Documented and undocumented technology transfer programs

Recently, the University of California, in an internal report on its technology transfer program, indicated that its commercialization rate was 0.5%–1 invention in 200 got to the point of a commercial product. There was no indication whether those commercial products were “successful”–that is, became broadly available to the public on reasonable terms. That 0.5% rate is an order of magnitude lower than what was reported for both federal government licensing and for licensing under the IPA program before Bayh-Dole.

In 1969, Willard Marcy of Research Corporation participated in a workshop looking at federal patent policy. Here is his take on Research Corporation’s experience, in response to the question why his organization does not pay a royalty to the government on its licensing of federally supported inventions:


Research Corporation’s selectivity was 10%. It screened and deferred to manage 90% of the inventions that it received. Compare that with present university practice, which is to claim ownership of everything, often whether inventive or not, patentable or not, which greatly increases the volume of assets to be screened for whatever it is that justifies continued university ownership. Notice that university screening after an ownership claim reverses the situation–the university has to find some reason to give up ownership, not to figure out why it would ever want to own a given asset. As a result, universities tend to give up virtually nothing that is reported to them–they take it all in. The effect is to forestall the markets (as we might call them) for work that would otherwise move through channels other than patenting (or pseudo-ownership, where no patent rights are available) and licensing–publication, teaching, consulting, laboratory exchanges, and the like.

Universities do not generally report their selectivity rates for patenting. But from informal information, I expect the selectivity rates to be closer to 30%–about a third of inventions reported receive a patent application. With the creation of “provisional” patent applications, the rate at which these initial patent applications are filed may be closer to 70%. A patent administrator at one major research university told me that they filed a provisional application on almost everything reported to them.

Research Corporation in 1969 reports that it licenses about 10% of what it selects for management. That’s a licensing rate of 1%. Again, universities now don’t report their licensing rate. They report the number of inventions received in a given year and the number of licenses granted–these are activity metrics–but they don’t connect up inventions (reported in past years) and licenses (for each invention). In short, the universities don’t report performance metrics.

Now the kicker. Research Corporation reports that only 10% of licenses generates at least $10,000–about $65,000 in 2016 dollars. Research Corporation doesn’t say whether it licenses mostly exclusively or not, so we can’t know if each modestly successful license is to a different underlying patent–but let’s say so. At this point we are at a 0.1% commercialization rate. The UC figure starts to look, um, good. Stanford reported in 2009 that in 36 years and 6,400 inventions, it had licensed 287 that had generated $100,000 or more royalty income–about 0.45% commercialization rate, assuming that income reflects sales and not reimbursement for a foreign patent portfolio (translation costs for a foreign patent application can run $10,000 or more).

What can we make of all this? At licensing programs regarded as best-in-class, the commercialization rates are about 0.5% or less–regardless of the degree of selectivity. Put another way, 99.5% of inventions never see a commercial version.

There are, then, two technology transfer programs at work, one formal and stated and the other undocumented and implicit. In the formal program, inventions are evaluated for commercial potential, patents are obtained, and the technology is marketed to industry or investors, licensed, and products show up to the amazement and benefit of the public (products!), corporations (profits!), and universities (royalties!). This formal program is what gets reported as “success” stories. It represents 0.5% or less of the total invention activity.

The undocumented program, however, is where there is cause for concern. The repeated refrain of Bayh-Dole advocates is that before Bayh-Dole, inventions just “sat on the shelf” and gathered dust, but after Bayh-Dole, things are a lot more rosy. Then there is the usual quote of statistics on the growth of patenting and a big (appearing) number reflecting total impact of technology transfer on the economy (typically without any information about how that big number was obtained). But if these figures from UC, RC, and Stanford are a good guide, the undocumented program of technology management leaves nearly all of research result assets “on the shelf.” And it does so in a worse way.

Under the federal program of taking title to patentable inventions, inventions were made broadly available. The federal government considered control only where there was statutory control–through the patent system. Everything else remained as it was, open for private use in whatever way might come about. For the federally held patents, companies did not even have to worry about obtaining a formal license. The federal government did not sue for infringement, did not have a profit motive to push it to attack the very industries it was aiming to support. If things “sat on the shelf” it was because there was not much interest in pursuing them. Only rarely was the problem that a thing lacked a monopoly with a private profit motive for development. After all, if development would be expensive and difficult, there likely would be more inventions to be made and more patent positions to be taken before a product could be produced–and any of these later patents might anchor new monopolies that “protected” the development pathway. And if the development would not be expensive or difficult, then most anyone could do the work to use or sell product based on the invention, and the issue becomes one of whether there would be a “market” for such work, or whether companies and professionals would simply practice the invention directly, working up their own implementations.

