I will start with a mostly unreadable diagram:
This is the rhetorical anatomy of the relationship between an invention and a patent, or a “claimed invention.” It is important to see the difference because people tend to talk about inventions and patents as identical, that an invention is “patented” and the invention is a “thing” (like a cotton gin or an MRI device or compound 169′ that became Xtandi). But the thing that’s conceived in a material form–which an inventor might think of as an “invention”–and the thing that’s claimed legally as the subject of a patent that grants the right to exclude others are often very different things.
Let me explain, using the notation of the mostly unreadable diagram. Here’s an instance of invention, a little circle with a line to the scrawl “invent.” The circle represents a “prototypal” invention, a conception in someone’s head of something that’s new, that will work, that wasn’t obvious before but perhaps is now. It might be a new method or a new compound or a new device or improvement. It’s an invention, something to be used, demonstrated, replicated. Unexpected by the common person, the one with ordinary skill, but possible for that ordinary person to use, if taught carefully how.
This sort of invention is the one that is “made” and “reported” on a university disclosure form. “We thought of this, and did this or made this, and it works like this–pretty cool, huh?”
Typically, once there is an initial invention, various refinements are evident. Not always, but often. There may be additional steps–or steps that on second thought aren’t necessary. The prototypal invention is often the entry point into a sequence of refinements, variations, and improvements that clarify the invention, allow it to operate more reliably, to be constructed or used more easily, to be adapted to desirable situations. We might then draw a box to represent this set of variations. The circles represent variations on the invention, with the arrow showing “development” of the initial prototypical invention to an alternate version. Perhaps the new version uses a different reagent, or involves one fewer steps, or uses revised software–debugged, perhaps.
Often, one of the first things that might happen when someone has discovered or invented something material is an exploration of variations. Some variations may be needed to get the invention to work–the prototype might be an emblem of the idea, but might not work all that well. That was true, for instance, of early nuclear spin resonance imaging systems. The idea was there, but an actually useful device for body imaging wasn’t.
In fact, that initial inventive conception, the spark of the idea, might not actually work at all. No matter, that. What matters is that the spark opens up a box of potential variations, some of which may work. Some variations are just there and are immediately possible; others have to be hunted for. Who knows which variation might have the desired properties–or what the desired properties themselves might be?
When Karl Paul Link’s students figured out the chemical formula for the agent in moldy sweet clover hay that prevented cow’s blood from coagulating–3,3′-methylene-bis[4-hydroxycoumarin]–they went on on synthesized a hundred compounds built on roughly the same structure. Compound 42 turned out to be warfarin.
In engineering settings, in particular, a given functional output might be enabled by any number of pathways of substances, mechanical parts, circuits, software, or procedural steps. Each pathway may have its own advantages (cheap, easy, sensitive, controllable, serviceable, reliable, durable, configurable, non-infringing, elegant…). One explores those pathways to see how they work. An invention, in such circumstances, might morph or migrate relatively quickly from an initial version to one that better meets the design specifications for a desired area of use.
Thus, the prototypal invention often gives rise to a set of known variations. If an inventor publishes immediately, then these variations may be uncovered by any number of other experimenters and designers drawn to the idea. Otherwise, often the inventor (or, a team including the inventor) do the work of documenting variations. One might see, then, in the choice about whether to publish immediately or delay there might be very different development pathways for a given invention–the stuff in the box. Further, there might be very different pathways if someone seeks a patent immediately on the prototypal invention, without considering the prospect of variations. If someone has a right to exclude others from the get-go, then what are the responses from the rest of the creative community? Hold that thought.
There’s more than just the produced variations, however. There are also variations that are known by formula or generalization. If sodium chloride works, then so also might any salt–potassium chloride, say. So we can define general classes of compounds–“any of these” combined with “any of those” to produce “anything in this range.” These are variations by definition. One does not have to produce each and every combination and claim it–it’s good enough to claim the general formula and argue that every combination that’s claimed represents a variation on the idea inherent in the prototypal invention. Here are the claims for Heidelberger and Duschinsky’s patent for 5-fluorouracil, from 1959:
In addition to some specific compounds, there are claims that are general–N-glycosides of 5-fluorouracil plus any “medicinally acceptable” bases. That’s a whole class of compounds right there. (Plus there’s a method claim involving a process to make 5-fluorouridine from 5-fluorouracil). You see the strategy. Here there is a mix of specific compounds and classes of compounds. All of them are variations on the idea established by the first inventive idea.
Now add an additional element–the unknown variations. An inventor might not be able to create every variation, or even create formulas for every variation. Instead, we introduce the idea of the “functional equivalent.” Despite the limitations of language, it’s clear that there may be various ways to accomplish the task of making or using the invention, and any of these are “equivalent” to the invention. And, hence, within the box of variations, even though they aren’t expressly known to the inventor. We can draw a bigger box of variations, then, and divide it into the known and the unknown variations.
