of course hard to say how the costs and benefits net out under such an analysis.
Under these circumstances, a recent innovation in U.S. patent law, the new “prior commercial use” defense under the America Invents Act (AIA), may prove helpful.8 This defense permits one who can prove early commercialization of an invention to avoid liability even though he or she otherwise infringes. By rewarding an infringer’s early action, it cuts down on some of the costs of the absolute liability standard, while leaving that standard as the general rule in patent infringement cases. It is difficult to say whether this combination of (1) absolute liability, plus (2) an exception for early commercialization, provides the optimal set of incentives for potential patent infringers. But at least the prior commercial use rule encourages activity that has independent social value, in the form of rapid movement toward the market. Given that there are, as I show, real benefits to the longstanding rule of absolute liability in patent law, this may be the best we can do.
By requiring only proof that the infringer is using an invention covered by the patentee’s claims, U.S. patent law in its current form dispenses with the need to prove actual communication. This means that inventors seeking patents can freely disseminate their ideas, without needing to memorialize carefully all communications with fellow researchers. And it means that other researchers have no incentive to protect against receipt of outside information.
This, together with the fact that copying can be subtle and “inadvertent,” constitutes a reasonable justification for the absence of an independent invention defense in current U.S. patent law. A recent exception to this – the prior commercial use provision of the AIA – represents a fair yet limited way to recognize the importance of independent invention. Because this exception requires actual deployment of earlier research more than a year prior to a patentee’s filing date, it will only apply in a small number of cases. This may be a good thing, as it preserves the general incentive not to engage in extensive information embargoes to protect against patent infringement liability.