Science at the Frontier and the Effect of the Linear Model

In Science the Endless Frontier, Vannevar Bush proposed federal funding to universities to expand the frontiers of science. Folks these days focus on the science part of Bush’s proposal and his advocacy for funding research at universities. They skip over the idea of “frontiers”–old hat Star Trek sounding stuff, now. They also don’t appear to consider the structure of Bush’s proposal. It’s enough that universities get money–and that a mandate for such funding is baked into the federal funding pie.

Whole careers–administrative careers perhaps more than scientific ones–depend on such federal funding. After all, one third of the federal total research allocation never gets to researchers, but instead goes to administrators and contractors to administrators. What would happen if the federal government stopped paying more than 15% in indirect costs for grants (which are in their way gifts of public money to support individuals proposing interesting scientific work) and asked researchers to write direct budgets for what they needed. You know, like the Gates Foundation does. Then we’d see the actual administrative costs required to manage a project. If the project required IP management, then the project would have to contract for it somehow, and show the cost in its accounting. If the project required an administrator to worry risk management or fight over whether federal or state mileage rates would apply, then the project writes a budget for the angst. If the project needs a new lab space, then the remodel goes in the budget, too.

But let’s consider the structure of Bush’s insight, rather than muse about how things could be different if the federal government was not supporting a bubble of administrators. Focus for a moment on the idea of “frontier”–a place at the edge, where the borders get iffy and the rules of conduct become a matter of internal governance, since the frontier is beyond the reach of regular government with its police and lawyers and courts, its threat of discipline and dismay in the presence of taboos. On the frontier, things get settled without appeal to a nanny state. (And that doesn’t mean there has to be gun fights and all that.) A frontier of science has some of these same attributes, but now we are talking a geography of mind rather than sagebrush-strewn high plains. At a frontier of science, one lacks the controls of conventional scientific thinking. Theory becomes conjecture or simply disappears. One has one’s wits, and has to learn anew what to look for, what to observe, what to study. The problems are not pre-formed, as they are in textbooks that make it appear that training in science consists primarily of showing that you can work standard problems to arrive at the right answers. As if. We don’t get much training in what to do when the textbook ends, the standard predictions of theory are useless, and we don’t know what we are looking at. How to make sense of what hasn’t been made sense of?

It’s not that difficult to get to a scientific frontier, even in this age of “information.”
Take 3d printing. Sure, the modality is new. But there’s a scientific frontier in the behavior of powdered materials in combinations. How do clumps of stuff at the scale of, say, 10 microns, behave, as a potential aggregate, especially when there is a mix of perhaps ten different compounds? No good scientific theory for that. But figuring out how such mixes behave is the immediate concern of someone messing around with new mixes for powder-based 3d printers. Could be an undergraduate. On the frontier. Anything she observes that’s useful is also publishable, because it’s quite likely no one has been there before, and there’s not much in any textbook to help her.

The reality is, we keep students firmly away from the frontiers of science for years and years of their “training” in science, until they are firmly conditioned to expect science to consist primarily of the obedient demonstration of what is already known, in the form of problem sets. Discovery is for elites, for the movies. It is easy to grade such “problems”–and the STEM folks have got it down “to a science” so they frequently use on-line grading software. They don’t have to even review the work personally. It doesn’t matter what thinking might go into working a problem. The method and the solution are one: convention, demonstration, obedience. Good metrics for a scientific curriculum is how quickly it gets someone to a frontier and how well it balances rubbing a student’s nose in the known and what opportunities it gives to train that nose to sniff for the unknown.

Textbooks don’t spend much time on the unknown. Richard Feynman’s Lectures on Physics is a refreshing counterpoint, as about every other paragraph, Feynman suggests something we don’t know, that he’s using an approximation because that’s the limits of what we know. Once, at a physics conference, Feynman got up to give a talk and started out with something along the lines of “We have heard many sophisticated talks at this conference debating this phenomenon, but it comes down to the fact that we are just too stupid to understand it.” Physicists walked out in protest–but Feynman had a point. When the physicist is all-knowing, then any bombast passes as theory. But if we are all too stupid to understand something, then we are at a frontier, and all that authority, those professorial titles, the connections to editors at elite journals, the government grants, the fawning doctoral students–all that is useless, far away, not really of any help. Just stupid us again trying to make sense of a world stripped of its conventions, of our habits of mind. May as well be working out a way to make a stone ax.