We have a box of variations separated into those that are “known” and those that are “equivalents.” Some one of these variations might be identified as the “best mode” (not beast mode) for practicing “the invention” at the time a patent application is filed. But there could be even better “modes” that become known later, all of which are anticipated by the claims of the patent application. The invention is the biggest box one can make. The job of the patent examiner is to carve down the sides of the box, while the job of the patent attorney representing the inventor (or owner of the patent right) is to keep the box as big as legally allowable. This is the start of the territory claimed by a patent–invention becomes multiple “things.”
But this is just the start. One can also then consider methods of creating “the invention”–any of the variations in the box, known or unknown. Often, a patent application will claim the invention and methods of making the invention. Similarly, there may be multiple applications of an invention, extensions with various features. If X is utterly new, then (so the logic goes), so is X+Y and X+Z and X+(Y+Z). Sometimes there will be multiple patent applications, each specific to a particular aspect of the invention. Sum the claimed inventions for each patent application and one gets the extent of the invention in its patent-claimed form. This brings us around to the full, largely unreadable diagram, now with “methods of making” and “applications, extensions, and combinations.” The full set constitutes the range of the “claims” established by one or more patents. It is this conceptual–and legal–object that is in play when one shifts from talking about a discovery (even one made in a “lab”) to managing patents on a discovery. These are rarely the same “discovery”–the patent-defined discovery is often much broader than the thing first produced and reported.
One of my favorite patent attorneys once apologized that we had all claims allowed on the first office action. I took the allowance as a sign that we had got a patent to issue quickly and cleanly, without a lot of fuss, delay, and expense–and thus had a solid patent in hand to license immediately (as we had a deal pending, where the patent ran alongside some software). From the attorney’s perspective, he was worried that he should have drafted broader claims–so that the patent examiner would have to carve them back to their legal maximum. Perhaps the claims already were at their legal maximum. Who would know? For our purposes, the claims were as we wanted them, and that was good enough.
What can we make of all this? First, the prototypal invention does not have to be very good. It might not even “work”–other than plausibly. The prototypal invention needs to be only good enough to define the class of variations, where at least one of those variations, known or unknown, does work. I’ve seen inventions that simply didn’t work–the data was fudged–but that the description of the invention for the patent application was sufficiently persuasive that the class of variations contained versions of the invention that did work. How about that? It’s like horseshoes–one does gets points for getting close. With a good patent attorney, one gets the same patent claim points for being close as for throwing double ringers. The prototypal invention–and even the best mode–is just an instance of a set. It is the set that gets claimed in the patent. It is the set that is the subject of licensing. It is the set that may be excluded from all other use while only a single variant–number 42 or 169′–is “developed” as a commercial product. All the rest of the variations claimed–known or unknown–are subject to enforced nonuse.
In terms of Bayh-Dole, which establishes as patent law that the patent system should be used to promote the use of inventions arising from federal support, we have the strange outcome that the patent system may permit a patent owner to claim a much broader territory of variations than the one the owner attempts to “commercialize.” Is it really Bayh-Dole’s purpose to establish as policy that owners of inventions made with federal support should obtain broad rights to all variations on an invention and exclude all use of those variations to support their financial interest in some one of those variations? Would people find it acceptable if this suppression of variations was expressly stated as an objective in the law? One might find oneself arguing that the exclusive rights of a patent on a subject invention should extend only to those instances that are indeed made commercially available (or otherwise used to provide a public benefit on reasonable terms). All other variations, after some period of time (say, three years or sooner if there’s no intent to attempt to use them), should be released for public access, even if those variations compete with the instance of the invention chosen for development, even if some of those variations are *better* (for some purpose) than the “commercial” version.
This situation is the case for some disease assays, where universities have patented (broadly) a method of detecting a disease state and exclusively licensed the patent (and the invention, broadly–even assigned) to a company, which creates and sells a specific implementation of the assay. A lab medicine clinic might be able to prepare a more selective assay–fewer false positives, better sensitivity–but the patent and exclusive license prevent the lab medicine clinic from doing so. They have to purchase the commercial version of the test, do without, infringe, or design around the invention (broadly, around the patent) and thus work to undermine the invention, to limit the scope of its benefit.
