Five Questions That Shape Federal Research Invention Ownership Policy

In the industry research laboratories of the early 20th century, the question was, which comes first, basic research leading to new scientific knowledge, followed by development efforts to create commercial products? or development efforts to create commercial products, which, when problems arise are shifted to basic research to find answers? If we look at the industrial revolution more broadly, the answer has been that science has come along behind technology changes. Both Matt Ridley and Nassim Taleb have remarked on this sequencing. Ridley in The Rational Optimist, and Taleb most recently in an Edge piece:

Textbooks tend to show technology flowing from science, when it is more often the opposite case, dubbed the “lecturing birds on how to fly” effect. In such developments as the industrial revolution (and more generally outside linear domains such as physics), there is very little historical evidence for the contribution of fundamental research compared to that of tinkering by hobbyists.

We can posit that innovation’s arrow is quirkier than a simple one way or the other, but the basic question has significant operational consequences for a company looking for new products.

1.  Which direction does innovation’s arrow point?

Where to put the primary effort? Tinkering by hobbyists doesn’t sound like serious boardroom talk (not that serious boardroom talk has been known to produce lots of innovation), nor does it sound like federal innovation policy, which strikes a grander register that does not include words such as “tinker” and “hobbyist.” Such policy prefers “innovation” and “blue-ribbon panel.”

Similarly, university policy statements on innovation feature such abstract language as:

As a state-supported institution of higher learning, the University … has a responsibility for and an interest in the advancement of knowledge and creative work that will enhance its educational mission and promote the economic and social welfare of the public it serves, particularly the people of the State…. This responsibility and interest are advanced by engaging in research, the results of which may, on occasion, have commercial applications which are patentable or copyrightable.


Creativity is evidenced in fields as diverse as science and technology, literature and the humanities, and the fine and applied arts; the discoveries resulting from creative work in these fields may represent valuable intellectual property. The University and its members have a common interest in protecting and capitalizing on intellectual property, using such legal vehicles as patents and copyrights that make possible the ownership and control of some creative works and that provide an incentive both to create such works and to make them public.

The purpose of these statements is to make a connection between research and IP, and then to set up a claim for institutional ownership as the best possible outcome of the connection. Absent is the idea that there are faculty, students, and staff messing around with stuff–and it’s their stuff, their insights, their connections that might matter. It is as if whole swaths of how new things come to be cannot even be mentioned. There is no proper “high style” of policy for them.  This parallels Eric Auerbach’s observation in Mimesis, that there are periods in which language takes forms that do not allow in their “rhetorical pomp” the mixing in of a “glaring depiction of reality.” The “mixed” style of big ideas and gritty tinkering has no place in university policy drafting. It would seem that a policy thought, if it is worth stating these days, has to be expressed in the style of rhetorical pomp that works to obscure the mess of life, and in our case, innovation.

It’s also pretty clear that a lot of corporate-sponsored internal research never makes it into product at the company doing the sponsoring. The classic example is that of Xerox PARC, where such things as the mouse, graphical user interfaces, ethernet, and postscript were developed but not made into Xerox products. It also appears that other companies have a similar problem–they find cool stuff, but it has a tough time crossing the “brain-blood” barrier from research to product development. Thus, people talk about “spillover” effects, where ideas and technology get out another way and get used in the practice community, and developed into products by other companies, often without a formal relationship (such as a license) with the organization that supported the original work.

If the arrow of innovation runs from research to development (following the linear model of innovation), then a whole lot of corporate research is wasted. Perhaps this realization is one reason why a lot of corporations just don’t do so much research. But what if the arrow goes another way, from development to research? That is, start a development project and drive it home until it hits a snag, and then work on the snags using research as needed? That would change everything about where to put the money. This, too, is just a simplification. In some settings, it appears that combinations of some researchers, and some developers, maybe in different areas of interest, might have interactions that are more productive than the groups on their own. Thus, one company’s developers bump up against researchers at a university doing something rather different and there’s all this sparking that might not happen with the company’s own researchers.

