Article information: Access to this document was granted through an Emerald subscription provided by Emerald Author Access

Transcription

1 European Journal of Innovation Management Emerald Article: Identifying collaborative innovation capabilities within knowledge-intensive environments: Insights from the ARPANET project Pierre Barbaroux Article information: To cite this document: Pierre Barbaroux, (2012),"Identifying collaborative innovation capabilities within knowledge-intensive environments: Insights from the ARPANET project", European Journal of Innovation Management, Vol. 15 Iss: 2 pp Permanent link to this document: Downloaded on: References: This document contains references to 67 other documents To copy this document: permissions@emeraldinsight.com Access to this document was granted through an Emerald subscription provided by Emerald Author Access For Authors: If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service. Information about how to choose which publication to write for and submission guidelines are available for all. Additional help for authors is available for Emerald subscribers. Please visit for more information. About Emerald With over forty years' experience, Emerald Group Publishing is a leading independent publisher of global research with impact in business, society, public policy and education. In total, Emerald publishes over 275 journals and more than 130 book series, as well as an extensive range of online products and services. Emerald is both COUNTER 3 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation. *Related content and download information correct at time of download.

2 The current issue and full text archive of this journal is available at EJIM 15,2 232 Identifying collaborative innovation capabilities within knowledge-intensive environments Insights from the ARPANET project Pierre Barbaroux French Air Force Research Center, French Air Force Academy, Salon de Provence, France Abstract Purpose This article aims to identify the capabilities supporting the development of collaborative innovation within knowledge-intensive environments. Design/methodology/approach Re-considering the history of the ARPANET project as a vivid example of collaborative innovation, the article presents qualitative research from a historical case. Findings Within this framework, the article shows that benefiting from collaboration in innovation entails that the innovative organisation is capable of achieving (at least) the following tasks: to leverage complementarities between internal and external sources of innovation (design capability); to codify, capitalise and disseminate knowledge outcomes (knowledge management capability); and to align product and organisations in a dynamic way (adaptive governance capability). Research limitations/implications This contribution is limited by looking at a single case. On the premise that model generalization depends on extensive empirical data, the current article should be considered as preliminary/exploratory research that aims at identifying the capabilities supporting collaborative innovation within knowledge-intensive environments. Originality/value The originality of this article is to look at a historical case to elaborate on a typology of collaborative innovation capabilities. Keywords Collaborative innovation, Capabilities, ARPANET, Organizational innovation, Partnership, Knowledge organizations, Innovations, Knowledge management Paper type Case study European Journal of Innovation Management Vol. 15 No. 2, 2012 pp q Emerald Group Publishing Limited DOI / Introduction Research on innovation and R&D practices has put particular emphasis on the role played by collaborative organizational forms to support the invention and commercialization of new products and/or services (Miller and Morris, 1998; Miles et al., 2000, 2006; Nobelius, 2004; Weaver, 2008). Scholars acknowledged that, in an increasingly interconnected and turbulent economy, innovation requires that the firm is able to interact and collaborate with others (including users, suppliers, rivals, etc.) to create, absorb, combine and integrate a variety of in-sourced and out-sourced knowledge (Cohen and Levinthal, 1990; Kogut and Zander, 1992; Chesbrough and Teece, 1996; Chesbrough, 2003; Dodgson et al., 2005; Chesbrough and Appleyard, 2007; Bierly et al., 2009; Miles et al., 2009). Following Dooley and O Sullivan (2007, p. 389), organizations operating in this collaborative era face new challenges in effectively structuring and managing

