[T]he matter with which the chemist deals is the same always: buteconomics, like biology, deals with a matter, of which the inner natureand constitution, as well as the outer form, are constantly changing
A rigorous examination of some common modeling practices in economics may suggest the desirability of moving away from mechanistic models and toward a more evolutionary and institutional approach to economic theory and policy.
Many of the more important standard models in economics are mechanistic. Often inspired by physics, such models represent economic dynamics as the unfolding of a process fully described, up to a stochastic error term, by a master set of equations or an evolution function. The microeconomic models of general equilibrium theory and the macroeconomic models of dynamic stochastic general equilibrium (DSGE) are important examples. In such models, economic dynamics are law-governed in a strong mechanistic sense: in principle at least, there is a set of equations that embodies the system's laws of motion up to an error term. The typical model used in policy evaluation, including macroeconomic policy evaluation, is thought to be a low-dimensional approximation to a possibly high-dimensional reality. If we could have a set of equations correctly describing the high-dimensional reality, it might not be analytically solvable or computationally tractable. The low-dimensional model, however, is a simplified approximation that is constructed to be tractable. Model calibration ensures that the approximation will hold good in future periods, allowing reliable policy recommendations to be based on the model.
This modeling strategy is generally satisfactory only if the predictions such models can generate are reliable enough and specific enough to guide policy. It is generally sufficient only if it is not necessary to describe in detail the institutions governing social and economic life and only if there are no particular limits to prediction in economics. We argue, however, that there is a sense in which economic dynamics are ‘creative’ and that this ‘creativity’ limits the usefulness of standard mechanistic models in economics. If we are right, then standard mechanistic models may have less to say about economic dynamics than their widespread use might suggest. In this case a greater attention to institutional particulars may be in order.
We question the cogency of the standard approach to economic analysis. If economic dynamics are ‘creative’ in a sense we will specify, then there can be no low-dimensional approximation to the high-dimensional unmathematizable reality. The underlying dynamics of the econosphere continually produce novelty and thus change the phase space, that is, the space of pertinent observables and parameters, within which the system unfolds. The phase space of economic activity is not stable. But without a stable prestatable phase space, tractable models will not generate the reliable counterfactuals required for the standard sort of policy analysis. Indeed, no set of equations, whether tractable or not, can be written or track this sort of creative dynamics even in principle.
Our discussion of creative dynamics resembles other discussions of similar themes, including Dopfer (2005), Hayek (1967, 1978), Hodgson and Knudsen (2010), Nelson and Winter (1982), Potts (2012), Shackle (1969), and Witt (2009). Witt (2009) in particular takes a very similar approach. Bunge (2009) emphasizes the incompatibility between novelty and a stable phase space, as do we. Our discussion may have at least three advantages over past treatments. It probably contains a more detailed consideration of the mathematical structure of standard methods than other similar critiques. In particular, we have given more attention to evolution in the phase space. Second, we link our critical treatment of standard models to the evolutionary logic of what Hodgson and Knudsen (2010) call Generalized Darwinism. Finally, our treatment locates novelty production within the process of evolution, rather than viewing it as exogenous or attributable primarily to individual creativity. This last feature of our analysis seems to be unusual or unique, and we will discuss the ‘positive’ implications of this move below. Briefly, this move lets us explain why ‘creativity’ would not come to an end even if it were only the recombination of existing elements. It also draws our attention to what we shall call ‘novelty intermediation’, which relates to the notion of ‘novelty-bundling markets’ discussed by Potts (2012) or the ‘market for preferences’ of Earl and Potts (2004). We will note that governments and businesses, as well as households may seek the services of novelty intermediaries.
We are conscious of the changes in recent decades that have weakened the hold of the old neoclassical orthodoxy of the post-war period. As Colander et al. (2004) have chronicled, mainstream economics has been greatly influenced by complexity theory (also see Durlauf, 2012). It has become more inductive and open. But mainstream economists still tend to be attached to mathematical methods that are not always well suited to a creative economy, as with DSGE models. Much of the mainstream matches our description of orthodoxy only too well, especially in macroeconomics. The tools of complexity theory have helped to push practice in the right direction, but further movement in that direction may be good if economic dynamics are creative in the sense we explain below.
We are also conscious of the variety of senses in which a process may be ‘law governed’ or in which one may speak of ‘economic laws’. Our negative results cast doubt on one sense of ‘law governed’. It says relatively little, however, about other senses in which economic processes might be ‘law governed’, some of which may be compatible with our view that economic dynamics can be ‘creative’ in the sense given below. Indeed, we affirm the existence of an evolutionary logic in the multiplication of goods over time. If evolutionary biology may be said to have ‘law-like hypotheses’, then so too does the sort of evolutionary economics suggested by our analysis. Such law-like hypotheses do not generally come in the form of specific predictions about the future, for example, that a particular sort of species will emerge in a given time and place. They often come in the form of if-then statements. Such if-then statements, however, are generally recognized as ‘laws’ or ‘law-like’ hypotheses.
It may seem unreasonable, even absurd, to reject ‘entailment’, ‘law’, and ‘cause’. In each case we do so only for a narrow and precisely defined sense of the term. Even though we reject only narrowly circumscribed senses of ‘entailment’, ‘law’, and ‘cause’, we are indeed suggesting a departure from customary modes of thinking about social processes. This departure does not somehow imply that the social world is a magical place where anything can happen and systematic thinking is futile. We affirm the existence of recognizable regularities in social processes such that a science of society is possible. The methods appropriate to that science, however, should not rule out the continual production of novelty and the resulting continual change in the space of observables and parameters within which the system unfolds.
If economic dynamics are ‘creative’ in some sense, then standard mathematical models cannot reliably predict the consequences of different policies, at least at the level of specificity such models often aim at. Our criticism raises the question of how economics should be done. We discuss what an economics for a creative world might look like. Our discussion is tentative and suggestive. We do not think we can predict the future course of science, which is, after all, a creative process. Nor would we pretend to dictate to others how they must think and what they must say. Having criticized orthodoxy, however, we have an obligation to show that the alternative to that orthodoxy is not silence.
The German historical school and the original American institutionalists claimed that, because of the historical specificity of economic institutions, there were few (if any) universal economic laws (Caldwell, 2005; Hodgson, 2001). But they failed to provide a developed theoretical framework to deal with this problem. Our analysis may contribute to a theoretical framework that is rigorous and scientific without assuming the sort of ‘universal law’ that fits standard modes of explanation in physics. Our analysis is more metatheoretic framework than applied theory. Institutions will feature strongly in any applied theory, however, that fits our framework. We develop this institutional point in the concluding section.
The Sun is new everyday (Heraclitus).
2. No entailing laws
Economic dynamics are generally treated as ‘law governed’ in the sense that there is a master set of equations or an evolution function describing the dynamics of the system.1 Economic theory, on this conception, can be loosely compared to a computer that has been programmed to execute this master set of equations or compute the intended function, or some approximation thereof. The economist feeds a description of initial conditions into the computer and the computer spits out the future path of the system. When the economist feeds in the hypothetical initial conditions a given policy would create, the computer spits out the future path the system would take if the policy were adopted.2 We may say that the equations of motion and initial conditions ‘entail’ the subsequent evolution of the system, and the equations or evolution functions give the ‘entailing laws’ of the system. Entailment in this sense is a strict condition much like logical entailment: everything that happens as the system unfolds was already implicit in the initial conditions and the assumed laws of motion.
This way of putting things may create the impression that economists are insensitive or unsophisticated about what is predictable. As we noted in the introduction, however, the typical model used in policy evaluation is thought to be a low-dimensional approximation to a possibly high-dimensional reality. The low-dimensional model is a simplified approximation that is constructed so as to be tractable. If economic dynamics are creative in the strong sense we explain below, then this strategy may not work. To help bolster this claim, it may be helpful to distinguish three levels of unpredictability.
First, consider the mathematical description of a double pendulum or of the logistic function, that is the discrete time, continuous space solution of the famous Lotka–Volterra prey-predator model. In both cases, there is a problem with implementing the system on a digital computer. The discrete computational trajectories that can be programmed are necessarily approximations that do not follow the continuous trajectories of the system's equations of motion. When we compute the discrete logistic function, a difference at the 16th decimal (an excellent approximation) is sufficient to produce, after only 50 iterations, different results whose distance covers the full phase space; thus, all information about the system is lost. It would be an error in such cases to think that the system trajectory can be computed: neither the continuous mathematical trajectory, nor the intended physical process is ‘followed’ or ‘shadowed’ by the digital computation, even vaguely. (The approximation, at best, goes the other way round, see Pilyugin 1999.)
