Let’s turn to Research Corporation now. For decades, Research Corporation was the dominant invention management broker for universities. Even where universities created research foundations to manage inventions, those foundations often contracted with Research Corporation to do the actual patent work. By 1976, a few universities were developing their own internal licensing operations–notably MIT, University of California (1960), and Stanford (1972). The Stanford approach, to be called the Reimers model, was just beginning to catch hold. Things heat up after Stanford starts its non-exclusive licensing program for the Cohen-Boyer gene splicing patents, just before Bayh-Dole becomes effective. Here’s a graph of Research Corporation patent activity. Not all Research Corporation patents are necessarily from universities, but many are.
During our study period, Research Corporation doubled its patenting activity, while universities and other nonprofits increased their patenting activity 15x. There’s nothing at all distinctive, however, about the onset of the Bayh-Dole Act in 1981. There’s an uptick in 1996 that’s significant, but that can hardly be the result of Bayh-Dole.
What about Research Corporation’s work with subject inventions?
The share of subject inventions has increased by 3x overall, and for a bit was as high as 4x. The obvious thing is that the activity is volatile–up, then down. And down in a big way from 2000 to 2005. Why? I don’t know, but it is not a change brought on by Bayh-Dole–unless, of course, Research Corporation realized that subject inventions, having been papered over by Bayh-Dole red tape and institutional ownership, were less marketable than other inventions.
During the same period, university and nonprofit share of subject inventions increased by 5x. So Research Corporation’s activity has increased at a slower pace than university patenting for patenting overall and patenting subject inventions. Perhaps that indicates selectivity–or that universities have become less selective in what they choose to patent. They patent most anything they can, which would mean that they have money to spend without much accountability. Such a thing happens when someone is gambling (or, as university administrators call it, “investing” for “public benefit”). One can be fiscally undisciplined, so it goes, if it is for a “good cause.”
Now here’s the interesting thing. Look at Research Corporation’s share of university patenting activity:
Research Corporation’s patent work falls from about 40% of university activity to under 5%. The share was falling before Bayh-Dole came into effect in 1981, and by about 1997 had leveled off at about 5% of total university/nonprofit patenting activity. Bayh-Dole did not start the decline. Perhaps university interest in managing inventions locally had more to do with it. Bremer and Latker made the argument in debating Bayh-Dole that middlemen closer to campus were better positioned to license inventions than were officials working for distant (in the case of, say, Wisconsin or Purdue) federal agencies. Local middlemen would work directly with inventors and have incentives tied to their relationships with the universities that hosted the research and with regional industry. This same argument easily can be shifted to apply to Research Corporation as well–though from what I have learned, Research Corporation did not anticipate this argument coming and focused rather on the idea that private patent brokers could do a better job licensing inventions than could federal agencies. Nothing here, however, indicates that Bayh-Dole itself altered the fortunes of Research Corporation, though the argument regarding local middlemen patent brokers very likely did.
It’s rather interesting, that a debate about the role of research in producing innovation in the public interest should end up being over which sort of middlemen are more effective at securing patent licenses–government middlemen, private but distant middlemen, or local middlemen? We are drawn to make comparisons among these groups and might fail to consider that none of these groups may be all that effective in securing patent licenses. Worse, the real problem is that licensing of patents by nonprofits is among the least productive approaches to new technology uptake. That was a Harbridge House report finding, and it makes some sense. The middlemen debate is about who can do a better job with the worst possible approach to new technology deployment. Bayh-Dole resolves that debate in favor of local middlemen as a matter of policy. The evidence, however, is that local middleman are doing a worse job of licensing and technology deployment (pick whichever) than were either the national middlemen (led by Research Corporation) or the federal middlemen (the federal agencies, allowing contractors to own when the contractors had commercial positions and capability; releasing most else into the public domain, so contractors had access along with everyone else).
However, it may well be that these divisions of federal and non-federal, and distant and local, have nothing much at all to do with research-based contributions to innovation. Perhaps middlemen selected by inventors are more productive than middleman that take what they want without limit or accountability. Perhaps middlemen located where the most likely licensees are located are more productive than middlemen sidled up close to inventor campuses. Perhaps middlemen willing to travel are more productive than middlemen out of circulation in Pullman or Ann Arbor or Merced. Perhaps middlemen with a great deal of industry experience in the area of the invention are more productive than middlemen recruited as post docs. Perhaps inventors working directly with companies and not involving bureaucratic middlemen at all are more productive–relying on legal counsel and advisors who work directly for the inventors. And perhaps academic inventors publishing the majority of their inventions for open use, or use through a commons or standard are more productive than those that deal with middlemen brokers aiming to shop each invention as a private monopoly to a single company or investor.
Think about it this way. Bayh-Dole spurred on a debate over which middlemen would be the most worthy to manage subject inventions, on the premise that middlemen are as essential as clowns. The premise itself is not established, but let’s run with it for a moment. There were two primary incumbents. One was the federal government, which managed way more patents than universities did–the government had about 1800 patents issue in 1976, while universities had about 400. The government had an interest in another 900, and less than 10% of those were from universities another 17% or so were from Research Corporation. Most of the federal government’s patents were released for general use–non-exclusively licensed. The second incumbent was Research Corporation, which managed 40% of nonprofit patent activity, but only a small share of its work was with federal inventions.