3 innovation. Basically, the development of collaborative innovation does not lessen the need for nurturing in-house innovative assets, but involves that the innovating firm is also capable to establish relationships with external partners so as to exploit a collection of internally and externally distributed, often fragmented, pieces of knowledge (Miles et al., 2000, 2006; Swan et al. 2007). Although the literature on innovation management and the theory of the firm have long acknowledged the role played by capabilities in providing the firm with competitive advantage (Wernerfelt, 1984; Barney, 1991; Teece et al., 1997; Dosi et al., 2008), there is a need to improve our understanding of the nature and logics of the capabilities which enable the innovative firm to reap the full benefits of collaboration in innovation management. How does the firm manage to identify internal and external sources of innovation, integrate them effectively, and foster the diffusion of innovation outcomes? To address the previous questions, this article adopts an exploratory methodology and presents a qualitative research from a historical case. The case focuses on the development of the first computerised communication network by the US Advanced Research Project Agency (ARPA) in the late 1960s: ARPANET. The ARPANET project involved the participation of distinctive communities of scientists, R&D firms and telecommunication companies under the heading of a public establishment: the Information Processing Technology Office (IPTO). By building on a historical case study, this contribution seeks to investigate the nature and logics of the capabilities deployed by the IPTO and its partners to create the ARPANET. The article begins by reviewing the literature on innovation and R&D management and insists on the knowledge-intensive tasks the innovative firm must be able to realise so as to develop collaborative innovations. It continues on by presenting the research methodology adopted to document the history of the ARPANET project. Next, the case study findings are introduced. This section discriminates between two themes. The first theme focuses on how the IPTO designed a specific problem-solving organizational form to grapple with a number of technical problems and coordinate a variety of internal and external resources, communities and organizations. The second theme is concerned with the role played by the Network Working Group (NWG), a multidisciplinary community made up with representatives of users, developers and government agencies, in codifying the core concepts embedded in the network and facilitating their dissemination and implementation. Finally, the main implications of the findings are drawn and a typology of collaborative innovation capabilities is introduced and discussed. ARPANET project Conceptual background According to Miller and Morris (1998), innovation is an expression of learning that occurred as a process of mutually dependent problem solving among members of diverse communities (Miller and Morris, 1998, p. 122). Managing innovation requires that the firm is capable of coordinating a variety of creative, productive and marketing resources, including money, technological artefacts, human skills, marketing knowledge and social capital (Dodgson et al., 2008), from a variety of actors, including mobile workers, user communities, component suppliers, rival firms and venture capitalists (Chesbrough, 2003). Within this framework, the sources of innovation are not easily accessible to the firm (Pénin, 2008). They are distributed inside and outside its own financial, technological, organizational and cognitive

4 EJIM 15,2 234 boundaries (Dooley and O Sullivan, 2007). It follows that the innovative organization must find ways to access, combine and integrate heterogeneous sources of knowledge so as to develop new products and services (Miles et al., 2000; Blomqvist and Levy, 2006). One way of achieving these highly demanding knowledge-based activities is to harness the power of collaboration considered as an effective mode of manaagement for knowledge-intensive innovation and R&D projects (Nobelius, 2004). Fundamentally, collaboration consists in a process where two or more parties work closely with each other to achieve mutually beneficial outcomes (Miles et al., 2006, p. 2). When transposed into the context of innovation management, collaboration designates a process involving at least two separated but interacting entities (individuals, teams, communities and/or organizations) that create and exchange knowledge, and share a common goal consisting in inventing and commercialising a new product, technology and/or service. As Ribeiro-Soriano and Urbano (2009) argued, collaborative entrepreneurship emphasizes the possibility of creating something of economic value based on new, jointly generated ideas that emerge from the sharing of information and knowledge (Ribeiro-Soriano and Urbano, 2009, p. 421). Blomqvist and Levy (2006) further argued that in dynamic and uncertain environments in which unusual situations demand coordination (...) knowing how to collaborate helps the firm to create and transfer knowledge in pursuit of innovation and better performance (Blomqvist and Levy, 2006, p. 40). How does the firm manage to grapple with these demanding knowledge-oriented activities? Investigating how firms grapple with joint innovation processes, scholars emphasised the role played by trust, commitment and communication (Blomqvist and Levy, 2006). Given that collaboration involve the sharing of tacit knowledge among various levels and dimensions inside and outside the firm, it follows that organizations and managers that are capable of creating conditions that build and sustain trust, including a commitment to equitable allocation of the returns on innovations, are more likely to be successful (Miles, 2007, p. 195). The establishment of an atmosphere of mutual trust, based on honesty and openness, enables individuals to freely collaborate in the process of innovation, sharing tacit knowledge and creating new knowledge out of combinations and new interpretations of the pieces of knowledge each possesses (Miles, 2007, p. 195). The foregoing led scholars to insist on relational attributes rather than on purely transactional factors (Blomqvist and Levy, 2006, p. 40), particularly when the pursuit of innovation requires coordination of distributed resources that go beyond the firm s boundaries. Within this framework, the capacity of the firm to build and manage truthful personal relationships with external partners is critical to ensure proper knowledge creation and integration (Tödtling et al., 2009). Complementing the previous perspective, scholars further suggested that participants cultural values and norms might have an impact on their ability to collaborate successfully. Focusing on large projects within defence and aerospace industries, Lawrence and Scanlan (2007, p. 514) explained that the cultural values shared by the actors involved in a project critically affect performance. Together with the sharing of a common goal, individual and/or organizational cultural compatibility is a condition for interpersonal relationships to develop and ensure fruitful interactions. As Edmonson and Nembhard (2009) noticed, collaboration is difficult because each profession has its own language, terminology, beliefs about relative importance of performance attributes,

