In Defence of the Linear Model. An Essay M. Balconi, S. Brusoni and L. Orsenigo Pecs, July , 2010 Motivation Widespread and criticism towards the Linear Model (LM) Everybody agrees that the LM is wrong and useless But then, why continuing to criticise it? Are some parts of the LM still useful? Pecs, July , 2010 Objectives 1)What is the LM? Can we identify the main propositions which constitute or are usually associated to the LM in the literature? - some brief historical remark on the origins, status and content of the LM - V. Bush and “Science: The Endless Frontier” 2) Which are the main critiques advanced against the LM and to which context do they apply? 3) Do these critiques really destroy the LM? 4) 3 main dimensions of the debate: - cognitive - organisational - normative 5) Conclusion Pecs, July , 2010 Preliminary Observations critiques of the LM encompass a variety of different – and often mutually incompatible – arguments and implications, which do not necessarily derive from the model itself the demise of the LM has opened the Pandora’s box of possible alternative “models” and normative prescriptions Pecs, July , 2010 CAVEAT Not an historical reconstruction Not a claim on the empirical validity of the LM But an essay: - not all the critiques are really destructive - often they are uncoherent or mutually contradictory - the LM – in a weak form – may well survive and be useful, at least in some domains of analysis and policy Pecs, July , 2010 To begin with • “at one time it was almost impossible to read a book or an article on technology policy or technological forecasting that did not begin or end with a polemic against the so-called linear model of innovation” • “The LM cannot be simply dismissed as a convenient strawman erected for the convenience of those expounding alternative ideas” (C.Freeman, The Greening of Technology, 1996) Pecs, July , 2010 Some background literature • Did the LM ever existed? • Or was it a straw man? • Is it a theory? Or a model? • A Folk Model? • What are its main components? Pecs, July , 2010 Alternative interpretations • D. Edgerton (2006): the LM did never actually exist, neither in theory nor in practice • A model of the relationships between science and society and specifically, innovation and economic growth • The term LM was rarely used before the '80s and almost always critically • It comes in various forms and it is never very well spelled out. But its common theme seems to be: – basic science is the main source of innovation – the innovative process is sequential – Innovation is a major source of growth • Critiques: – the innovative process is “irrational and cannot be programmed in advance (Price and Bass 1969) – Innovation rarely rests on scientific research Pecs, July , 2010 Godin (2006) • The model, whatever its name, has been the very mechanism used for explaining innovation in the literature on technological change and innovation since the late 1940s. • The model postulates that innovation starts with basic research, then adds applied research and development, and ends with production and diffusion: • The model has been very influential. Academic organizations as a lobby for research funds (National Science Foundation 1957) and economists as expert advisors to policy makers (Nelson 1959) have widely disseminated the model, or the understanding based thereon, and have justified government support to science using such a model. As a consequence, science policies carried a linear conception of innovation for many decades (Mowery 1983a), as did academics studying science and technology Pecs, July , 2010 Godin (ctd) • The LM did not arise from the mind of one individual. Rather, it developed over time in various steps: • First were natural scientists (academic as well as industrial), developing a rhetoric on basic research as the source for applied research or technology (from F. Bacon, to M. Holland to W.R. Maclaurin,..); • second were industrialists and consultants from business schools, having been interested in science studies long before economists and studying the industrial management of research and the development of technologies; • third were economists, bringing forth the concept of innovation”: production and diffusion: • The result is a “rhetorical entity”, which gained strength and became entrenched in discourses and policies with the help of statistics and methodological rules for collecting data (Frascati Manual, 1963 Pecs, July , 2010 Industrialists • Maurice Holland, Director, Division of Engineering and Industrial Research, National Research Council: series of papers and a book on the importance of research for industrial development: research as a modern method of accelerating industrial evolution (1928-1933) • K. Mees (Eastman Kodak) describes the work of the development laboratory as a sequential process: development work is “founded upon pure research done in the scientific department, which undertakes the necessary practical research on new products or processes as long as they are on the laboratory scale, and then transfers the work to special development departments which form an intermediate stage between the laboratory and the manufacturing department” (Mees 1920) • R. Stevens, vice president at Arthur D. Little: United States National Resources Planning Board report titled Research: A National Resource in 1941. Pecs, July , 2010 Economists • W. Rupert Maclaurin: developed Schumpeter’s ideas, analyzing technological innovation as a process composed of several stages or steps. Maclaurin constructed one of the first taxonomies for measuring technological innovation in the literature, that led to current indicators on high technology • economists bringing forth the concept of innovation”: production and diffusion: Y. Brozen, 1951, Usher 1954, Carter and Williams 1957, Ruttan 1959, Machlup 1962, Schmookler 1966, Scherer 1965, Mansfield 1968 