Second, and more generally, non-linear equations in a pregiven phase space can produce non-analyticity. The derivatives may diverge so that there is no solution. This situation creates bifurcations and homoclinic trajectories, so that we cannot predict or follow the continuous trajectory even in principle. Minor fluctuations or perturbations below observability may break the symmetries (forcing the planet in one direction of the bifurcation or away from the unstable trajectory) in a totally unpredictable way. This holds even apart from considerations of what is computable on a given digital computer. This case has been known since Poincaré's analysis of the ‘three body problem’, in which two planets orbit a sun and the dynamics are entailed by Newton's equations. Note that all processes that contain some ‘interactions’ (e.g., the interacting weights in a double pendulum or two planets revolving around a sun) are soundly described by non-linear mathematical systems that present computational problems.
Third and finally, there is our case of a system without a pregiven or prestatable phase space, which we claim characterizes both biological and economic evolution. In such cases both observables and parameters change and we cannot even write entailing equations.3
In short, ‘algorithmic’ means ‘deterministic and predictable’. There are plenty of deterministic systems of equations that are not predictable and thus not algorithmic: typically, the non-linear systems we mentioned above. We claim that the econosphere is not even deterministic let alone predictable. The lack of pregiven and prestatable phase space means that we cannot even write the equations that would determine the system.
In the more standard conception of economics, it becomes an essential job of the economist to identify the right set of equations to support prediction and control of the economy. This approach to economics seems to have failed in the Great Recession, provoking Queen Elizabeth to ask a group of economists, ‘Why did nobody notice it?’ (Pierce, 2008). The British Academy's response to the Queen suggests that mainstream economists are still thinking in terms of entailing laws. ‘Everyone seemed to be doing their own job properly’, they told the Queen, ‘and often doing it well. The failure was to see how collectively this added up to a series of interconnected imbalances over which no single authority had jurisdiction’ (Besley and Hennessy, 2009). Rather than questioning the dynamics of the econosphere, this answer questions the organization of economic authorities. If we had had a better organization amongst ourselves, the whole thing could have been prevented.
For economics to be law governed in the sense we have indicated requires that there be a set of equations entailing system dynamics. As we have argued elsewhere (Felin et al., 2013) there can be no entailing laws for the econosphere. Our argument is that of Longo et al. (2012), for biological evolution.
In law-governed systems, we can prestate the configuration space or phase space, which is given by the set of pertinent observables and parameters. In standard economic models, the system's laws of motion determine the paths of certain variables such as prices and quantities. Typically, each of the n endogenous variables corresponds to an axis in n-dimensional Euclidean space. The system unfolds within that n-dimensional ‘phase space’. At least some of the endogenous variables of a law-governed system must be observable if the model is to be tested or calibrated. And the endogenous variables must be known and listed ahead of time for the analyst to write down the equations. Time plays at most a limited causal role in such a system. Ergodic systems ‘go everywhere’, so that time and history play no role in the system's dynamics. Spin glass models such as Minniti (2004) exhibit non-ergodicity. But all possible states exist in a stable phase space and are not qualitatively different from one another. Time matters in a weak sense, but the possible paths the system can take are all predetermined before the dynamics unfold. At most, as in Statistical Mechanics, the phase space may increase in dimensions, but the observable properties in the new dimensions are pre-given.
The notion of general equilibrium from economics and phase space from physics share many similarities. Both view systems as bounded, where the space of all possible actions or positions can be delineated a priori. The assumption in this sort of economics is that actors omnisciently calculate and compare all possible actions, including future ones, while in thermodynamics and statistical mechanics the assumption is that the trajectories – positions and momenta – of particles cannot be accounted for within the phase space except in statistical terms. Both general equilibrium and phase space assume that all future goods or states can be mapped and calculated (Arrow and Debreu, 1954). While general equilibrium theory has of course looked at the ‘process’ (or temporal factors) through which markets get to equilibrium – uncertainty, tatonnement, search, and learning (for a recent review, see Balasko and Geanakoplos, 2012) – nonetheless the notion of equilibrium remains central and strongly wedded to phase space-like mathematical formalizations.
But in economic evolution the phase space itself changes continually in ways that cannot be prestated. The Turing machine ultimately has yielded the World Wide Web, selling on the Web, Web browsers, and iPads – none of this was prestatable in Turing's time. Because of this continual change in the phase space, time matters in economic systems.4 For example, the list of traded goods has grown radically over time from zero for our pre-human ancestors, to a small handful at about the time biologically modern man appeared (Marwick, 2003: 78), to the multitudes of the modern global economy. Beinhocker (2007: 9, 456–457) estimates that the Yanomami have about 300 distinct goods, whereas there may be about 10 billion distinct goods for sale in New York City. This multiplicity of goods is important enough to have its own name. The Latin for trade is cambio. Thus, we might call the increase in the number of traded goods ‘cambiodiversity’. Cambiodiversity is analogous to biodiversity; it is diversity in the things traded. And it is increasing cambiodiversity that has caused an increase in wealth. Adam Smith (1776) understood this connection when he said, ‘It is the great multiplication of the productions of all the different arts, in consequence of the division of labour, which occasions, in a well-governed society, that universal opulence which extends itself to the lowest ranks of the people’ (I.1.10). More recently, Hidalgo et al. (2007) have produced evidence suggesting that cambiodiversity is correlated with wealth.
The continual multiplication of goods to trade means continual change in the space within which economic dynamics unfold. Recall that we cannot calibrate or estimate our model unless at least some of the endogenous variables are observable. In economics the observables are prices and quantities or aggregates of them. If we are tracking prices and quantities, new goods change the set of variables we are tracking and change, therefore, the phase space. Tracking aggregates does not solve the problem. The arrival of new goods, and new uses for old goods, implies the need to change our aggregation method, which may imply a change in the behavior of our aggregates and, therefore, a change in the system's laws of motion.5
There is an evolutionary logic to this multiplication of goods over time. At a given moment humans in an exchange network trade certain goods. Those goods in that network create what Kauffman has called an ‘adjacent possible’. Certain things are possible and others not from a given space or state. Powered heavier-than-air flight was not part of the adjacent possible 100,000 years ago or even 150 years ago. By 1900, however, we had relatively light, reliable, and powerful internal combustion engines. It seemed clear to many would-be innovators that such engines could be mounted, somehow, to wings to create a powered flying machine. Thus, powered heavier-than-air flight was part of the adjacent possible for the Wright Brothers. At any moment, the econosphere's adjacent possible contains opportunities for profitable innovations as well as false opportunities that may tempt action. There are many such opportunities, only some of which will be acted on. Once that has happened, the attempted innovations – some successful and others not – become what we want to call ‘enabling constraints’ that will change the system's adjacent possible, enabling a different set of profit opportunities. Only some of the profit opportunities in the new adjacent possible were present in the old adjacent possible.
There is some tendency for this process to create larger exchange networks, more goods, greater complexity, and greater wealth. The tendency is not an iron law. Institutional changes in ancient Rome, for example, brought about a collapse of trading relationship and, eventually, the payment of taxes in kind. Presumably, the number of traded goods shrunk correspondingly. Nevertheless, there is an evolutionary logic behind the general trend toward increased cambiodiversity. Adam Smith (1776) identified several reasons for the growth of cambiodiversity. The division of labor increases the ‘quantity of work’ (I.1.5) in part because specialization stimulates technological change (I.1.9). Moreover, the process can be self-reinforcing because ‘the division of labour is limited by the extent of the market’ (I.3). More, and deeply important, can be the emergence of ever new economic opportunities created by an ever-increasing cambiodiversity itself.
Biological evolution shows similar properties. As shown by Gould (1998) and mathematically explained in Longo and Montevil (2012), phenotypic complexity grows over time, with no aim or direction. This growth takes place along random paths, produced, in an unbiased way, by an asymmetric diffusion of bio-mass over complexity. The asymmetry informally corresponds to already-occupied niches, beginning with the original bacteria, as with species that ‘reproduce with variation’ (the fundamental principle by Darwin). This creates new niches, as we discuss below.
In his meditation on Adam Smith's ‘dictum’ that ‘the division of labor is limited by the extent of the market’, Young (1928: 533) noted that the extent of the market is endogenous to the market process. Each refinement in the division of labor increases wealth, which is an increase in the extent of the market allowing a further refinement in the division of labor. (Endogenous technological change is a part of this self-reinforcing process.) Thus, ‘Adam Smith's dictum amounts to the theorem that the division of labour depends in large part upon the division of labour’. Far from empty tautology, Young explains, this point shows how economic growth can be self-sustaining.
Not only new or adventitious elements, coming in from the outside, but elements which are permanent characteristics of the ways in which goods are produced make continuously for change. Every important advance in the organisation of production, regardless of whether it is based upon anything which, in a narrow or technical sense, would be called a new “invention,” or involves a fresh application of the fruits of scientific progress to industry, alters the conditions of industrial activity and initiates responses elsewhere in the industrial structure which in turn have a further unsettling effect. Thus change becomes progressive and propagates itself in a cumulative way. The apparatus which economists have built up for the analysis of supply and demand in their relations to prices does not seem to be particularly helpful for the purposes of an inquiry in these broader aspects of increasing returns. (Young, 1928: 533).