5 approaches to learning, mechanisms for information exchange, goals, and reward structures (Edmonson and Nembhard, 2009, p. 128). It follows that the collaborative organization set up to invent and commercialise new products and services must be culturally consistent, the notion of cultural consistency being rooted in the level of compatibility between distinctive cultural values. Here again, Lawrence and Scanlan (2007) explained that when innovation requires large-scale collaborative organizational forms, the desirable cultural features revolve around openness, honesty, knowledge sharing and trust (Lawrence and Scanlan, 2007, p. 514). Dooley and O Sullivan (2007) went forward indicating that cognitive and geographic distance among participants involved in joint projects is likely to affect the value of collaboration. Distance, Dolley and O Sullivan explained, can seriously impede effective knowledge exchange (Dooley and O Sullivan, 2007, pp ). Considering the importance of culture and distance issues, it appears that collaborative innovation management requires that the firm is capable of balancing the tension between diversity and proximity. On the one hand, diversity fosters knowledge creation and integration (Love and Roper, 2009); on the other hand, proximity reduces potential communication failures to occur among partners (misunderstanding, disagreement, ambiguity, conflict) and ensures cultural consistency. This tension characterises inter-individual and inter-organizational relationships within multicultural, cross-functional work environments. The foregoing view of innovation as a collaborative process is in line with the literature on innovation capability (Romijn and Alabaladejo, 2002; Swan et al., 2007). The later is defined as a specific type of dynamic capability enabling the firm to acquire, integrate, combine and recombine a variety of socio-material resources (Teece et al., 1997; Eisenhardt and Martin, 2000) so as to develop new products and services. As Branzei and Vertinsky (2006, p. 79) explained, successful innovation capabilities encompass firm s abilities to acquire and assimilate external knowledge, transform it into novel, unique competencies and ideas, and then harvest these ideas by first generating and effectively commercializing new or improved products. Here again, proximity and trust have been identified as influencing factors on the ability of the firm to build up personal relationships, collect information about markets and technology and nurture fruitful knowledge exchanges with external partners (Romijn and Alabaladejo, 2002, p. 1055). In this view, the capacity of the firm to:. access; and. absorb knowledge (Cohen and Levinthal, 1990; Zahra and George, 2002) is critical. ARPANET project 235 Investigating collaborative modes of innovation occurring in the UK and US biomedical industry, Swan et al. (2007) particularly insisted on the role played by integrative and relational capabilities in that they foster the acquisition and absorption of knowledge. On the one hand, the authors argued, integrative capabilities facilitate the translation of basic research into commercial applications through the movement of scientists and their enhanced labour market mobility (Swan et al., 2007, p. 531). On the other hand, relational capabilities facilitate innovation by (...) supporting collaborative product development projects (Swan et al., 2007, p. 531). Together, integrative and relational capabilities shape the way firms acquire innovation capabilities.

6 EJIM 15,2 236 This view of innovation as a collaborative process necessitating specific knowledge-oriented processes and capabilities led scholars to focus on the properties of the relational structures and organizational forms supporting interactions among partners. In particular, research efforts have been directed towards addressing the impact of networks properties on the capacity of the firm to create, share and disseminate knowledge (Van der Walk et al., 2011). Network-centric organizational forms deserved particular attention from scholars, in particular those interested in addressing the influence of the spatial location of basic research, R&D, production, and distribution activities on innovation performance (e.g. Maggioni et al., 2007). By focusing on networks properties (e.g. cohesion, density, frequency of interactions, centralization, see for example Cowan and Jonard, 2004), scholars demonstrated that specific modes of interaction are determinative for the capacity of the firm and its partners (including universities, research institutes, suppliers, and customers) to create and commercialise new products and services (Nieto and Santamaria, 2007; Tödtling et al., 2009). Therein, innovation is seen as a process which results from various interactions among different actors, the later being shaped by consciously designed organizational forms which facilitate the accelerated flows of information, resources and trust necessary to secure and diffuse innovation (Zeng et al., 2010, p. 182). It should be noted that students of innovation in complex product systems (CoPS, Prencipe et al., 2003) long acknowledged that the architectural properties of the organizational form (e.g. network, cluster, modular structure) deployed by the firm to innovate shapes how it encourages participation, coordinates contributions and, at the same time, controls the resulting innovation outcomes (Sosa et al., 2004). By investing time and resources to design organizational forms aligned with product architectures, it has been argued that the firm is likely to manage interdependencies between dispersed knowledge sources and reduce integration and communication costs (Ethiraj, 2007). According to Dodgson et al. (2005), the ability of the firm to engage in technological innovation resides in the combination of creative leadership in design and development linked with effective integration of other productive functions (Dodgson et al., 2005, p. 24). The authors also insisted on the capacity to manage complexity, and the ability to fully engage the users of innovation in the process of its realization (Dodgson et al. 2005, p. 24). Hence, the capacity of the firm to design organizational forms properly is determinative for its ability to support collaboration among heterogeneous participants involved in large, knowledge-intensive projects (e.g. users, suppliers, university, and financial institutions). It follows from the previous literature review that managing collaborative innovation requires that the firm is capable of achieving many different tasks. It must be capable of accessing, diffusing and integrating heterogeneous pieces of knowledge thanks to the establishment of relationships among various partners. It must also be capable of coordinating multicultural, cross-functional teams made up with heterogeneous individuals and organizational units (Edmonson and Nembhard, 2009). Finally, it must pay attention to designing proper organizational forms and governance mechanisms so as to reduce the likelihood for communication failures to occur and improve coordination (Prencipe et al., 2003). These tasks and related capabilities are usually considered separately, scholars focusing their research efforts on one particular task or dimension (e.g. absorptive capacity, cross-functional teams, network structure, or governance issues). The next sections build on a historical

