We are working through an AUTM effort to get readers–and especially people wishing to influence federal policy makers on matters of inventions and patents made in faculty-chosen and led research–to self-delude themselves about Bayh-Dole. AUTM cites a GAO report from 1998 to back up its “talking points” about Bayh-Dole, so we have done a dive into that report to see what how AUTM’s case holds up. As you might expect by now, AUTM uses its expert position to cheat at Bayh-Dole, figuring folks won’t notice.
Here’s another finding from the GAO report:
Despite the perception that Bayh-Dole is working well, none of the federal agencies or universities we contacted evaluated the effects of Bayh-Dole.
Isn’t that an amazing finding? No-one had evaluated the effects of Bayh-Dole. But they said it was a wild success. And AUTM cites this study to support their claim that Bayh-Dole is a wild success. This is the best they can do!
The only available nationwide data on the effects are those published by AUTM.
So AUTM cites a study that finds there is no support for the success claim–just an appearance of success, and then cites AUTM, with this to say:
While limited in application because they apply to all inventions
regardless of funding source and are based on a survey of participants, AUTM’s statistics nevertheless indicate that universities are increasing their licensing activities and that revenues from licenses are growing.
That is, the AUTM statistics conflate all inventions with subject inventions. In the Bayh-Dole era, universities and related foundations and institutes have acquired around 120,000 patents (many patents in related families, many more inventions claimed and controlled but not patented), and only about 40% or 50,000 of those patents recite federal funding. It is irrelevant to an evaluation of Bayh-Dole whether universities are increasing licensing activities or making more money from licensing. Bayh-Dole does not have a stated objective to increase the staffing or activities of university licensing offices, nor that its goal is for universities to make more and more money from licensing. Growth of university licensing offices can point to more administrative effort–and that would be counter to an express policy and objective of Bayh-Dole. Growth of university licensing revenues could mean a suppression of free competition and enterprise–again, running directly against an express policy and objective of Bayh-Dole.
Bayh-Dole’s stated performance standards are “use of the patent system to promote the utilization of inventions arising in federally supported research or development” and “utilization” is re-stated in the form of “practical application”–utilization with “benefits available to the public on reasonable terms.” Bayh-Dole reporting considers “utilization.” The implementing regulations expand this reporting to include date of first commercial sale or use and gross royalties received. Bayh-Dole march-in can be triggered for nonuse or use that fails to meet needs. An evaluation of Bayh-Dole starts with an examination of subject inventions, not all university-owned inventions. That evaluation looks for practical application–(i) use; (ii) benefits of use; (iii) available to the public; (iv) on reasonable terms. Within this evaluation, what is the date of first commercial sale or first use? What are the gross royalties received for each subject invention? Not in aggregate, not conflated with all non-federally supported inventions. Not conflated with all non-inventions (such as software, biomaterials, defect-ridden mice, data sets, instructional videos). But no one bothers to do this Bayh-Dole evaluation. Wouldn’t they want to, to have firm proof of Bayh-Dole’s amazing success? Funny, no rush to evaluate would appear to be strong evidence that there’s nothing good to find there. Funny, substituting proxy data that’s irrelevant and deeply ambiguous would appear to be strong evidence that folks are covering up something they don’t want anyone to see.
AUTM also believes that the activities involving inventions have added to the economy in general.
Here we have an evaluation of Bayh-Dole by reciting the “belief” of a lobbying organization fronting for university licensing officers, the memberships for whom are typically paid by university administrations. What does AUTM’s “beliefs” have to do with anything? “Credo, quia absurdum est” might handle the mysteries of religious experience, but is that what we are reduced to with Bayh-Dole?
The GAO mentions AUTM’s belief about the economic impact of licensing activities:
The AUTM survey report concluded that the economic impact of licensing activities undertaken by academic institutions, nonprofit organizations, and patent management firms is extensive.