Pecs, July , 2010 Hounshell (2004) • The LM can be considered as a system of belief, a heuristic which simply states that the new knowledge generated by investment in fundamental, unfettered research will, at some point in the future, yield radically new inventions and technologies. • the linear model was very real in the United States at the end of the Second World War and up to the early seventies; • it made the case for the United States government’s funding of scientific and engineering research at universities • and for R&D strategies of companies like DuPont, who established fundamental research programs for the first time in the American history (e.g. new nylons). Pecs, July , 2010 Science The Endless Frontier (V. Bush, 1945) One would be hard-pressed to find anything but a rudiment of the LM model in Bush’s manifesto. Bush talked about causal links between science (namely basic research) and socioeconomic progress, but nowhere did he develop a full-length argument based on a sequential process broken down into its elements or that suggests a mechanism whereby science translates into socioeconomic benefits. Bush was making an argument for science policy, not for innovation (Edgerton): - support to public funding of academic research - basic science should be unconstrained and it will lead to innovation Bush participated into the rethoric that basic research leads to applied research (Godin) Pecs, July , 2010 The LM according to V. Bush 1) technological innovation and economic development are based on new scientific knowledge - Examples: health care and defence, where discoveries (such as penicillin and radar) often arose from remote and unexpected source - XXth century basic research has become ‘the pacemaker of technological progress 2) A distinction is drawn between basic and applied research, based upon the interest in practical ends and a continuum of activities are identified between these two research orientations 3) the centres of basic research are identified with colleges, universities and research institutes ‘where scientists may work in an atmosphere which is relatively free from the adverse pressure of convention or commercial necessity’ 4) Since science is considered a proper concern for government (…the new frontier..), government had to support basic research 5) Government can promote industrial research also by providing suitable incentives to industry to conduct research, and by strengthening the patent system. In addition, ways should be found to spread the benefits of basic research to industries which do not now utilize new scientific knowledge. Pecs, July , 2010 Before and beyond ‘Science: The Endless Frontier Maclaurin: He carried out an important meso- and micro-level study of the process of innovation and the development of the radio industry, that he considered ‘a direct outgrowth of a revolution in the science of physics and its applications to the study of electricity’ (1950) In this type of industry, ‘science and technology’ can be broken down into five distinct stages: fundamental research, applied research, engineering development, production engineering and service engineering’ (1947) “In ‘studying the determinants of investment in any advanced economy, it will be significant to assess the variations in the following factors: the propensity to develop pure science, the propensity to invent, the propensity to innovate, the propensity to finance innovation, the propensity to accept innovation. (1953) The important thing is that some scientists have been willing to speculate deeply and widely without immediate practical objectives in mind’. (1953, pp. 98 and 99). science is not the only source of ideas for innovation, as ‘new uses’ may emerge for established products and the most likely source for such ideas is not the pure scientist, but rather the industrial innovator Pecs, July , 2010 Maclaurin (ctd) a taxonomy of industries on the basis of the rate of technological progress (industries with very high, high, medium and low progress), where progressiveness depended on whether research and engineering were ‘directed primarily to refinements in existing products rather than to radical improvements or the creation of entirely new products or processes’ (1954) Maclaurin’s account of the LM adds a clear articulation of the sequence of phases, but also unambiguously limits the interpretative scope of the model to science-based industries, such as radio. Pecs, July , 2010 C.C. Furnas (1948) “it is the ‘cross-over’ of science and invention which lay at the heart of the new ‘industrial revolution’ which was becoming visible in the aftermath of WWII (Furnas, 1948, p. 1). he stressed the increasing role played by corporate R&D laboratories, without however implying that ‘no advances can be made in industry without a formal research organization. “Research organizations have not had in the past and will not have in the future a monopoly on sources of new ideas and improvements’ Pecs, July , 2010 Summing up A number of contemporaries to Bush did provide fine grained analyses of the unfolding of the innovation process, e.g. identifying limits to the applicability of the LM across different industries, and also pointing out feedbacks across different stages of the innovation process. Nevertheless, memory of this broader – and deeper – discussion has faded. What we are left with is a common sense understanding of the LM A straitjacket, which, in depriving the arguments of any historical references and any nuance and subtlety, makes them appear an inflexible and oversimplified model of the innovation process. The LM referred to by the literature (the LM in Strong Form) Basic research Applied Research Development Production Marketing Diffusion Pecs, July , 2010 The conventional presentation: the process and the cognitive dimension Since prior scientific research is the unique source of new technologies, innovations