Judicial patent decisions are retroactive. When the Supreme Court changed the standard for assessing obviousness in 2007 with KSR v. Teleflex, it affected not just patents filed after 2007, but also all of the existing patents that had been filed and granted under a different legal standard—upsetting existing reliance interests. But in a terrific new article, Patents, Property, and Prospectivity (forthcoming in the Stanford Law Review), Jonathan Masur and Adam Mortara argue that it doesn't have to be this way, and that in some cases, purely prospective patent changes make more sense.
As Masur and Mortara explain, retroactive changes might have benefits in terms of imposing an improved legal rule, but these changes also have social costs. Most notably, future innovators may invest less in R&D because they realize that they will not be able to rely on the law preserving their future patent rights. (Note that the private harm to existing reliance interests from past innovators is merely a wealth transfer from the public's perspective; the social harm comes from future innovators.) Moreover, courts may be less likely to implement improvements in patent law from the fear of upsetting reliance interests. Allowing courts to choose to make certain changes purely prospectively would ameliorate these concerns, and Masur and Mortara have a helpful discussion of how judges already do this in the habeas context.
The idea that judges should be able to make prospective patent rulings (and prospective judicial rulings more generally, outside habeas cases) seems novel and nonobvious and right, and I highly recommend the article. But I had lots of thoughts while reading about potential ways to further strengthen the argument:
1. The article is mostly from the perspective of cutbacks in patent rights that retroactively invalidate patents. Masur and Mortara have some discussion of how strengthening patent rights harms the reliance interests of nonpatentees, but I think they could also make the argument that rule changes that strengthen patent rights should generally be non-retroactive because past innovators won't have relied on the stronger rule. For example, if diagnostics continue to be generally unpatentable for 20 years, and then patentable subject matter doctrine is revised to make clear that they are patentable, it generally wouldn't make sense to say that the rule change applies to any diagnostic patents that managed to slip through the USPTO in the intervening years due to sloppy examination—they should still be invalidated under the old rule because those investment / patenting decisions weren't made in reliance on the new rule.
2. The current draft considers reliance interests of patentees and nonpatentees based on the law not changing, but I think one could also argue that people in patent-related fields have reliance interests that judges will continue to use patent policy levers to change patent law and get rid of patents that seem to be harming rather than helping innovation. That is, if there seems to be a problem, courts will fix it, even if that means retroactively invalidating patents that are holding back an industry. So some innovators might rely on the knowledge that patent law will change.
3. More generally, the extent to which retroactive patent rulings harms reliance interests seems like an empirical question that will vary by field. For example, my intuition is that the prospect of business method patents did not actually have that much affect on business method R&D, and that even before the recent patentable subject matter changes, it wasn't hard to figure out that this was a controversial field in which patenting carried a higher-than-normal risk of future invalidation. And I doubt that the effective elimination of many business method patents has created much concern for patentees in many other fields (e.g., mechanical inventions). And the early empirical work I've seen suggests that the effective elimination of many diagnostic patents hasn't really affected R&D. So I'd love to know more about cases where retroactive changes actually seem to have created reliance problems, along with some concrete examples in which the authors would advocate for making a change in patent law non-retroactive.
4. Masur and Mortara argue that changes to patent law should not be classified as judicial takings, including because legal change would then be "prohibitively expensive." But I think this discussion misses the point that the money that would need to be paid to inventors (and raised through public finance) would be offset by a savings to society through elimination of the "shadow tax" those patents would otherwise have provided through above-marginal-cost prices. Government takings of patents would thus not be so different from other proposals for changing the allocation aspect of patents, such as Michael Kremer's patent buyouts or the Ayres–Klemperer proposal for duopoly pricing, as Daniel Hemel and I discuss in a forthcoming article. There would be additional administrative costs, and yet also administrative cost savings due to less patent litigation / licensing, so it's not obvious how that cuts. That said, I agree that applying takings law to patents is probably a bad idea—but it's because for the most part, I think the major changes that have retroactively invalidated a bunch of patents haven't created huge reliance problems.
As noted above, I think all of these issues could be addressed in ways that bolster Masur and Mortara's arguments, or even developed in a separate piece—I'm sure the article will spur many interesting conversations among scholars of patent law and judicial decisionmaking more generally.