I feel like Charlie Chaplin in a pie factory. Before I could work through an op/ed by Niels Reimers in the Mercury News last April (2021) that the Bayh-Dole Coalition has dredged up to contest the use of march-in to address unreasonable terms–apparently terms are never unreasonable–Niels Reimers has gone and published another op/ed, also in the Mercury News. Reimers reuses some of the same copy in the second as the first, but changes the worry from march-in for covid vaccines to march-in for the prostate cancer drug Xtandi.
Niels Reimers started the Stanford TLO in 1970, back when it was one of a very few university-based patent management offices. Since then, Reimers has been one of the big influencers of university IP practice, having worked with both the University of California and MIT to refashion their approach to be similar to that of Stanford’s TLO. It’s tough to take on an old pro in university technology transfer, one who has shaped much of university TLO practice (for good or ill). But when old pros go political and swap getting things done that serve the public for trying to prevent the public from getting the benefit of its statutory bargain, then it’s time to respond.
Gosh, this is going to be long–a series of articles–but there’s some really good stuff to point out, when Reimers grossly distorts Bayh-Dole, which fortunately for us, he does often. There’s a short form if you want to save time and skip the really good stuff.
Reimers is wrong.
Reimers grossly distorts Bayh-Dole.
March-in addresses price by introducing free competition.
March-in is third among Bayh-Dole remedies for price-gouging and price discrimination.
Private investment won’t catastrophically drop if the public gets reasonable terms.
Bayh-Dole, even enforced, has defects so bad no regulatory twiddling will save it.
Let’s work through both op/eds to see what Reimers has to say. By using Bayh-Dole, Reimers claims, the government will undermine Bayh-Dole. Pretty twisted stuff to come.
We will start with the earlier op/ed, and then come back around to deal with the later one.
Reimers makes his primary point upfront:
If the United States is to remain the world leader in private-sector innovation, we have to preserve Bayh-Dole.
I know–it really sounds like to preserve Bayh-Dole means to put it in a jar and add formaldehyde, which would be a good thing. But no. To preserve Bayh-Dole means, here, not to use it. In particular, Reimers argues that the federal government should not use the Bayh-Dole provision of march-in to protect the public from unreasonable use of inventions made in work receiving federal funding. Under march-in, the government requires the owner of an invention subject to Bayh-Dole to introduce competition to address unreasonable terms (such as price) and insufficient availability (for inventions directed at public health).
Let’s look at a few claims Reimers makes, and then (a few articles ahead) discuss the big things that Bayh-Dole has got wrong, some of which cannot be fixed by legislative tinkering.
The law allowed universities and non-profits to retain the patents on discoveries made with federal funding.
Reimers repeats a claim in a form that the Supreme Court rejected. Bayh-Dole allows nonprofits and small companies to retain title to inventions that they have acquired. Reimers leaves out the whole acquisition step, as if somehow Bayh-Dole hands title to universities and nonprofits, and it doesn’t. Without acquisition by a contractor, an invention made in work receiving federal funding is not a subject invention and not controlled by Bayh-Dole. Bayh-Dole actually makes it more difficult for a university or other organizational contractor to claim ownership of inventions made in federally funded work. Reimers instead makes it appear that Bayh-Dole gives universities a special right in such inventions and patents. The Supreme Court in Stanford v Roche was clear on this point. Reimers declines to follow the Court.
In exchange for royalties, universities can then license those patents to businesses, which take big risks in commercializing those ideas.
Generally, this is not true, either. Yes, in theory, a university might license an invention and receive royalties, but in practice it doesn’t work this way. Reimers wants you to mistake a nice sounding theory for fact. In practice, universities take control of inventions and do mostly nothing with them. In the Bayh-Dole era, universities and their affiliated foundations have received over 50,000 US utility patents citing federal funding. And they have in the same time obtained over 70,000 more US patents that don’t cite federal funding. Given that federal funding averages about 60% of a university’s extramural research budget, but only about 45% of its patenting, it would appear on the face of it that federal funding works against university patenting. But that’s not the big part of this story. While in theory universities could license the inventions they’ve taken, they mostly don’t. Reimers doesn’t tell you that.