Young describes a system in which regime change is continual. Each change in regime is a toothing stone for the next. Each regime change enables the next. This enablement represents a complex mix of interactions between enabling institutions, actors, and entrepreneurs. The overall process of increasing cambiodiversity is similar to the process of increasing biodiversity. Consider two more or less parallel examples.
In biology, the swim bladder is an adaptation giving some fish neutral buoyancy. The bladder, which probably evolved from the lungs of a lungfish, contains air and water in the right mix to produce such buoyancy. Apparently, water got into some lungs producing a sac with air and water that was then poised to become a swim bladder. This ordinary case of evolutionary adaptation brought about a new function in the biosphere, neutral buoyancy, which then changed the process of evolution, giving the system new species of fish with new swim bladders and new proteins. More importantly for our story, this new function and adaptation, created new evolutionary niches for other organisms. A worm or bacterium might evolve to live exclusively in swim bladders. The swim bladder thus changes the future possible evolution of the biosphere. It is not possible to list such biological innovations ahead of time as they result from the superposition of the unpredictable new ‘niche’ and the variation of a bacterium or worm produced by random changes, possibly at the molecular level (mutations or the like).
Following Darwin's two fundamental principles, reproduction with modification and selection (negative: excluding the incompatible; positive: enabling the compatible), we understand the phenomenon as follows. Random variability is at the core of reproduction; a ‘hopeful monster’, in Goldschmidt's sense, with a new and a priori, ‘pathological’ bladder, turned out to be compatible with a contingent environment and was thus positively selected. Given the swim bladder's existence, a new adjacent possible empty niche, or opportunity, emerged, and a variant of a worm or a bacterium then became a possible parasite or symbiote.
Plants provide a paradigm for the extended creativity of evolution. Random specific and interspecific hybridation continually creates new species. Now, most (or all) species have specific fungi or prokaryotes as parasites or (often essential) symbiotes. They are specific as they did not exist before the new plant species. More generally, plants largely influenced animals’ evolution; for example, around 140–100 million years ago, the formation and diffusion of angiosperms with their flowers enabled pollinator insects, by forming previously non-existing niches.
In short, new species and new niches enable yet more species and niches. Moreover, this randomly increases the complexity of the ecosystem, increasing both the diversity of the biosphere as well as phenotypic complexity. Most of plants and animals’ plasticity is based on genome and epigenetic effects; the behavioral plasticity of humans is almost exclusively due to a huge, neotenic, highly adaptive brain.
Something similar happens in economic evolution, where human brain and history matter crucially. Consider the emergence of BookScan, a Japanese company that scans books and libraries into iPads. The founder, Yusuke Ohki, wanted to reduce the clutter created in his apartment by the 2,000 books of his personal library. He solved the problem by scanning them into his iPad. Ohki realized he had stumbled into a business idea and founded BookScan, which quickly took off and generated competitors in the Tokyo market (Eki and Alpeyev, 2011).6 The makers of iPad were probably not thinking about crowded Tokyo apartments when they perfected the tablet computer. And yet their innovation unintentionally created an economic niche Ohki was able to enter. Apple's innovation changed the set of economic niches in the econosphere's adjacent possible in very much the way the swim bladder changed the set of biological niches in the biosphere. This transformative effect of the iPad exists in some degree for more or less any economic innovation, just as the argument in Young (1928) suggests. In this sense economic dynamics are ‘creative’. At each step, new opportunities, new economic niches, are created in the adjacent possible. Successful innovations are entries into these newly created niches. The process itself continually generates new possibilities. The process itself generates novelty; it is ‘creative’. In this sense we may speak of the ‘creative dynamics’ of the econosphere.
Our emphasis on creative dynamics is an emphasis on emergence. We are suggesting strong parallels between emergence in the biosphere and emergence in the econosphere. We emphasize the emergence of complexity and, especially, novelty. Harper and Lewis (2012) review the variety of meanings ‘emergence’ has in modern economics. They note the link to complexity, and they note that only some conceptions of emergence, such as those common in evolutionary economics, imply the emergence of novelty. See also the rest of the volume, which they edited, for important ‘New perspectives on emergence in economics’.
Positing that the new goods are all somehow ‘out there’ in a stable, high-dimensional prestatable phase space waiting to be created cannot solve the problem. In this solution, we would model the system as tracing out a relatively low-dimensional path in a very high-dimensional but prestated phase space. The trouble is that we cannot list elements in this imagined high-dimensional phase space. Consider Beinhocker's estimate of about 1010 goods in New York today. This number of realized goods is much smaller than the number of possible goods existing in the imagined high-dimensional phase space. And yet to list all of them at the rate of one per second would take 317 years even if we did not stop to rest along the way! Even if we had all the time in the world, or rather much more than that, there is no algorithm that can list all the future economic possibilities, let alone compare them.
To provide an example: consider the uses of a screwdriver. No algorithm can list all possible uses of a screwdriver. One use, for example, is to serve as an antenna for a short wave radio. This usage we can list today. But to list it in 1850 would have required that one anticipate the emergence of radio transmission. The number of uses of a screwdriver is indefinite and these uses are not orderable. Therefore no effective procedure, or algorithm, can either list all the uses of a screwdriver or, in general, provide a means to find a new use.
Because we cannot prestate the ever-changing and emergent phase space of economic evolution, we have no settled relations by which we can write down the equations of motion of the ever new economically relevant observables and parameters revealed ex post by the market process. Nor can we prestate the emergent opportunities of the system as boundary conditions. Thus, we could not integrate the equations of motion even if we could somehow have them.
If the above is true, no laws entail the evolution of the econosphere. And if by ‘cause’, we mean what gives a differential effect entailed by law, then we can assign no cause in the evolution of the econosphere. In this sense, the economy is not a causal system. The past does not ‘cause’ the future so much as it enables some futures and disables others. The main reason for this is that, in theories of the inert (classical physics), the default state is inertia and, thus, nothing happens with no causes; while the ‘default state’ in biological evolution is Darwin's ‘reproduction with variation’ and, in human history, it may be fair to consider ‘activity with variation’, as a default state (Longo et al., 2014). Thus, ‘enablement’ is a key notion in our analysis. Past innovations enable future innovations, but do not cause them. Even more so: new niches are enabling constraints that enable innovations, which create new niches.
In the strict sense of ‘cause’ just given, novelty and innovation are uncaused. Other economists have expressed similar notions. Knight (1921: 221) considered ‘the possibility that “mind” may in some inscrutable way originate action’. Shackle (1972: 122) also described choice as ‘originative’. Mises (1966: 18) took a similar view when explaining the principle of methodological dualism. Hodgson (1993: 219) describes choice as an ‘uncaused cause’. If there are no entailing laws, then some events will be less than fully entailed without being, therefore, arbitrary. Shackle (1979: 20) has captured the point well: ‘The anarchy of Nature is as fatal as the determinacy of Nature to the notion of choice as the source of history’. Our focus is somewhat different because novel possibilities emerge in the adjacent possible even before they are discovered by (possibly creative) minds. Like Knight, Shackle, Mises, and Hodgson, however, we articulate a carefully delimited sense in which economics must allow for uncaused causes.
Although the economy is not a ‘causal system’ in our sense, cause and effect still operate in the econosphere, just as they do in biosphere. Darwinian principles let us predict broad patterns such as the emergence of organisms resistant to pesticides and antibiotics. And it can explain particular past events, such as the evolution of the mammalian eye. Similarly, in economics we may be able to predict certain broad patterns and to explain past outcomes. But relatively specific predictions may be out of reach.7
Our summary paraphrases Longo et al. (2012):
In law-governed systems such as physics, we can prestate the configuration space or phase space. Dynamics are geodesics within such prestated phase spaces – at most the dimensions of the phase space may change, as in statistical physics, but the properties of the observables are prestated. That is, the path of each variable, which depends on the paths of all other variables, minimizes some metric in a predefined space.
In economic evolution, the phase space itself changes continually and in ways that cannot be prestated, since the enablement relation allows the formation of new observables.
Because we cannot pre-state the ever-changing phase space of economic evolution, we have no settled relations by which we can write down the equations of motion of the ever new economically relevant observables and parameters revealed ex post by the market process. Nor can we prestate the emergent opportunities of the system as boundary conditions. Thus, we could not integrate the equations of motion even if we could somehow have them nor can we write down a well-defined mathematical function describing (determining) the dynamics.
If the above is true, no laws entail the evolution of the econosphere.
If by ‘cause’ we mean what gives a differential effect entailed by law, then we can assign no cause in the evolution of the econosphere. In this sense, the economy is not a causal system.
The past does not ‘cause’ the future so much as it enables some futures and disables others. Thus, ‘enablement’ is a key notion in our analysis.
We might note that some attempts have been made to frame new or changing observables in ‘second-order dynamics’, that is dynamics of phase spaces, a very interesting approach to ‘viability’ (see, for example, Nation et al., 2002). However, while part of the ‘first-order’ observables serve as parameters of the ‘second-order’ systems, where they determine the dynamics of second-order observables, the space of the latter is functionally pre-given, in order to allow the mathematical writing of equations and evolution functions.