We can now come back around to considering the effects of institutional patenting on the diffusion of an “invention.” In a competitive environment, other companies have to decide whether to seek a license to the patent, to design around the patent (and so to come up with a variation that’s outside the box), to infringe (better, then, never to read the literature so the infringing isn’t “willful”), or to find something the original company might need and offer to cross license rather than beat each other up over infringement claims. An industry standard may arise where a number of companies decide it is better to contribute to a common platform than to fall into a frenzy of infringement suits and counter suits. Everywhere a patent excludes use in a competitive environment, it invites attacks on the invention–not merely attacks on the patent (for misuse, fraud on the patent office, invalidity, and the like). “Progress” in such a setting takes place because everyone excluded from the practice of the invention (in any of its variations, preconditions, and applications) has an interest in rendering the invention obsolete or irrelevant–undermine, design around, challenge, block, destroy. That may be progress in a fun Schumpterian microcosm of creative destruction, but it’s an especially odd use of the patent system to promote the use of inventions made with federal support.
Now consider the situation when a university becomes the owner of a patent and works with a default of exclusive licensing. Universities make a show of permitting “research” uses for other universities and nonprofits (though their exclusive license agreements rarely reserve rights so broadly and they do not as standard practice grant a royalty-free non-exclusive license upfront for such uses). Even the “Nine Points” document proposes that
Universities should reserve the right to practice licensed inventions and to allow other non-profit and governmental organizations to do so
In the Nine Points meeting, I proposed that universities should allow all organizations, regardless of tax classification, to use inventions for research purposes. This proposal was strenuously opposed by representatives from a number of universities and so never made it into the final form of the document. There are three forms of research use to consider: 1) “evaluation of” allows one to make and use an invention to determine if the invention operates in a manner consistent with published claims. To refuse to permit such use by companies amounts to limiting the operation of peer review to nonprofits. 2) “research on” allows one to study how an invention works, to use the invention to get at its manner of operation, the principles behind that operation, and thus to do science. It was long after Warfarin was introduced as a blood thinner that folks figured out just how it worked. 3) “research with” allows one to use the invention as a research tool, to use it to pursue investigations of one’s own. “Research with” may even compete with the researchers at the institution holding the patent rights for funding, for priority, for status. Is it an institutional conflict of interest to use one’s patent positions to prevent others from using one’s invention as a research tool? Is it adverse to public policy to do so?
For technology transfer, a broad patent (that is, a patent that defines a broad invention with many variations possible) held for exclusive licensing creates problems for even the university community. Even if a university follows the First Point of the nine points to consider and reserves rights for nonprofit uses, that defeats research with an intent to transfer results for public (or commercial use). Any other university may host research that practices the claimed invention, but that research cannot be transferred outside those universities without a license from the university that owns the patent–other than to an exclusive licensee of that university. Of course, if the research finds alternatives–and so avoid the patent–then we are back to the competitive situation in which the purpose of all others is to undermine the claimed invention, since there’s no purpose in trying to use it. Any other nonprofit does research, then, to block or disrupt the work of any other institution that refuses to grant broad licenses–licenses that permit all other nonprofits to transfer their results for public use (and commercial use, and development of commercial products). We can call that “progress”–but we could also call it “waste of resources.”
If a university cannot transfer the results of the research it hosts other than to a monopoly, then one wonders just what justifies the research. Sure, if the second university in develops a patent position in improvements or applications, then it can prevent the monopoly from practicing those improvements or applications until the monopoly agrees to pay up–stacking royalties or squashing the royalties in the stack, forcing the monopoly to decide which patents might be worth more than others. (Just because a variation has been claimed does not mean it is obvious how an improvement or application might work–the first patent owner can prevent use of the improvement or application, but the owner of the patent in the improvement or application can prevent the owner of the original patent from practicing the improvement or application). Thus, even with a “research” exception for nonprofits, a default exclusive licensing program prevents all other nonprofits from transferring their new technology that relies on that research exception. The only good for such a research exception is to design around or to block a line of development and force a cross license for a share of the action with whoever holds the monopoly.
These, then, are technology transfer consequences of the invention box, which is broader than the prototypal invention and broader than any given variation developed for commercial use or sale under an exclusive license (or assignment). If a university licenses exclusively a broad invention, then it cuts off technology transfer of research involving that invention at all other organizations. They can do research and publish that research, but no one outside of nonprofit research organizations dare use those results if they depend on a nonprofit research license. Patents on such results are useful only in blocking the original–and only then if they are directly in line with where the original invention’s commercial variations are headed. Otherwise, they are patents on inventions that cannot be used by the public–and that’s by design of the administrative choice of using exclusive licenses to broad inventions, covering all the possible variations but not requiring the commercial development of all those possible variations.
This is an aspect of university technology transfer that universities don’t report–how many other organizations in the past year have been prevented from practicing a patented invention? How many research proposals have avoided including a patented university invention because no rights for its use will be available, but for a licensed monopoly interest? These are not easy figures to gather. They are the figures of dark alleys, where valuable things have been lost, not the figures of pavement under street lights, where it is easy to gather “data” and comment on it but there’s actually nothing of value to discover.
