Designing Public–Private Transactions that Foster Innovation
Most university technology licenses are extremely conventional. The university selects one partner and gives it an exclusive patent license. The partner—who now has a monopoly in the university’s technology—promises to pay royalties. However, this “exclusive licensing” model is only one possible transaction. Many alternative business models are possible and some have already been tried. We need to figure out which of these new ideas are wise and which would be utter disasters.
I will start by reminding you that patents have important drawbacks for society and that it is often preferable to leave discoveries unprotected—that is, putting them into “the public domain.” I will then look at why universities focus so heavily on exclusive licenses. Finally, I will discuss 10 alternative licensing models that can often do a better job of spreading knowledge.
INTELLECTUAL PROPERTY VERSUS THE PUBLIC DOMAIN
It is easy to forget that intellectual property (IP) rights are, in fact, monopolies. They create incentives by letting the inventor stop others from using his invention. This creates an artificial scarcity in knowledge in exactly the same way that a baker’s cartel creates an artificial scarcity in bread. Legislators and judges have always known this. In fact, the first patent statutes were at least as concerned with limiting the IP monopoly as creating incentives. That is why the English Parliament called its first patent statute “The Statute of Monopolies.” Thomas Jefferson agreed. When he set up the American patent system, he said that his central task—which I think is our central task, too—was to draw “a line between the things that are worth to the public the embarrassment of an exclusive patent, and those which are not.”1
In addition to monopoly, patents have two other potential drawbacks. If I am allowed to patent a particular idea, how do I make money from it? The most obvious way is to hire someone to develop it, i.e., to turn it into products. But if I want to make a profit, I must keep my costs down. So I am not going to hire everyone in the world. Instead, I will I hire one person or maybe two, and see what they develop. This approach works well when development is obvious and straightforward, but that is not always the case. In fact, there are two reasons why it may be better to leave the idea in the public domain. First, suppose that the product made from the idea is patentable. In that case, the whole world ends up racing to develop the idea. This means that society gets the product faster. This is a very valuable benefit when the underlying idea represents a fundamental advance like, for
example, the laser. If we believe that universities produce more than their fair share of fundamental advances, we should put their discoveries into the public domain.
Second, some ideas are “embryonic,” i.e., the strategy for developing them is not immediately obvious. Again, many university inventions fit this profile. Suppose that a biologist discovers that a certain protein binds to the outside of a cell but has absolutely no idea what it does. Turning that kind of idea into a workable product is fraught with peril. If the idea is patented, the patent holder will hire one or two people to attempt development. They could easily fail. But if the idea is not patented, 500 people may decide to take a shot at the problem. If just one of them succeeds, society will receive a huge payoff.
WHY UNIVERSITIES LIKE EXCLUSIVE LICENSES
If patents and exclusive licenses have such drawbacks, why do universities favor them? There are at least four reasons. The first reason is ideological. Congress passed the Bayh–Dole Act because it believed that industry would not develop university inventions unless it received IP rights. There is some evidence for this position. If you go out and ask companies if they would have done a particular investment without licensing rights, they sometimes tell you—and it may be an honest answer—that they would have said “no.” But there are also many cases where companies were already using the invention before the university got wind of it and demanded royalties. Now the university has gone from disseminating knowledge to taxing it. At least potentially, Bayh-Dole has become a drag on innovation.
The second reason that exclusive licenses are popular is that they have a powerful constituency. There is the licensing officer, who is praised for bringing in cash. There is the faculty entrepreneur, who would like to get rich. And there is the university administrator or state legislator, who hopes to make the university at least partially self-supporting. All of these groups want to maximize revenues. But the best way to maximize revenues is to act like a monopolist. In other words, stick with the exclusive licensing model.
The third reason that exclusive licenses are popular is that universities tend to produce embryonic ideas. University technology officers and faculty members frequently say that they have no way of knowing whether an individual idea will be valuable or not. So they obtain as many IP rights as they can. Now it is one thing to patent an idea because you plan to develop it. That is an investment. But it is quite another to patent an idea in case it later turns out to be valuable. That is a lottery. The result is overpatenting and a shrunken public domain.
The final reason is that many people think that “exclusive licenses” and “start-up companies” are evidence of economic development. This is misguided in my view and comes from the fact that it is much easier to count licenses than to track the number of people who use the public domain. Nevertheless, the fact remains that many universities subsidize patents and exclusive rights transactions. At the same time, hardly anyone subsidizes the Creative Commons or other efforts to expand the public domain. So the net effect is that society is paying people to shrink the public domain.
