We have been talking university technology transfer metrics.
First, that there aren’t any metrics. No one bothers to collect them or report them. Instead we get proxies of activity–number of patents, number of licenses. Once one has patents, then one creates a need for licenses. An invention might be taught–or might be developed independently–but once you have spent your $15K to get a patent, then for anyone else to legally (without threat from you to shut them down) practice that invention (howsoever they have got to it), they must get a license from you. Thus, counting patents identifies how much has been held back from broad access. The licenses attached to these patents–if ever–indicate how much of what administrators have held back has then been released, and with what scope and conditions. To get back to the open access that things started with, every invention that has been patented must be licensed. Merely counting patents (how much held back) and licenses (some stuff released) does not get at what the administrators have been doing. It merely records activity and proposes that others make a connection between the two counts.
Such metrics are worse than nonsense. They are deceiving. And it’s worse than deceiving, because the university administrators often deceive themselves that higher numbers of patents and licenses means (to their muddled minds) that they are doing a better job, a more important job. In short, the number of patents and number of licenses report bureaucratic make-work, not technology transfer or technology change relevance, or anything meaningful.
Those proxies–patents, licenses–don’t track anything close to transfer, nor do they track Bayh-Dole’s objectives of invention use (with benefits available to the public on reasonable terms), maximum participation by small companies, collaboration between universities and companies, free competition, and use of American industry and labor. They distract from the metrics that might guide management and policy–if discovery and technology change need management and policy. At some point, one wonders, just how far do we go down the road that the thing missing from the improvement of our lives is more bureaucrats involved trying to make money from patent positions?
Second, let’s pound the point that technology transfer isn’t inventions, patents, licensing, and money but rather the movement of technology from one place where it is known to another place where it becomes known and used. Even the use of the term “technology transfer” has come to mean, in bureaucratic circles, the proxies of activity and not any actual movement of technology from one place where it is practiced (or at least imagined as practice-able) to other places where it is also then practiced. Apparently, there’s not now any current term for such a thing, since “technology transfer” now means “whatever it is that university offices do that they call technology transfer.” In a way, then, we don’t even have words to point to the movement of technology and differentiate movement from independent discovery and use of similar things. Terms such as “teaching” and “showing” and “sending out” and “explaining” seem dowdy next to “exclusive rights” and “innovation ecosystem” and “public-private partnerships” and “return on the public’s investment.”
Beneficial technology change is the public “return” on the public’s investment in technology-directed research. Not enhanced institutional reputations or big pots of institutional money (extracted not just from the public but from–mostly and most. intently–those members of the public suffering some health problem that speculators have turned into a betting pool). Helping an industry get fat exploiting public health distress is no decent role for a university. Doing so as a patent parasite is all the worse. Believing that such university parasitics are somehow necessary or important for technology change borders on psychosis.
I’ll state some things coarsely.
Technology change does not require technology transfer.
Technology transfer does not require patents and then exclusive licenses.
Patents and licenses do not necessarily improve technology transfer.
Patents and licenses do not necessarily enhance technology change.
Bureaucrats managing patents and licenses do not make things better.
It’s not that we have this swirl of discovery and invention and people trying things or abandoning things and in all of this people still get sick and there’s no effective treatments so that if we add bureaucrats at nonprofits managing patents and exclusive licenses things will somehow become better. Bayh-Dole merely says, add the bureaucrats as the default without holding them accountable. It’s bad enough to add bureaucrats as the default. Worse to claim as a matter of policy that holding bureaucrats accountable will prevent them from doing the wonderful things that we are promised they will do.
Thus, to be remarkably worse about it, Bayh-Dole keeps reports of bureaucrats’ licensing activities a federal secret, does not require federal agencies to ask for reports, does not require federal agencies to enforce public protections in the default patent rights clause or to create new clauses to better protect the public–from the bureaucrats–and there’s no public right of appeal, either, to put things back on track. Whatever one thinks about exclusive patent positions as a driver of technology change, and whatever one thinks about university bureaucrats getting involved in such exclusive patent positions, Bayh-Dole–especially as practiced, especially with NIST’s clueless regulations–has to be the worst possible way of forming federal policy on the matter.
If you want metrics to ascertain how the bureaucrats are doing with their patents and licenses, you are a long, long way from looking at technology change and how anything discovered via research contributes to technology change, among all the other things that contribute to technology change. And if you let the bureaucrats come up with their own metrics, you will find that they prefer to report things that make them look good, make their policies look successful. According to bureaucratic norms, metrics should not report the past–they should report aspiration as if it were fact.