Again, these are not Bayh-Dole licensing activities–but all licensing activities, including licensing of non-inventions, licensing of non-federally supported inventions, licensing of inventions that did not result in utilization, licensing of inventions that resulted in benefits not available to the public or not available to the public on reasonable terms. So AUTM reports “extensive” economic impacts. But here’s the kicker:
For the reasons discussed earlier, we did not verify the accuracy of the model or the projections made by AUTM for economic impact,
Nor has anyone else, ever. It’s just bluster. A close analysis of the little that has been published about the models used by AUTM ends in laughing fits.
nor did we attempt to determine what portion of this impact was attributable to Bayh-Dole.
What a surprise–because it would be next to impossible. And besides, the impact of Bayh-Dole practice is not merely a calculation based on money reported by universities. It cannot possibly be that–any evaluation has to consider the economic impact of putting tens of thousands of the most significant findings of federally sponsored research in the public interest behind patent paywalls, never licensed or when rarely licensed are never utilized, and even if rarely utilized are even more rarely result in benefits to the public on reasonable terms. We might reasonably hypothesize that every patented and unlicensed subject invention has an adverse economic impact–people have to do without, design around, risk infringement, take work offshore where it won’t infringe, create products lacking important features, delay development of standards and interoperable systems and products, and generally avoid the results of federally supported work. Show this hypothesis is not true. That’s where the burden lies. Only drunken economists construct their models on positive data of activity.
GAO, on the AUTM data:
The underlying data are based on unverified information
reported on a survey.
Quantitative bullshit made to sound good so people ready to delude themselves may do so freely and openly.
Furthermore, the projections are based on one approach for measuring economic impact; other approaches might yield
substantially different results.
A generous description. “Measuring” isn’t apt, though. The approach merely constructs a big number from a pattern of estimates and assumptions. “This big number reported as licensing income by all universities must mean an even huger amount of sales income and if all that sales income is used to pay for tech workers then this many jobs are created, and those workers spend what they have in the economy with this sort of standard estimate of that impact.” Something like that. Beyond silly. Most instances of licensing income don’t involve sales–upfront fees, milestone fees, annual fees, reimbursement of legal fees, realized equity, payments in tied to the value of equity at a liquidity event, purchase of future value of a royalty stream, settlements. If you wanted to clear away the rubbish, Bayh-Dole should stipulate that no nonprofit, for any exclusive license, could charge a dime for anything other than a royalty on income earned from sales. None of that other stuff–nothing upfront, no equity interest, no reimbursement of legal fees. Then we’d see university officials keenly interested in practical application, not in the growth of licensing activities. Notice how nearly all university diagrams depicting the “technology transfer process” end at “licensing” or “commercialization” not at “public benefit.”
Also, the projections cover the impact of all licensing activities, not just those attributable to federal funding and not just those attributable to universities.
Meaning–there’s nothing here useful to evaluate Bayh-Dole. Or, rather there is: universities are so uninterested in Bayh-Dole that they don’t bother to report Bayh-Dole specific activity or outcomes.
The GAO report goes on to give a list of inventions that the ten leading research universities reported as
successful inventions they believed showed that university research subject to the Bayh-Dole Act was having a positive impact.
They cite six inventions, only one of which–organ storage solutions–appears to have been a subject invention under Bayh-Dole. The GAO report is from 1998 and the top research universities report on research conducted no later than 1985. Bayh-Dole came into effect in July 1981. In the 17 years to the GAO report, ten leading research universities can report only one invention that illustrates public benefit under Bayh-Dole. There rest are pre-Bayh-Dole. Where’s the research from the early 1990s resulting in inventions with public benefit? Wouldn’t we expect a broader sampling of benefits from Bayh-Dole?
The same trick happens in almost any list of inventions published by advocates of Bayh-Dole. They slip from Bayh-Dole to general university licensing–so they mix Bayh-Dole inventions with others, and they slip from an evaluation of Bayh-Dole to the benefits of what universities have licensed, and they slip from actual history to a fake history in which university licensing arose from Bayh-Dole and that before Bayh-Dole universities were unable to license federally supported inventions–all deeply untrue and misleading.