Energy and Innovation

Pascal Zachary is the New York Times's man in Silicon Valley, and he's done a quick tour of SV companies working on solar. This is not Zachary above, however, but T.J. Rogers, CEO of Cypress Semiconductor and also the head of its new solar company, SunPower Corporation.

Call me shallow, but I would feel less like this were less of a publicity stunt if Rogers weren't dressed like the missing member of a barber-shop quartet.

The article mentions a couple of promising tech areas - thin-film and the more original solar thermal. The technology is incredibly interesting.

1.

The business statements are incredibly boring. I mean "boring" in the clinical sense of the experience that comes from disavowing a painful truth. Boring-but-true statements #1 and #2 are:

• "Affordable solar development is also still dependent on government subsidies."
• "so much of our effort is going into short-term victories that I worry our pipeline will go dry in 10 years."

There is nothing explicit at least in SV culture or the "California ideology" of high-tech entrepreneurship that supports the government subsidies - call them public investments - that allow for long-term victories. I juxtapose this to the organizational capacities of Asia and Europe, and I worry about us.

2.

Most U.S. readers don't think this contrast between markets and government reflects reality. At least two generations of sociologists, B-school theorists and journalists have discovered the network, and consider it the crucial third term. High-tech business pulls people and resources from everywhere - government labs, universities, large corporations, start-ups, NGOs, and takes whatever it needs. In the U.S., networks are thought to make government direction obsolete, whether it's called investment, industrial policy, or something else.

A case in point is a good piece by Jason Owen-Smth and Woody Powell in the book Cluster Genesis, whose title starts making the book's core point that both research and development take place in clusters that combine disparate institutional forms and continuously and dynamically evolve. The authors look at "strategic alliance networks" in biotech in Boston and the Bay Area, identify five types of organizations and four kinds of connections among them (R&D, financial, licensing, and commercialization). The core point is that clusters and their geography form the infrastructure of biotech development. Although big firms tried to "cherry-pick" the best people and ideas, and everyone predicted biotech shake-outs and consolidation, this trend "faced significant obstacles imposed by deeply collaborative R&D efforts and a mobile scientific work force." As the sector evolved, "the pattern of dense inter-connection deepened, suggesting that the original motivation of exchanging complementary resources had shifted to a broader focus on mining innovation networks to explore new forms of collaboration and product development."

The network is thus more fundamental than financial markets or government planning, in this view, and in some sense the latter categories don't really make sense since they never operate as such. Owen-Smith and Powell find that Bay Area biotech has relied more on venture capital than Boston-area firms, which had more partnerships with academic and public-sector organizations. And yet although the Bay Area relied more on private capital, and Boston on public science, both regions did quite well, generating successful firms and valuable products.

But there are important differences in the innovation systems too. The "private" Bay Area system produced almost twice as many patents per firm as Boston's public. But another measure (variance in forward citations) is higher in Boston, suggesting that that region may engage in more "'exploratory' innovative research." In other words, the VC-fueled Bay Area folks may patent everything in sight, including incremental improvements as a defensive play.

Another important network indicator is "prior art" citations in patents, or backward citations. The "open" public-science structure of Boston-area biotech made 71 percent of its citations of prior art to non-biotech firms while that number was only 45 percent in the Bay Area. Owen-Smith and Powell interpret this to mean that the more commercially-driven, VC-based Bay Area cluster consisted of firms that were interested in their own research and that of their competitors, and less in research from outside their industry sector. When the authors compared a firm from each area that had developed competing therapies for the same condition (relapsing multiple sclerosis), they found this pattern replicated in the patent citation patterns of the respective firms.

The same pattern appears in a third indicator - product type. The Bay Area cluster produced more therapeutics more quickly - about twice as many. But the Boston cluster focused on treatments for rare diseases covered by "orphan designations" that offered "tax breaks and regulatory assistance to organizations that develop such medicines." There was a difference in strategy in the two cases, and although the authors don't use the term, Boston is closer to the outcomes one would expect of "public science" - addressing a clear need even when markets will offer modest or minimal rewards.

Thus "network" isn't so much a third term as a compound of the other two. Good empirical work redivides it into two familiar tendencies. The first cluster - the Bay Area's - is composed "largely of competitors and investors [who] are primarily concerned with speed and with commercial development, hence they pursue a more focused innovation process that relies heavily on internal R&D and attention to the efforts of direct competitors." On the other hand, there are "firms that are embedded in networks anchored by public research organizations and that lack strong investor involvement" and thus don't only "pursue immediate commercial returns." The latter, Boston model relies on external sources of knowledge and favors "more exploratory efforts at discovery." Governing ties in Bay Area firms tend to be more exclusively financial. Finally, the Bay Area model seems to be suffering some kind of lock-in, as the need for access to early capital sends firms back to the same small group of VC folks - with their same strategies, connections, and goals - again and again and again.

