Joe Lane and Benoît Godin are out with another paper that follows up on their Science Progress discussion. In their new paper, they argue that innovation arises along three related areas of activity–scientific research, engineering development, and production. Each of these areas need to be funded. It is not the case that pouring money into research will result in a scatter of discoveries just waiting for speculative investors to turn them into products. The engineering development step–perhaps one could still call it “applied research”–is needed.
But in a way, applied research and engineering development really are not the same thing. It may be that the rhetoric of “research” is part of the problem. Lane and Godin call it the “research myth”: the idea that basic research is the font of discovery that will transform society, and if not society, then the economic parts of it in favor of the nation or region that sponsored the research. And, sure, there are research discoveries. But that’s not the point of the myth. The point of the myth is that basic research somehow holds the prize position and has come to monopolize the discussion about government funding and where innovation comes from. Any number of studies shows that the linear model of innovation–basic research to applied research to product development–is mostly rhetoric. There are many ways that innovation arises, as Eric von Hippel and Steven Johnson, among others, have shown. The problem, as Lane and Godin point out, is that we don’t have a good narrative frame beyond the linear model, and so it is invoked.
For university tech transfer, this is pivotal stuff. Tech transfer offices love to put up diagrams that show basic research leading to development to products and economic vitality, along with riches for administrators who in a generous moment even share some with the inventors. From the tech transfer point of view, then, the claim is the more research funding, the more inventions, and the greater the chances of finding that one big hit every 15 years that makes an office appear “successful” and allows whatever is the prevailing theme of the day to claim credit for whatever has happened.
From a larger perspective, tech transfer is merely the handwaving at the far end of the pitch for research funding from the government. Since the pitch for research funding is tied to practical things, like inventions, and useful things, like cures for disease, there has to be some part of the deal where inventions get used or turned into products, and cures are practiced. That’s the part for tech transfer, generally. University licensing offices construe this to mean, file patents and seek exclusive deals, and if one is forced to it, then accept non-exclusive deals but grump and complain about it. The idea is to make money, and make as much of it as one can, figuring that all that’s needed is a lucrative 20-year patent run every 15 years or so. For a typical university licensing shop seeing 150 or 200 inventions per year, one needs only 1 in 2000 or 3000 inventions to have a financial success. The rest can go rot, for all the licensing office cares. Of course, other inventions will be fawned over, but the reality is a few deals float an office for decades, and the rest can be thrown in to make licensing activities look a lot more important and effective than they actually are.
Here’s the thing. It doesn’t matter what the TLOs do, actually. They are there to give it an effort. Whatever the folks in the TLO think, they are just the rear turret gunner in a big operation to keep the research money flowing. The real issue is to maintain the government’s fixation on basic research funding. If the TLOs are successful, that’s great. If they are not, then that’s not a fault of research, but of the TLOs or more likely of the “funding gap” created by the TLOs trying to leap past engineering development to find speculative investors who can take research findings directly into the marketplace. The funding gap, in all this, is not the challenge of the startup in finding support prior to product sales, but rather the gulf in understanding of technical details between the hype of research potential and the realities of engineering development. My argument has been that the “funding gap” is mostly, in the work of TLOs, an artifact of the mismatch of licensing efforts with the conditions under which an area of technology might develop. Put it bluntly: the effort to find speculative investment to make a product from a research-originated “invention” creates the funding gap that then is blamed for the lack of regular successes in the little linear model of university licensing for profit.
University TLO practice is often like the character Pigpen in the old Peanuts cartoon strip–they bring along with them a lot of the problems that they ascribe to the marketplace, or to companies, or to investors. They are their own little dust cloud of irritation. The university administrations, however, cannot possibly care. As long as there is a show of effort, then the problems in the transfer of technology don’t have anything to do with how research is funded. Remember, administrators only need 1 in 2000 for success, just one big hit every decade or two. Everything else is for show. The last thing university administrators will do is to advocate for the government to re-think the distribution of innovation funding so that there is money flowing to engineering development and industrial production, well outside the reach of university grant proposals. And that means: the university administrators are not in it for the innovation in society. They are in it for the status that comes from winning federal basic research grant awards. Bragging rights based on money. Pretty simple, and simplistic. There are no calls from university administrators to move government funding to, say, manufacturing, or to support more non-institutional developers. Instead, there is the cult of university startup companies based on basic research. Let’s look at one of the vocal leaders of this cult, Utah.