Whew–spend a couple days on the frontier and you know a whole lot better what you need to do to survive there, to do something useful there, even to thrive there. We don’t much give learners that experience in the sciences. Instead (starting now in grade school) science is a matter of working problems and learning to observe what has already been observed. “Labs” are exercises in following instructions and handling lab equipment. Not at all about trying something new. Play-acting. Pretend. A STEM fantasy land, but nicely organized into “learning objectives.” No wonder many creative folks dislike the experience and choose something else–art or literature or history, say, where they can get to a frontier quickly and see what it is they need to develop to do the work they imagine for themselves. Recruitment isn’t based on the mystery for which help is needed. It’s based on “workforce development” (make money in this field, it’s a gravy train of federal handouts!) or “diversity” (we need token people to help us meet our statistical goals, and because you are a “non-traditional” person in our field, you will think “non-traditionally” and that will be so nice–until we pound it out of you with a few years of obedient problem-solving.

One senior research biologist I know returned from a trip where he served as a judge for a national high school science fair competition. He was downcast. “Why?” I asked. “Because there were brilliant students there. I would hire them in an instant to work in my lab.” “And why is this so bad?” “Because I know that in the next four years, their college work will beat most of them down to dullness.” Rock that STEM curriculum. If a curriculum doesn’t get people to the edges of things in a matter of weeks, and demands years of calculus, trig, chem, and physics problem-solving, then it creates, conditions folks for “lamplight” science. Work the problems that have already been worked. Use the random distributions that were taught, not some new distribution that you need to create to account for what you observe.

As a freshman physics major, I was required to take a course called “Intro to Experimentation.” We met in a lab. The professor would come in, discuss an idea, show us a piece of equipment, and suggest we use it to help us observe. “Here’s a plastic wafer containing a bit of cesium 137. It’s radioactive. So be careful handling it. Here is some equipment that detects radioactivity. Have at it.” Then he leaves us for three hours to mess with stuff. At first, we feel like idiots. Then we realize that’s what we are, idiots on the frontier, until we learn to cope with what we don’t know. That was a good exercise. There was nothing we were “supposed” to find or do, other than to get our bearings with regard to the utter unknown (for us). And there is no good theory that accounts for when a particular radioactive source will decay, other than a description of the random distribution of the times between decays (Poisson, for some reason). Take most any literature class and one gets this same experience every day. What did you observe? What do you make of it? What are you, now, for what you have done with that text? For Middlemarch read wafer of cesium 137 and there you are. It’s just that if literature were taught like STEM classes, they would tell you what you are supposed to find in Middlemarch, and show you the steps you must reproduce to prove you know how to recite the convention down to the finest details, and then you would be given ten more reading problems just like it, on which to display the same method and come to the expected “right” result.

For Bush, one problem was how to extend the frontiers of science, extend the boundary. Another was to make sure more people were trained in science, especially the part about working at the frontier, where new things could be observed and characterized and ideas about them developed. The interplay of theory and experiment, and all that. These were the two key goals for Bush in proposing that the federal government support university research–get onto the frontiers, and do it with students around, so they get training on those same frontiers. Our now conventional STEM curriculum saves that frontier work largely for graduate students, who most likely will be handed a “problem” as part of a team’s work on something. Not really much curiosity-driven work, even for undergraduates, even for high school students, though there is plenty of areas for that, if one were allowed to go after them. Most of the math taught in schools was created by folks under the age of 25. Those are the critical years. But most STEM students spend those years working problems to demonstrate their obedience to the conventional views of science, to which they become so invested that they lose sight of what’s not known, what’s inadequate, what’s just plain wrong. Conventional explanation is the enemy of curiosity. Why be curious if there’s a textbook answer?

Perhaps this is one reason why Peter Thiel argues that bright folk should skip college and get into a startup. Startups can be on a frontier, too. Often it’s not a frontier of science, but rather of business economics–a new business model, a new gadget, a new market. But the feeling is much the same, developing a sense of how to behave beyond the “conventions” of business that get put into textbooks.

Bush wanted universities to have some work (not all) at the frontiers of science, with students there, too. Perhaps that was too much to ask.

Here’s how his approach would have worked.