Even scholars dedicated to network theory with excellent data point out the constant danger of market failure and the rarity of a strong foundation of Boston-style public science.

3.

There's also something about the sheer exteriority of this understanding of innovation - big and small companies, capital flows, people from famous corporations who start new ones. What about the invention process?

I happened to read this passage in a London Review of Books piece about a new book on gravitational waves.

Near the end of the 17th century, Edmond Halley examined records of medieval and ancient solar eclipses back to the time of Ptolemy. He discovered that when he used the position and trajectory of the Moon to determine retrospectively when solar eclipses should have occurred, the times calculated differed from the actual ones by up to an hour. Halley deduced that in the past the Moon must have moved across the sky from east to west more slowly than in his own time. This was a far-reaching, even heretical assertion. For the Moon to have changed its motion in such a way would imply that its course through the heavens did not repeat in periodic orbits. Such ‘secular’ changes in its orbit could eventually cause the system itself to disappear, and the Moon to fall into the Earth or escape into space. For many philosophers, to theorise that the cosmos could decay in this way was a slur on the Almighty, as it implied that God was such an unskilled craftsman as to have constructed a system of stars and planets that could fall into ruin and disorder. Nonetheless, Halley was right, as even the fundamentalists were eventually forced to concede. The question now became: what causes the secular acceleration of the Moon?

This passage amazed me. I can't imagine running these complicated hand calculations for days and weeks and maybe months, coming up with a one-hour difference over decades and centuries, and not thinking oh well, I screwed up a line of my math. How do we still find and nurture these people - the ones with the sheer courage, the mind-boggling stubbornness, to decide they are right, everyone else is wrong, there's a one-hour gap, then come up with a new theory, one that would not be bourne out for hundreds of years?

How do we remake a business system with enough space and time to allow this kind of research? How likely is it to happen in California?

The Shock of the Old: A Review

This book's full title is "The Shock of the Old: Technology and Global History since 1900" (Oxford, 2007). Its author, David Edgerton, a historian at Imperial College, manages not to use the word "nanotechnology" a single time in 270 pages. Yet the book is directly relevant to nano-issues. In addition to insights about innovation that I discuss below, the analysis can be taken as an antidote to the pressure on all nanoscale fields to develop faster than any previous wave of technology.

Edgerton's title is obviously saying, "the old is important too." The book has a huge number of examples of old technology and low technology that continue to affect society long after they supposedly peaked. Hybrids spring from every page: there are the traditional Thai long boats joined to V-8 car engines - remember the flying-boat chase scenes in the Bond film Tomorrow Never Dies? Edgerton always stresses the overlooked efficiency of old tech in new situations. For example, he traces Rwanda's "spectacularly fast genocide" in 1994 in part to the machetes stockpiled in advance: "most victims were killed machetes (38 per cent), clubs (17 per cent) with firearms accounting for only 15 per cent of deaths."

Edgerton is right that we underestimate the role of the old. But his second and most important theme concerns why we do this. His explanation is that we make the mistake of centering our histories of technology on innovation rather than on use. We date advancement and progress from the moment a technology appears or is first applied, and downplay the long and winding road of adoption, imitation, diffusion, improvement, recycling and hybridization. And yet it is this long haul that decides the impact of a technology on society, and not its exciting first revelation.

Important things follow from retelling the history of technology as the history of use.

• it changes technological time, slowing it down, stretching it out, and shifting its major impacts, usually to a much later date.

• it allows us to follow the social uptake of technology, and tell the stories of its actual deployment.

For example, steam power "was not only absolutely but relatively more important in 1900 than in 1800." Similarly, "the world consumed more coal in 2000 than in 1950 or 1900." If we date steam power from its earlier appearances in Britain in the 1700s, we will identify it with the "dawn of the industrial revolution," see its importance as ended by other energy forms (oil, electricity), and miss steam's long and influential presence in later decades. If we do look at use rather than invention or first adoption, we can, to take another case, recognize the continuing importance of coal to China's current round of spectacular industrialization, and appreciate the extent to which it will "win" by using the old.

Seeing the power of the old doesn't mean we can ignore the new or the processes of innovation that create it. But we will redefine innovation when we keep the old in view.