What are the lessons to be learned here for policy and practice? First, that the “invention” in that press release or fakographic is not the tingle in the mind of the inventor reflected in some laboratory apparatus to be “developed” into a commercial product; rather, the “invention” is likely a relatively broad swath of variations, pre-conditions, extensions, improvements, and applications. And only a tiny subset of these will ever be developed for commercial use or sale. The rest will be prevented from such use, and any other public or research use, by the combination of patent rights and an exclusive licensing practice.
Second, the First Point of the nine points to consider is not nearly sufficient. Universities should grant a license upfront for all research uses–evaluation of, research on, and research with–regardless of the tax classification of the user. Nonprofit, for-profit, government, amateur or professional–doesn’t matter. But that research use, on its own, serves primarily to permit competition with the invention–not to make a similar product but to make a product that renders the original invention obsolete. Exclusive licensing has this property, and a non-exclusive license for research use only makes this property even stronger.
The practice point, then, has to be extended: universities should grant a non-exclusive license upfront for all making, having made, and using. The licenses need not be “royalty free” but they ought to be fair, reasonable, and non-discriminatory. Getting such a license should be as easy as registering one’s interest and perhaps paying a simple fee. Making and using covers a great deal of new ground that is also competitive with the commercial interests of any monopoly for sale of the invention (in its broad form but implemented in one or more narrow forms). But now the competition is between DIY use and a commercial product that offers use. If one can make and use an invention (in some variation) without requiring a commercial product, then the commercial product has to find its market among those that cannot make and use or do not want to spend the time and money to do their own work. That’s an interesting form of competition, that requires a commercial product to be better than DIY or directed at those who choose not to DIY.
But, really, there’s a third practice lesson here. A university should grant its non-exclusive license for use with the addition of an option to sell, so that anyone with the capability to DIY should also be assured of access to a license to sell also. Call it a DIY Plus license. Doing this permits all other universities that host research involving an invention to assure anyone they work with that they can transfer technology that they develop and also know that the original university holding a broad patent will grant a license as well, on fair, reasonable, and non-discriminatory terms. Such an option on DIY use ensures technology transfer. Anything less limits technology transfer and returns us to an interest in blocking use or obsolescing the original invention–utterly opposed to Bayh-Dole’s central policy of using the patent system to promote the use of such inventions.
Finally, there’s a fourth practice lesson. Universities should grant licenses under their patent rights that become as narrow as the actual commercial product that is used or on sale. While a patent license might allow a given company to spend time lingering over the box of chocolates represented by the invention box, the company will have to choose, and choose quickly–say, within three years of the patent issuing. Everyone else will be on notice that they can work on DIY immediately, and anything that isn’t already well under development within three years will be available for a DIY Plus license. There might be a delay in moving from use to commercial sale, but there is no uncertainty with regard to whether rights will be available for any of a number of variations within the invention box.
This DIY Plus approach is called “crossing the commons.” First one creates a commons of DIY use. From that use one sees the outlines of commercial practice emerge. Where that practice might be made more efficient by commercial products (even ones based on standards), then there’s a time to license the sale right. Use comes first, then sale, then royalties in a baby carriage.
One might find, then, that a patent license in these cases will look much more like a copyright license–thin, directed at a specific implementation rather than merely to a “field” or “territory.” Any exclusive licensee still has a monopoly position in the marketplace–but only for the particular implementation that they have chosen to develop. Others will get their chance to develop functional equivalents, but with a three-year delay, rather than having to wait for 20 years, as is the case now with the fragmented nanotech field, which university patent practice has pretty much ruined. Thus, while the first licensee in can prevent others from copying the patented technology used in their product, it cannot prevent others, three years on, from making competing versions using other variations in the invention box.
For the original university patent owner, this is a wondrous thing. Instead of relying on a single licensee to develop a commercial product (often, this never happens, especially with startups), the university has available to it the entire commercial marketplace. Its royalty income can be as broadly distributed as industry use. Further, this approach does not create incentives for others to work around the original invention but rather to work with it. This is huge, as it turns a university from being a threat and pain in the industrial butt to being an interested but reasonable partner. And last, this approach does not interfere with the technology transfer practice at other universities. Rather it augments that practice and invites participatory research, even research that creates competing results because a DIY Plus approach permits the original university to participate in the promotion of even functional equivalents and new applications.
If one wanted to worry the point, then the DIY Plus option would only be available to other universities that offered DIY Plus privileges back to the original university. Otherwise, in the words of the WTF license–er, well, go read the FAQ for yourself. You’ll find what I mean, expressed as clear as one can make it.