In Longo et al. (2012; also see Longo and Montevil, 2014), the construction of the phase space in mathematical physics is closely examined: it depends on the invariants that are observed in trajectories. Since Boltzmann and Poincaré, these were transformed into a ‘background space’ for the complete equational determination (typically: momentum, as observable, with respect to position, as parameter; energy, as observable, with respect to time, as parameter). This is exactly the conceptual operation that those authors claim to be unfeasible in biological evolution and that we claim to be unfeasible in economic analysis, since these very observables dynamically change and new unprestatable ones are continually created. When the double jaw of some vertebrates was transformed into the bones of the middle ear of later vertebrates (Gould's preferred example of ‘exaptation’, adaptation ex post), a new unprecedented observable, most probably a consequence of a random molecular or epigenetic event, is formed, around 250 million years ago, and changed the niche and the ecosystem.
When those arts that proceed from design come into competition and theircraftsmen work in rivalry, without doubt the good intellects, exercisingthemselves with much study, discover new things every day inorder to satisfy the various tastes of men (Giorgio Vasari).
3. Novelty intermediation
If economic evolution is ‘creative’ in the sense we have indicated, then novelty is ubiquitous in economic affairs. Novelty can be squelched by a sufficiently rigid institutional system, such as the Stalinism of the old Soviet system or the manorialism of medieval Europe. But even a squelched system produces some novelty as illustrated by Sputnik or the flying buttress.8 More open systems generally produce more novelty.
We are, of course, not the first scholars to note that economic systems generate novelty. Adam Smith (1776) attributes the greater productivity of the division of labor in part ‘to the invention of a great number of machines which facilitate and abridge labour, and enable one man to do the work of many’. He notes that ‘the invention of all those machines by which labour is so much facilitated and abridged, seems to have been originally owing to the division of labour’. He further notes the emergence of ‘philosophers or men of speculation, whose trade it is not to do any thing, but to observe every thing; and who, upon that account, are often capable of combining together the powers of the most distant and dissimilar objects’. In this passage Smith adopts the theory that creativity is recombination. Campbell (1960) and Koestler (1956) are frequently cited examples of combinatorial theories of creativity. In his review of ‘creativity research’, Mumford (2003) emphasizes the importance of ‘conceptual combination’ in ‘creative thought’. Witt (2009) adopts a combinatorial view, describing novelty generation as a process in which ‘conceptual inputs. . .are recombined’ (Witt, 2009: 371).
The combinatorial view of creativity may raise a puzzle. Why don't we run out of elements to combine? This concern has been expressed in the economics literature. Cowen (2011) argues, for example, that we have reached a ‘technological plateau’, creating ‘The Great Stagnation’ (location 64).9 Schumpeter's notion of the ‘mechanization of progress’ (Schumpeter, 1947: 131) might be another example. Schumpeter imagined that ‘capitalist evolution – “progress” – either ceases or becomes completely automatic’ (p. 134). A ‘completely automatic’ process may be ‘innovative’ in some constrained sense, but it seems fair to say that a mechanized system cannot produce ‘true’ novelty.
Our argument shows that there is no particular reason to expect potential novelty production to decline as we try out more and more combinations. If the system's dynamics are ‘creative’ in our sense, then it constantly generates new elements (and new uses of old elements) that might be combined and recombined in novel ways. Thus, even if recombination ‘can easily be imagined to be carried out mechanically or by an alpha-numeric algorithm’, as Witt (2009: 364) suggests, the elements being ‘mechanically’ recombined are new and could not be listed algorithmically in advance. Witt makes a further point that there is no algorithm that could describe or replace the ‘interpretive operation’ that discriminates between combinations ‘that do make sense and those that do not’. Thus the problem is not so much the lack of things to be combined, but rather the specification of a mechanism for explaining which of the many elements ought to be combined (Felin and Foss, 2011).
While generally benefiting from novelty, the people in systems with more novelty will have more difficulty coping with it. Following Earl and Potts (2004) and Potts (2012), we should expect to see market mechanisms for coping with novelty. Earl and Potts (2004) discuss the ‘market for preferences’ and Potts (2012) discusses ‘novelty bundling’. In both cases they largely restrict their attention to the phenomenon of businesses helping households to cope with novelty. We will extend their insights to markets more broadly and speak of ‘novelty intermediation’. Households, businesses, and states must cope with novelty and may be willing to pay a specialist to help them do so. A novelty intermediary has specialized knowledge about an area in which novelty seems to matter, digital technology for example, and updates that knowledge frequently. This intermediary transfers such knowledge to its clients or otherwise helps them to cope with novelty in the area of its specialization. Most economic theories and models in which novelty matters will anticipate in some degree the notion of novelty intermediation, but Earl and Potts (2004) and Potts (2012) may be the first clear and reasonably complete statements of the basic idea.
Potts (2012: 295) illustrates novelty bundling with ‘fashion magazines such as Vogue’, which present readers will novel combinations of fabrics, clothing items, hair styles, and so on. Earl and Potts (2004: 622) point to ‘product review websites such as Amazon.com’. They note that novelty production creates a ‘framing problem’, which has been ignored in standard microeconomics. In standard microeconomic theory ‘the frame given by preferences and prices is always unambiguous’. But consumers coping with novelty cannot be expected to have a complete set of preferences across all possible bundles of market goods. Some of those goods are new and it is not clear what combinations will best satisfy the consumer's preferences. Analytically, Earl and Potts view ‘high-level preferences’ as given and known to the consumer. But a ‘market for [low-level] preferences’ is necessary if the consumer is to translate high-level preferences into a market plan. In the market for preferences experts help consumers negotiate novel products and novel product combinations. They argue that ‘the fundamental knowledge problem that an evolving economy poses for boundedly rational consumers is solved to a significant degree by specialized learning rather than by individuals learning in a self-contained manner across the whole compass of their lives’ (p. 631).
Novelty intermediation is not about the diffusion of innovations. It is about finding the innovation in the first place. In our model of creative dynamics, novel possibilities emerge in the adjacent possible. Novelty intermediaries know how to discover the new knowledge required to innovate. There is a structure to the intermediary's knowledge of some changing field and this structure of knowledge allows it to discover the novel possibilities of the adjacent possible, including novel combinations of elements, some of which may be old and others, perhaps, new. It is not correct to equate this discovery process with the diffusion of innovations because it is about discovering the innovation rather than either adopting the innovation or the rate at which a given innovation spreads. Nor does novelty intermediation deal with the emergence of novel possibilities in the adjacent possible. It deals with process of discovering such possibles and turning them into actuals.
Firms and governments are no less in need of novelty intermediation than consumers. It seems reasonable to guess that if economists more fully absorb the ideas of novelty production and creative dynamics they might see familiar phenomena in the new light of novelty intermediation. From this perspective we might begin to reconsider, for example, the role of venture capital (VC) in rapidly changing industries. Using data on European firms in ‘medium and high-tech manufacturing and services industries’ Croce et al. (2013: 495, 503) produce evidence that ‘non-financial value added provided by VCs’ is ‘the main driver of the better performance of VC-backed firms’. We may conjecture that the ‘non-financial value added’ is often novelty intermediation, at least when the VC firm specializes in a rapidly changing industry. The VC may be bringing more than capital and general expertise to the startups. They may be guiding the startups through the novel possibilities of the industry's adjacent possible.
Presumably, VC firms in Silicon Valley have been providing novelty-intermediation services to startups. The description given by Florida and Kenney (1988: 311) seems to support that view. ‘Between 1968 and 1975’, they report, ‘approximately 30 new or reconstituted venture capital operations were established in the Bay Area’. This period saw ‘the emergence of the limited partnership’ VC firm ‘with professional venture capitalists managing capital provided by passive outside investors’. Importantly, ‘Over time, a growing group of former entrepreneurs, past employees of venture firms, and outside personnel were able to attract financial resources and launch limited partnerships’. The emergence of VC partnerships that included former entrepreneurs seems to be evidence that such firms were acting as novelty intermediaries.
Renaissance art patronage might also be reconsidered in the context of novelty intermediation. Nelson and Zeckhauser (2008) take it for granted that the artists are the experts hired by the patrons who are not experts. But they also point out that some patrons were well informed and had good judgment. We suggest viewing patrons as novelty intermediaries providing a ‘non-financial value added’ to artists. Patrons were ‘combining together the powers of the most distant and dissimilar objects’ by simply associating certain artists with one another. Gilbert (1998: 395, 396, 398, 405, 436) notes that humanist experts might give artists advice or instruction on iconography and symbolism. In some cases patrons provided such expert advisors (see Gilbert, 1998: 395, 396, 405). Krautheimer and Krautheimer-Hess (1956: 171) say ‘calling upon’ a ‘scholar to submit a learned program’ for a commissioned work was ‘established custom’.10
According to Hollingsworth (1994), ‘One of the most forceful myths of the Renaissance is the idea that its artists freely explored their ideas and created their masterpieces for enlightened patrons eager to acquire these works of genius’. Rather, ‘it was the patron who was the real initiator of the architecture, sculpture, and painting of the period’. The patron ‘played a significant part in determining both form and content. Fifteenth-century patrons were not passive connoisseurs: they were active consumers’. In a remark significant for us, Hollingsworth says, ‘It was the patron, and not the artist, who was seen by his contemporaries as the creator of his project and this gave him the strongest possible motive for controlling its final appearance’ (p. 1, 2). At the time, patrons were recognized as creative actors in the arts. Hollingsworth's model of the active patron strengthens the view that patrons may often have been acting as novelty intermediaries and experts in the adjacent possible of relevant areas.