7 case study to elaborate on an integrated conceptualization of the capabilities involved in the management of collaborative innovation. 3. Research design 3.1 Methodology This contribution adopts an exploratory methodology based on a qualitative research from a single case (Eisenhardt, 1989). In particular, a historical case study is developed focusing on the ARPANET project. The latter provides a vivid example of what collaboration means for innovation management within knowledge-intensive environments. As such, the development of ARPANET had necessitated a number of public agencies, research laboratories, R&D firms and telecommunication companies to collaborate under the heading of a public establishment: the US Department for Advanced Research Projects Agency (ARPA). The project exemplifies how innovation management depends on the ability of the corporate sponsor (i.e. ARPA) to encourage interactions among participants, both formally and informally, so as to solve a variety of technical problems, codify and disseminate knowledge, and promote the diffusion of innovation outcomes. By opting for a historical methodology, the objective is to deconstruct the innovation capabilities on which the ARPANET project relied, identify their main attributes and analyse their main implications for researchers and practitioners. Building on the case study findings, we aim at elaborating on a series of conceptual arguments focusing on the project s main organizational and technological attributes, and compare these arguments with existing theory on collaborative innovation (Glaser and Strauss, 1967). The unit of analysis therefore is the ARPANET project, and we investigate its historical development from its inception as a scientific concept emerging within the early 1960s computer science community, to its material incorporation into some reliable technological artefact. We proceed to an in-depth analysis of its technological and organizational attributes, paying particular attention to the functions, competences and objectives assigned to the various individuals and organizations participating in the project. This also leads us to focus on the nature of their interactions and explore how they create and exchange knowledge. This methodological position is neither unique nor singular. Weick (1993) adopted a similar perspective to identify the sources of resilience in organizations by re-analysing the history of the Mann Gulch disaster. Hargadon and Douglas (2001) also applied a historical methodology to examine how the social and institutional embeddedness of Thomas Edison s system of electric lighting (a radical innovation) influenced its adoption and further commercial exploitation and development. In the same vein, Scranton (2007) developed a historical case study of the US military Jet-propulsion industry in the World War II and early Cold War periods to elaborate on the concept of dynamic innovation. Recently, Lenfle (2009) explored the organizational logics supporting the development of the Manhattan project and introduced a theoretical framework which reconsiders the concepts of explorative innovation project and management. Several limitations need to be addressed before starting with the presentation of the case study. First, as the current article is limited in looking at a single source of evidence, it remains impossible to draw on general insights from the case study findings. On the premise that generalisability depends on extensive empirical data, this ARPANET project 237

8 EJIM 15,2 238 research should be considered as exploratory in its attempt to identify the capabilities supporting the development of collaborative innovation. In addition, practical as well as theoretical implications that might be drawn from our case study are circumscribed to specific technological, socio-organizational and historical contexts. Their relevance is thus definitely grounded in unique data and do not encompass the diversity of innovation projects in other industry and historical contexts. 3.2 Data analysis To document the case, a variety of data from books, academic publications, archives, interviews and documentations published by the various institutions, organizations and researchers that participated in the development of the ARPANET, has been collected and analysed. The analysis of the data followed a three-step process of classification, mapping and synthesis (see the Appendix). First, data sources have been classified according to their form and content. This preliminary classification provided elementary associations between the various data sources and their descriptive, illustrative, comparative and/or analytic content. Second, connections among the previous classified data have been established by employing three criteria: chronology, technology, and actors. This allowed for a detailed mapping of the data according to their form, content and major topics. Third, the previous structured data have been scrutinised (hand-made analysis) resulting in the identification of the following themes: participants skills and responsibilities, technical challenges and scientific problems, product and organization design issues and knowledge management processes (e.g. codification and dissemination). These themes shaped the way the presentation of the case-study findings have been organised (see the Appendix) 3.3 A brief history of the ARPANET In 1958, the US Department for Advanced Research Projects Agency (ARPA) was created by the US Department of Defense (DoD) to promote applied research in computer technology. With the intensification of the Cold War, it was expected that this emerging technological field could deliver decisive command and control (C 2 ) capabilities to the US military and maintain the technological superiority of the US services over the Soviets. Indubitably, one of the most outstanding accomplishments of the ARPA had been the development of the first wide-area computer network: the ARPANET (Kleinrock, 2008). The concept of social interactions that could be enabled by computer-mediated networking appeared at the very beginning of the 1960s. In 1962, Joseph Licklider and David Clark, researchers at the Massachusetts Institute of Technology (MIT), envisioned in a memorandum untitled On-Line Man Computer Communication a globally interconnected set of computers through which everyone could quickly access data and programs from any site (Licklider and Clark, 1962). Very rapidly, Licklider became the first head of the computer research program at the ARPA, called the Information Processing Techniques Office (IPTO). While at the IPTO, he argued for the importance of the concept of networking in remote computer-mediated communication (Leiner et al., 1997). In 1964, Paul Baran at the Rand Corporation (Santa Monica) and Donald Davies at the National Physical Laboratory (Middlesex, UK) investigated the conditions for deploying highly reliable computer networks by exploring the concept of distributed communication (Baran, 1964; Davies and Barber, 1973). Two years later, working with