Of course, if a company were to dip in to get help from university researchers, it would be important to have a model in which the researchers are poised to assist the company’s developers. But university models are set up on the expectation that a) university research takes place in isolation from commercial interests; b) universities should own anything that results (except in clinical trials and for artistic works); c) the purpose of that ownership is to require a licensing arrangement for resulting IP, making it appear that the company has to acquire the IP as if it had nothing to do with the statement of the problem in the first place. So, university IP models create sparks of a different variety, and companies realize that the universities are configured for a shakedown after the first date, and that the university is the wrong side of town for companies that have any self-respect.

This first question, of the direction we expect of innovation, and thus of where to put resources, is answered on the federal level by the title of Vannevar Bush’s report to the president: “Science the Endless Frontier.” For Bush, the answer is Science, and the driver for it is government investment in research conducted independent of government, at places like universities. The government picked up on a part of Bush’s vision, and in 1950 established the National Science Foundation out of his proposed National Research Foundation.

2.  Should the research be independent of government?

The federal government did not follow Bush’s recommendation that the government stay out of the research business itself. For the military, futzing into nuclear weapons and lasers and stuff was indeed a new and possibly endless frontier, but the economics of building weapons, anti-weapon weapons, weapon detectors, anti-weapon weapon detectors, and the like does not translate much to civil economic vitality, other than for those making all these weapon-things, beating their plowshares into swords, so to speak. Clearly, for really new weapons, one needed to go beyond what folks in the weapons factories were imagining. For such mind work, science clearly had to come first, just as it had in making the atomic bomb. No amount of tinkering with conventional bombs would get there. A new set of binoculars would not make the leap to radar or sonar or lidar.

In a sense, this problem of research independence is at the root of a discussion of government interest in owning the results of research, and what that ownership means for innovation, economic vitality, and all those good things of rhetorical pomp, along with tinkering, delight, and the messy stuff of living. It also sets up a basic problem for innovation. If history shows us that tinkering can produce decades of innovation, supplemented by science that cleans up and explains afterwards, but history also shows us that sometimes you just gotta have bright scientific folks proposing something very different, then it would appear you can’t have just any single approach. You have to have at least two, working at cross purposes, working from different premises and with different protocols. Universities could not then have just one patent and invention policy–they need at least two each. Same for the government. Bayh-Dole is just one bit of it, or rather, the Department of Commerce’s work is not done yet, and we need a further set of standard patent rights clauses for when tinkering development is leading, not science.

The answer to the government’s military-industrial complex was the rhetoric of “dual use”: for new technology, there could be a military use and a civilian use, and the civilian use would foster economic growth. There have been instances of dual use. Nuclear reactors, say, making weapons-grade uranium while generating electricity for cities and towns. But really, building the idea of a national economy around finding new uses for swords once you have them seems a bit off.  Extending the concept to “private use” and “government use” also doesn’t answer the bell. Take the SBIR program for instance. Small companies answer to specific agency calls for innovative research in areas of interest to the government, but the companies don’t then go off and make products from that work–instead, they come back and compete for another round of grants. It’s easier to get another SBIR award than it is to “commercialize” the results.

In addition to these basic issues of the direction of the arrow of innovation and the independence from government by those involved, we can consider three additional questions that fill out the shape of the problem.

3.  What is the purpose of acquiring title?

As Sean O’Connor has pointed out, the idea behind President Truman’s executive orders (10096) normalizing emerging federal patent policy appears to have been that the government acquired ownership of inventions made by government employees for the purpose of providing access to everyone. The idea was to remove the inventor’s personal interests from the picture. Why not then just force things to the public domain? Well, there was always the problem of international competition. Why should other countries get to ride the coat-tails of US government investment in research? Why shouldn’t they pay their own way, or contribute back something of equal value? These questions no doubt were part of the mix that went into a previous executive order (9865), focused on the government management of foreign patents. The Department of State was responsible for negotiating with foreign governments over access to inventions in which the US government holds foreign patents, and the Department of Commerce was to manage patent rights obtained from foreign governments:

Nationals of the United States shall be granted licenses on a nonexclusive royalty free basis except in such cases as the Secretary shall determine and proclaim it to be consistent with the public interest to issue such licenses on a nonexclusive royalty free basis.

Pretty interesting–the question is whether the royalty-free license should be non-exclusive or exclusive.