Mateer (2000: ix, x) reinforces Hollingsworth's model of Renaissance patronage. It is ‘demonstrably a romantic myth’ that ‘artists worked in an atmosphere of creative and intellectual freedom, producing masterpieces that were received passively by enlightened patrons’. This romantic view is inconsistent with ‘the hierarchical structures of a typical Renaissance court’ in which ‘every aspect’ of life ‘was designed as a manifestation of its “prince” and an embodiment of his, or her, values’. The princely patron ‘would naturally pay close attention to the development of any commissioned work and expect to participate in the creative process as an active collaborator’ (p. ix).
As discussed by Flap (1990), patronage itself might be seen as an ‘institution in its own right’. Thus the whole system of interaction between artists, entrepreneurs, and patrons represents far more than just a simple exchange relationship, but rather a complex institution that enabled the generation of novelty.
Long (1997: 3) emphasizes the political and, especially, military functions of patronage. She says, ‘in the fifteenth century the practice and representation of rulership came to be closely associated in particular ways with technological power and the mechanical arts’. At that time, Long explains, ‘Rulers, princes, and military captains who wanted to consolidate their power achieved legitimation through the remodeling of urban space, the construction and decoration of great palaces, and the creation of painting, sculpture, and other material artifacts’ (p. 4). Importantly, the practice of ‘military leadership came to be closely associated with armaments and techniques’ (p. 5). The successful Italian condottiero such as Ludovico Sforza, the Duke of Milan, had to be familiar with the arts and with their corresponding technologies. Long relates a fanciful story from Filarete's Treatise on Architecture in which the architect, the lord, and the lord's son are dining together as equals. The lord's son becomes so taken with the architect's knowledge that he pleads with his father to let him study under the architect. Long says, ‘He wants to know how metal is melted, [and] how furnaces for melting bronze are made’ (p. 33). While fictional, the story strengthens the view that some patrons such as Ludovico Sforza had to have some technical knowledge of the arts, and that they sometimes acted as novelty intermediaries to defend and expand their power, including their military power.
The view that some patrons were producers and novelty intermediaries is bolstered by recognizing, as Long (1997) does, the multiple roles of artisans in the political and military life of an Italian city in the Renaissance. Architecture has an important military function. The brothers Giuliano and Antonio da San Gallo received many commissions for fortifications and war preparations (Vasari, 1568: 696–710). Vasari (1568: 74) recounts the story of how the sculptor Nicola Pisano (1250–1278) found an ingenious method to safely ‘throw to the ground many towers’ in Florence. The political and military importance of this work of demolition is reflected in Vasari's remark that these towers were important in the many ‘brawls that were often taking place between the Guelphs and the Ghibellines’ at the time (p. 74). Leonardo's war machines are well known. He may have been the inventor of the wheel lock system of firearms ignition (Foley et al., 1983). Whether Leonardo's designs for war machines were unoriginal, as Gille (1966) suggests, or innovative, as Foley et al. (1983) suggest, it remains true that ‘Leonardo labored on military devices for Ludovico Sforza’ (Foley et al., 1983: 399).
Gilbert (1998: 392) opposes the view that ‘Renaissance patrons usually kept creative control over works they commissioned’. But if patrons were novelty intermediaries, then they would be collaborating with the artists and not giving them orders. As Gilbert (1998: 397) notes, Sassetta's 1439 contract for the altarpiece of the church of San Francesco in Borgo San Sepolcro includes the stipulation that the artist will compose and arrange the figures and stories of the work ‘as seems fitting to us and the master together, etc’ (Banker, 1991: 54).11 This passage reinforces the view of patrons as collaborators who would likely have, as Gilbert (1998) documents, varying degrees of involvement and control depending on such varied considerations as the patron's knowledge, the purpose of the work, and the artist's stature.
Vasari's (1568: 465) description of Francesco di Giorgio Martini (1439–1502) is revealing.
Francesco was a very able engineer, particularly in connection with military engines, as he showed in a frieze that he painted with his own hand in the said palace at Urbino, which is all full of rare things of that kind for the purposes of war. He also filled some books with designs of such instruments; and the Lord Duke Cosimo de’Medici has the best of these among his greatest treasures. The same man was so zealous a student of the warlike machines and instruments of the ancients, and spent so much time in investigating the plans of the ancient amphitheatres and other things of that kind, that he was thereby prevented from giving equal attention to sculpture; but these studies brought him and still bring him no less honor than sculpture could have gained for him. For all these reasons he was so dear to the said Duke Federigo, whose portrait he made both on medals and in painting, that when he returned to his native city of Siena he found his honors were equal to his profits.
Notice that Francesco's military knowledge, like his knowledge of sculpture, was linked to ‘the ancients’ and the study of their achievements. Vasari does not clearly distinguish between the sort of things we might now label ‘engineering’ or ‘technology’ and what we might now think of as ‘high art’ or ‘pure art’. The ‘painters, sculptors, and architects’ were united in their practice of ‘design’ (disegno), which meant, at root, drawing. The person who could draw, could design a beautiful fresco, decorations for a public festival, a new bridge for commercial traffic, or war machines. Some patrons added value to the artists they sponsored through at least three forms of novelty intermediation. First, they had collections of ancient artifacts to which these artists gained access. Second, such patrons brought together complementary artisans who could learn from and emulate one another. Third, they brought humanists and artisans into collaboration. The efforts of these artists, in turn, added value to the political projects of such patrons, including war making.
Vasari tells a famous story about Giotto that illustrates our suggestion that patrons were sometimes innovation intermediaries. According to Vasari (1568: 102–103), Pope Benedict (the IX), who was planning ‘to have some paintings made in S. Peter’s’, sent one of his ‘courtiers’ to investigate the renowned artist. When the courtier asked for a sample of Giotto's work, the artist responded by drawing a perfect circle freehand and giving it to the courtier, who was embarrassed by the seeming slight. Upon seeing the drawing and learning that it was done without a compass, the Pope ‘recognized by this how much Giotto surpassed in excellence all the other painters of his time’ (Vasari, 1568: 103). In this story, the Pope was not interested in the particular design Giotto might have made for the paintings to be commissioned. It seems instead that the artist's general technique was of interest, which suggests the patron's role as producer as well as consumer of artistic innovation. Land (2005) notes the similarity of Vasari's story to two earlier stories told of the ancient artist Apelles. In each of these two stories, however, we have a competition between two artists, and there is no role for a patron. What is significant for us in the story of Giotto's O is the central role of the patron who is interested not in the final product, but in the process of production.12 Presumably, the story is false. Land says it is ‘clearly fiction’ (2005: 6). But the role played by the patron in this story does bolster the view that such patrons may have been novelty intermediaries.
The economic theory of professions as developed by Savage (1994) and Langlois and Savage (2001) anticipates the notion of novelty intermediation at least somewhat. Professions such as medicine may exist when the professional must combine expertise based on general principles with judgment based on specific circumstances. ‘As a result’, Langlois and Savage explain, ‘professionals do not standardize the application of their routines. . .but only the “toolkit” of routines from which they draw’ (2001: 155). Professions may exist when general knowledge is more or less static. The general principles of the profession will then be embodied in a static set of routines characterizing the profession. Traditional Chinese medicine (TCM) seems to fit this description of a profession with a static knowledge base.13 If the professional knowledge base is dynamic, Langlois and Savage explain, ‘professionals must also employ judgment in the creation of new routines’ (2001: 155).
It seems reasonable to guess that many economic phenomena may be subject to reinterpretation as, in part, novelty intermediation in the context of creative dynamics.
[T]he bold, creative, innovative entrepreneur, too, is at a yethigher level of abstraction also engaged in arbitrage(Israel Kirzner).
4. Toward an entrepreneurial economics
It is not yet clear what an economics for a creative world will look like. We have described a creative world in which the model of explanation borrowed from physics does not apply. Our discussion of creative dynamics suggests that mathematical methods are less powerful than economists have generally imagined in the post-war years.14 In particular, well beyond the familiar limits of computability, the phase space construction, as practiced in mathematical physics and standard mathematical approaches in economy, does not apply in the econosphere.
