Faux Bayh-Dole and Stanford v Roche

I have been tracing the history of two versions of Bayh-Dole. One version is based in the law as written and reflected in the implementing regulations and the standard patent rights clause. The second version shows up immediately after the passage of Bayh-Dole in statements by Norman Latker, Howard Bremer, Joseph Allen, and others.

In the legal version, Bayh-Dole applies to federal agencies and directs them to use a standard patent rights clause, to be prepared, and sets out the conditions under which a federal agency can require the assignment to the federal government of an invention made with federal support. In the faux version, Bayh-Dole applies directly to universities and gives them title outright to inventions made with federal support. Any paperwork involving assignments is then something of a scavenger hunt to satisfy patent office rules that for some reason were never changed when Bayh-Dole came into effect.

Stanford v Roche

The Supreme Court in Stanford v Roche had to decide between these two versions. In one version, patent law had been changed to vest title of inventions with universities, not inventors. In the other version, federal rules regarding agency contracting were changed to limit what agencies could require as a condition of awarding money for research proposed by university faculty and small businesses. Over 70 universities signed on to an amicus brief advocating for the vesting version of the law. Senator Bayh submitted his own amicus brief arguing as well for the vesting version:


The Supreme Court considered all the arguments and found that there is in fact nothing in Bayh-Dole that requires inventors to give up their rights in inventions.

Nowhere in the Act is title expressly vested in contractors or anyone else; nowhere in the Act are inventors expressly deprived of their interest in federally funded inventions.

The Bayh-Dole Act’s provision stating that contractors may “elect to retain title” confirms that the Act does not vest title.

The Bayh-Dole Act does not confer title to federally funded inventions on contractors or authorize contractors to unilaterally take title to those inventions; it simply assures contractors that they may keep title to whatever it is they already have. Such a provision makes sense in a statute specifying the respective rights and responsibilities of federal contractors and the Government.

Only when an invention belongs to the contractor does the Bayh-Dole Act come into play. The Act’s disposition of rights—like much of the rest of the Bayh-Dole Act—serves to clarify the order of priority of rights between the Federal Government and a federal contractor in a federally funded invention that already belongs to the contractor. Nothing more.

The Fight Over the Regulations

There was a big fight between the federal agencies and the universities over the drafting of regulations to implement Bayh-Dole, including the standard patent rights clause. One outcome of that fight was a weakening of march-in requirements. Another, however, was the (f)(2) requirement. That requirement could have been parallel with that of the earlier Institutional Patent Agreements–requiring universities to obtain assignment of all inventions made with federal support.


They had that text available if they had wanted to use it. But they didn’t. Instead, they drafted this:


This written agreement requirement is nothing like the IPA’s–other than there is a written agreement required. In (f)(2), universities are required to protect the government’s interest by delegating authority to inventors rather than taking rights from them, as the IPA required. Inventors are to be required (that is, delegated the responsibility and released from any conflicting obligation to the university) to disclose inventions, sign paperwork to permit the filing of patent applications, and most importantly, to sign paperwork to establish the government’s rights in their inventions. There would be absolutely no need for this last requirement if universities owned outright subject inventions.

But further, there is no authority in Bayh-Dole for universities to take title or to have title vest in them or even for there to be a “presumption” that title is to be disposed this way, it would appear there is also no authority in Bayh-Dole under which anyone could draft a standard patent rights clause that went ahead and vested title with universities anyway. Vesting title is not something that could be done at the federal contracting level; it is a matter of federal law (or state law, in efforts to displace common law). The drafters of the standard patent rights clause could have required universities to require assignment, but they did not even do that–since there was nothing in Bayh-Dole that mandated it, to do so would have involved creating a government-wide federal policy out of thin air, with nothing but whim and politics behind it. At least someone involved felt the pressure of staying within the limits of the law as it was written. Hurrah for you, whoever you are!

