John F. Padgett and Walter W. Powell. 2012. The Emergence of Organizations and Markets. Princeton, NJ: Princeton University Press.
Innovation in the sense of product design is a popular research topic today, because there is a lot of money in that. Innovation, however, in the deeper sense of new actors—new types of people, new organizational forms—is not even much on the research radar screen of contemporary social scientists, even though “speciation” (to use the biologists’ term for this) lies at the heart of historical change over the longue durée, both in biological evolution and in human history. Social science—meaning mostly economics, political science and sociology—is very good at understanding selection, both at the micro level of individual choice and at the macro level of institutional regulation and lock-in. But novelty, especially of actors but also of alternatives, has first to enter from off the stage of our collective imaginary for our existing theories to be able to go to work. Our analytical shears for trimming are sharp, but the life forces that push up novelty to be trimmed tend to escape our attention, much less our understanding. If this book accomplishes anything, we at least hope to put the research topic of speciation—the emergence of new organizational forms and people—on our collective agenda.
One reason, but not the only reason, for this state of affairs is the current hegemony of methodological individualism. If one starts with axioms about actors, then necessarily those become reified exogenous assumptions by the analyst, not dynamic objects of emergence. No theory (including our own) can derive its own axioms. If the axioms of methodological individualism are actors, then those become the black hole of that theory—the part of the theory untouchable to itself.
A second reason for our collective blind spot lies even deeper. We are who we are. Therefore, it is difficult for any of us to see history, including our own, from a perspective other than our own. In particular, humans have a deep cognitive bias, and maybe even self-interest, in seeing ourselves as special. Darwin did not eliminate this bias. Even though now all agree that humans are products of evolution, social scientists (like the rest of humanity) seem fixated on focusing on those aspects of ourselves that seem to differentiate us from that evolution. Consciousness, rationality, language, culture—these are the usual candidates for glorifying and admiring ourselves. We certainly do not wish to deny the existence of any of these fascinating features of mankind. We just wish to interpret these as one among many forms of life. Humans definitely do, but they are not the only ones to produce, to communicate, and to cognize (i.e., information process) in ways that reconstruct themselves through time. This is the alternative axiom of our book: to understand the emergence of novelty, human and otherwise, through thinking about dynamic constructive and re-constructive processes in life.
This axiom might lead some readers to misapprehend us as socio-biologists, evolutionary psychologists, or evolutionary game theorists. Nothing could be farther from the truth. We do not operationalize our axiom by putting genes or pseudo genes (like “memes” or “strategies”) at the foundation of everything. Instead we define “life” as a network-oriented biochemist would define it, not as an atomistic social Darwinian would define it—namely, “life is autocatalysis.” Autocatalysis, in turn, is “a set of nodes and transformations, the interactions among which produce nodes and transformations that collectively reproduce the set” (in the face of fluid turnover and death among elements in the set). To understand the meaning and multivocality of this mouthful, substitute particular instantiations: (a) Biochemical autocatalysis or life is a self-reconstructing ecology of molecules and chemical reactions, in which chemical interactions among the molecules create other molecules that are already represented in the set. (b) Economic autocatalysis or life is an economy of products and production rules (a.k.a., technologies), whose production chains produces other products and technologies, which collectively reconstruct the economy. (c) Social life is a set of people and social relations, the learning interactions among whom reconstruct the types of people and social relations already there. (d) Linguistic life is a set of words and conversations, in which words and the syntactic ways they are assembled reproduce through conversational use.
Autocatalysis is a fundamentally relational and processual view of life, in which the objects that carry life—organisms, people, organizations, languages—are demoted from being Enlightenment-like autonomous agents to becoming transient carriers (almost Petri dishes, albeit sometimes very sophisticated ones) of the reproducing transformational dynamic of life that flows through them all. Three important corollaries follow from this view.
From the perspective of social network theory, the first corollary is that autocatalytic networks are not “pipes,” passively delivering unchanged lumps of material or information from one place to another. Networks instead do production or communication work, transforming the flows through them. What flows through the networks analyzed in this book? Products, words, and people. We call the first type of reproducing flow “production autocatalysis,” the second type “linguistic autocatalysis,” and the third type “biographical autocatalysis.” These three types of autocatalysis overlay and become intertwined into causal feedback (a.k.a., self-regulation) through multiple networks, both in social and in biochemical life. Which networks play such transformational roles? Our theoretical networks are constructed out of production rules and relational protocols.
Production rules or skills are the extant ways in the population under study in which flows of products, words or people are transformed through interaction. Think of this as technology in the case of products, conversation in the case of words, and learning in the case of people. Relational protocols are the extant ways in which interactions in the population under study are formed and broken. Think of this as like markets in the case of products, syntax in the case of words, and social opportunity and roles in the case of people. (Perhaps the word “institutions” could have been used in the last example, but that word has come to have so many disparate meanings that it has lost some of its analytical edge.) “Network evolution,” in this view, means change in the set of production skills and/or relational protocols that collectively reconstruct one another. More observable and measurable changing network topologies and distributions of agents are rooted in this.