President Truman’s second Executive Order, focused on the inventions of government employees, lays out a model for government practices in the management of inventions:

(b) Under arrangements made and policies adopted by the Chairman, all inventions or rights therein, including licenses, owned or controlled by the United States or any Government agency shall be indexed, and copies, summaries, analyses and abstracts thereof shall be maintained and made available to all Government agencies and to public libraries, universities, trade associations, scientists and scientific groups, industrial and commercial organizations, and all other interested groups of persons.

It would be great if Bayh-Dole had included such a requirement, so everyone could see what was being invented and licensed, along with summaries and analyses. By contract, Bayh-Dole makes reported development, utilization, and licensing information protected from public disclosure. How much difference thirty years makes!

The idea behind government ownership of inventions was to use the patent system to make those inventions available broadly to the public, while working out how foreign interests might be advanced–export as well as balance of innovation trade, as it might be called. This approach makes perfectly good sense. Public domain would leave US government investments exposed to foreign free-ridership as well as blocking improvement patents domestically. By taking patent positions, the government could ensure access by the public to these inventions and negotiate with anyone holding improvement patents that would block free enterprise and competition. A similar approach is taken today in establishing standards–everyone can get fluffy about the public domain, but if patents exist, then the patent system has to be used both to support private initiative and to resist that initiative when it would dominate public access.

4.  Should the rights to inventions be owned by the government or licensed?

Truman’s executive order 10096 lays out three cases for the government’s interest in inventions made by government employees: (a) the government owns; (b) the government obtains a license; and (d) the government leaves stuff alone. The structure of the guidance is to state claims (a) and (b), then qualify and limit those claims, and then define (d) as what yet lies outside those claims.

Here is (a), the government owns:

(a) The Government shall obtain the entire right, title, and interest in and to all inventions made by any Government employee (21) during working hours, or (2) with a contribution by the Government of facilities, equipment, materials, funds, or information, or of time or services of other Government employees on official duty, or (3) which bear a direct relation to or are made in consequence of the official duties of the inventor.

This is a carefully marked text, tied to working hours, to official duties, and to the contribution of resources. If the criteria for (a) are not robust to justify assignment, or (b) the government does not have an interest in obtaining title, then the government will require a royalty-free non-exclusive license with the right to sublicense. This is much the same requirement as we find in Bayh-Dole today. Here is (b) (with my emphasis):

(b) In any case where the contribution of the Government, as measured by any one or more of the criteria set forth in paragraph (a) last above, to the invention is insufficient equitably to justify a requirement of assignment to the Government of the entire right, title and interest to such invention, or in any case where the Government has insufficient interest in an invention to obtain entire right, title and interest therein (although the Government could obtain some under paragraph (a), above), the Government agency concerned, subject to the approval of the Chairman of the Government Patents Board (provided for in paragraph 3 of this order and hereinafter referred to as the Chairman), shall leave title to such invention in the employee, subject, however, to the reservation to the Government of a non-exclusive, irrevocable, royalty-free license in the invention with power to grant licenses for all governmental purposes, such reservation, in the terms thereof, to appear, where practicable, in any patent, domestic or foreign, which may issue on such invention.

There follows a division that limits government interest in title to inventions made by inventors assigned to invent, active in research, supervising such research, or serving as liaison with such researchers. For any other inventors, the license in (b) applies. One might see in this how different the government’s approach was from the present comprehensive policies put forward by universities, which show none of this refinement–all inventions are claimed, there are no equitable release points in practice, and there certainly is no expectation of acquiring patents so that inventions can be made broadly available on a royalty-free basis.

Section (d) finishes off the logic:

(d) In any case wherein the Government neither (1) pursuant to the provisions of paragraph (a) above, obtains entire right, title and interest in and to an invention nor (2) pursuant to the provision of paragraph (b) above, reserves a non-exclusive, irrevocable, royalty-free license in the invention with power to grant licenses for all governmental purposes, the Government shall leave the entire right, title, and interest in and to the invention in the Government employee, subject to law.