The University of Utah managed to persuade the state to dedicate $93m over 5 years to the USTAR program, which was going to stimulate the Utah economy with innovation that would lead to startup companies and jobs. Well, here are the numbers: 4 startups and 13 jobs from that $93m. All the University could do is write a report estimating the “contribution” to the Utah economy of spending the $93m. It does get a big-sounding number–$755m in direct and “indirect” contributions. Most of that comes from multipliers in a standard economic model, without any attempt to verify. And most of the expenditure contribution is for construction of research buildings. Even $755m sounds big until one sees that Utah’s GDP is about $140b/yr, so that $755m over 5 years ends up being just above rounding error on the Utah economy, a 0.1% blip. It would have been that same blip regardless of how the state had spent its $93m, based only on an “economic contribution” model. There is no economic “impact”.
Although the report expressly disclaims that it estimates “economic impact” (see p. 3, first column), the University of Utah ignores this its press release. And the University of Utah conveniently conflates the USTAR state-funded program with the University’s own startup company puppy mill, which over six years or so created 125 or so paper companies, only a few of which received a round of investment worth reporting, and a few more have lived for a while off SBIR funding. Most of the companies are moribund, many have the tech transfer office as their business address. Yet the University of Utah pounds its chest about company creation, as if the public is to believe that these companies are all creating jobs and new products in the Utah economy, when in fact they appear to be a net drain on public support and a diversion of state money to a particularly ineffectual kind of spending at the University of Utah. Why? The rhetoric of research, and an effort to keep state money flowing to the University, rather than to other parts of the Utah community.
In a time of short funding, I guess one should not expect university presidents to advocate for helping the needy when there’s good money to be had for research. I might also note that across the country, while instruction has been hammered at universities, it has been business as usual in research, even though many of the largest research operations at universities are losing money–lots of it. At the University of California, on at $3b annual research budget, the loss was over $700m in a recent year. Extramural research is losing the universities money, not resulting in the claimed for outputs, and yet it is instruction that is being cut and tuition that is going up. It’s all part of the rhetoric of the basic research myth.
Engineering development, by contrast, is rather like practice-based innovation. Matt Ridley makes a case for practice as a source of innovation in The Rational Optimist. The rule of thumb may be a more perceptive guide than a scientific theory for much of what we consider to be innovation in such contexts. Based on Ridley’s account of the developments in the weaving industry during the industrial revolution, a whole lot of innovation simply did not rely on science. Science came along much later to supply explanations, but the perception of how the system works, or could work, and the means to explore those possibilities, arose from nuanced perception of those engaged in practice.
One might say that beyond the soulcraft of shop class, there is also the potential for innovation. In Railroaded, Richard White talks about how the shops building locomotives in the 1870s and 80s were essentially “open innovation”–each locomotive had practice innovations that its crew came to understand. When the crew took a break, the locomotive also went out of service. Each locomotive had its own crew, with the crew’s distinctive working understanding.
I flew back from Chicago a year ago in the seat next to an engineer for a major natural foods manufacturer who was installing a new factory in northern Illinois. He said that for the large industrial ovens, each was distinctive. The day crew could get the ovens working perfectly, but the night shift would come in and change up just enough to undo the efficiencies. Much the same thing as the locomotives a hundred and fifty years ago. Once a machine or device is in service, it diverges from the specs and becomes its own thing. Those that come to know it are able to get the best out of it, anticipate its shortcomings, and keep it in a condition of productivity. From that nuanced knowledge, they also see how they might improve it, or replace it altogether with something different. That knowledge does not come from science or research: it comes from practice epiphany. You don’t write proposals to the government to have a practice epiphany. We have no structure by which the government would fund such stuff.
It may be, then, that if one is going to look at a three-activity model, it is not that funding should be at parity across the three areas identified by Lane and Godin, but that for any given area of activity, one has to ask what areas would benefit from assistance, and fund accordingly. For different industries, the mix of funding levels across the three areas might be rather different. And for all that, if Ridley is correct, then R (for Research) does not point to D (Development) so much as successful D points to R and unsuccessful D also points to R, but for different reasons. The science often comes afterwards. Science research often does not lead the dance, but follows it. Funding this follower science, however nice, would be at cross-purposes to an innovation intent. Increasing the funding for follower science would be like trying to increase the swelling after a bruise. Folks need to fund the leaders–and for innovation that means, often, arranging for it to be the case that folks with ideas get a chance to lead, no matter what the credentialed consensus thinks. Consider information technology, where the leaders of the major companies, Gates and Allen, Jobs and Wozniak, Brin and Page, Zuckerberg–these folks were practitioners, most without college degrees, and now computer science departments are all abuzz in the space these folks created. Why doesn’t the government give grants to folks so they can avoid university STEM course hell for a few years and do something with their lives, like Peter Thiel suggests.