What the (f)(2) agreement does is patch Bayh-Dole to make it work. If Bayh-Dole had been a vesting statute, then there would have been no need for (f)(2) or even for a requirement that universities make all their inventors assign to the university. Bayh-Dole does not do that, even with its clever use of “retain” as in “elect to retain title.” Thus, the problem (one of many, really) in Bayh-Dole is that there’s no way for a federal agency to reach to individual inventors to gain assurance of at least a non-exclusive right in any subject invention for government purposes, nor for requesting assignment of subject inventions if an inventor does not otherwise assign rights and trigger the election to retain title apparatus. The (f)(2) agreement does this, by requiring universities to flow down obligations under the standard patent rights clause to individuals. The individuals then become parties to the funding agreement and by the definitions in Bayh-Dole therefore become “contractors” and when they invent, because they own rights in their inventions, those inventions are then inventions “of a contractor” and are therefore properly subject inventions from the moment they are created. That’s an astute solution to the problem.

The Supreme Court saw that “retain” could not carry the meaning that Stanford and the horde of university amicae had placed on it:

It would be noteworthy enough for Congress to supplant one of the fundamental precepts of patent law and deprive inventors of rights in their own inventions. To do so under such unusual terms would be truly surprising. We are confident that if Congress had intended such a sea change in intellectual property rights it would have said so clearly—not obliquely through an ambiguous definition of “subject invention” and an idiosyncratic use of the word “retain.”

No Recourse for Inventors

The Supreme Court went further with its concern. If Bayh-Dole vested ownership of subject inventions directly with universities, then a university could simply declare an invention to have been made with federal support and take it. Bayh-Dole, the Court noted, provides no recourse for inventors to challenge such a determination. That would “be deeply troubling.” But the absence of such protections makes sense if there’s no vesting:

But the lack of procedures protecting inventor and third-party rights makes perfect sense if the Act applies only when a federal contractor has already acquired title to an inventor’s interest. In that case, there is no need to protect inventor or third-party rights, because the only rights at issue are those of the contractor and the Government.

What is still deeply troubling is that university administrators exploited exactly this argument in claiming ownership of inventions made with federal support, even when their own patent policies gave them no authority to do so. The way they worked it was to rely on policy provisions that made the terms of approved contracts take precedence over patent policy. If the university contracted so that it obtained ownership of inventions, then heck with whatever the patent policy might say. This exploit is still in use, now often backed by research policies that stipulate that administrators must insert clauses into research agreements that require the university to take ownership of any inventions made by its own personnel in that research.

The Stanford Patent Policy

This exploit was also at the root of the Stanford v Roche situation. Here’s Stanford’s patent policy at the time:


Inventors own their inventions unless there’s a contract that requires otherwise. Agreeing to follow the policy is not contemplated as such a contract–there has to be “other arrangements” “specifically agreed upon in writing.” Stanford policy then requires such agreements when there’s a research contract that requires it:


And further, even if such an agreement is made, Stanford still requires itself to waive inventions to inventors if the sponsored research agreement permits it to do so:


Now into this policy environment comes the problem with Roche. In 1988 Stanford post-doc signs a patent agreement with Stanford and then is sent for nine months to Cetus to study its PCR technology. To gain access, he signs a patent agreement with Cetus promising to assign (and “hereby assigning”) to Cetus any inventions made as a result of his access the company’s proprietary technology. The post doc returns to Stanford, joins a federally funded project working on the same thing that Cetus was working on, and promptly invents. Stanford files a patent application, then gets an assignment from the inventors (why? if title vests), and sometime later notifies the government that it elects to retain title. In 1991, Roche acquires Cetus’s PCR assets, develops a kit using PCR to evaluate the efficacy of HIV interventions, and starts selling. Stanford asks Roche to take a license to its patents, and Roche declines, arguing that it already has rights through the post doc’s assignment of rights. In 2005, Stanford sues for a quarter billion in damages and the case works its way through the courts, culminating in the Supreme Court decision in 2011.