Our picture of creative dynamics in the econosphere does not automatically provide the sort of clear policy prescriptions some readers may desire. In particular, it does not allow us to call for ‘more regulation’ or ‘less regulation’ of ‘the market’. It is, perhaps, a truism that we do not need ‘more regulation’ or ‘less regulation’, but ‘right regulation’. And yet truisms must not be forgot. Designing wise regulation is a vast topic, given the creation of partially prestatable adjacent possible strategy spaces in the evolving econosphere. Preventing regulation from serving special interests rather than the common good is also a vast topic. We hope and expect that our analysis may inform policy when combined with close institutional analysis and an interpretation of economic history. In other words, facts matter and details matter. The current necessarily rather broad essay cannot enter into the level of factual detail required to produce relatively specific policy recommendations. Our main result is not ‘positive’ in that sense, but entirely ‘negative’ in the sense developed by Longo (2008).
Our central result that there can be no entailing laws of economic dynamics is a negative result. We think it would be an error to view negative results as somehow cause for disappointment. Longo (2008) has noted the long-term value of negative results, which destabilize our expectations for ‘positive’ theory without necessarily indicating where to go from there. Negative results have marked the beginning of new scientific thinking on several occasions. Poincaré's Three Body Theorem, Heisenberg's uncertainty principle, and Gödel's theorem, all opened up new ways of thinking (Longo, 2008). Limits clarify what is feasible and what is not feasible with the existing tools. They may show us new directions by their very nature, at least if the analysis establishing limits has a sufficiently precise scientific content. Negative results, by setting or showing constraints and limits of current knowledge, enable new ideas, new forms of thinking. By the very methods used, Poincaré's result started the modern Geometry of Dynamical Systems and, later, Chaos Theory. Gödel's 1931 theorems allowed us to define Computability and enabled Proof Theory. It may take time, however, for those new directions to emerge clearly.
Longo (2008) notes that negative results run counter to the ‘culture of results’ in today's scientific funding agencies. He says,
The resistance [to negative results] may not only be of a philosophical nature, but may also stem from this “culture of results” more than “of knowledge,” a culture which increasingly claims to completely direct science. The accountability obligation, increasingly required by the managers who rule the scientific financing, is of an industrial type and imposes its paradigms: one must beforehand clearly set out the projected methods, the expected results (the “deliverables”. . .) in order to be able, at the end of the project, to compare them with the results effectively obtained.
We cannot be sure where our argument will take us. It seems reasonable to suggest, however, that our discussion points toward an economic theory in which ‘entrepreneurs’ solve the ‘frame problem’ of social systems that allow relatively dispersed decision making. We are not aware of any earlier work forging a tight logical link between entrepreneurship and the frame problem. It may be helpful, therefore, to spend some time carefully developing the concept of the frame problem.
In a creative world, the system must adapt to unforeseen and unforeseeable changes that represent new possibilities and new opportunities in the adjacent possible. By definition, no prestated algorithm can anticipate such changes. The system, therefore, must be capable of adaptive, non-algorithmic change. The system must, as it were, continually change its frame of analysis. We may define the frame problem as the problem of modifying the system's implicit frame of analysis to adapt successfully to non-algorithmic change in the adjacent possible. Our ‘frame problem’ is similar to the ‘framing problem’ of Earl and Potts (2004: 622) noted earlier.
It is true, of course, that the system as such does not make decisions and cannot, therefore, have a ‘frame of analysis’ in any strict sense. Rather, each decision maker within the system has its own frame of analysis or even multiple frames that it applies in different contexts. Denzau and North (1994) call these frames ‘mental models’. At any moment in time, there is an interlocking set of ‘mental models’ or ‘frames’ guiding the actions of the several decision makers in society. It is this set that we call ‘the system's implicit frame of analysis’. As Denzau and North (1994) note, these ‘frames’ or ‘mental models’ may be shared, which may help to justify our decision to call the inter-locking set of frames guiding action a ‘frame’ as well. Indeed, following Koppl and Langlois (2001) we note that all interactions in society may be construed as Wittgensteinian ‘language games’. (see Wittgenstein, 1953). Koppl (2002: 210–211) gives a set-theoretic analysis of the potentially intricate structure of interlocking ‘language games’ in society. The ‘system's implicit frame of analysis’ is not only the set of operative ‘mental models’ but also, in its public aspect, a set of Wittgensteinian language games, which implies some degree of commonality in the frames of interacting decision makers. I cannot sell you an apple for a dollar unless our ideas of ‘apple’, ‘dollar’, and ‘trade’ are more or less the same. Entrepreneurial discovery implies change in the individual entrepreneur's personal frame of analysis. It implies also, however, a change in the public ‘language games’ the entrepreneur will play with other actors in society and, we may therefore say, change in ‘the system's implicit frame of analysis’.
Biological evolution solves the frame problem all the time. For example, swim bladders created the new unprestatable function of neutral buoyancy in the water column as well as a new niche. The problem is solved in market economies too. For example, consider Yusuke Ohki's entrepreneurial invention of a new use, hence new business opportunity, for iPads in the context of small Japanese apartments in which the accumulation of too many books may quickly create crowded living conditions. These actuals are exactly the enabling constraints that created the adjacent possible business opportunity Ohki seized. In doing so, and in finding a new use for the iPad, he solved the frame problem as well.
The term ‘frame problem’ seems to have been coined by McCarthy and Hayes (1969). The problem in its original sense arose in artificial intelligence. It is the problem of how to code a sequence of actions without re-coding at each such step a complete description of the system's environment. They give the example of a computer that must be told, ‘if a person has a telephone he still has it after looking up a number in the telephone book’. According to Shanahan (2009), ‘The first significant mention of the frame problem in the philosophical literature was made by Dennett (1978: 125)’. The puzzle according to Dennett, Shanahan explains, ‘is how “a cognitive creature. . .with many beliefs about the world” can update those beliefs when it performs an act so that they remain “roughly faithful to the world”’? Our definition is related to both of these uses, but identical to neither.
By definition, an algorithmic system cannot change its frame of analysis. It seems reasonable to suggest, therefore, that systems concentrating decision making in a small number of ‘experts’ may be less adept at updating their frames than systems distributing decision making across a relatively large number of persons. In the extreme case of a social system with a controller, the controller is responsible for all frame changes. Even if it is able to solicit advice broadly, it is by definition the controller that decides when, whether, and how to change frame. Constraints on the controller's cognition become constraints on the range of frame changes it can consider. If a frame change is chosen, it must then be translated into changes in the states of the system's other agents. Thus, all details of the frame change must pass through the controller and are correspondingly subject to its cognitive constraints.
With more dispersed decision making, by contrast, there may be no one person responsible for the system's overall frame of analysis. While the lack of control may create the possibility of inconsistencies, it may also reduce the epistemic burden of frame changes by distributing them to dispersed autonomous persons.15 We may use the label ‘entrepreneurs’ for persons who notice new opportunities available in the adjacent possible. Each innovation, whether successful or not, is an adjustment in the system's implicit frame of analysis. Over time these adjustments may accumulate and become very large frame changes. In this sense we may say that entrepreneurs solve the frame problem for social systems, or at least for systems that allow dispersed decision making. The frame adjustments we are describing are non-algorithmic. Just as we cannot list all uses of a screwdriver, we cannot list all innovations that entrepreneurs will notice and act on over time. And yet entrepreneurs notice them and act on them. Thus, frame change is non-algorithmic at the level of both the individual and the system.
The entrepreneurial agents of distributed human systems are individually non-algorithmic just as the system is non-algorithmic. The system is non-algorithmic because it generates new uses, thence new possibilities that cannot be listed ex ante, as it happens in biological evolution and niche creation. Both evolutionary and human agents cannot be mathematically analyzed in a pre-given phase space – even though the math may conveniently be available – as the dynamics of inert bodies have been investigated in physics. A fortiori, they are non-algorithmic because they learn and have protensional (preconscious) or intentional (conscious) actions; that is, they make bets on the future and act accordingly. By this, they alter their models of the world to incorporate at least some of the novel possibilities created by the system's evolution. Such learning is necessarily non-algorithmic because what the agent can learn was not listed and not listable ex ante. Protensional actions are not algorithmic because they are based on learning and on betting on novelty.16
We might note that these future-oriented ‘bets’ and associated actions are not necessarily random, nor given by an analysis of past experience, but might in fact be informed by the theories, hypotheses, and conjectures that economic agents have about future possibilities (Felin and Zenger, 2009). In other words, economic actors have generative capacities to postulate future states that are not listable a priori. This generative capacity points out that humans are scarcely algorithmic input–output machines (though treated as such by some), but have unique abilities to bootstrap novelty, even with highly limited or impoverished external inputs (Felin and Foss, 2011). Note that the human construction of communication by semantic laden language, writing, and gestures provides the enabling conditions for future novelty, and leads to what Wilhelm von Humboldt in the early 19th century called the ‘infinite use of finite means’. Importantly, this generative capacity extends well beyond the setting of language – language simply providing us a ‘window into the mind’ – as this generativity provides the foundation for the more general emergence of novelty in social and economic settings. In all, without getting bogged down in the details of this generative capacity, we simply note that its existence highlights the non-algorithmic nature of human activity, economic activity and entrepreneurship being a clear example.