  1. Talented folks at the edge of knowledge at universities pursue their interests, supported by long-term federal grants. They broaden the frontiers and report back publicly what they observe, creating a pool of new ideas and new phenomena for others to draw on. They also train more people how to do such science. (A’s hire A’s; B’s hire C’s–folks who play professional hockey know how good Sidney Crosby and Joe Pavelsky are–they don’t fuss over it; fussing over talent is for Dunning-Kruger syndrome folk, who don’t know enough to know they don’t know enough).
  2. Another set of folks look at an area of established practice. For instance, the military. Medicine. Communications. They consider problems and opportunities in that area of practice. They are folks who get insights about how something might get done because they know the frontier work along with conventional science. They browse various areas of practice to put together new ways of doing things, drawing on both conventional and frontier scientific work. These folks work outside the established order, in teams of scientists, engineers, and gadgeteers, to create a “coordinated attack” on a given problem or opportunity. Their aim is to obsolesce a class of issues in favor of something new. Cures or prevention say, rather than finding a better way to cope with a chronic condition.
  3. In developing prototypes for their new practices, these folks will benefit from expanded scientific frontiers. They are not doing “applied” research in the form that Bush understood. They are making something new. A digital computer, say, in an age with a mania for improving the mechanical adding machine. “Applied” research is research posed by a problem coming from industry, from healthcare, from business–how to solve a manufacturing problem, a drug delivery problem, an accounting system problem. Find a solution that’s safe, profitable, reliable, useful. For that, one typically turns to conventional science, if science is needed at all. Conventional science provides theories about what has been observed. Frontier science seeks to observe what isn’t known, what we don’t know we don’t know.

Do you see the space Bush wanted to work? He had worked it during World War 2. He saw that combining frontier science with competent industry folk and folks who could build things meant that one could build prototypes directly for a purpose without staying in the path of improving what one already had. As H.S. Harrison would have called it (in A History of Technology, vol I–on-line but only for university folk), Bush wanted to generate “primary mutations”–events that changed mere objects–new things observed in the world, at the frontiers–into tools, into artifacts with human purpose, or “free mutations” that transform an artifact for a new purpose, or “cross mutation” that swaps into an artifact characteristics from some other tool. For this, he wanted to broaden observation, to see more, to see better, to recognize. There did not have to be an application for every such observation. That’s wasn’t the point. In fact, surrounding the need to observe with such a context–to solve some pressing industrial or political problem–would necessarily constrain observing to the conventional.

So Bush wanted a kind of unconstrained pursuit of curiosity, by the best and brightest, not by just anyone, not by the lazy and mediocre and not by folks going through the motions to live the happy life of the unaccountable academic–the kind of academic freedom Feyerabend railed against. The freedom that Bush sought was a freedom from establishment conventions, even the establishment conventions of “science”–and so he asked for funding for science at the frontiers, not for more of what’s “normal” or even what’s chronically “at the frontiers” for the establishment. His problem was how to create within established orders–government, university–a lasting program that got to frontiers of knowing not selected by those established orders. In this, it appears Bush failed, and as the money grew, so did the interest of established orders in co-opting it, making sure it served their purposes, and so was credible, accountable, proper, reasonable, and the like.

Established orders do have their frontiers, and on those frontiers they plod along with the problems they have defined, and try to find ways to solve those problems or at least mitigate them or prevent them from becoming a liability. In this way, the established order preserves itself, preserves the status of its leaders, preserves the expertise that comes with knowing how to maintain the order and benefit from it. Established orders do not commission research aiming to throw that order into disarray, to disrupt it, even creatively. Established orders commission research that improves the stability and efficiency of the order and maintains the status of the order–its income, its leadership role in society, its pecking order. They don’t rush to any other frontier to do research. Other research would be goofy, laughable, purposeless, useless, feckless, dangerous, irresponsible, irrational, criminal. Thus, the cell phone companies don’t press ahead with health issues–no surprise why they don’t.

Now we can critique the “linear” model from a different direction. In the linear model, when an invention is made in research, the next step is to find “practical application” for the invention. For that, so the model goes, one needs “applied research” to fit the discovery to the context of an industry, say, that sees a purpose for the discovery. That is, the linear model suggests that the proper next step is to shop the discovery to an established order–industry, or a company in an industry, usually a leading company or one with tons of venture capital.

To shop a discovery to an established order, so the theory goes, one must turn it into an invention and obtain patent rights for it. That is, the discovery must be owned. Then it must be licensed–that is, there must be a formal contract between the patent owner and anyone who might think to use the discovery-turned-invention-turned patent-turned contract. The selection criteria then move from anyone capable of using the discovery (for whatever purpose, playful or profitable) to companies, and among companies only those willing to try to turn the discovery into a product, or incorporate it into existing product, and among those companies, only ones willing to take an exclusive position, and willing to pay a royalty on sales–not just a one-time fee, certainly not no fee at all.

This approach has proven to be difficult or impossible. Often, the null set is all that shows up. A company may want to use the discovery, but screw the licensing. A company may want a non-exclusive license, but the university says screw that until we have proven there’s no exclusive licensee out there. An exclusive licensee might show up, but not want a university putting its nose into company business via an audit clause on royalty calculations. So it’s easy for there to be no companies at all, or just companies backed by speculators willing to say anything at all needed to sew up a position on a discovery, figuring that they can always sell it off to someone who wants the rights more than they do. That is, make money by getting in the way of use, betting eventually on some such stuff, the use will become important, and the speculators will get their investment back with a nice return. Maybe not from every position, but all it takes is one or two a decade to feed the inner troll.