The Promise to Assign

The argument has been made that this case was about whether a later present assignment (“I hereby assign”) defeated an earlier promise to assign (“I will assign”). Even though the wording was debated in the case, that’s not what was on the line, ever, other than as an argument made to try to sway the court. But the court did not need to address that argument otherwise. The federal appeals court had got things right–when Stanford got around to requiring assignment of the invention, the post doc had no rights to assign, having already committed them to Cetus, with Stanford’s approval. Nothing in Stanford’s policy or agreements with the post doc gave Stanford any standing to require or even to expect ownership of any inventive work the post doc did while at Cetus. Here’s the agreement that the post doc signed, as submitted by Stanford in its court paperwork. We will take the first few sections one at a time.


The preamble references federal funding agreements and indicates that these may “impose and set forth certain obligations and requirements with respect to rights in copyrightable materials and inventions.” The agreement is specific to “agreements” and does not consider ownership by operation of law. Thus, at least at the outset, this agreement does not specify that inventions made with federal support simply vest with the university. It’s more complicated than that.


Now comes a statement of consideration–because I have the benefit of one or more of these theories of consideration, I agree to what follows. Keep in mind, however, that with regard to inventions, Stanford by policy makes no claim to inventions unless there’s a contractual requirement, as referenced in the preamble. This present document is not some “other arrangement” as anticipated by the patent policy. It is a document that confirms obligations should they arise in research contracts, bound up by at least some of the consideration listed.

We then get a conditional restricted to a sponsored agreement (“under or in the course of an [sic] Contract or Grant”).


The agreement requires two classes of things to be provided to Stanford. The first class has to do with copyright:


The second class has to do with inventions. Here is what the post doc has promised to provide to Stanford:


Stanford requires an invention disclosure with enough detail that it could be used as the basis for a patent application. The second paragraph brings us to a promise to assign:


Read carefully. The agreement to assign is limited to those things required by “Contracts or Grants.” The language here is interesting–“I agree to assign or confirm in writing to Stanford and/or Sponsors . . . .” The promise to assign is not directly to Stanford–it is to whomever the Contract or Grant might require assignment. Furthermore, the use of “confirm in writing to Stanford” is an odd usage. It makes more sense if it is written to cover in the abstract the faux version of Bayh-Dole, that it is a vesting statute and that Stanford already owns inventions made with federal support, so all the inventor must do is “confirm” it. But “confirm” may here be simply an ambiguous nod toward works made for hire under copyright law.

The entirety of the obligation presented here, however, has to do with the scope of Stanford’s obligations to sponsors of research as provided in funding agreements with those sponsors. There is no outright demand for assignment to Stanford but for a funding agreement that requires it. This is all entirely consistent with Stanford patent policy. The consideration stated–broad in scope and consisting of a variety of theories–intends to make the agreement binding, as between Stanford and anyone signing it.

Stanford’s expectation to own inventions made with federal support rests on there being in federal funding agreements a requirement that Stanford own–either by vesting or obligation to obtain assignment from inventors on behalf of the government.

It is possible that Stanford administrators considered their agreement to be responsive to the (f)(2) requirement in the standard patent rights clause authorized by Bayh-Dole. But these two agreements do very different things. The (f)(2) agreement requires a university to make potential inventors conditional parties to the funding agreement, so they commit to protect the government’s interest independent of anything a university might do or fail to do. With (f)(2) complied with, the government has a direct relationship with inventors through the funding agreement, just as it also stipulates it has with subcontractors. Indeed, if a federal agency permits inventors to retain rights to their inventions, the implementing regulations require the agency to treat them as if they are small business contractors, not as the folks at the bottom of a university pecking order. And even then, the requirements placed on university inventors are even fewer than those otherwise placed on small business contractors.