Innovation begins to look different in Edgerton's chapter on nations. I will start a list of the core changes:

1. high levels of research and development (R&D) spending correlate with higher levels of national wealth and growth in gross domestic product (GDP).
2. high levels of R&D do NOT correlate with high levels of innovation OR with high levels of economic growth.

We should pause to think about these points, since they undermine the central justification for R&D spending, which is that this spending always boosts the national economy.

Edgerton offers some examples of his claims. "In the 1980s Italy overtook the United Kingdom in output per head . . . while spending much less on R&D than Britain did" (109). Similarly, "Spain was one of the most successful European economies in terms of rates of growth in the 1980s and 1990s, and yet this is a country which spends less than 1 per cent of GDP on R&D." Edgerton notes that the USSR spent as much or more of its GDP on R&D than did the US in the 1960s and 1970s, and yet is regarded as "having contributed practically nothing novel to modern industry" (110). Or take China vs. Japan. Japan has long had one of the highest rates of R&D spending in the world, but "while China has transformed itself and flooded the world with manufactures, the much more innovative Japanese economy has been, by comparison, stagnant" (109). This kind of data - which has been known by specialists since the 1960s - leads to further claims:

3. imitation as at least as important as innovation to economic growth.

4. "global innovation my be the main determinant of global economic growth, but it does not follow that this is the case for particular nation states" (113).

We can restate (3) to make it more compatible with Edgerton and with the work that we're doing here at the CNS:

3a. uptake and use of technology are more important than innovation as such to economic growth.

Innovation has not been demoted here so much as it has been yoked to diffusion, spillovers, transfer, and technology sharing. Technological innovation has meaning (and economic value) only as a social process.

This general principle does not help explain a central mystery in this book: why the US has been the great exception to the rule that innovation and growth do not go hand-in-hand? "By mid-century . . the USA was a clear leader in industrial research and innovation by any standard: it dominated both world production and world innovation. As such it was wholly atypical" (112).

Edgerton is disappointingly unwilling to dig more deeply into the American Anomaly. We must explain it, though, because the belief that effective R&D leads to greater economic growth continues to drive technology policy. Policymakers could accept points 3a and 4 above, and respond as follows: R&D certainly does improve economic growth (4), and the problem is with freeriders who take your great new stuff without paying you properly. Thus (3a) just means that we need stronger intellectual property protections, including international agreements like TRIPS, to prevent countries like Italy and Spain and China from borrowing - i.e. stealing - their way to the top.

Edgerton would reject this interpretation, since he rejects the "linear model" that lies behind it (see below): he would deny that most innovation comes from official R&D lab sources, that imitation is so different from invention, and that use is theft. I'm getting ahead of myself here, but I mention this now to suggest that the stakes are very high, and that it is not enough for historians to say the linear model (in which R&D leads to growth) isn't always true or generally true. They must also do more to spell out a non-linear model that will encourage policymakers to do more than just patch the linear model with stronger IP - which is where most policy rests right now.

But back to the book: Edgerton does say that the US enjoyed high rates of growth in earlier periods because it applied itself to borrowing and adapting technology first developed in Europe: this continued, and may explain the "golden age" growth of the 30 years following World War II. Historians of US technology would also suggest that the US benefited from massive Cold War military investment and from its long experience with the highly skilled coordination of large-scale engineering projects. The most famous of these was the Manhattan Project (actually the "Manhattan Engineering District") that produced the atomic bomb during World War II. As Edgerton points out, this "builds on decades of experience in large-scale research and development" (199). Technology sharing and ambitious, well-coordinated projects are a major part of the US economic story, and they lead to a further point.

5. technological and economic development depend on advanced infrastructure and coordination, which historically arise more from sophisticated institutions than from markets.

Evidence of this last point comes in the form of Edgerton's stress on the role of very large firms in the innovation process. Major advances have continuously come from companies like BASF, Hoechst, Bayer, AGFA, General Electric, AT&T, IBM, Du Pont and Eastman Kodak: "all these firms were already very large, innovative in 'science-based' technologies, and employed an abundance of scientists and engineers" (193). They created internal R&D operations, and these generally remained productive for decades at a time. "At least fifteen out of the twenty-three firms listed as the top R&D spenders in 1997 (and 2003) were formed before 1914" (194).

Taking points 1-5 together reminds us that there has never been such a thing as "closed innovation," in which development took place inside one institution. Analysts like AnnaLee Saxenian (Regional Advantage), Clayton Christensen (The Innovator's Dilemma) and Henry Chesbrough (Open Innovation) have made much of the new dependence on innovation networks that no company or even nation can control. The history of technology shows that there is nothing new about the sheer dispersal, the boundary-crossing, the institutional mixing and sharing, or the global scale of invention. A whole range of motives, participants, organizations, and sectors are always involved in any major technological wave. And this insight leads to a further major point:

6. "Most invention has taken place in the world of use (including many radical inventions) and furthermore has been under the direct control of users" (187).