If not (a) or (b), then title and rights stay with the inventor. Pretty basic stuff, really. The problem comes as the government expands its support for private research, such as at universities. Should the government own such inventions, or just have the right to practice them? That’s not so clear as the case with government employees, where one is directing the work.  The Stevenson-Wydler Act works through the employee situation, pushing the case for government ownership in terms of “commercialization” (my emphasis):


(a) IN GENERAL.–If a Federal agency which has the right of ownership to an invention under this Act does not intend to file for a patent application or otherwise to promote commercialization of such invention, the agency shall allow the inventor, if the inventor is a Government employee or former employee who made the invention during the course of employment with the Government, to retain title to the invention (subject to reservation by the Government of a nonexclusive, nontransferable, irrevocable, paid-up license to practice the invention or have the invention practiced throughout the world by or on behalf of the Government). In addition, the agency may condition the inventor’s right to title on the timely filing of a patent application in cases when the Government determines that it has or may have a need to practice the invention.

One can see in the logic much of the original Truman executive order, with the odd final addition that somehow the government might require an inventor to file a patent application so that the government can practice the invention, as if one could not do that without a patent. Very strange.

In each of these developments, the purpose of taking ownership of a federally supported invention cannot be simply to ensure the government has a right to practice the invention–a license would do equally well. Thus, it is not simply ownership that is in play, but the purpose of ownership above and beyond freedom to practice. For the government, that purpose appears to be to reserve rights that otherwise would create a monopoly just when the public would be best served by competition, allowing for exceptions when a monopoly is the best way to get something done without interference.

To get at this problem, we turn to final question:

5.  When title to an invention is acquired by the government, is it to promote use or to promote commercialization?

(Or, more cynically, to promote private enrichment of whomever happens to come away with ownership of patents? But let’s say for the sake of argument that this is *not* what the ownership issue is all about–though in some ways, this is *exactly* what it is about!)

If one looks at the Truman approach, one would have to answer that use takes precedence. The presumption is that the government takes title to ensure access rather than to promote private investment, and if private investment is desired, then the government can move its management to paragraphs (b) or (d) and not worry it. If there’s commercialization to be had, it can be had without a monopoly position, for the government’s purposes. In this, Sylvia Kraemer’s discussion of the problems in the US airplane industry of the early 20th century (in the context of NASA patenting) is relevant.  Companies all had staked out their bit of turf and while they fussed over each other’s infringements, other countries were developing new airplanes. In a sense, it is absurd to think that the airplane would only be developed if there was a monopoly interest involved.

Or, more to the point, that once an airplane was shown as possible, it is absurd to think that any one monopolist will be able to invent and develop all the variations and improvements that are possible, nor should the patent system be seen as a basis for reserving the economic benefits of all such variations and improvements to some one monopolist. The variations are simply too great; the lines of improvement too robust.  If one aims to control such stuff, rather than merely to have access to it at need, one all but kills the innovation. That’s what happened with nanotechnology, and what is happening now with 3d printing. It didn’t happen with the internet, but that was due to diligence on the part of those involved. It may be that at some point the economic benefits fall to a few firms–even the internet appears to be going that way–but the aim of early innovation dynamics is to keep things open for long enough for the variations to be explored, the improvements made apace, before the forces of control figure out how to come to dominate the creative and practice folk.

Kraemer offers some context for Truman’s need for an executive order on patent ownership:

The first year in which patents were issued to the U.S. Government was 1931, when it obtained 28 patents. Until 1944, at the height of WW II, the largest number of patents acquired by the government in any one year never exceeded 62; the average number of patents acquired annually during the 14 years between 1923 and 1943 was only 38. All this changed during World War II: between 1944, when the government acquired 106 patents, and 1950 when it acquired 622, the Federal government became the owner of 1,542 patents, bringing the total number of patents in its portfolio in 1950 to 2,489. During the next decade the government acquired well over 9,000 patents.

By the late 1970s, the number would be over 25,000. Clearly, the shift to government held patents was well underway by 1950, but truly blossomed, if that is the right word, along with government support for research, in the 1950s.

President Eisenhower in his farewell address worried about the military-industrial complex, the distortion of university research by government funding, and the isolation of the independent inventor. President Kennedy established a foundation for government interest in inventions that expanded ownership claims, differentiated between inventions that could be practiced directly and those that required a manufacturing step (for which discussion, see Sean O’Connor’s paper) and permitted exclusive licensing in some cases.  Later, President Johnson presided over a commission on the patent system that President Nixon then relied upon to endorse Kennedy’s approach while calling for greater flexibility. These are the run-ups to Bayh-Dole.