By contrast, the Stanford agreement is between Stanford and the potential inventor. It is a matter of state contracting, not federal contracting. Stanford construes this agreement as an employment agreement and recites California Labor law at the end. But the (f)(2) agreement has nothing to do with an employment agreement, even though it is restricted to certain employees. It is not a condition of their employment by the university. It is a condition of their participation in the “planned and committed” activities of the federally supported project. California Labor law has nothing to do with it. The (f)(2) agreement is a federal agreement, not an agreement enforced by state law. The Stanford agreement does, however, provide for the possibility that an inventor would assign directly to a sponsor and not to Stanford. There is nothing in this agreement that assumes that all ownership must move through Stanford in order to comply with the terms of a sponsored research agreement. That makes good sense, actually. It’s a pity that more universities don’t use that approach.

The next sections do some mop up work on details:


Waive compensation under Atomic Energy Acts. The matter of commissioned copyright works. A statement about prior commitments. In paragraph 5, “confirm” (unlike the ambiguous usage in paragraph 2) reflects the possibility of work made for hire–where authorship and hence ownership of copyright would be with the university by federal law. Perhaps the folks behind the clever wording in Bayh-Dole sought a similar “invent for hire” provision for their special garden in patent law. It’s just that they didn’t dare (or weren’t competent) to draft that openly. While there is a bit of an argument that an employing or commissioning party can specify original expression and so be considered the author, it is more difficult to image that one could so specify an invention and not then be the inventor. As it is, in American law, even hiring someone to invent only gives the employer standing to claim ownership of the invention–the inventor is still the one who does the inventing and owns the invention until the invention is assigned to another.

Then we get a commitment not to create conflicts with this agreement and a statement of scope with regard to inventions.


Did the agreement the post doc signed with Cetus create patent obligations in conflict with this agreement? Absolutely not. At the time that the post doc signed the agreement with Cetus, there were no grants or contracts that placed restrictions on him through this agreement. Stanford’s patent policy also confirmed that there were no university claims on his work at Cetus. Even a university patent policy that claims everything that can be owned and most of the stuff that can’t be owned, too, does not reach to inventions made off-campus, not on university “time,” and not within the scope of any university contract and not with any university-controlled money.

In paragraph 7, we get a construction with an echo of Bayh-Dole concerning the scope of interest in inventions: “all inventions made, conceived, or first actually reduced to practice.” The use of “made” is interesting. To make an invention that’s patentable is to conceive it and reduce it to practice and recognize that it is inventive–in addition to the invention meeting other statutory requirements. So the list is expansive, setting the scope to include inventions merely conceived–that would be just ideas, thoughts toward an invention–and as well things first actually reduced to practice, regardless of whether the invention has been made and constructively reduced to practice in the form of, say, a patent application. The “or” between conceived and first actually reduced to practice comes from Bayh-Dole, as does the focus on “first actually” rather than simply what is made. This odd usage shows up in the standard patent rights clause in the definition of subject invention. So it figures that Stanford would set the scope sufficiently broad to capture subject inventions within their patent agreement.

But in Bayh-Dole an invention has to be first patentable–a separate definition. That means that an invention has already been “made”–conception, reduction to practice, recognition (plus the other statutory requirements). Then the invention is sorted as to whether it is also a subject invention. Was the conception or first actual reduction to practice specified in the the “planned and committed” activities of the federally supported project? Then it’s also a subject invention. Did the conception or first actual reduction to practice of the invention “diminish or distract” from the “planned and committed” activities? Then it’s a subject invention. Otherwise, it’s not.

It’s not a subject invention because it uses equipment purchased under a federal grant. It’s not a subject invention because it is closely related to work under a federal grant. It’s not a subject invention because work on a federal grant created the conditions for an invention to be made. The scope of a subject invention is a matter of federal law, not administrative or inventor whim.

There is no provision in Bayh-Dole for an administrator to assert an invention is a subject invention and have it be so, nor for an inventor wanting to make performance under a federal grant look more productive to assert an invention is a subject invention. The documentary record determines if an invention, once made, is also a subject invention. But it is standard practice in universities to expand the definition of subject invention to anything “arising” with the use of federal funds. That practice is, shall we put it, outside the law.