This is a fantastically important idea. It puts practitioners of every kind at the center of innovation throughout history. It puts use at the center of invention. It puts the street and the shop next to the state-of-the-art academic lab. It puts imitation at the heart of invention. It truly displaces the "linear model" (from bench to bedside, from lab to market, from specialist to customer, from agent to recipient, from producer to consumer, from smart to dumb). It discredits the basic categorical distinctions on which that model generally rests.

Once technological development is defined through use, we can push Edgerton's point for heuristic purposes and say:

7. The history of technology is the history of everybody. That is, of everybody's uses of it.

which implies:
7a. the importance of laboratories to technology has been greatly overrated.

or more precisely and helpfully:

7b. There is no "downstream" (public) to try to push "upstream" (scientific laboratories), because in the history of technology, there is no "upstream." In other words, at different points in a technology's history, everyone is upstream.

or:

7c: technology develops variously all over a global field, one that mixes technique, infrastructure, know-how, facilities, social frameworks, and social needs. Tech development must be studied this broadly.

This means that the study of the history of technology must become as radically interdisciplinary as technology itself. Economists and historians need to work together regularly. Institutional sociologists need to be there too. So do specialists in cultural and artistic change, which are part of the same process. The intellectual task needs to be seen in all its profound difficulty before it can be resized and broken down enough for progress to be made on its parts, correctly interrelated to one another.

Our innovation group at CNS is particularly interested in the institutional capacities that link research and use. We seek to develop the non-linear development model. Edgerton's book confirms and extends our existing thinking. It teaches a few other things about innovation in general:

• "the twentieth century was awash with inventions and innovations, so that most had to fail. Recognising this will have a liberating effect. We need no longer worry about being resistant to innovation . . . Living in an inventive age requires us to reject the majority that are on offer" (210).
• "we are free to oppose technologies we do not like."
• we are free "to research, develop, innovate, even in areas which are considered out of date by those stuck in passe futuristic ways of thinking."
• while our technology has been highly innovative, our technology policies have not.

Though the non-linear model does not emerge here, Edgerton does build pieces of the foundation for more innovative policies about innovation. Ironically this means seeing innovation as not-so-central to technological change, seeing innovation as common, and seeing the rejection of innovation as essential to sustaining it.

It also means seeing that innovation is often opposed to innovation. In my favorite single sentence in the book, Edgerton says, "calling for innovation is, paradoxically, a common way of avoiding change when change is not wanted" (210). Edgerton is thinking of climate change: perhaps we call for new technology as an alternative to creating the new social arrangements that would truly renew the impacts of technology.

Any valid non-linear model will need to consider the function of publicly funded basic science in government and university laboratories. The issue is given additional urgency by the fact that the American research university system has been the key element in creating the American Anomaly mentioned above, in which the US translated innovation into greater economic gain than did other countries. As Edgerton's work implies, the difference probably does not flow from a linear tech transfer system in which academic research results are rapidly transferred to product developers in the private sector. The decades of high economic growth preceded the modern university-industry tech transfer system, and decades of work in research economics have shown the importance of the public funding of basic research to economic development. The research university mostly likely helped the economy by staying apart from it, and concentrating on basic research aimed at the far horizon, whose payoffs would arrive in 20-50 years.

The Internet is a classic example of the long development process for a transformative technology, which was underway in government and university labs at four decades before its commercial emergence. Nano is another example: as the work of CNS researchers Patrick McCray and Cyrus Mody - among others - has shown, key nanoscale techniques, concepts, goals, and materials have been in development since the 1970s, if not before.

If Edgerton's first theme is the power of the old, and his second is putting use before invention, his third is that innovation is the frequent enemy of progress. Innovation today occurs during "the expansion of a new kind of poor world, a world which has been almost continuously at war, and in which millions have been killed and tortured" (211-12). Technology has done well by war and killing (the titles of two of Edgerton's chapters on R&D). Edgerton adds on the book's final page, "Technology . . . has been responsible for keeping things the same as much as [for] changing them."

The social and cultural study of science doesn't say that the third theme dominates the first two. But it does reject their segregation, and tries to "see technology whole" in its relations to the world overall. It asks us to interpret technology through its social and well as economic effects. Finally, it says to all of us in the nanoenterprise: history will refuse to made any exceptions for nano.