9 Thomas Marrill (Computer Corporation of America, Cambridge Massachusetts) on Leonard Kleinrock s (UCLA) packet switching theory (Kleinrock, 1964), Lawrence Roberts (MIT) empirically demonstrated the feasibility of communications using packets rather than circuits (Roberts and Marrill, 1966). The result of this experiment was the realization that time-shared computers could work well together, running programs and retrieving data as necessary on the remote machine. In late 1966, Roberts went to ARPA[1] to develop the computer network project. He published his plan for the ARPANET ( Multiple Computer Networks and Intercomputer Communication ) at the Association for Computing Machinery (ACM) Symposium on Operating Systems Principles (Gatlinburg) in October The IPTO was responsible for a program entitled Resource Sharing Computer Networks which objectives were to develop technologies and obtain experience on interconnecting computers, and improve applied research in computer and resource sharing (BBN, 1981, II-2). In particular, this program aimed to fund and coordinate research efforts in distributed computer-mediated communication networks, and to select prime contractors for the nodes and the overall network design (BBN, 1981, II-10). In July 1968, a competitive procurement was prepared and issued for the selection of prime contractor. Bolt Beranek and Newman Inc. (BBN) was selected in January 1969 for designing the Interface Message Processor (IMP, BBN, 1981, II-11). The IMP appeared critical since they enabled distinctive hosts to communicate with each other thanks to a telephone line. The BBN team consisted of seven computer scientists (including William Crowther and Robert Kahn) placed under the supervision of Franck Heart. In September 1969, after BBN finished with the development of the IMP, Leonard Kleinrock s Network Measurement Center at UCLA was selected to be the first node on the ARPANET. Kleinrock assembled a research team of computer science graduate students (including Charlie Kline, Steve Crocker, Jon Postel and Vint Cerf) to prepare for connection as the first node on the experimental network (XNET). One month later, Douglas Engelbart s project on Augmentation of Human Intellect at Stanford Research Institute (SRI) provided a second node. Two other nodes were selected to complement the network architecture: Glen Culler and Burton Fried at the UC Santa Barbara (UCSB) and Bob Taylor and Ivan Sutherland at the University of Utah. The first four host sites formed what is called the experimental network (XNET). In December 1970, the initial ARPANET Host-to-Host protocol called the Network Control Protocol (NCP) was completed by the Network Working Group (NWG). The NCP was the first transport layer protocol of the ARPANET, later to be succeeded by the TCP (Kleinrock, 2008, p. 13). Remote computers were added quickly to the ARPANET during the following years, and work proceeded on completing a functionally complete Host-to-Host protocol and other network software. In the early 1970s (1971/1972), network users could begin to develop applications, including commercial applications. In October 1972, a public demonstration of the ARPANET was organised in Washington during the first International Conference on Computer Communications (ICCC). Robert Kahn (SRI) installed a complete ARPANET node at the conference with about 40 active terminals permitting access to dozens of dispersed computers. Each host site was invited to participate in the demonstration. An air traffic control simulation involving many geographically distant interconnected sites was successfully organised during the conference. The ARPANET demonstration ARPANET project 239