Shackle (1969: 3–13, 47–61, 278–279) contains a penetrating discussion of the ‘listing problem’ in economics, which is closely related to our analysis: because choice is ‘originative’, we cannot list all possibilities for choice or, therefore, the future. We might say that non-algorithmic learning is true learning. In a creative world, distributed systems and the entrepreneurs in them are both capable of non-algorithmic change and adaptation, by learning and acting in a changing environment. In short, opportunities are yielded by actuals that create enabling constraints yielding adjacent spaces of what is now possible. Successful innovations are discoveries of appropriate non-algorithmic responses to new possibilities.
From whence come opportunities entrepreneurs may notice? From whence the new partially unprestatable adjacent possible opportunities? We are arguing that, in general, opportunities arise from actuals that are the enabling constraints ‘creating’ the adjacent possible. In this view, the newly emerging opportunities are unprestatable; they emerge non-algorithmically. Even more: we can have no entailing mathematical theory of the emergence of opportunities in the biosphere or economy, since current equational approaches require the predefinition of a phase space. This non-algorithmicity and non-equational determination is a challenge to the construction of a scientific economics in part because it implies that our most cherished modes and models of ‘science’ do not seem to apply to economic evolution. Moreover, this is so already for evolutionary theories. Our alternative to more traditional models of ‘scientific’ economics, then, becomes a kind of entrepreneurial economics.
A central task in entrepreneurial economics is to ask what institutions and laws (conceived, perhaps, as webs of enabling constraints that create adjacent possible opportunities) promote favorable entrepreneurial opportunities. In a classic article, Baumol (1990: 893) says,
[W]hile the total supply of entrepreneurs varies among societies, the productive contribution of the society's entrepreneurial activities varies much more because of their allocation between productive activities such as innovation and largely unproductive activities such as rent seeking or organized crime. This allocation is heavily influenced by the relative payoffs society offers to such activities. This implies that policy can influence the allocation of entrepreneurship more effectively than it can influence its supply.
We interpret Baumol's account of productive and unproductive entrepreneurship as a call for institutional analysis. We ask what institutions promote ‘favorable’ entrepreneurial opportunities. Here are some rather hum-drum examples that may suggest what we have in mind. Productive new-venture entrepreneurship (the founding of new for-profit enterprise) is promoted by transparency, simplicity, and openness in the legal system. Productive political entrepreneurship is promoted by representative democracy with constitutional constraints. In contrast, powerful actors may try to block innovations that threaten their economic position, their rents. In effect, our power structures themselves have become self-sustaining autocatalytic sets, adapting non-algorithmically to regulations to promote their own survival and power. Hence, these constitutional constraints should be designed to limit the power of special interests and prevent the consolidation of power into the hands of an elite. How to achieve this wisely requires careful innovations, including legal innovations. Finally, productive social entrepreneurship is promoted by freedom of association and movement. It is promoted by a democratic and voluntarist system that encourages critical, innovative thinking and adaptive forms of decision making.
Our rather hum-drum examples of how institutions influence the direction of entrepreneurship are meant only to suggest some of the questions that might be asked in an entrepreneurial economics, and what sorts of answers might be given. We do not mean to suggest that we can predict the future of economic theory. As we noted in our introduction, economic theory is characterized by the same creative dynamics we have described for the biosphere and econosphere.
We have made some suggestions about the future of economic theory. In particular, we have discussed novelty intermediation and the move toward a more entrepreneurial economics. But we would not pretend to have a complete system of political economy to offer. Indeed, a truly ‘complete’ system would seem ill suited to a creative world. Our perspective may support a somewhat skeptical attitude toward economic experts. Berger and Luckmann (1966) long ago noted the risk that ‘universal experts’ may be rigid and willfully obscure. Experts, they note, ‘claim ultimate jurisdiction of the stock of knowledge’ which is their specialty (p. 117). Knowledge and expertise are fundamental, yet the rate of obsolescence of knowledge in a creative world is high. And, more importantly, the required updating and reframing are probably best performed in decentralized and democratic manner that cannot be readily reconciled with expert claims to epistemic monopoly.
It seems likely that the tools of high theory should continue to evolve in the direction taken by complexity theory. In particular, computability theory seems to give us a set of tools useful in sorting out what is feasible and what not feasible in both theory and policy. Network theory seems well suited to an economics for a creative world. It is an increasingly important area and we need to work out what we can and cannot know about networks. Economic history is clearly important, as is the related idea of close description of institutions. Our description of creative dynamics in the econosphere seems to suggest the importance of institutions and institutional change, which tends to elevate the importance of thick description, economic history, and applied economics relative to the sort of high theory that dominated economics in the post-war period.
Our hints for the future of economic theory all point to a more institutionally rich analysis. The less that abstract theory has to say, the more work will have to be done by close institutional analysis. As we said in the introduction, our analysis is more metatheoretic framework than applied theory. But in any applied theory that fits our framework ‘entrepreneurs’ (whether new-venture entrepreneurs, political entrepreneurs, or social entrepreneurs) will act in a specific institutional context. As Baumol (1990) emphasized, different institutions will encourage different sorts of entrepreneurial actions. Thus, for our framework to meet empirical reality, the institutional context must be specified. It must be given in sufficient detail to support general claims about the direction of entrepreneurship.
If we live in a creative world, it may be time to reconsider our methods, our theory, and the role of economists in a democratic society. We have made some broad suggestions along these lines without pretending to have an algorithm for any future economics. We do not have all the answers. We will have provided a helpful prolegomena, however, if our arguments help to promote a larger discussion within the economics profession.
K. Arrow and G.Debreu (1954), ‘Existence of an Equilibrium for a Competitive Economy’, Econometrica, 22: 265–290.
Y. Balasko and J.Geanakoplos (2012), ‘Introduction to General Equilibrium’, Journal of Economic Theory, 147: 400–406.
J. R. Banker (1991), ‘The Program for the Sassetta Altarpiece in the Church of S. Francisco in Borgo S. Sepolcro’, I Tatti Studies in the Italian Renaissance, 4: 11–58.
W. J. Baumol (1990), ‘Entrepreneurship: Productive, Unproductive, and Destructive’, The Journal of Political Economy, 98 (5): 893–921.
E. D. Beinhocker (2007), The Origin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics, Cambridge, MA: Harvard Business Review Press.
P. Berger and T.Luckmann (1966), The Social Construction of Reality, New York, NY: Anchor Books.
R. Blundell and T. M.Stoker (2005), ‘Heterogeneity and Aggregation’, Journal of Economic Literature, 43: 347–391.
M. Bunge (2009), Causality and Modern Science, 4th rev. edn, New Brunswick/London: Transaction Publishers.
B. Caldwell (2005), Hayek's Challenge, Chicago, IL: University of Chicago Press.
D. T. Campbell (1960), ‘Blind Variation and Selective Retention in Creative Thought as in Other Knowledge Processes’, Psychological Review, 67 (6): 380–400.
A. J. Cohen and G. C.Harcourt (2003), ‘Whatever Happened to the Cambridge Capital Theory Controversies?’, The Journal of Economic Perspectives, 17 (1): 199–214.
D. Colander , R. P. F.Holt, and J. B.Rosser, Jr . (2004), The Changing Face of Economics: Conversations with Cutting Edge Economists, Ann Arbor, MI: University of Michigan Press.
T. Cowen (2011), The Great Stagnation, e-reader edn, New York, NY: Dutton.
A. Croce , J.Martí, and S.Murtinu (2013), ‘The Impact of Venture Capital on the Productivity Growth of European Entrepreneurial Firms: ‘Screening’ or ‘Value Added’ Effect?’, Journal of Business Venturing, 28: 489–510.
D. Dennett (1978), Brainstorms, Cambridge, MA: The M.I.T. Press.
A. T. Denzau and D. C.North (1994), ‘Shared Mental Models: Ideologies and Institutions’, Kyklos, 47 (1): 1–13.
K. Dopfer (2005), ‘Evolutionary Economics: A Theoretical Framework’, in KurtDopfer (ed.), The Evolutionary Foundations of Economics, Cambridge, MA: Cambridge University Press, pp. 3–59.
S. Durlauf (2012), ‘Complexity, Economics and Public Policy’, Politics, Philosophy & Economics, 11: 45–75.
P. E. Earl and J.Potts (2004), ‘The Market for Preferences’, Cambridge Journal of Economics, 28: 619–633.
J. Felipe and F. M.Fisher (2003), ‘Aggregation in Production Functions: What Applied Economists should Know’, Metroeconomica, 54 (2/3): 208–262.
T. Felin , S.Kauffman, R.Koppl, and G.Longo (2013), ‘Economic Opportunity and Evolution: Beyond Bounded Rationality and Phase Space’, Strategic Entrepreneurship Journal, forthcoming.