can be considered as practical applications of basic scientific research New knowledge acquired through basic research trickles down, almost automatically, to applied research, technology and innovations, even within short time spans the innovative process can be represented and conceptualised as sequence of steps In the sequence there is no feedback from later steps to earlier steps Pecs, July , 2010 The actors and the organisational dimension • 1) There is a clear division of labour along the sequence between different types of agents who specialise in the various relevant stages: – scientific research is conducted in universities and public laboratories – Technological development is carried out by firms: – universities contribute to applied research (innovation) primarily through the conduct of research and teaching. Direct interaction with industry is not perceived to be a fundamental mission of universities • 2) universities and firms respond to different types of motivations and incentives. – Universities: public interest, the welfare of the society, individual prestige, fame and career, ‘publish or perish’. – Firms are driven by the quest for profit. Pecs, July , 2010 The conventional presentation: normative prescriptions • basic research – and therefore the agents performing it, typically universities - should be funded by public sources • new knowledge has to be placed in the public domain. • applied research – typically performed by business firms – should not in principle be supported by the government, at least to the extent that its output can be appropriated and protected by imitation. Pecs, July , 2010 The critiques to the LM: the cognitive dimension 1) The relationship between science and innovation • the distinction between basic and applied research is not clearcut (e.g. Stokes, Dasgupta and David) • most technological improvements are unrelated to basic research and they often anticipate science …after the WW2 incremental technological innovation remained extremely important (Kline and Rosenberg) • not only is technology independent of new science, but it also provides essential inputs to scientific research (problems to be solved, instrumentation) • the conventional time orientation and direction of causation of the model should be reversed in many cases • users of products and processes are the developers of many important innovations that are later produced and sold by manufacturers (von Hippel, the mountain bicycle (Lüthje et alii, 2005)) Pecs, July , 2010 ‘An Overview of Innovation’, Kline and Rosenberg , 1986 There is a tendency to identify technological innovation with major innovations …The fact is that much technological change takes the form of very small changes, such as minor modifications in the design of a machine… Most innovation is done with the available knowledge already in the heads of the people doing the work …It is only when those sources of information fall short of solving the problem that there is a need for research in order to complete a given innovation…The notion that innovation is initiated by research is wrong most of the time… According to KR the initiating step of most processes of technological transformation in today’s world is typically design rather than research Pecs, July , 2010 This stream of critiques does not destroy the LM, but drastically reduces the sphere to which it can be applied However: it is often argued that in the XX century the emergence of major new technological paradigms has frequently been directly dependent and directly linked to major scientific advancements others claim that in the two or three last decades the role of science as a major source of innovation and as a driver of the expansion of high tech industries has further increased Pecs, July , 2010 We don’t know, but: in knowledge intensive sectors scientific advance - remains extremely important and sometimes the initiating point of the process of innovation, often with long temporal and cognitive lags Basic research increases research productivity (Nelson, 1959; Mowery and Rosenberg, 1998). basic research does not necessarily coincide strictly with “pure science”: e.g. vaccines and most biomedical research. If the distinction between basic and applied research is blurred, why not simplifying the LM, instead of complicating it? (Stokes, user inspired basic research) Pecs, July , 2010 • user-developed innovations: niche products and highly sophisticated customers (mountain biking, surfing enthusiasts or surgeons etc.), in situations where the interplay between technical performance and practice is paramount. And in most cases, these customer-driven innovations rely on well established science (e.g. applications of new materials to surf boards) • science and technology are not perfectly malleable to economic and social signals (Dosi, 1982). Pecs, July , 2010 • According to a “weak” LM: basic research (and scientific advances) are neither necessary nor sufficient for innovation to take place, but remain very important Pecs, July , 2010 The critiques to the LM 2) Bottlenecks, feedbacks, interconnections • knowledge does not flow smoothly among different stages of the innovative process and among different organisations and institutions or geographical areas (tacitness, need of incentives) • But the LM may easily accommodate for the existence of impediments to the flow of knowledge. In fact, one might argue that it is exactly the use of a linear representation of the innovation process which has enabled researchers to identify bottlenecks. Pecs, July , 2010 • Technological progress is interactive in nature CHAIN LINKED MODEL • Given the fundamental role of design in triggering innovation, KR (1986) criticise the sequentiality of the process of technological change, stressing that the activities involved occur simultaneously and/or with continuous feedback among them • A