10 EJIM 15,2 240 lasted for three days and proved the feasibility of packet switching theory and reliability of the network (Roberts, 1985). In 1974, a common language was developed through the participation of a team of scientists at Stanford (SRI) under the supervision of Robert Kahn and Vinton Cerf, which would allow different packet networks to interconnect and communicate with each other. This was known as a transmission control protocol/internet protocol (TCP/IP). In 1977, the transmission control protocol (TCP) is used to interconnect three networks (ARPANET, PRNET and SATNET) in an intercontinental demonstration (Kleinrock, 2008, p. 13). This gave birth to what shall become the INTERNET, a connected set of networks using the TCP/IP standard. 4. Case study findings Next sections discriminate between the following themes to organise the presentation of findings: organization design, participants role, competences and contributions, and codification of knowledge. 4.1 Organization design Organization design refers to the deployment of an organizational form that fits with the characteristics of the innovation project. It involves the selection of appropriate sources of knowledge with respect to problem-solving activities, the establishment of formal and informal linkages among them and the definition of specific modes of interaction and coordination. Fundamentally, the organizational solution adopted by the Office (IPTO) had been supported by two pillars: (1) Decomposition. (2) Collaborative relationships. Decomposition allowed for treating interdependent technical problems as if they were independent, while collaborative interactions fostered knowledge exchanges and contributed to the establishment of a dominant design for the network. In addition, IPTO promoted direct and informal interactions among participants and did not impose systematic central control over local developments The implementation of a problem solving organization. As a complex and technically demanding project, the development of the ARPANET required a proper definition and attribution of both technical and organizational authority and responsibility. Therein, the selection and division of tasks among participants had been primarily determined by their respective expertise and capabilities regarding a number of key technical problems. Seven technical problems needed to be handled before connecting the first four sites of the network: topology, error control, host interfaces, switching node performance, remote control, routing and host protocol (BBN, 1981, II-13-19). Each problem required interactions and active collaborations between representatives of the host sites (research centres at universities), R&D companies (NAC, BBN and Honeywell), and government agencies (IPTO, RML and DSS-W). Table I presents the seven technical problems and the solutions initially adopted to cope with them. By identifying major technical problems, the IPTO made a first step towards selecting the individuals and organizations (i.e. designing an organizational form) capable of conceiving local solutions to the various domain-specific problems and at

11 Challenge Formulation Solution Topology To design network architecture (arranging M links among N nodes) subject to given constraints (e.g. maximum/average time delays, reliability) at the lowest cost Leonard Kleinrock (NMC-UCLA) proposed a design algorithm, the Concave Branch Elimination (CBE). NAC submitted a competing solution called Cut Saturation Method (CSM). The latter solution had been adopted ARPANET project 241 Error control Host interface Switching node performance To minimise errors in data (packets) transmission from one site to another To allow a logical match between switching nodes and the host site, each using distinctive language (the word length of the interface and the computer hosts vary) Allowing remote computers to share resources and data requires effective switching nodes in terms of reliability and computing speed The solution consisted in comparing input data with output data according to a cyclical procedure. A message was then transmitted to the transmitting node is there was no error. If there was an error, no message was send to the transmitting node and the packet was redirected The solution adopted by BBN consisted in implementing a dedicated hardware unit in each host machine, called special host interface, to enable dialogue between host sites and the IMP BBN elected the Honeywell 316 computer to be integrated into IMPs because it provided users with high capacities at modest costs Remote control The increasing number of computers in the network requires designing a technology for managing, updating and debugging remote computers A Network Control Center (NCC) was established and software was developed which made it possible to examine or change the operating software in any node of the network from the NCC Routing How to streamline the decision process by which each node decides to route information to reach any particular destination node? A distributed traffic routing algorithm had been developed which estimated on the basis of information from adjacent nodes the globally proper path for each message to be transmitted. This enabled adaptation to varying traffic loads and potential lines and node failures Host protocol To design a common language called the Host Protocol so as to facilitate communications among heterogeneous machines Source: Adapted from BBN, 1981, II The NWG developed a program called the Network Control Program (NCP) which function was to establish, cut and switch connections between host machines and the network, and to control information flows Table I. Seven technical problems and solutions

12 EJIM 15,2 242 the same time integrating them properly. Each participant involved in the project held particular skills and expertise within a number of scientific, managerial, engineering, industrial and organizational areas The selection of heterogeneous but complementary participants. The ARPANET project had necessitated interactions among a range of heterogeneous communities of scientists, R&D firms and telecommunication companies under the heading of a public establishment (ARPA-IPTO). In addition, a number of government agencies assisted the IPTO within specific domains IPTO and government agencies. The ARPANET project required a joint effort of many government agencies, all responsive to the Information Processing Techniques Office (IPTO) at the ARPA. These agencies offered critical support capabilities which enabled the IPTO to supervise the development of the network, control costs and play a leading role in the emergence of modern IT industry. The effective coordination between government agencies (IPTO, DECCO, RML and DSS-W) facilitated the development of expertise in key organizational and technological domains which, in turn, enabled the implementation of effective governance mechanisms. As such, the capabilities offered by the IPTO and other government agencies reflected the internal resources injected by the U.S. Department of Defense (DoD) to develop the project. More precisely regarding organizational and technical leadership, the key participant in the ARPANET project was the IPTO. Representing the ARPA, this Office made certain key architectural decisions [...] the IPTO set policy of the network, made decisions about who would join the network (BBN, 1981, III-26). As Leonard Kleinrock indicated, I think the IPTO was a prime mover for the United-States in the advancement of computer technology through advanced thinking and [...] what I shall say [...] heroic funding of the things they thought were worthwhile. Their motto was, high risks, high payoff [...] It was one of the great experiments in science, I think. It completely changes the way things are going on now commerce, government, industry, science, etc. (Interview with Judy O Neill, Charles Babbage Institute, Center for the History of Information Processing, 1990, April 3). A number of government agencies provided the IPTO with a bundle of management and technical support capabilities. Therein, three government agencies played a critical role. A procurement agency, the Defense Science and Security-Washington (DSS-W) supported the IPTO in providing expertise in contractual negotiations and interactions with contractors. The scope of DSS-W expertise also included technical monitoring of component contractors. In addition, the Range Measurements Laboratory (RML) at Patrick Air Force Base (Florida) offered procurement and technical support capabilities which complemented those of the DSS-W. Finally, the Defense Commercial Communication Office (DECCO) monitored the contractual arrangements made by the IPTO with the telephone companies which provided the telecommunication infrastructure of the network. In particular, DECCO acquired wide band facilities which enabled the rapid growth of the network in the 1970s Computer science community. In the mid-1960s, few research centres and/or companies had successfully connected remote computers for the purpose of experimenting with shared resources (BBN, 1981, I-5). Only the Western Data Processing Center at UCLA and the Bell Laboratories had been capable of enabling similar computers to perform load sharing. In this context, the most efficient way to