LOOKING AT THE ALTERNATIVES
I want to describe three categories of alternative transactions. I will argue that the first category consists of transactions that are unambiguously good for society. The second category consists of transactions where the “core” value of scientific data—the right to use, modify, and republish information—is unimpaired. I argue that this situation is always preferable to a conventional exclusive license. Finally, I will describe a variety of transactions that are usually preferable to an exclusive license. Even though these need to be judged on a case-by-case basis, we should probably encourage them. One way to do this is to put such transactions into some type of “favored” or “safe harbor” category.
Situation 1: No Public Funding
The first category involves situations in which the experiment cannot be done with public money or charity. In other words, nothing will happen without the private sector. Of course, you should still try to negotiate the least
restrictive license you can. But suppose that your negotiating partner insists on an exclusive license. Should you agree to give the private sector a monopoly right? Many people are surprised to learn that the answer is “absolutely, yes.” A monopoly price is bad, but it is still preferable to not getting the information at all. You should do the exclusive license.
The clearest historical example I know about is cyclotrons in the 1930s. It was the first big physics research program. It absolutely could not have been done with government money. E.O. Lawrence made a deal with the Research Corporation. The Research Corporation figured that if it got enough cyclotron patents it could make cheap versions of radium and corner the medical isotopes market. Now, as most of you know, World War II and the Manhattan Project intervened, so that medical isotopes became plentiful. But suppose they had not. Suppose the Research Corporation had cornered the market. Even if the new isotopes were only a few pennies cheaper than radium, society would have still been better off. Would they have been as well off as if the government had funded the whole thing? No. But at least we would have had the information.
There is a point here about the federal science agencies. We have to reinvent the way that agencies do things. In this example, I assumed that the government was never going to fund cyclotrons. In many cases, however, scientists do not really know whether the government is willing to fund their projects or not. Only the agency knows. So my suggestion requires candor: The agency must be willing to say “We’re never going to fund this— go out and make the best deal you can.”
Situation 2: Tangential Applications
It turns out that scientific data have uses besides “doing science.” Can we sell those uses while leaving the core ability to consult, modify, and republish data in the public domain? And, if we do, can we earn enough money to support the database? The dot coms have produced several business models that are worth considering.
The first model is to sell companies the right to use the database as content or a traffic builder. The basic idea is that posting data on a website “attracts eyeballs” to the host’s other products. I will discuss an example of such a deal below.
The second model is to run an alert service. Suppose you have a database and you keep putting new data into it. If you know that somebody is interested in a particular category of data, you might send them an e-mail alert whenever relevant information is added. That service is worth something.
The third model is to sell advertisements. In the general dot com world ads are incredibly lucrative. Although ads are less lucrative in science, some journals have said that they can generate about $250 per published paper.
Finally, there is data delivery. Companies sometimes pay hundreds and even thousands of dollars to receive data in special formats. For example, you can sell them a CD-ROM version that they can archive or make special arrangements to deliver the data behind their firewall.
I can see the emergence of a principle where we should always choose transactions that leave people free to do science. Such deals are always preferable to traditional exclusive licenses. They are also preferable to some recent experiments designed to make academic databases self-supporting. The most famous example is probably the SWIS/PROT database in biology. It imposes substantial restrictions on academic users’ ability to modify and republish entries.
Situation 3: Safe Harbors
The third category consists of business models that are usually preferable to exclusive licenses. Unlike the first two categories, you cannot say that such terms are always preferable. However, it still makes sense to create some type of favorable presumption or “safe harbor” that such deals ought to be approved.
The first type of business model involves selling updates. Dr. Bretherton talked about the SeaWiFS (Sea-Viewing Wide Field-of-View Sensor) experiment, where commercial fisheries receive data 2 weeks before the scientists do. NASA paid approximately 40 percent of what this mission normally would have cost. Frankly, it sounds like they got a good deal. Updates have also been tried in biology. For example, Cardiff University’s Human Gene Mutation Database has been very controversial. Celera’s commercial subscribers receive Human
Gene Mutation Database updates 1 full year before the general public does. Biologists are up in arms about that, and maybe they should be. But now we are arguing about how long the embargo should be. The answer will almost certainly vary from case to case.
The second model is to provide an enhanced version of a basic (no-charge) database. Companies will often pay a lot of money for extra bells and whistles. In principle, the money can help fund your basic service. My background paper for this conference describes a deal between a corporation called Synx and a biology database called the Human Genome Database.2 Synx would have received the right to produce a premium version of the Human Genome Database; in return, it would have paid royalties and delivered certain software to make the public-domain version stronger.