O’Connor argues that Bayh-Dole basically took the Kennedy platform and made it less flexible and responsive to government and public purposes, in favor of non-federal ownership when non-federal owners are pre-approved for that ownership. O’Connor, in assessing whether Bayh-Dole makes for a good export for adoption in India, suggests:

The problem instead seems to have been that interpretations of the Kennedy Patent Policy erred on the side of favoring government title. Thus a default of government ownership seems to have arisen, which required a petition process by extramural researchers who wanted to retain the title. Any reversion to something similar to the Kennedy Patent Policy, therefore, should likely include a default rule that leaves title with the researcher.

For Bayh-Dole, the idea is that the researcher who invents will partner with an organization that manages inventions. The overlay by university administrators is that this partner must be the university (or the university’s research foundation). So much for flexibility, even if the government offers it. What is strange about all this is that there is nothing inherently a problem with either the inventors owning their inventions or choosing partners to help them with their inventions. The researchers and their partners could, for instance, aim to place their inventions in the public domain (or offer royalty-free non-exclusive public licenses), just as the expectation was in the Truman years. For economic development purposes, such an approach could be just the thing, as could be done with make/use commons approaches. It is just that this sort of expectation is never considered.  Researchers and universities alike are expected to file patents only to pursue monopolies and sudden wealth. This is what is meant, apparently, by “commercialization.” There is never any discussion of quality control, standards development, choice of early partners, access through cross-licensing to other technology, and the like. It is as if inventions must be patented, and patents must create monopolies, and scholarly exchange or public purposes be damned, but for the “public purpose” of “making money from licensing”.

Consider then the discussion of Bayh-Dole that poses the issue in terms of commercial incentives. Here is a discussion by authors who were close to the implementation of Bayh-Dole (one was Chief Counsel for Technology, US Department of Commerce, 1989-91), writing in 1992:

There remained no incentive for private sector institutions and universities to pursue research with important private sector applications.  As in the past, commercialization of federally-owned technology was not a priority.

and furthermore:

There was no incentive for government inventors or institutions to create commercially viable technology because there was no legal basis to gain a piece of the resulting monetary rewards. As a result, commercially viable technology was not being created, and the wealth of federal inventions that were available for licensing were not being transferred for use in the private

There is no support given for the first assertion, about incentives, perhaps because there is no evidence for the lack of incentives, or perhaps because the assertion is a lawyer’s rhetoric and there’s no need for observation. There is a footnote with regard to the second statement, but it is to the legislative debate, as if whatever someone asserts in a legislative debate is fact.

Work it through. Why is a “legal” basis (meaning a patent licensed under contract for royalties) needed to “gain a piece of the resulting monetary rewards”–especially for universities? Universities can gain “monetary rewards” by providing training in new technology, by offering research services, and by receiving donations from happy alumni and boosters, among other things. That somehow by changing ownership disposition universities would then create “commercially viable technology” makes no sense whatsoever in the context of federally sponsored research. For the mission-directed federal agencies, the research is to serve their missions (presumably by drawing on research when needed, not just “research for research’s sake” or “research as an industry”), and commercial “viability” is not necessarily one of these, especially for basic science (origin of the universe, dark matter, or earth-impacting asteroids, say) or health (prevention, say again–why does a discovery about prevention have to be “commercially viable” to be any good?).

The argument here is a rhetorical plumping of the case for bureaucrats to get their thumbs on creative work. The argument runs that university bureaucrats have better thumbs than federal bureaucrats. After 30 years, there is no accounting of Bayh-Dole subject inventions by university administrators. There is no reporting of a list of the inventions claimed, the patents filed and issued, the licenses granted, the dates of first commercial sale or use, the gross income, the jobs created. There is no accounting for the cost of operations, the amounts recovered from licensing, or the net return to universities after all costs are considered. There is nothing to show productivity or innovation.  It is a gaping black hole of non-data. The AUTM licensing survey does not report any of the essential metrics required by Bayh-Dole, nor does it differentiate federally funded inventions from others.   Again, there is no information to show that Bayh-Dole has been “successful” or that American “productivity” or “innovation” has benefited by the law. Of course, we have yet to see Bayh-Dole as it stands be implemented by American universities–they are fixated on the faux Bayh-Dole, under which they nominate themselves to own everything and sell off what they can, which isn’t, actually, very much–perhaps way less than 5% of what they claim ownership of.