13 develop the techniques needed for an effective network was thought to be involving the research talent [...] in prototype activity (BBN, 1981, II-2). Two research centres played an active role in the initial development of the experimental network (XNET): the Network Measurement Center (NMC) at UCLA and the Network Information Center (NIC) at Stanford. The NMC had the responsibility for much of the analysis and simulation of the ARPANET performance, as well as direct measurements based on statistics gathered by the IMP program (BBN, 1981, III-39). Leonard Kleinrock at UCLA was involved in the definition, development and testing of all protocols and procedures. His competency was unanimously acknowledged within the computer science community. He also enjoyed personal relationships with Lawrence Roberts and a number of other key individuals involved in the project, in particular with scientists and engineers from R&D companies (NAC and BBN). The NIC, under the heading of Douglas Englebart at SRI (Stanford), was responsible for collecting data, codifying knowledge and developing the technological tools for storage and dissemination of technical documentations. As Leonard Kleinrock explained, SRI was an important member of the community [...] they were the second node of the network [...] Documentation went through them, as did the network RFC notes (Interview with Judy O Neill, 1990, April 3). The NIC had developed an on-line computer program to provide network users with electronic and hardcopy documentations. The latter included information about resources available at the various host sites, network protocols, languages and procedures, and a number of working papers and technical reports. More generally, the computer science community had been responsible for designing and implementing the hardware and software necessary to connect themselves to the network and to access to other hosts resources in the network. The academic community therefore contributed primarily by offering high level technological skills and user expertise, and sharing critical component knowledge in a number of technological domains, including microcomputers and software programming. Subsequently, research host sites participated in the ARPANET project by offering user and developer capabilities. They articulated the preferences, needs and constraints characterising the (future) users of the network. These skills and expertise represented key external resources which the IPTO needed to access so as to solve the multiple scientific problems that came along with the exploration of computer-mediated communication technology R&D companies. During the initial stages of the project what delineated the division of technical responsibilities among research centres, R&D companies and Telephone Companies was the specification of a key component of the network: the Interface Message Processor (IMP). Although Telephone Companies (AT&T and General Telephone) provided the physical architecture of the network (e.g. circuits, data sets and lines), representing the most costly element of the project budget, individual host sites could not communicate without reliable interfaces. Amazingly, the ITPO designated two research-oriented companies as prime contractors for developing the IMP and configuring the network: BBN and NAC. Bolt Beranek and Newman Inc. (BBN) was responsible for the development of the interfaces (IMPs). BBN subcontracted part of the work on IMPs with Honeywell and Lockheed. These companies offered the system hardware (Honeywell 316 computer ARPANET project 243