A second corollary of autocatalysis is repair. Grab a chunk of an autocatalytic network and throw it into the trash, and the rest of the network may well be capable of reconstructing the destroyed section, through interaction among the remaining parts. This is what resilience is all about in living systems. Because they are autocatalytic systems, bodies can often (not always) repair injuries by themselves. The same is true with natural ecologies. The same is true with human economies. The same with human brains. The same with human cultures. In none of these examples does resilience always mean 100% return to exactly what was there before—though it sometimes can mean that. Often consequential hybridity is introduced in the process of resilient reconstruction from damaging perturbation. But the core point is that without autocatalysis there would be no resilient system there to reconstruct itself in the first place. In the face of death and destruction, either catastrophic or routine, living objects—organisms, people, organizations, whatever—cannot persist through time without the continuous reconstruction and repair of themselves by the autocatalytic networks of life in which they participate.
The third corollary of autocatalysis is a distinction between innovation and invention— Schumpeter revisited and updated by biochemistry. Unperturbed autocatalysis doesn’t lead to novelty. On the contrary, it leads to perpetual fat and happy stasis (in the sense of Prigogyne’s “far from equilibrium” dissipative systems), as long as you feed it. Autocatalytic systems to a first approximation are powerfully resilient, not innovative. They wouldn’t be alive if that weren’t true. That property notwithstanding, a consistent finding in this book—both in agent-based simulations and in our many empirical chapters—is that interaction among and spillover across multiple networks are the keys to understanding relatively rare speciation events in living and evolving systems.
Agent-based models in the first section of this book yielded the surprising finding that autocatalytic processes automatically generate multiple overlapping production networks that interpenetrate each other through multifunctional parts in common. In competition, multiple autocatalytic networks that intersect to reinforce each other are more reproductively stable than single networks operating on their own. This multiple-network topology mimics the “overlapping and redundant control loops” so familiar in genetic regulatory chemical networks. Sociologists will be used to this issue as “the problem of social differentiation” of society into multiple domains. In autocatalysis, differentiation of domains is not as big a deal to explain as Durkheim and Parsons, in their more static formulations, thought it was. “Social differentiation” is just a corollary of life—nothing uniquely human about that.
The reason multiple-network architectures are so important for explaining path-dependent speciation is that one differentiated network becomes a self-sustaining pool for potential “innovations” in another network, with which it is intertwined. Actually in the second network, the production rule or relational protocol in question is boringly routine. Transposed to the first network, however, it could become “innovative” if its incorporation reproduces, which it usually does not. This is a biochemical-network way of rediscovering the “exaptation” idea of Stephen Jay Gould. In autocatalytic multiple-network architectures, multiple networks preserve pools of potential innovation for each other, even as they usually police against transgression.
In this context, we define “innovation” as the transposition of production skills and relational protocols from one autocatalytic network domain to another. In our empirical observation, micro transpositions like this happen all the time, but almost always they goes nowhere—that is, the innovation is not picked up and reproduced—precisely because autocatalytic systems are powerfully resilient. Put more precisely, they are powerfully resilient in their constitutive cores. But around the periphery of “parasites” or “free riders” that are frequently observed in the agent-based models to tail off from autocatalytic cores, autocatalytic networks don’t care much about inconsequential innovation. Creativity at the micro level of individual agents, one might say, is greatly oversold in our Enlightenment culture: it is easy in the periphery of an autocatalytic network, where that doesn’t matter much; but it is incredibly hard in the core of the network, where its ramifying consequences are profound.
“Inventions,” from our perspective, are innovations that tip autocatalytic networks—a much more serious matter. At the micro level, an innovation is no different from an invention; the difference is that inventions cascade out to affect the network arrangement of other rules and domains, whereas mere innovations do not. Think of the difference between the incessant churning of consumer styles (innovation) and the widget or organizational form that changes the way whole industries are organized (invention). The former is just a creative idea; the second is system tipping. For this reason, we focus in our empirical cases on analyzing inventions—their reproductions, their tipping cascades, and their structural vulnerabilities and cleavages—not on mere innovations, which too often are analyses of individual actors out of historical and social context. [Not that our empirical cases aren’t filled with individuals who made a difference in their “agency”; it’s just that those individuals are always situated deeply in social and historical network context. The reactions of that context, as much as the actions of individuals themselves, are shown to determine the difference between innovation and invention.]
The empirical cases of invention in this book span a very wide scope: the biochemical origins of life (chapter 2); the medieval invention of international finance (chapter 5); the Renaissance invention of the partnership system (chapter 6); the Dutch invention of the joint-stock company and stock market (chapter 7); the nineteenth-century construction of Germany (chapter 8); the twentieth-century invention of the central-command economy in Communist Russia and China (chapter 9); emerging post-Communist markets in Russia (chapters 10 and 11) and Hungary (chapter 12); the invention of the biotechnology firm (chapter 13) and its development in regional districts (chapter 14), in career systems (chapter 15) and in normative rules (chapter 16); the evolution of inventor networks in Silicon Valley and Boston (chapter 17); and finally organizational evolution in the open-source computer industry (chapter 18).