The third model involves offering academic users a different, cheaper subscription rate. This strategy is ambiguous because it can be used to maximize revenue. For example, Celera’s academic licenses generate a big percentage of the company’s total income stream. On the other hand, you can also imagine setting prices to minimize the cost of academic subscriptions subject to the database breaking even. Should you do that? There is a large group of people who say that data should always cost exactly nothing. On the other hand, no such rule exists for other inputs you need to do science. For example, hardly anyone complains that Nature charges for a subscription. So it is a plausible question: Do data prices become reasonable when they are comparable to journal subscriptions?
The fourth model involves disappearing IP rights. Just because patents last 20 years does not mean that exclusive rights have to. During the 1980s, Harvard did a deal with Monsanto where the exclusive right disappeared long before the patent did.
The fifth model involves fixed-fee contracts. For reasons I will not get into, many of the economic distortions caused by monopolies go away if royalties are fixed in advance. What is really pernicious is the so-called running royalties, where I get a slice of your sales. So to the extent that universities negotiate fixed fees, that is a good improvement.
The last model is nonexclusive licenses. These are tricky. If a university charges high enough royalties, it can end up licensing everyone in the world and still enforce a monopoly price. So nonexclusive licensing is only a solution if the university also charges royalties that are something less than the full-blown monopoly price. One interesting feature of the Novartis and Monsanto deals is that they both stipulated that the corporate sponsor would receive a nonexclusive right to discoveries. A cynic might say that this did not matter, because if I already have a nonexclusive right I am in an excellent position to demand exclusivity. After all, the university knows that this is the best way to maximize revenue. So these deals could be a kind of cat’s-paw arrangement. But what if the university committed itself to give out at least one nonexclusive license? Variations on this theme are worth exploring.
THE POLITICS OF VOTING “YES”
At this point, I want to make a confession. I once negotiated a traffic builder agreement between a worldwide community of 600 mutations biologists and a company called Incyte. The deal we made was that Incyte would pay the community $2.3 million to build a worldwide database of human mutations information, which currently does not exist. In return, Incyte would have received one—and only one—exclusive right. It would be the only commercial company allowed to post the database on its Web site. There may be a more minimal version of a right that you could give to a company in exchange for real money, but it is very hard to imagine. Politically, I think that this is a bit of a test case. If you cannot get a community to agree to this deal, you probably cannot get it to agree to anything.
So what happened? The community held a meeting to discuss Incyte’s proposal—and, after protracted wrangling, decided not to hold any vote at all. As far as I can tell, there were two reasons for this. First, most members were hesitant: “Incyte’s proposal looks fine to me, but I just saw this 20 minutes ago. Maybe there’s a catch here.
See S.M. Maurer, “Promoting and Disseminating Knowledge: The Public/Private Interface” at http://www7.nationalacademies.org/biso/Maurer_background_paper.html.
Give me six months to think about this and I’ll get back to you.” Of course, six months later the deal was no longer on the table.
The other problem was that members were afraid that they would be punished for voting “yes.” “Even if it is a good idea, this deal is so novel that somebody’s bound to criticize me in the pages of Nature.” This problem only got worse after the representative from the National Institutes of Health (NIH) gave her opinion. She completely ignored the Incyte proposal and suggested that her agency might consider a grant application instead. People asked themselves, “If NIH won’t venture an opinion, why should I?”
WHAT AGENCIES CAN DO
Both of the foregoing problems involved a failure of leadership. Society needs to decide—in advance—what types of transactions it wants. And it needs to give people the confidence to say “yes” if somebody goes out and obtains a suitable offer. Neither of these things is likely to happen as long as saying “yes” requires a personal decision by 600 individual members.
I believe that funding agencies have a role here. However, it is not a traditional one. Deciding whether a particular transaction is in society’s interest is not like peer review. You cannot answer it by appealing to scientific merit. Instead, the agency has to decide whether it has enough money to fund a particular experiment and, if not, how many rights it is willing to give the private sector to get the job done. You can even imagine a day when NIH shows up at negotiations between the private sector and the academics and says, “I’ll chip in some money if you loosen these restrictions.” That will require some heroic changes, but it is not fundamentally unreasonable. In fact, NIH has already become much more willing to write regulations that tell people to make sure that any private-sector deals include particular provisions. So they are getting into a hortatory mode.
Finally, the idea of doing new types of private–public deals is not a theoretical subject. My own experience is that industry will stand in line to talk to you. As soon as we talked to Incyte, Celera got jealous and asked to get involved. In the end, we had three or four firms offering to help. So these deals are feasible. It can be done. The only question is whether individual scientists feel that they have a green light to do them. If you look at the statistics, individual scientists originate most grant proposals, put together the most exclusive license deals, and create most start-up companies. So my final advice is that the funding agencies need to decide which deals are desirable, announce some clear guidelines, and then stand back. The scientists will do the rest.