Given the difficulty of this approach, universities have turned to doing the “applied research” themselves. The front for this is the startup company. The startup is often a shell company covering more university research. Then they go get “seed” funding for the shell, often from state economic development funds that were designated to help real, private companies, or from SBIR/STTR program funds. The shell company then contracts with the funds back to the university labs, where the “applied” research takes place.

Even then, the “applied” research may not be sufficient to discover an actual product or to show that the product can be produced. But it’s a nice added source of research income, and so it’s worth the gesture even if it doesn’t go anywhere. Where industry and leading investors don’t see a market, universities persist in pursuing the “linear” model to try to develop the discovery to some magical point at which industry or investors will give the discovery a second look. This is mostly lipsticking pigs, but now the pigs are called “early-stage technologies”–as if discovery is an “early stage” of the linear model and all that’s needed is money, lots of money, to buy data and employees and markets and customers and demand. As if, with “research” and “marketing” one can somehow “make” commercial products out of discoveries–if only the greedsters in industry and luddites society in general would play along and make the model “work” and make administrators fixated on the model look brilliant.

Yes, the linear model is also burned into Bayh-Dole. The idea that presents in the law is that the next step after discovery must be a patent and then practical application, which university administrators call “commercialization” and mean actually “a royalty-bearing license with upfront fees.” There is nothing to indicate why the law is drafted this way–other than that it sounds reasonable, if one learns to think that way. (Notice, however, that universities ignore the parts of Bayh-Dole that don’t suit them–they do not, for instance, systematically report first commercial sale or use–only how much money from how many licenses they have granted on how many patents on how many reported inventions–the implication is that licensing for money is the primary goal, not “benefits available to the public on reasonable terms”).

But just because the linear model is burned into law doesn’t mean that the linear model is at all effective. And for all that, Bayh-Dole doesn’t require the use of patents, and doesn’t require universities to own patents or even to deal with inventions, other than to report them to the government. It’s the university administrators who have decided to adopt the linear model, to interpose institutional ownership of and licensing of patents as essential steps in the model, not because doing so shows productive gains in encouraging use of discoveries, but because there is a norming myth in practice that patents are a way to make people pay–and isn’t that the goal? So shouldn’t universities make people pay, too, and try to make them pay using the best tactics available? And aren’t patents and an overwhelming preference for exclusive licenses, and the threat of infringement among the best of those tactics?

My thought is that Bush would argue this is entirely the wrong idea. But since Bush isn’t here to make the argument, I will: this approach to the linear model is entirely the wrong idea. First, because it’s a rotten way of getting value from patents. Second, because it’s a rotten way to exploit universities. Third, because it makes a hash of what Bush was trying to do. The first two reasons ought to get administrators’ attention–you aren’t making the money you wanted from your tech transfer program because you hired dutiful and fast-talking people fixated on wasting money on a model that performs poorly if that’s what you say you want; you are damaging your university’s standing in the community by pissing off industry and entrepreneurs and making it appear that you are making scads of money when you mostly aren’t, and when you do make money, you blow it on a total eclipse of the sun, not on something socially valuable, such as lowering tuition for students.

But let’s look at the third reason–Bush’s insight about innovation outside the established order. Bush did not argue that the next step following discovery was applied research. His experience went another direction. He started with the John von Neumanns and asked them what they would do about some problem faced by the military–how to detect submarines, or how to save wounded soldiers dying of infection. These folks then could draw on their knowledge of frontier science, industry, and making stuff to propose approaches that come at the problem outside the context that the established order would present for itself, in terms of do-ability, costs, priorities, favored personnel, and the potential disruption to the order itself. What Bush wanted was to have frontier science available to these coordinated efforts to develop a response to a problem or opportunity in an established order. For that, the discovery does not require a patent, does not have to be first turned into a commercial product, does not have to become a product at all. All it has to do is be available at need for use by someone leading a team working on a coordinated response. That is, the discovery needs to be reported, needs to be characterized, needs someone to “science the shit out of it.” That’s all. Not ownership, not license, not picking monopoly holding companies, not “commercialization” of the discovery, not making money on a patent position as a precondition.