constellation of concomitant tasks, instead of a sequence • This challenges the very notion of linearity Pecs, July , 2010 It is (correctly) destructive only of the strong form LM (and it applies mainly to incremental innovations) Is the LM is a model of the innovation process performed within individual firms? Or a model which applies at the macro level and considering the long run?. When considering science-based sectors such as biotechnology, we do not find so often the occurrence of concomitant tasks In sectors where the outcomes of basic research take a decade to reach the market, feedback from users will impact on current or future research projects, but cannot influence the research carried out a decade earlier (drugs, …) Before applying something, this something needs to exist Pecs, July , 2010 At a more theoretical level: 1) A process may exhibit feedback loops but remain linear: many linear systems exist in theory and in practice. 2) The fact that various components interact does not imply that they are completely and fully interconnected, and thus need to unfold in parallel. A system or a network can be partially decomposed in subsystems, linearly connected to each other. 3) There can also be different structures that cannot be classified simply into the two extreme forms. 4) Fully connected systems are very unstable systems and partitioning pays off in terms of stability, predictability and sheer manageability (project management builds on a linear sequence). 5) Danger: everything depends on everything else 6) Perhaps the LM can still be usefully applied at least within specific subsystems in some technologies Pecs, July , 2010 The critiques to the LM 3) The organisational/institutional dimension Systems of innovation literature (national, regional, sectoral) • there is a large variety of organisations, both public and private, that contribute to the generation of technological innovation • a large variety of institutions (the financial system, laws and practices governing labour markets, etc.) • the relations and interactions among the various actors are crucially important Not at odds with the LM: even within a system, significant relationships among agents may remain linear Pecs, July , 2010 the critiques of the LM may have gone too far: they focus attention too much on relationships rather than on the properties and characteristics of the individual components (nodes) of the system (network). A network is a network is a network ….. Systems and network theories are a language: everything can be represented as a system or a network Thus, it it is necessary to specify very carefully and in detail, the structure of any system or network under observation Often, networks and systems have a a hierarchical nature Pecs, July , 2010 The critiques to the LM 4) The normative dimension The LM does not bear any strong and obvious normative implication (If anything, the LM suggests the public support of basic, unfettered, research) Critiques: -At the micro level (firm’s strategies and organisation) -At the macro level : From science policy to innovation policy Pecs, July , 2010 The micro level • Decentralisation of R&D • Promoting dense knowledge flows within the firms and across different types of organisations • But: – Not necessarily at odds with the Weak LM – Need to strenghten integrative capabilities Pecs, July , 2010 The Macro level • The third mission of universities (Triple Helix, the European Paradox, …) • Implication: closer and flexible interaction among universities, firms and intermediate organisations should be promoted: transfer of knowledge • institutions should be created to facilitate these exchanges • Basic research should be exploited more aggressively for economic and social applications • Not necessarily at odds with the LM basic research comes first, but it does not trickle down Pecs, July , 2010 • Different arguments for justifying (alternative) normative prescriptions: Basic research too remote from application different sets of incentives (Dasgupta and David) Tacitness knowledge as information, partial appropriability Mode II: scientific research has become increasingly multidisciplinary and involves different types of institutions, techniques and methods (Gibbons et al., 1994). But: Continuing relevance of disciplinary based research: who pays for the overheads? See David vs. Kealey: these prescriptions can be sustained or criticised on the basis of the LM Pecs, July , 2010 Conclusions a) Basic research and science continue to be a fundamental – although certainly not unique – source of technological advance b) Frictions in the knowledge flow can be easily accomodated for in the LM c) Systems can be linear or linearly decomposed; the structure of the relevant networks must be clearly identified d) Time is irreversible e) Multiplicity of agents in the innovative process can be easily accomodated for in the LM f) As such, the LM does not imply strong normative prescriptions g) But critiques to the LM are used to support widely different suggestions and they are often mutually inconsistent; perhaps, the advocacy of stronger linkages among agents is even more compelling in the context of a weakened LM Pecs, July , 2010 Conclusion (2) • Either the LM is dead (or it never existed): then stop criticizing it • Or it is still alive in a weak form: why and where? • The LM in Weak Form might still be useful: – for understanding a subset of technologies, industries, activities – at a sufficiently high level of aggregation and/or over sufficiently long time horizons – As a conceptual tool for understanding and managing complex structures and relationships • Alternative models are often as generic as the LM – It is important to specifify and strengthen them Pecs, July , 2010 • “I come to bury the Linear Model, not to praise it” • Adapted from W. Shakespeare, Julius Ceasar, Act 3, Scene 2 Pecs, July , 2010