Patent law promotes innovations that are different from what the world already knows. This may seem a truism: What is innovation other than the search for new things?
But mere novelty is not the aim of innovation policy—improvement is. Put more plainly, what we want is “better”; what we get is “new.” Often, in fact, we get innovations that are new purely for the sake of being new, and not better at all. Sometimes they are worse. This Essay explores how patent law promotes what I call divergent innovation—innovation that develops significant or minor changes to existing technology rather than learning more about that technology—drawing from examples in the field of biomedical innovation.
Patent law can drive divergent innovation even when other incentives suggest focusing on developing an older technology. Known drugs provide a useful case. If a known chemical would make an effective drug, companies typically still won’t develop it, because it isn’t new, and therefore, isn’t patentable.
Instead, they may make minor changes (or seek out a totally different chemical), even if those changes might make the drug worse, because the patent incentive is so important to the innovative process.
Patent doctrines focus incentives on the search for new and different innovation without emphasizing improving technology or increasing welfare. Novelty, nonobviousness, and utility doctrines all drive innovations’ newness when innovators seek patent protection.
And in a mirror image of patent incentives to create new innovations, innovators “invent around” patents on existing technology not because inventing around will improve a product but to avoid the cost imposed by existing patents.
Divergent innovation may bring benefits, but it also brings costs that differ from the cost of patent exclusivity itself. I identify three here. First, the process of inventing around existing patents is itself costly. To develop a new drug similar to an existing drug, the developer needs to undergo the entire process of preclinical work, clinical trials, and Food and Drug Administration (FDA) approval, at a cost of many millions of dollars. Second, divergent innovation can lead to incompatible technologies, the absence of standards, and the loss of network effects and economies of scale. Third, when the path of innovation is deliberately forked, we lose the ability to draw from existing stocks of knowledge about a particular technology because the new technology is different for the sake of being different. Those new clinical trials mean that we know less about both the new drug and the old drug than we would have known if we had simply developed more knowledge about the old drug. Even when divergent innovation effectively moves technology forward, the costs of divergent innovation should be weighed against its benefits.
The costs of divergent innovation have gone understudied partly because divergence fits so neatly within a patent-driven market theory of innovation. Patent law relies on the market to sort out the value of inventions. Patents are only worthwhile if the protected goods are valuable in the marketplace, so the market will work to sort out the valuable innovations from a mélange of patented inventions.
As John Duffy puts it, “[P]atent law has no aversion to awarding commercially worthless property rights.”
But firms, with their private knowledge about markets and consumers, can predict market value (to some extent), and use that information to drive their innovation investment decisions. Under this account, while patent law does not require superiority for individual inventions,
patent law and markets together should lead to overall improvements over time.
Unfortunately, markets aren’t always great at identifying innovating improvements. Consumers select goods for many reasons besides quality.
We might therefore expect the costs of divergence—the source of the variety from which markets pick the winners—to be more problematic where market mechanisms do a poor job of incentivizing, selecting, and adopting superior innovations.
Markets are especially bad at selecting superior biomedical technologies. Efficient markets require informed consumers who can choose goods. But biomedical technologies, which make up a trillion-dollar annual industry with tremendous health implications, are often “credence goods,” meaning that their users cannot evaluate the innovations’ quality independently.
Even with FDA regulation, information about biomedical technologies’ quality is frequently poor or unavailable.
Finally, patients, doctors, and insurers split the consumer functions of selecting, paying for, and benefiting from goods, each with their own incentives. Since market mechanisms for selecting superior technologies underperform for biomedical innovations, the costs of divergent innovation matter more. These costs are particularly significant because biomedical technology encompasses drugs, which are presented as the exemplar of an industry where patents are truly important and work as designed.
Patents create incentives within a broader innovation ecosystem, where some policy tools, like patents, promote divergence, but others discourage innovators from diverging. For biomedical technology, consider two among many:
Market approval by FDA and health insurer reimbursement decisions each create substantial innovation incentives. Each can penalize divergence. At FDA, specialized processes ease market access for new medical devices that resemble existing devices; if the technology is substantially different, getting premarket approval can be much harder.
Similarly, winning insurer reimbursement for a new technology is easier if the insurer is familiar with the technology, such that the innovator need not make their case from scratch.
The interaction between these incentive systems can lead to bad outcomes: Patent incentives pushing for newness can be partially counterbalanced by reimbursement and regulatory incentives pushing against too much newness, resulting in a spate of technologies that are different enough to bring the costs of divergence, without being sufficiently different to bring substantial benefits.
Good innovation policy depends on understanding innovation, and the costs of divergence are a part of that. This Essay makes a general claim: Patent law drives divergent innovation, and that divergence carries costs. As to this picture of innovation in general, the only prescription I offer is that scholars and policymakers take divergence costs into account when analyzing innovation incentives, benefits, and costs.
This Essay also makes a specific claim: Divergence is especially costly for biomedical innovations, which patents and other incentives can drive toward an unhappy medium of differentiating, proliferating, nonsuperior technologies. This specific problem may be amenable to solutions. Within patent law, strengthening the nonobviousness requirement could reduce close imitators of existing technology. Outside patent law, FDA or insurer requirements for superiority could do the same.
This Essay proceeds in five Parts. Part I describes the ways in which innovation paths can diverge or not. Part II lays out how several patent doctrines can drive divergent innovation. Part III provides three cases to illustrate divergent innovation and the costs it can bring: me-too drugs, including statins; insulin pumps for diabetics; and epinephrine auto-injectors like the EpiPen. Part IV places patent law into a broader context in the case of biomedical innovation, addressing how the incentives provided by FDA approval and insurer reimbursement can drive innovation to follow the path of existing technology. It also considers how these combined incentives can perversely lead to innovation landing in an unhappy middle: close enough to existing technology that we derive little social benefit from diversity but far enough from that technology that we see the costs of divergence. Part V describes potential solutions to the problems specific to biomedical innovation, located either within or outside patent doctrine.