Stuff wouldn’t “sit on shelves” in university labs because there would be folks funded to know about it, to find it, to recognize how it might be used outside the field in which it was developed. This for Bush was the insight–find people who could make stuff that appeared “unrelated” become meaningful in a new context. Filing a patent on such stuff creates a barrier to use. Insisting on licensing creates a further barrier. Licensing preferentially on an exclusive basis creates an even higher barrier. Locking a discovery down for use in the field in which it was found runs against the important objective of innovation outside an established order–instead, the patent-license-commercialization approach works to prevent a frontier discovery from becoming readily used in an unrelated area.

The linear model is not merely “one approach among a number, but one that’s well regarded and widely practiced”–it’s an approach that damages the opportunity for “unrelated” access to frontier science. It’s an approach that saves frontier science for established orders, for the expected rather than the unexpected use. It’s just that what university administrators think are “expected uses” through “commercialization” turn out to be stupid, dull, economically impossible, bureaucratically compromised uses, and so the licensing for 99.5% of inventions claimed by universities goes nowhere, if nowhere is defined as practical application so that the benefits are available to the public on reasonable terms. Universities may make money. Speculators may make money. Trolls may make money. But no one tracks those benefits available to the public because, well, that’s just fluff language to make people feel good about the folks making the money.

Bush’s argument is that coordinated development of prototypes with the potential to transform an established order’s practices–call that “innovation”–is greatly enabled when the developers have access to frontier science. And as frontiers of science are broadened, there’s no telling what new things may become available. For this to happen, we need

1. research at the frontiers, beyond institutional science, beyond the pragmatic context of the needs of industry or government.

2. coordinated teams of frontier scientists, industry experts, and gadgeteers able to make things, led by people with insight into how to put the frontier ideas to work in combination with conventional science and technology to do unexpected things (at least, not expected by the status quo).

3. ready access by those teams to the discoveries of frontier science. If there are going to be patents, then those patents are made readily available, if not with a public license the moment they issue, and not just for “non-commercial research use.”

4. steady funding for frontier research and training in frontier science, outside of institutional interests, but not to reward the lazy and the mediocre, but for the John von Neumanns, for the Richard Feynmans, for the — see, where are the current exemplars of this sort of researcher–ah, they might be there, but they are lost in the funding of just about anything.

5. funding for coordinated teams to exploit frontier science along with conventional science, working outside target established orders, as if skunk works, disinterested in what a target established order announces as its highest priority problems or best practices or leading expertise.

This is the approach Bush used during the war. This is the approach Bush proposed for funding through a National Research Foundation, something enough outside the established order that it could see its way to fund people willing to work outside conventional science and outside problem definitions supplied by industry or government–that is, people working on the frontier of science, with the optimistic expectation that may well be endless.

The present university implementation of linear model patent commercialization works to suppress Bush’s insight. Bayh-Dole is just a dull gravestone marking the formalization of government policy around the idea that public science should be reserved for institutional administrators as leverage to money-grub speculative investors. Bush had a very different idea, one largely objected to, laughed away, mistaken, ignored, lost. But to my mind, Bush had–let’s say, has–a compelling insight. He was successful with his work for the military, and prescient regarding communications. As for medicine, he was right about the contributions coming to medicine from unexpectedly related areas–chemistry, then computer science, next maybe additive manufacturing, using cells.

I have yet to see a compelling argument that ownership of frontier science by patents leads to better innovation. I have yet to see a compelling argument that institutional licensing must be a precondition for the diffusion of discoveries to the benefit of the public. I have yet to see a compelling argument that making institutional money-making from patents a priority will as if guided by an invisible economic hand create a wonderful, new world. Yet in practice, these are the operative principles that underlie most university licensing shops, their policies, and the expectations of the senior administrators that permit the whole scheme to continue on its present path. The few licensing officers who do things differently, do things with a humanitarian priority, or support the engagement of research teams with broader practice communities–those folks come under fire and get driven out, and senior administrators sit on their hands about it, let it happen.

We could say it’s a “structural” failure–that the federal policy is a failure that puts frontier research behind patents to be held for two decades when not licensed to monopolist speculators. Of course, federal policy does not do this–it merely sets up the conditions that allow this approach to be practiced by university administrators. And it does not matter one bit whether those administrators think they are being sincere or mercenary or opportunistic or just doing what they see everyone doing, as if they were still in high school–what matters is that the effect of their choices is to ensure that Bush’s insights–the ones that led the move for expanded federal funding of university research–would remain suppressed. We get, instead, a complicated, expensive, delaying fixation on a version of the linear model that demands institutional ownership of frontier discovery, institutional contracts as a means to allow public use, and a widespread default monopoly position that demands commercial products rather than creating a space outside established orders for coordinated teams to draw readily on science at the frontier.

This entry was posted in 3D Printing, History, Innovation, Policy, Social Science, Technology Transfer. Bookmark the permalink.