Consider the Circumstances

Now come back around to the situation involving the post doc, Stanford, Roche, and the federal government. When the post doc returns to Stanford, he has made two agreements, one with Stanford involving the terms of Grants and Contracts, and one with Cetus regarding the months of access he has had to the company’s proprietary technology. Stanford then assigns the post doc to a federally supported project. A post doc cannot simply join a project on his own and get paid for it. In approving the post doc’s work on the federally funded project, it is Stanford that brings these two agreements potentially into conflict. If Stanford believed these agreements were in conflict, it could never have allowed the post doc to join the federally funded research. He would have had to make a new agreement, disclose the prior conditional assignment to Cetus, and Stanford would have had to resolve the conflict before allowing the post doc to do any work on the federally supported project. But that didn’t happen.

One there’s an invention, and Stanford wants it, then Stanford has another problem, this time with Bayh-Dole and its own patent policy. To have a claim on the invention under its patent policy and the patent agreement Stanford had the post doc sign, the funding agreement has to require that Stanford own the invention. Thus, to defeat Roche’s claim that Roche has rights to the invention by the post doc’s assignment to Roche, Stanford had to show that the post doc had previously promised the invention to Stanford–as a requirement of the standard patent rights clause in the federal funding agreement. This Stanford could not do, though it and a host of universities and front organizations for universities all claimed that Bayh-Dole vested ownership of subject inventions with the universities that hosted the federally supported work. The Supreme Court rejected that idea and with it rejected the faux version of Bayh-Dole.

There was never, ever, a time in all of this when a prior promise to Stanford to assign was trumped by a later present assignment that assigned promised inventions to someone else. It was not a battle of magical words, even if a Supreme Court wondered about the significance of magical words in a dissenting opinion. Even if Stanford had used the tight-fisted present assignment language that they have in their policy now, it wouldn’t have mattered, since Stanford’s own patent policy (at the time, since changed to ugly) required assignment only when a research agreement required it. Stanford treated its commitment of resources to its faculty, staff, and students as a subvention–a gift in support of their work, and not a procurement exchange under which Stanford harvests all their ideas in exchange for inducing them to use university resources.

The Continuing Exploit

That hasn’t stopped university administrators, however, or even federal agencies. NASA for one requires universities to obtain assignments from their employees–an old IPA-style requirement, not authorized by Bayh-Dole. The NSF recommends that universities demand assignments (making it appear that doing so is necessary to comply with Bayh-Dole, but only in some mystical way since Bayh-Dole openly does not require assignment and the Supreme Court confirmed that. Apparently the Court can read and a host of administrators cannot or can but choose to ignore what they read.

The fixation on institutional ownership of federally supported inventions runs deep. It is the result of failed prophecy. The early advocates of Bayh-Dole, the “founding fathers” as they permit themselves to be called at Bayh-Dole Central, made the case that Bayh-Dole was some form of vesting statute. They couldn’t quite agree how or why, but they were sure it was. It might have changed the presumption of ownership, or it gave a first right of refusal, or it gave a first right to take ownership, or it prevented inventors from assigning to anyone other than the university, or it gave universities the right to take title, or it vested title in universities when the notified the federal government that they elected to retain title, or it gave universities title when they elected title regardless of the timing of any notice, or it simply vested title outright in the universities. One would think that if a law were written to provide for universities to obtain title to inventions, the folks benefiting from the law would be able to come to a single story about how the law operated. But no, they can’t. Because they are just making it up or repeating what they have been told by people they assume are authoritative.

But those folks, for all their efforts and prestige, aren’t authoritative. Perhaps they intended vesting. But that’s not what the law does. Perhaps they screwed up. Perhaps they thought they were being clever and pushing through ambiguous language that they could fix in the implementation process, where there might not be so much scrutiny. Perhaps they thought they could just repeat their claims until people “fixed” the text of the law to confirm to their claims. Perhaps they thought a Supreme Court ruling could make that “fix” happen. Vesting is a crappy strategy. It has nothing to do with inventors and everything to do with bureaucracy. It is inventor loathing.