Third, technology transfer is not the only way in which technology change comes about. There’s independent discovery, and there’s assisted discovery in which one puts another in a position to realize how to do something new rather than show how directly. And there are various drivers for technology change. One driver is that something new works and others want it–solves a problem, opens up new opportunities, provides something customers want. Another is regulatory–a standard expects it, a law requires it, liability attorneys demand it. A third is fashion. People want what other people have. And tied to that desire comes interoperability and network externalities–as things scale, other benefits of change emerge–new supply chains, common training, production scale, custom production equipment, technology clustering for parts and maintenance. Any and all of these drivers (and more–spite! greed! lulz!) may be in play for any given area of technology practice.
To get at measuring things that ought to be managed–and there’s a leap or two there if one is interested in technology change–then we need to get clear on the idea that research produces technology change and that the means by which research does produce this change should involve patents held for exclusive rights by university bureaucrats.
The first volume of A History of Technology (edited by Singer, Holmyard, and Hall) describes technology change “From Early Times to Fall of Ancient Empires.” There’s not a patent nor a bureaucrat around to catalyze, facilitate, or take credit for technology change. Maybe there were bureaucrats doing all those things in ancient times to make sure the next tribe over had the latest in flint arrowheads next time they came raiding. We just don’t have any record of it. Technology also changes without research, without patents, without bureaucrats. It’s not the way bureaucrats want things to be, of course, but there it is. Pound that point for a bit.
There’s a huge difference between discovery and use of some new thing and a patent that claims not only that particular use but also anything similar and as many other applications or methods as can be thought of, along with the implication that all functional equivalents are also fair game for exclusion. It’s one thing to imagine an inventor as a person out working on something no one else could possibly care about and using a patent get repaid for the effort once it becomes clear that there really was something to care about. It’s another to recognize that many people are working in the same areas, with the same platform technologies and same ideas, and any number of them are able to come up with new things that are, technically, patentable, but which will become apparent–even if in some alternative implementation–to most anyone else.
A recent paper by Carl Shapiro and Mark Lemley points out that over 95% of all IT patents that are litigated involve cases of independent development. It’s not that one companies steals from another and refuses to pay, but rather that the patents on IT “inventions” claim stuff that most any company working in the area will come up with in any 18 month window of activity, especially if that activity is directed toward what companies expect will become part of a new standard (footnotes removed):
Empirically, between 95% and 99% of patent defendants in the IT industry are not in fact copying anything. They are independent inventors. Indeed, as we have seen, it is quite often impossible to know whether someone else invented the same thing you did at around the same time until years after the fact.
In some areas of technology development, then, a patent may easily claim not only what will become apparent soon enough to any number of companies–if they haven’t got the idea already–but that these companies become aware not by reading the patent literature but rather by looking at what needs to be done and applying the usual strategies of engineering to attempt to do them. A patent, then, excludes not only this independent development of just the same thing but also anything remotely close, including methods of making, compositions of matter, methods of use, applications, improvements, functional equivalents.
Because patent applications aren’t published for 18 months, and because the claims in an application may not be allowed as first drafted, no one knows for three to four years just what claims may be allowed. In that time frame, companies are not going to sit on their hands and wait to find out. They are out working on their own variations, developments, improvements, prototypes, applications. They may not even care to use the patent system, but you can see now that they pretty much have to–they have to be first to file to preclude someone else from filing first and knocking them out or taxing them three years in the future, and for anyone else with a patent already positioned to do that, a company acquires patents as a form of currency to counterclaim or offer by way of cross-license, or to dedicate to a standard with the expectation that others will do so as well and that the standard will exclude those companies not willing to cooperate for the standard.
The worst case of all, then, in this scenario is the nonprofit obtaining a broadly claimed patent and insisting on licensing that patent exclusively–with all of its claims–to a single company, with provisions that prevent the company from dedicating the patent to a standard, cross-licensing to lower exclusionary barriers, or otherwise ignoring infringement. The nonprofit’s position is not so much to transfer technology as it is to deal in the right to prevent new technology from emerging, in exchange for payment. All this gets wrapped up in abstractions about invention, patent, license, and money–but such stuff is not technology transfer, or if it is, it is the worst sort of transfer for an industry that has active research and development programs, standards, and roadmaps for development.