T. Felin and N.Foss (2011), ‘The Endogenous Origins of Experience, Routines, and Organizational Capabilities: The Poverty of Stimulus’, Journal of Institutional Economics, 7: 231–256.
T. Felin and T.Zenger (2009), ‘Entrepreneurs as theorist: On the origins of collective beliefs and novel strategies’, Strategic Entrepreneurship Journal, 3: 127–146.
F. Fisher (1987), ‘Aggregation Problem’, in J.Eatwell, M.Milgate, and P.Newman (eds.), The New Palgrave: A Dictionary of Economics, vol. 1, London: The Macmillan Press Limited, pp. 53–55.
H. D. Flap (1990), ‘Patronage: An Institution in its Own Right’, in R.Wippler, K.-D.Opp, and M.Hechter (eds.), Social Institutions: Their Emergence, Maintenance, and Effects, New York, NY: Aldine de Gruyter, pp. 225–243.
R. Florida and M.Kenney (1988), ‘Venture Capital and High Technology Entrepreneurship’, Journal of Business Venturing, 3: 301–319.
V. Foley , S.Rowley, D. F.Cassidy, and F. C.Logan (1983), ‘Leonardo, the Wheel Lock, and the Milling Process’, Technology and Culture, 24 (3): 399–427.
C. Gilbert (1998), ‘What Did the Renaissance Patron Buy?’, Renaissance Quarterly, 51 (2): 392–450.
B. Gille (1966), Engineers of the Renaissance, Cambridge, MA: The M.I.T. Press.
S. J. Gould (1998), Leonardo's Mountain of Clams and the Diet of Worms, New York, NY: Harmony Books.
D. Harper and P.Lewis (2012), ‘New Perspectives on Emergence in Economics’, Journal of Economic Behavior and Economics, 82: 329–337.
F. A. Hayek (1967), Studies in Philosophy, Politics, and Economics, Chicago, IL: University of Chicago Press.
F. A. Hayek (1978), New Studies in Philosophy, Politics, Economics and the History of Ideas, Chicago, IL: University of Chicago Press.
C. A. Hidalgo , B.Klinger, A. L.Barabási, and R.Hausmann (2007), ‘The Product Space Conditions the Development of Nations’, Science, 317: 482–487.
G. Hodgson (1993), Economics and Evolution: Bringing Life Back into Economics, Ann Arbor, MI: University of Michigan Press.
G. Hodgson (2001), How Economics Forgot History: The Problem of Historical Specificity in Social Science, London: Routledge.
G. Hodgson and T.Knudsen (2010), Darwin's Conjecture: The Search for General Principles of Social and Economic Evolution, Chicago, IL: University of Chicago Press.
M. Hollingsworth (1994), Patronage in Renaissance Italy: From 1400 to the Early Sixteenth Century, Baltimore, MD: The Johns Hopkins University Press.
J. James (1992), ‘Evidence for Flying Buttresses before 1180’, Journal of the Society of Architectural Historians, 51 (3): 261–287.
A. P. Kirman (1992), ‘Whom or What does the Representative Individual Represent?’, The Journal of Economic Perspectives, 6 (2): 117–136.
F. H. Knight (1921), Risk, Uncertainty, and Profit, Boston, MA: Houghton Mifflin.
A. Koestler (1956), The Act of Creation, New York: Penguin Books.
R. Koppl (2002), Big Players and the Economic Theory of Expectations, London/New York: Palgrave Macmillan.
R. Koppl and R. N.Langlois (2001), ‘Organizations and Language Games’, Journal of Management and Governance, 5: 287–305.
R. Krautheimer and T.Krautheimer-Hess (1956), Lorenzo Ghiberti, Princeton, NJ: Princeton University Press.
E. Lefèvre-Pontalis (1920), ‘Étude historique et archéologique sur l’église de Saint Germain-des-Prés’, Congrès archéologique de France, LXXXIIe session tenue à Paris en 1919, 82: 301–366.
N. E. Land (2005), ‘Apelles and the Origin of Giotto's O’, Source Notes in the History of Art, 25 (1): 6–9.
R. N. Langlois and D. A.Savage (2001), ‘Standards, Modularity, and Innovation: The Case of Medical Practice’, in R.Garud and P.Karnøe (eds.), Path Dependence and Path Creation, Hillsdale, NJ: Lawrence Erlbaum, pp. 149–168.
P. O. Long (1997), ‘Power, Patronage, and the Authorship of Ars: From Mechanical Know-How to Mechanical Knowledge in the Last Scribal Age’, Isis, 88 (1): 1–41.
G. Longo , M.Montévil, and S.Kauffman (2012), ‘No Entailing Laws, but Enablement in the Evolution of the Biosphere’, Physics ArXiv, 1201.2069v1: http://arxiv.org/abs/1201.2069.
G. Longo , M.Montévil, C.Sonnenschein, and A.Soto (2014), ‘Searching for Principles for a Theory of Organisms’, unpublished manuscript.
G. Longo and M.Montévil (2012), ‘Randomness Increases Order in Biological Evolution’, Computation, Physics and Beyond, 7160: 298–308.
G. Longo and M.Montévil (2014), Perspectives on Organisms: Biological Time, Symmetries and Singularities, Berlin: Springer.
B. Marwick (2003), ‘Pleistocene Exchange Networks as Evidence for the Evolution of Language’, Cambridge Archaeological Journal, 13 (1): 67–81.
D. Mateer (2000), ‘Introduction’, in D.Mateer (ed.), Courts Patrons and Poets, London: Yale University Press in association with The Open University, pp. ix–xi.
J. McCarthy and P. J.Hayes (1969), ‘Some Philosophical Problems from the Standpoint of Artificial Intelligence’, in B.Meltzer and D.Michie (eds.), Machine Intelligence 4, Edinburgh: Edinburgh University Press, pp. 463–502.
M. Minniti (2004), ‘Entrepreneurial Alertness and Asymmetric Information in a Spin-Glass Model’, Journal of Business Venturing, 19: 637–658.
L. Mises (1966), Human Action: A Treatise on Economics, 3rd rev. edn, Chicago, IL: Henry Regnery Company.
M. D. Mumford (2003), ‘Where have We been, Where are We Going? Taking Stock in Creativity Research’, Creativity Research Journal, 15 (2–3): 107–120.
J. Nation , I.Trofimova, J.Rand, and W.Sullis (eds.) (2002), Formal Descriptions of Developing Systems, Dordrecht/Boston/London: Kluwer Academic Publishers.
R. Nelson and S.Winter (1982), An Evolutionary Theory of Economic Change, Cambridge, MA: Harvard University Press.
J. K. Nelson and R. J.Zeckhauser (eds.) (2008), The Patron's Payoff: Conspicuous Commissions in Italian Renaissance Art, Princeton, NJ: Princeton University Press.
G. P. O’Driscoll and M. J.Rizzo (1985), The Economics of Time and Ignorance, Oxford and New York: Basil Blackwell.
S. Y. Pilyugin (1999), Shadowing in Dynamical Systems, Berlin/New York: Springer.
J. Potts (2012), ‘Novelty-Bundling Markets’, in David E.Andersson (ed.), The Spatial Market Process, Advances in Austrian Economics, vol. 16, Bradford: Emerald Publishing Group.
D. A. Savage (1994), ‘The professions in theory and history: The case of pharmacy’, Business and Economic History, 23 (2): 129–160.
J. A. Schumpeter (1947), Capitalism, Socialism, and Democracy, 2nd edn, New York/London: Harper & Brothers Publishers.
G. L. S. Shackle (1969), Decision Order and Time in Human Affairs, 2nd edn, Cambridge: Cambridge University Press.
G. L. S. Shackle (1972), Epistemics and Economics, Cambridge: Cambridge University Press.
G. L. S. Shackle (1979), ‘Imagination, Formalism and Choice’, in M.Rizzo (ed.), Time, Uncertainty and Disequilibrium, Lexington, MA: D.C. Heath.
A. Smith (1776; 1981), An Inquiry Into the Nature and Causes of the Wealth of Nations, R. H.Campbell and A. S.Skinner (eds.), Indianapolis, IN: Liberty Classics.
J. L. Tang , B. Y.Liu, and K. W.Ma. (2008), ‘Traditional Chinese Medicine’, The Lancet, 372: 1938–1940.
G. Vasari (1568; 1996), Lives of the Painters, Sculptors and Architects, London: Everyman's Library.
U. Witt (2009), ‘Novelty and the Bounds of Unknowledge in Economics’, Journal of Economic Methodology, 16 (4): 361–375.
L. Wittgenstein (1953), Philosophical Investigations, G. E. M.Anscombe and R.Rhees (eds.), New York, NY: The Macmillan Company.
D. H. Wolpert (2001), ‘Computational Capabilities of Physical Systems’, Physical Review E, 65: 016128(1–27).
A. A. Young (1928), ‘Increasing Returns and Economic Progress’, Economic Journal, 38 (152): 527–542.