At Stanford, where Reimers ran the TLO for many years, they licensed about 20% of their inventions. And only a few of those 20% that got licensed ended up as commercial products, something like 257 out of 6,400 over 36 years, by one account, or about 4%. And Stanford is considered to be one of the top university licensing programs in the country. That means that over 6,000 inventions at Stanford were effectively suppressed by an institutional claim of exclusive rights–and companies then had to design around those inventions, ignore them, exclude them from use, exclude them from working standards. That’s a horrible outcome–other than if one aims to make a pile of money from only a handful of inventions. It’s rather like dredging up all the oysters in the bay, looking for a few that have pearls, and making it appear like the destruction of the whole bay is worth a handful of pearls. It’s the old “rape the land” argument applied to research enterprise.
Reimers says that the companies taking university patent licenses “take big risks.” Maybe, sometimes. But much of the time, there are no such big risks or the biggest risk a company takes is created by the license deal itself. In the case of Roche, which Stanford sued for infringement, the company had already developed its new product and was selling it. The big risk was fighting Stanford’s illegitimate claims to an invention underlying Roche’s product, which turned out to be jointly invented by Stanford personnel while working at Roche.
It’s not true, in general, that a company takes “big risks” in using something that’s patented in its commercial operations. A university can also increase the risks involved in working with a new technology by holding out for an exclusive license. If a number of companies have access to a new technology, they can compare notes, share data, and reach an understanding on a platform–a common set of specifications that they can all then use. The risks get distributed among the companies involved. The “big” risk gets mitigated. But if only a single company gains access to a new technology, then that exclusivity itself enhances the risk. The company has to do the work on its own. Other companies are motivated to design around or undermine the company with exclusive rights. The exclusive license itself increases the risks. The university’s IP practice increases the risks. Bayh-Dole IP practice increases the risks even more.
And keep in mind that just because a university grants to a company an exclusive license does not mean that the company will “commercialize” the licensed invention. Often, it doesn’t. It just takes the license, fusses around for a while, and then drops work or shifts the work over to a similar invention not within the scope of the patent license. Take the Sonicare toothbrush. The invention at the University of Washington was for the use of ultrasound to clean teeth. UW dutifully licensed its rights to a startup company. The company then hired the lead inventor as a consultant and designed around the UW patent, discovering that the sound didn’t have to be “ultrasound.” So it’s “Sonicare” and not “Ultrasonicare.” But, hey, UW counts the license and the startup, even though the UW invention didn’t technically get used. Do you think that’s what Congress intended when it passed Bayh-Dole: “use patents on inventions made in federally supported work to get money out of companies even when they don’t use that invention because once they have exclusive control of the patent, they can use the patent to suppress competition that would use the invention made in federally supported work.” In other words, companies use patents on federally supported inventions to suppress the use of those inventions in favor of what the companies have invested in. UCLA’s claims to a nicotine patch (with federal funding, just not, as it turns out, despite UCLA’s assertions, subject to Bayh-Dole) worked pretty much this way.
This public-private collaboration has spawned thousands of startups.
Spawn is a good verb here. But Reimers has changed things up without giving you any notice. The “public-private collaboration” has to do with universities and licenses generally. Remember–60% of university patents are not Bayh-Dole. Universities were licensing inventions–or at least their affiliated foundations and Research Corporation were–long before Bayh-Dole. Stanford’s first two big license deals–voltage controlled oscillators for electronic music and gene-splicing methods–neither were Bayh-Dole. Bayh-Dole has not “spawned” public-private collaboration, and there’s a good argument that Bayh-Dole practice has done just the opposite. In the two main areas in which one might expect to see public-private collaboration–research tools and research consortia–federal agencies have to work around Bayh-Dole to get things to happen. The NIH publishes guidelines on unique research resources, pleading with universities not to license such resources exclusively–even when the NIH ought to determine exceptional circumstances under Bayh-Dole and preclude exclusionary practices for research tools. But the NIH won’t do that, in part because Bayh-Dole’s requirements for exceptional circumstances are so clunky that the NIH is afraid to try. Same for the NSF, with its Cooperative Research Center program and related multi-company/university consortia. Instead of using Bayh-Dole, the NSF requires universities to adopt a consortium membership agreement that dictates non-exclusive access to any invention made in the consortium’s research. University TLOs then turn the procedures into a heaping pile of bureaucratic pain for everyone involved. But Reimers won’t tell you that.