Now, however, the universities claim ownership of almost everything, not the federal government claiming ownership only of patentable inventions made with federal support in areas of public welfare in which the contractor does not have a capability to develop and an established commercial position. The universities claim ownership across the board, often now through a “present assignment” of all future work, without regard to formal intellectual property laws. Where the federal government at least had standing policy to make inventions available “through dedication or licensing” to the public, universities have no such policies. It’s license exclusively or nothing. If one tries to license an invention non-exclusively for internal use, especially early in its reporting history, the answer is likely to be “no way.” Until efforts to license exclusively for productization have failed, few universities are willing to contemplate a non-exclusive license (other than as required in a sponsored research agreement) that would reduce the “apparent value” of the patent.

We find that 99.5% or more of reported inventions not only “sit on the shelf” but are restrained to do so by university policy and practice. Clearly, this is a portfolio approach to asset management, not an agent approach. In a portfolio approach, many assets are aggregated but only a few may need to perform to achieve financial success for the portfolio. In venture capital, for instance, ten investments may yield only two or three companies that are “successful” and of these only one might be “really successful.” The broader investments that suffer losses are part of the strategy of risk-taking that accounts for the fact that investors, even professional investors, are not able to pick winners beforehand. (See the Bessamer “anti-portfolio” for a fun read on what one fund passed up). Similarly in prospecting for customers one might build up a list of leads, and from that qualified leads, and from that conversions to sales and to continuing customers. From a pile of possibilities, one screens and closes and retains and expands–a few loyal customers may be all that’s needed.

The portfolio approach differs markedly from the agent approach. Where a portfolio might benefit from a broad catchment–even venture capital firms want to see potential deal flow in volumes–an agent approach has to be highly selective because there’s an expectation that for each asset under management, the agent will work a deal if at all possible. If you are an athlete and you engage a sports agent, you expect that if you have any talent at all, the agent will get you signed with a team. If an agent takes on too many clients, and most of them never get a deal, the word gets around and athletes choose a different agent. In some industries–music comes to mind–some agents merely aggregate possible “talent” and then sell their agency contracts to other agents or publishers. The agent makes money on the transfer of the collective pool of talent (a portfolio approach) rather than on income from successfully representing any individual act (the proper agent role). If the situation becomes one in which the use of a given agent is mandatory, regardless of the agent’s performance or operating model, then the agent clearly has it good–it can make money any way it wishes–but those compelled to use the agent are just grist for the machine. That is, they are the 99.5% that do not have to be put in play for the agent to be financially successful.

That’s where we are with the present approach. It’s not a matter of whether the program is compulsory or not–Research Corporation was largely supplied with voluntary invention submissions–but whether the program is selective, and what happens to the work that is not selected. Now, however, the work is first selected, and then dropped, so that nothing can happen with the dropped work without university permissions. Even if there’s no statutory ownership position available, so that those outside the university could exploit work, those within the university would violate university policies and come within the scope of discipline for doing so under the university’s broader policy claims to “ownership”–meaning the right to control, relative to any employee, anything claimed under policy, even when not patentable.

It is the 99.5% of university research assets that are withheld that is an unannounced problem. These materials should be in the public domain, should be available for private exploitation, for research, for internal company use, for exchange, and for building out private property positions in improvements and developments. But the university withholds these assets as part of its portfolio strategy to create income from a very few deals. For these withheld materials, there is no program of technology transfer. Most anything done with these materials ends up being a technical violation of policy–end-runs, work-arounds, defiance, indifference, covert.

It is this second, informal, unannounced technology transfer program that must be addressed by university policy. It must be given its own standing, operating on both NIPIA (non-IP intangible assets) and IP (where an exclusive license has not been obtained by the time a patent issues, or in any event within a year or two of issuance). If such a program had standing to release inventions (taken broadly) through “dedication or licensing” to the public domain, to create commons, to build consortia, to assist an open community, to provide assistance without licensing demands, to participate in standards, then a university would begin to see the value creation (if we must use such terms) that comes about through engagement of research and use and shared development communities rather than from excluding those communities in favor of engaging speculative investors and monopolists. One might even come to find that the NIPIA program exceeds the value of the monopolist program–both in terms of money and reputation.

In fact, that’s what we did at the University of Washington for a decade. By the time the founding team broke up, we had created a program parallel to that of the patent licensing program. Two offices, one focused on open–led by software, digital media, and engineering-like stuff–and one focused on exclusive patent licenses. The way we put it–we handled most everything that wasn’t in a cage or refrigerator. The result–we were generating $3m to $6m per year being mostly open on a budget of under $1m, and the patent group was generating about $3m a year on a budget of about $3m a year. The patent group didn’t like the idea that there was a competing operating model that outperformed them each year by about 3x. Eventually they found a way to shut it down. Monopolists don’t like competition or co-existence. “It’s confusing” and “inefficient” and “inconsistent.” Thus, calls for a “uniform” patent policy end up being calls for an arbitrary patent policy, a monopolist patent policy. And one that sequesters 99.5% or more of what it gathers in.


This entry was posted in Bozonet, Commons, Freedom, Metrics, Policy. Bookmark the permalink.