In the case of warfarin (see my articles here and here), the research capability to isolate a compound and create scores of variations is good stuff, but the science for how any of these compounds actually worked came decades after warfarin was both an effective rat poison and a life-saving blood thinner. If that’s the case, then the idea that R points to D has to be abandoned–R does, sometimes, point to D, but a lot of the time, it is D doing things that advance R. Even more so, practice-based work–D, but “fooling around” D as distinct from “engineers whipped by managers whipped by investors who demand a product” D, appears to be every bit as productive in identifying innovation as anything R can produce. And the advantage of fooling around D is that if it is already embedded in practice, then it is already likely way closer to use and product than stuff popping up in basic R contexts.
The observation that matters in all this is that when there is an innovation–a new product on the market or a new thing being used–it is always possible to construct a narrative that looks back to find D, and looks back further to find R. One can then tell the story forward, as if there is R, then D, and then bingo, I (an innovative product). It is just that this forward story is really *told backwards*. That is, one starts with the innovation and creates a creation myth for it, a narrative that confirms what people want to hear, especially if what they want to hear is a tale of the linear model. It is just that one has a hard time going from this R to D to I story to make it happen that way in real life.
In real life, it appears that the epiphanies and determination to do things newly–and even outside of expectations and approved practices and what blue ribbon panelists espouse as the decreed roadmaps–arise at all points in our activity, in “value chains” of connected companies. Research is often a side issue, and scientific research even more so. If the “science policy” is for science to somehow imitate the backwards story of R to D to I, then no wonder we are stagnating and spending a lot of money for show, and to hurry the pulse of speculative investors hoping to arbitrage hope and hype into a blossoming of stock value–at least long enough to sell their shares to the next wave of desire.
If we try to tell a story about innovation forward into a new future, we get science fiction and exploration, not planning. Planning is about doing things that have been done before. Narratives about an innovative future are not for the managerial sorts, no place for administrators or processes. Innovation, and the future that is unlike the present, is the stuff for outliers, fooling around, and chance encounters. For that, it would be good to get rid of administrators, contracts, processes, and plans–or, reduce these in importance so that people can mess around more, and fuss with paperwork less. That’s another reason I advocate for universities to release their research talent from institutional compulsory ownership claims to inventions–so there is a chance to revive the networked, non-market economy of ideas and practice that has been so productive over the past 100 years.
The debates 100 years ago in the Westinghouse and GE labs over which comes first, research or development would still appear relevant. Do you start with research and hand discoveries to developers, tasking them to “make it so”? Do you start with development and take it as far as you can with rule of thumb, calling on research only when you get stuck? If the latter, then you may want expert researchers to be fiddling nearby, but willing to drop what they are doing to help when you need it. For that, a federal grant is terrible if it requires one to keep on task, with program officers whipped by auditors ready to whip investigators to ensure that the research gets done good and proper-like, without any diversions that would distract from the planned and committed activities under the grant, even when developers are asking for assistance. Worse if they have to wait while another government grant is written up….months later.
Andrew Nelson and Cyrus Mody made an interesting study of the Stanford electronic music program. The program consisted of two groups of developers–composers and engineers. Each group messed around a lot in their own area and some in the other group’s area. (See the section titled “CCRMA’s Roots: ‘Fooling Around’ and ‘Chance Encounters'” starting on p. 17). Some of that “fooling around” led to the VCO inventions that Stanford licensed to Yamaha. The composers could describe a concept they were working with, and the engineers would pipe up with something they had that was responsive. Or the engineers would show something they had worked out with sound generation, and composers would play the new stuff into a composition. One of the most famous of these is the song Daisy, which sparked the scene in 2011: A Space Odyssey, when HAL is being undone:
That is, CCRMA would expose musicians to the methods and values of scientists and engineers, allowing researchers to “produce fundamental knowledge” for the “science of music” funded by the NSF – exactly the model established by Terman. At the same time, CCRMA would assist scientists and engineers in the “application of a rapidly developing computer technology to the art” of music “in a rich interdisciplinary environment” – exactly the kind of humanistic collaboration and civilian application that would bring diversified funding and approval from campus activists.
In other words, the basic science funding provided the musicians with access to engineering expertise that was willing to be interrupted by chance encounters, and the engineers found that the musicians had ideas and capabilities that could drive the basic science.
We found the same thing going on (and worked at it, thanks in part to insights from conversations with Andrew Nelson) in the Solheim Open 3d Printing laboratory at the University of Washington. Students from the art department not only could push for developments in 3d print material systems, but they could step up and make contributions directly–but they could never have done so had not there also been an expert group of mechanical engineers fooling around with equipment, controllers, materials, and modeling systems. Those capabilities, in turn, allowed the artists to conceive of new ways of doing things, and new things to create. See the work of Charlie Wyman, Laura West, Philomène Longpré, Ron Rael, Meghan Trainor, and Mark Ganter among others.