So, not all public-private collaborations involve inventions made in federally funded work. And Bayh-Dole did not start up public-private collaborations all of a sudden, but rather intrudes and disrupts all sorts of such collaborations.
Startups sounds nice, and thousands of startups must sound nicer.
At Stanford, this kind of technology transfer gave rise to wildly successful companies, including Google.
Let’s get very basic. Stanford took ownership of the page rank invention, and then licensed the rights *back* to the inventors when they started Google. When a university demands to own an invention and then licenses that invention back to the inventor’s startup, that’s not “technology transfer.” That’s bureaucratic redirection. The inventor already has the technology. What the inventor lacks, because university administrators have demanded it, is the right to make, use, and sell the claimed invention. It’s rights arbitrage to demand that the inventor license back an invention and pay the university for that right. There’s nothing enabling about it, other than administrators have demanded the right, and then released the right once they have been assured they will get paid for releasing the right.
It’s also not much of a collaboration at that point–it’s more of a shakedown, just as a company gets going it has to deal with a 35 page fine print exclusive license agreement that university administrators have packed full of all their institutional desires (money, lots of it) and fears (liability, again, lots of it).
And the University of Pennsylvania licensed the IP from early research into messenger RNA technology, which, after decades of research and billions in private capital, later led to the Moderna and Pfizer COVID-19 vaccines.
A nice thought, but the underlying history is not quite what Reimers wants you to think. Penn refused to license the mRNA invention–a substitute for one of the amino acids that allows synthetic mRNA to evade the human immune system–to the inventors’ startup. Instead, Penn exclusively licensed to a small Wisconsin biotech firm for $300K plus royalties, that apparently sat on the rights for five years before granting two non-exclusive licenses for $75M each, plus royalties. One sublicense went to a start up out of Harvard–which became Moderna, and the other to BioNTech, where one of the Penn inventors ended up. It looks like a case of patent arbitrage, in which Penn moves the patent rights to a company that then does non-exclusive deals rather than developing the invention itself. The company did what Penn could have done except Penn was fixated on granting an exclusive license. So Penn gets something like $15M and Penn’s licensee nets $135M for brokering deals. Nothing here about companies having to have exclusive rights before they will develop or use anything. And the deals were done in the last decade or so, and the billions spent mostly was from the federal government.
Reimers’s claim for the necessity of Bayh-Dole is simply wrong. The private investment in mRNA technology absolutely did not depend on patent exclusivity from an invention made in federally funded work. The licenses for development were non-exclusive. If anything, there would have been more private investment to more companies–and the development of standards and shared data–if Penn, or the federal government, had made the inventions available on fair, reasonable, non-discriminatory terms. Penn may even have got way more money than it did from its exclusive license, if what Penn really, really wanted was a lot of money.
Broadly and bluntly, Bayh-Dole has nothing specific to say about startups. The closest Bayh-Dole gets is its statement of policy and objective, 35 USC 200, where it expects that the patent system will be used to “promote free competition and enterprise.” Perhaps “enterprise” includes the initiative to start a new company. It’s pretty remote, I know, but that’s it. Starting companies is not an objective of Bayh-Dole. Much of Bayh-Dole appears to be concerned about small existing companies. Bayh-Dole says this: use the patent system to “encourage maximum participation of small business firms in federally supported research and development efforts.”