The Logic, All the Way Down

Admiral Rickover, in opposing Bayh-Dole, nailed the logic. He argued that contractors should not be given title to inventions made with federal support. If companies (and he was most concerned about big companies) got title to subject inventions, then, “if we carry this idea to its logical extension, companies should give their employees the rights to inventions developed in the course of their employment. Few do.” If Bayh-Dole were to vest inventions with universities, then the same logic argues that universities should let employees own their inventions–just as Stanford did, back when Stanford was brilliant and enlightened. It should not matter that institutional money or facilities were provided. If the federal government is gifting money on behalf of faculty-proposed research (a subvention), how is it that university bureaucrats turn that money into their own private procurement? Why should not universities pass through that subvention? Why, if one uniform patent policy is required of federal agencies should not the same uniform patent policy–one of subvention–be adopted by universities as well?

Even Howard Bremer, another of the “founding fathers” of Bayh-Dole, worried during the writing of regulations to implement the Act that federal agencies were trying to treat their funding as procurement. Here’s an excerpt from a letter he wrote in April 1981, in which he worries about the federal agencies:



Bremer goes on:


In other words, not only were universities supposed to have “as much choice and as much time as they needed and wanted” but that the march-in rights should be gutted to make sure this was the case. In a keynote address in 1992, Bremer elaborated on what the universities did (my emphasis):

The passage of the law was not, however, the end of the battle. It took over a year to settle the controversy. During the course of the legislative effort, an almost adversarial relationship had developed as between the universities on the one hand and the Departments of Energy and Defense and NASA on the other hand. The nature of that relationship became very clear when those agencies combined to write regulations which actually controverted the law. As a consequence, much greater attention was given to the regulations by a university group which built into the regulations protection against both arbitrary exemptions to the law and to the exercise of march-in rights. 

So it was the universities that gutted the march-in provisions in the implementing regulations, making them unworkable.

But there’s no particular logic to the idea that a university bureaucracy without public oversight is so very much better for innovation than a federal bureaucracy. As Admiral Rickover argues, if the federal government is making a subvention, and it should not be construed as a procurement, then the universities should be required to pass through this subvention to their inventors and not turn federal funds into their own self-interested procurement.

Dodge City

Some time ago, when I was working through these materials for the first time, I wrote to Robert Hardy at COGR asking for a clarification regarding Bayh-Dole. Hardy never replied. I later found an email on the internet from Joseph Allen to Hardy and Howard Bremer:

three amigos 1

three amigos 2

Nothing whatsoever in Bayh-Dole or the standard patent rights clause requires a university to have a “technology transfer capability,” established or otherwise. The IPA was a negotiated master agreement based on such an established capacity; Bayh-Dole is federal law that does not consider a contractor’s technology transfer capacity. Joseph Allen was just making things up, even in a private communication.

The faux Bayh-Dole has stripped inventors of their rights and handed those inventions to universities and their invention management foundations with virtually no public oversight–reporting only on request, exempt from public disclosure, with such tangled march-in procedures that no one has ever been able to use them. Talk about turning inventor-loathing institutional self-love into national innovation policy. Who would have thought that a university bureaucrat’s thumb in every innovation opportunity would be the recipe for creating economic prosperity from federally supported research?

The funny thing, in a strange sort of way, is that the folks not to be trusted were the ones who have aimed to prevent university inventors from enjoying their Constitutional rights in order to feed their self-serving patent and licensing industry. The Supreme Court, when it had a chance to review their work in creating the faux Bayh-Dole Act, tossed them out of Dodge City. No doubt it is an easier life to ignore statutes and regulations and just make things up in a bureaucratic fantasy world, especially if one can use the color of federal law and a moralizing mantel as the premises for taking personal property. One wonders who the “bad guys” in all this actually are. The ones looking to the law, or the ones who have spent three decades suppressing it?

This entry was posted in Bayh-Dole, Stanford v Roche. Bookmark the permalink.