It’s difficult to shake a prevailing myth. Such things appeal to our System 1 pattern recognition thought systems. The linear model is one such myth, and the idea that basic research needs to have a lot of government funding is another. Lane and Godin call for parity in funding. I would push it and say, perhaps, even, that there can be too much government funding–uncritical government funding–and that can hurt research as well as innovation outcomes. When the ARRA stimulus money became available, NIH and NSF did not create new programs to get it into circulation–they merely lowered the quality threshholds on existing program reviews and funded stuff that wasn’t scored so well. John Ioannidis has just published a study asking how much of the most cited (and by extension, most influential) research in bioscience has been funded by the NIH. The answer is about 40%. The issue, as always, is what is the question? It may be that the NIH is funding too much research and not enough engineering development–or perhaps put it another way, the government is providing more money to the NIH than it can effectively use.
Here is an info graphic from Technology Review that Dan Collin at Kickstart Digital poked to me. It shows how NASA funding occupied a major part of the federal budget in the 1960s and 70s, and how that funding gradually gave way to NIH funding, so that whereas aerospace and Tang and Space Food Sticks were all the rage, now it is biotech, so that 47 states have named biotech startups as the key to their “innovation economies”. Well, obviously, to the extent that everyone is bidding for federal research pork dollars. This is not an austerity argument–but rather it is a discussion about the role that basic research has in an innovation environment. This discussion is also not an argument against science. I am quite fond of science, but I do raise the question whether credentialed scientists optimized to win federal grants are the primary source of good science–and I’m thinking that getting a credential in science does not endow one with imagination or social purpose or luck or determination. One can bring these things to science, but science credentialing really does leave a lot to be desired.
The present rage for STEM money–for science, technology, engineering, and math–is misplaced. Merely more engineers is not going to spark innovation, but may lead to Neal Stephenson’s quip that the story of our age is that we employed the best and brightest to make spam filters. And this is not an argument against basic research. But it is an argument that raises the question whether throwing more and more money at basic research, in the form the government is doing–what with peer-reviewed grant applications tied to announced areas of interest, conforming mostly to what the status quo consensus of peers (scientists) think is worthy of study–doesn’t get at much of anything new, really, or radical. As Carl Sagan warned in Contact (sorry about the ad first), it may be the folks fooling around that are best positioned for the chance encounter, not the career-minded politically savvy folks at the NSF.
It may well be that R needs D way more than D needs R. As for the I of Innovation, it appears to draw from many sources, and humping up R may do more harm than good. After all, why should anyone independent work one more day on 3d printing the day the government decides to throw $100m/yr at it? That R will swamp out all the independent work. The universities will fill the media with hype about “potential” only to be stymied by the naughty “funding gap” that really is another name for “crappy operating model, but it only needs to work once a decade or so.” The D community will be suppressed. Investors will be recruited to skip the D and go directly to attempting to create lucrative products based on R, supporting high-paying jobs for credentialed scientist types. A dreamy world that seems not to happen.
We don’t have live with the linear model myth, and we can value research without overstuffing it with funding and expectations that it has never been so good at. Fifty years ago there were tremendous debates over government’s role in science. Vannevar Bush argued that the government should not run its own labs. Eisenhower warned about the stifling effect of government funding on independent curiosity. Feyerabend argued that the separation of of state and science was critical to the activity of science. But you would never know it now, considering the rhetoric of university research administrators and technology transfer officers. Their view is, it would be great if federal law stripped faculty of their invention rights, and handed those to the state (in the case of public universities) or the institutional administrators, to be used as grist for profit-seeking ventures, putatively to create “innovation” and “jobs” and “public benefit” but in actual practice, really, just cover for “making money” and “getting bragging rights that will bring more money to universities in the form of research grants and state subsidies for economic development”.
For there to be innovation in innovation practice, there will have to be a break with the myth of the linear model and the idea that there is a necessary arrow from R to D to I. Sometimes there is. Often there is not. And it may well be that the primary arrow, as far as R is concerned, goes from D to R, or more broadly from the imagination to research, where development and engaged practice fuel the imagination more broadly and effectively than a line of research can do for itself. For that, then, if the government is going to fund with an eye for innovation rather than playing at making the linear model look good, then the funding is going to have to be managed some other way. Michael Crow at ASU thinks that the federal research system is “limited by design.” Agreed.
We need the discussion that Lane and Godin have begun. There may be a science of science policy, but it will be best informed, I believe, by a policy of practice and engagement–or as Lane and Godin would have it, of an innovation policy that treats “innovation” for what it is–a change in an established order, potentially at odds with the proprietors and happiest beneficiaries of the established order. For that, perhaps something a little more like a Shanzhai approach to development would do better–a bit of open, a bit of freedom, a bit of control, and a bit of access by anyone, sorted by what works and who is good at making things work. Again, this is no place for planning and administrators looking for a deal a decade. It’s gritty and direct and opportunistic. Stuff for which university policy statements are pretty much worthless, unless the policy is to keep out of the way.