14 EJIM 15,2 244 and Lockheed SUE minicomputer), as the building blocks of the IMP core design concepts (ARPANET Study Final Report, 1972). The foregoing gave BBN a special position within the project. Although the IPTO participated in key managerial decision processes regarding the setting up of a network policy or the selection of host sites and prime contractors, it was Bolt Beranek and Newman (BBN) that provided day-to-day operation and maintenance. BBN carried out much of the day-by-day business of the network [...] without need for daily IPTO supervision (BBN, 1981, III-26). Assuming the responsibility for maintaining orderly network operation, BBN implemented a Network Control Center (NCC) which assumed the central tasks of scheduling and monitoring network operations. As quoted in the ARPANET Study Final Report (1972), the control center appears highly effective in the area of problem identification, diagnosis and restore action. The foregoing stemmed from the healthy relationships between BBN, Honeywell and Lockheed which shared the responsibility for the maintenance of the networks and conceived a reliable diagnosis system. In 1975, after the ARPANET became technologically mature, the day-to-day maintenance and regulation of the ARPANET had been transferred from BBN to the Defense Communication Agency (DCA). Regarding network topology, the IPTO selected another R&D company called NAC. The Network Analysis Corporation (NAC) had been created in the late 1960s by Howard Frank, Ivan Frisch (Berkeley) and Steve Carr (SRI) as a profit-oriented R&D company specialised in network optimization. Leonard Kleinrock, who supervised the NMC at UCLA, remembered: In about 1971 or so Larry [Lawrence Roberts] was at my house, and I suggested that he meet Howie [Howard] Frank to assist in the topological design problem. So I put them together. And it was a click. Then Larry gave NAC the contract for doing the topological design of the ARPA network (Interview with Judy O Neill, 1990, April 3). NAC was thus selected by the IPTO and awarded a contract for designing the network topology. This task involved the selection of host-sites, the establishment of links between them, and the reconfiguration of the network whenever a new node joined the network. According to the ARPANET Study Final Report (1972), along with BBN, NAC has a very important role in the planning and engineering of the ARPANET. Beside topological optimization tasks, NAC also developed software to assist the IPTO and the NMC in planning the evolution of the network topology Telephone companies. The telecommunication infrastructure of the ARPANET had been supplied by traditional telephone companies operating at local, national and international levels. DECCO negotiated with the relevant companies and managed to obtain the desired service at reduced costs. In the case of a circuit from UCLA to RAND, for example, most likely the service would be procured from General Telephone, the dominant telephone company in the Los Angeles area (BBN, 1981, III-32). At the national and international levels, two companies played a central role: AT&T and Bell Systems. The latter utilised their Long Lines division with which DECCO negotiated specific procurements. National telephones companies also supplied the components necessary to make up the service from the regional telephone companies (BBN, 1981, III-32). Although the telephone companies procured and operated the telecommunication infrastructure and provided formal and informal assets to coordinate the local and the global levels, they did not participate in the innovation process per se. As Leonard Kleinrock argued:

15 It has been said that the telephone industry, or the communications industry, had absolutely nothing to do with the development of the ARPA network [...] To first order, that was correct [...] IBM dropped out [...] AT&T was not involved as an organization [...] It took them decades to come up to the technology that the data processing guys developed in the ARPANET (Interview with Judy O Neill, 1990, April 3). The telephone companies provided external resources to the IPTO but did not contribute to exploring new architectural and component knowledge. Rather they participated in the exploitation of the network and supported its rapid growth and extension The definition of formal and informal modes of interaction and coordination. The ability of the IPTO to coordinate heterogeneous expertises had been determinative for the success of the ARPANET project. The contractual arrangements made by the IPTO to select and coordinate the participants involved in the project had privileged decentralization which in turn enabled theoretical experimentations and creative exploration. At the same time, the IPTO supported direct interactions and informal dialogues among participants. As Leonard Kleinrock remembered: We dealt with BBN directly. When we had a problem with BBN, we complained to Larry [Lawrence Roberts] and he would step in and make sure things were fixed up. It was not a formal relationship that required all kinds of paperwork to go back and forth. It was peers, and researchers, and developers. It was a friendly and efficient environment in that sense (Interview with Judy O Neill, 1990, April 3). The objective of the IPTO was to reap the full benefits of cross-functional, interdisciplinary teams by encouraging informal dialogue and knowledge exchanges. Again, Leonard Kleinrock explained that the culture of those early days of the ARPANET community was one of open research, shared ideas and work, no overbearing control structure, and trust in the members of the community (Kleinrock, 2008, p. 12). Even when telecommunication infrastructure required close attention, participants (i.e. BBN and telephone companies, like AT&T) interacted directly through informal contacts which enabled effective maintenance with a minimum network disruption (ARPANET Study Final Report, 1972). Consequently, the ARPANET technology could be considered as a natural outcome of the progressive R&D atmosphere that was necessary for the development and implementation of the network concept (ARPANET Study Final Report, 1972). Although the resulting informal and cooperative mode of management appeared to be aligned with the decentralised problem-solving organizational form adopted, its practical application did not come without tensions. The software development and, to some extent, maintenance practices of BBN had been criticised by the network community for it did not prevent from network failures and long term inoperability (ARPANET Study Final Report, 1972). Despite remarkable achievements regarding hardware (e.g. IMP) and software development (routing algorithm, host-to-host NCP protocol), there was a pressure for more effective (centralised and authority-based) coordination policy to avoid duplication of efforts and characterise the network as an entity. In other words, there was a need for cross-administrative and inter-organizational governance that would promote the coordination and cooperation necessary to achieve technical and commercial successes. This need ARPANET project 245