In The Null Patent, Sean Seymore (Vanderbilt Law and Chemistry) makes the creative proposal that the patent system could help solve the problem of unpublished negative scientific results. Negative results are very important: they can save other scientists from wasting time on dead-ends, and they can help identify false positives in studies based on statistical significance. (There is generally a 5% chance that a "significant" result is just based on random error, which is a problem if only significant results are published, as xkcd aptly illustrates.) So it would certainly be good for science if more negative results were published. But what do patents have to do with this?
Seymore notes that private efforts to encourage publication of negative results have been largely unsuccessful, which means that government intervention may make sense, as the NIH is already doing with clinicaltrials.gov. But Seymore argues that the PTO is the right agency to manage this, both because it has already demonstrated the ability to create an accessible and searchable database, and because these negative results would benefit patent law. He proposes that researchers should be able to publish negative results in a patent-like "null patent," which would resemble an actual patent but without claims. In fact, he notes that something similar already exists in the Statutory Invention Registration (SIR) program (which is set to be eliminated by patent reform).
These null patents, Seymore suggests, would create an important prior art database. A 30-year-old null patent showing a failed attempt to make compound X would be per se nonenabled and would thus not defeat the novelty of a future patent on X. But that null patent might be patent-defeating on obviousness grounds if the reason for the failure was a problem that has since been solved, so that a PHOSITA could use the null patent to easily make X today. (It isn't clear to me, however, that this situation would occur more often than having the null patent on X be evidence that it isn't obvious how to make X.) I suspect the place null patents would be most useful is in showing that some patents that rely on constructive reduction to practice are not enabled, which is a significant problem—my recent survey of nanotech researchers found only 38% of patent-reading respondents think the inventions in patents are reproducible.
If the PTO started allowing null patents, would scientists participate? My survey also showed that a surprising number of nanotech researchers read patents and find useful technical information in them, so I think scientists might be willing to read null patents as well. But would anyone write them? Seymore makes the interesting suggestion that federally funded scientists be required to participate as a condition of getting grant money, which would probably work, although there would be problems defining what constitutes a negative result that must be published. (His suggestion that scientists account for "100 percent of the research effort" is impractical, since much scientific effort is spent on things that aren't "results" at all, like running code that has bugs or doing experiments incorrectly.) Seymore also suggests offering expedited patent or journal review to researchers who submit a null patent. Even with these incentives, I think it would be hard to convince many researchers that writing null patents is worth their time—but as Seymore notes, you don't need everyone to participate for this to be valuable.
I think we are unlikely to see null patents anytime soon, especially given the PTO's strained financial state and Washington's budget issues, but that doesn't mean it's a bad idea—though I would want to be more convinced that the PTO is the right agency to handle this. But null patents are an intriguing and novel idea, and I think Seymore's discussion of the role of failure in patent law is an area ripe for follow-up research.