When Bayh-Dole opened up federally funded inventions for non-federal management (in its way restoring the IPA program, changing it in significant ways, and making it government-wide), the real question was whether universities would push more federally funded inventions to Research Corporation or buy into the idea that local middlemen were better and attempt to emulate the Wisconsin Research Foundation (and do their own patenting through an affiliated foundation) or emulate MIT, Stanford, and the University of California, and create local “central” offices located inside university administrations. Because Bayh-Dole was made to appear to push the idea that ownership of inventions vested in the institution that hosted the research, it appeared that rights ran through local university administrators. (The research host institution, the first contractor identified in Bayh-Dole’s standard patent rights clause, could “elect to retain title” to inventions made with federal support, and this expression was shortened to “elect title” which was read to mean to “take title” or “to already have title.” None of this was true but that’s how things were depicted, and university administrators slurped up the idea rather than challenged it or read the law or implementing standard patent rights clause.)
Research Corporation had, well before Bayh-Dole, advocated for the development of “technology transfer” offices on university campuses. Personnel from these offices would “walk the halls” and help inventors document their inventions and then communicate these invention “disclosures” to Research Corporation for review. This was Research Corporation’s pro-active way of bridging the distance and organizational separation between inventor, university, and (often) remote patent broker. But perhaps you see the problem that develops. The university administration comes to have an enhanced interest in the inventions that it sees. Rather than having inventors send their invention disclosures directly to Research Corporation, or even directly to an affiliated research foundation, the disclosures are redirected to the university administration, which among other things hopes to make money when inventions are patented and licensed. If Research Corporation continues to operate with a highly selective approach–consistent with being an agent working on behalf of inventors, with universities as a designated beneficiary–then university administrators will see a significant number of inventions rejected for management by Research Corporation. It is these inventions that aren’t selected that become the foundation for the argument that a university ought to “be like Stanford” or “Cal” or “MIT” and have its own licensing program.
Thus, technology transfer offices became offices of technology licensing. The decision, it appears, had little to do with innovation in the public interest and a great deal to do with whether remote middlemen had their own university in mind in making decisions about what inventions to take under management. In effect, university administrators argued for less selectivity in the inventions to take under management. I watched this practice develop from the time I became involved in university patent licensing in 1989. The prevailing practice in the University of Washington Office of Technology Transfer was to identify inventions and pass these inventions to the Washington Research Foundation for possible management. WRF had been set up on the model of Research Corporation, and it operated with similar selectivity.
If a “technology manager” at the UW OTT wanted to manage an invention that the WRF had declined to handle, then the technology manager had to show the director that there was at least one (and usually there was only one) company willing to pay the patenting costs before any patent application could be filed. Only rarely–a few times a year–might a patent be filed “speculatively” with the idea that sometime later a company would be found to license the invention. Thus, invention reports were sent out first in a “non-confidential” form–enough information to identify the general area in which the invention operated and its basic results without revealing how it operates. If a company expressed interest, then it would sign a non-disclosure agreement and be given the details of the invention. If the company was willing to pay the cost of the patent application, then the university would file the application.
You might then also see how the logic of income might work on the mind of a university administrator. If technology transfer offices could license the inventions rejected by Research Corporation or by even a local patent licensing broker working on the selective agent model, then those same offices could do even better business if they also handled the “best” inventions. Given that Research Corporation and other such patent brokers often took 40% of royalties plus their direct costs, university administrators figured they could do just as well taking only 15% or 20% of royalties plus costs. When a big hit multi-million dollar deal came about (anywhere in the country, rarely, but even once every five years was enough), the difference between 40% and 15% became painful to university administrators.
If local patent brokers were more productive than remote ones, then patent brokers working for a university administration would surely be more productive than ones working outside the university. University patent brokers would have the immediate interests of the university in mind, not the overall success, if not the mere survival, of the foundation or company offering patent brokering services. That, of course, was the theory, but it wasn’t what happened in practice, since the “interests of the university” is a difficult concept to pin down and depends on whatever it is that university administrators assert on behalf of the university, which does not have any interests but for someone asserting them. What’s good for a university may not be good for university patent licensing, but if university patent licensors are the ones writing patent policy, then things flip and the claim is that what’s good for university patent licensors will be, eventually, also good for the university.
Not only, then, did university administrators challenge the remote management of inventions but also they challenged the agent model with its selectivity. Taking inventions “in house” meant that the patent broker worked “for the house” and not for a client such as an inventor or even for a particular university’s invention. University licensing offices could not afford to use an agent model. State universities rejected the agent model citing state “ethics” laws that prevent a state employee from having an economic interest in a contract involving the state. Generally, an agent model requires selectivity, and that in turn can have political ramifications when the inventor is a center director or a department chair or school dean.
For a remote patent broker, this is not such a big deal, because the question is whether the invention is a match for the broker’s licensing methods and experience. For a local patent broker, however, this can be a huge deal, because powerful faculty and administrators can make or break a licensing operation run as a university administrative program. For reasons such as this, university administrators adopted a portfolio approach rather than an agent approach. File patent applications on many things, and see which ones get licensed. When the provisional application was introduced in 1995, university administrators found even greater motivation to file first and ask questions later.
Only a few inventions need be licensed to claim success. One big hit deal every two decades is sufficient for financial success. The big hit pays for the patent work for all the others (even if Bayh-Dole’s standard patent rights clause restricts cost recoveries for subject inventions to subject inventions and not all inventions under management by a university). In a portfolio approach, universities claim inventions broadly, file applications for many (often about half), license less than 20%, and see only 1 in 200 become a commercial product. This looks a great deal like a “sales funnel”–except it deals with products not potential qualified buyers. But the rationalization sticks–from a university portfolio approach, what matters is the financial and publicity benefit of a few deals, not that there’s meaningful work done for each invention taken under management.
Here then is a possible account for the decline in Research Corporation’s share of university patenting and in the increase, generally, of university patenting. The arguments about middlemen that were presented to pass Bayh-Dole ended up applying as well to Research Corporation and eventually as well to licensing offices representing university systems, such as those at the University of California and the State University of New York.