Think of it: the whole point, in 1950, of federally owned patents was to make certain technology available for public practice. There was no incentive to “commercialize” because the focus of federal patents was “use” not “commercialization.” If the public used inventions, then they would figure out what the economics of that use might be, and could make products as they chose. That is, commercialization comes about not merely because there are patents in monopolists’ hands, but also because patents have been taken from such hands, as was required in the nascent American airplane industry.

The advocates for Bayh-Dole are fond of citing that only 4% to 5% of federally owned patents were “commercialized.” These same advocates do not point out that historically, that’s about the same percentage of patents that are commercialized by the general population. Given that the government’s purpose in obtaining patents was directed at a lot more than commercialization via monopoly positions, 5% starts to look much better–and it also raises the question of how American industry was doing where the government held patent rights on behalf of broad practice, standards formation, and competition based on build quality, timeliness, availability, and post-sale support. You know, stuff that doesn’t require a monopoly position on rights. The higher “commercialization” rate cited by Bayh-Dole advocates for university and research foundation licensing arises from a pattern of selectively choosing inventions suited to such an approach. Even here, some of the classic inventions that found the story of university technology licensing, such as the Cohen-Boyer gene-splicing patents, were non-exclusively licensed to industry. (And of course, the Cohen-Boyer patent application was filed in 1974, well before Bayh-Dole, and Stanford’s licensing followed to a large degree the federal government’s own preference for non-exclusive licenses available to all.)

It may well be that as university administrators became thirsty to expand their “success” in licensing, they petitioned more agencies to release inventions to them. But in doing so, it appears the administrators also abandoned selectivity and forgot about non-exclusive licensing on a royalty-free basis. By taking in more and more inventions for management, university programs have overstuffed themselves. The current rage for present assignments drafted on comprehensive, compulsory claims to all inventions faculty make, as folks like Parker Tresemer take as necessary, and administrators at the University of California and University of Washington take as their natural rights simply creates a stream of invention disclosures that no central office can possibly evaluate, manage, and act upon in the public interest. At best one can create a portfolio model in which a deal a decade creates the appearance of financial success and all the other inventions can go rot, for all the administrators care. The result is an accumulation of patents in the hands of university administrators that is worse than the federal government holding title, worse that allowing researchers to hold title individually, and worse than the real Bayh-Dole Act anticipated with its pre-approved agent approach, now long buried by the faux Bayh-Dolers.

Walter D. Valdivia makes an interesting case in a recent article for how the values and terminology regarding Bayh-Dole has drifted over the years. He argues that there has been a “public value failure” in the implementation of Bayh-Dole. I agree. He attributes the drift to “neoliberal ideas” which may well be, though I don’t see much added in labeling the debate this way. But I would add that the issue should not be whether universities should own (in order to sell of patent rights to speculative monopolists) or inventors should own personally (presumably to do the same thing, just without the assistance of bureaucrat employers), or government should own (state control of scholarship and all that) but rather how patents support access to inventions, practice of inventions, and development of those inventions–variations, improvements, products, all of that.

For different owners of patents, different values and purposes. Some to feed the commons, some to bring competitors together to form standards, some to motivate investment and product development, some to protect lines of improvement and variation. One does not get all this from either institutional ownership or private inventor ownership or government ownership. Ownership taken in the abstract is not the resolving issue–and often the mere fact of demanding a claim to ownership must be made is part of the problem, especially as scholarship is concerned. A policy would be preferable that pointed out the variability of ownership premises rather than trying to establish one outcome over others (say, under the idea that “title certainty” is more important than “opportunity”).

Where Bayh-Dole states as an objective (if not challenge) the use of the patent system to promote the “utilization” of federally supported inventions, university administrators (and others) have substituted “commercialization,” which is throughout the actual text of Bayh-Dole a subsidiary interest. The primary interest is “practical application”–which gets a definition–not commercialization, which at any rate, in the statement of objectives for Bayh-Dole, is assigned to industry, not universities:

to promote the commercialization and public availability of inventions made in the United States by United States industry and labor;

That is, even in the statement of objectives, commercialization happens *after* industry has got access to federally supported inventions, not as a *condition* of that access.


This entry was posted in Bayh-Dole, History, Metrics, Policy, Technology Transfer and tagged , , , , , . Bookmark the permalink.