By closely examining the evolution of the Linux kernel, we have come to realise that management and structural organisation dynamics have played a major role in ensuring the rapid growth that Linux has enjoyed ever since its inception in 1991.[1] Of course, this is not to say that community ethics and norms are of secondary importance.[2] On the contrary, one of the most significant accomplishments of the Linux community, with regard to R&D organisation, lies in its successful effort to bridge the gap between a global developer community and various commercial players. The beneficial effects of relations of interdependence in the R&D realm have been stressed before[3], although the diffusion of such ideas has been limited due to the increasing corporate drive to pursue innovation through institutions of proprietary knowledge (such as patents and copyrights), which effectively reduce the incentive to share knowledge, and thus reduce the potential appeal of collaborative R&D endeavours to business managers.
Crudely speaking, the success of Linux is not a mysterious cyberspace tale that bears little relevance to the real world of R&D and technological innovation, as so many corporate managers reckon; instead the underpinnings of the success of Linux can be traced to the emergence of a new commercial actor: community-managed projects. In order to experiment with an open and evolutionary R&D paradigm, in which virtually every unit becomes actively engaged in innovation, Netscape's decided to open the R&D model of the Mozilla project. This process of radical decentralisation, as exemplified by the Linux development model, can be a realistic option for implementation across a wide spectrum of industries.[4]
In fact, Japanese companies have been staggeringly successful in leading collaborative R&D projects. It is widely acknowledged that Japanese organisations employ a longer-term focus than their western counterparts, and that intra-industry sharing of information among clusters of closely-knit lifelong business networks is the norm in Japan. After all, the optimisation of dynamically re-configurable supply chains around the timely supply of production inputs and the equally well-timed synchronisation of supplier-buyer processes is being commonly practiced in Japan since the 1950s. The necessary alignment of operating processes across the group of industry partners, coupled with the free exchange of personnel, ideas, products, and services within the subcontracting network, culminated to the Just-in-Time phenomenon, which preceded the explosion in outsourcing practices. This mindset has certainly helped Japanese organisations explore win-win opportunities in the pursuit of technological innovation, however, Japanese business networks are usually hermetically closed to the outsider and the only way to participate in those networks of business intelligence is through official procedures and shared context provided by history, geography, and other social and economic bonds. Therefore, there also remarkable downsides to the Japanese version of collaborative R&D, most noticeable downturn of which is the case of managerial corruption creeping into the system.
Interestingly, the approach of technology fusion[5] that Japanese companies such as Fanuc, NEC, Sharp and Toray have adopted for organising their R&D projects, unveils striking parallels to the R&D organisation of the Libre Software community. Technology fusion refers to the process of merging existing technologies in order to produce hybrid, yet entirely new technologies. For example, the fusion of optics and electronics gave birth to fibre-optics. What makes this approach so special is that technology fusion rests upon cross-industry relationships of mutual commitment and reciprocity, and managerial reluctance toward partnering with external R&D entities is not an option as it destroys the very same capabilities on which the model of technology fusion thrives. In other words, not only companies have to tightly co-operate with companies competing within the same market, but most importantly they have to forge partnerships with companies in otherwise unrelated industries in order to tap into the (primarily technical) creative synergies that technology fusion seeks to mould.
First, everything starts with market demand articulation. If the customer wants a new product or service, then that is the moment an R&D project is initiated. Once a market need has been clearly defined, the R&D attention shifts to gathering knowledge. But intelligence gathering is no easy and error-free given that companies have to gather knowledge from many different industries and diverse sources such as industry and academic publications, professional associations and clubs, universities and educational research labs, government-run research facilities, and independent technology houses, to name but a few. Hence, the process of collecting and evaluating potential candidates becomes an extraordinarily complicated task. The only way to enhance the effectiveness of the research exploration is to distribute the task across the tightly co-ordinated web of partners, and to involve as many partners as possible. The diversity (in terms of diverse sets of capabilities) that flows into the network ensures that a large enough array of technologies will be considered and their potential will be critically evaluated by those units possessing the most appropriate skills. In doing so, an emergent self-selection of tasks takes place as each unit focuses on its own area of expertise. As the R&D network gains access to more and more domains, the bonds of the R&D network are strengthened and this invariably sets a collaborative strategy in motion. It should come as no surprise that those units (or clusters of units) that finally come up with the best-positioned technologies are the ones which emerge as the leaders of the project. This is justified both in terms of their involvement in the network of peers, but it mostly reflects their ability to convince the remaining partners of the feasibility of their chosen re-combination. In practical terms, if ten units explore the potential of ten different technologies, and the third and fourth unit manage to persuade all other units that the two technologies they had been investigating are the ones to marry, then those two units will lead the project. We need to remind ourselves that for such an R&D approach to work, everything needs to be based upon relationships of mutual commitment and reciprocity. Otherwise, knowledge will inevitably fragment into clusters and will not be shared with a detrimental effect on innovation.
The final stage in the technology fusion approach is joint, cross-industry R&D. Having identified what the market wants and having concluded which technologies will be fused, the network of partners proceeds on to the actual development process. All partners contribute funds and specific skills to the project and all partners benefit from the development effort. This is a true win-win R&D proposition: it reduces risk, encourages innovation and growth through collaboration, increases organisational learning, and seeks to bring useful technologies to the market.
By now, the parallel to the libre software development process should be self-evident. Most libre technologies initially aim at 'scratching a developer's personal itch' rather than just being dynamic and confusing for the sake of technological novelty and marketing hysteria. Furthermore, self-organisation around community ethics and norms is common in the libre software sphere and as I argued above similar community dynamics (mutual commitment, reciprocity, network of peers, sharing of risks and benefits) play a pivotal role in the organisation of technology fusion. In a similar vein, emergent task-ownership and leadership is characteristic of collaborative R&D with all stakeholders being involved in the decision-making process, and much of the development route being decided in an evolutionary bottom-up fashion.
Indeed, as the commencement of several R&D projects demonstrates, the focus when trying to reproduce the key success factors of libre software development should be on management and organisational aspects rather than obscure technical issues.[6] I am involved in one such collaborative effort, the Organis Project, which seeks to provide autonomously run clusters of coding talent with the essential infrastructure resources for the R&D process to get off the ground while laying out the organisational guidelines along which decentralised development over the Internet will be harnessed. The scope of the project is not confined to software, and we are confident that libre software development is a harbinger of an evolving type of peer production activity that slowly but gradually rises to prevalence.
This presentation will delve into the organisational and management aspects of libre software development and will suggest ways to apply its R&D organisational logic to other areas of production activity. The presentation will also address the problem of bridging the gap between a non-profit oriented global developer community and commercial organisations. However, I should mention that the element of intelligence gathering, and in general the research component of R&D, will be analysed with reference to weblogs such as slashdot, newsforge, and openflows, as they represent an immensely valuable resource embedded in the libre software community. Weblogs' role in such collaborative R&D efforts will be discussed and thoughts for the future will be shared.
Notes
[1] See Dafermos, G.N. 2001. Management and
Virtual Decentralised Networks: The Linux Project, First Monday, at
http://www.firstmonday.dk/issues/issue6_11/dafermos/
[2] Eksteen J. and Jackson C.A. 2000. Open
source, the Internet and national R&D organization: Where to?, at http://www.isoc.org/inet2000/cdproceedings/posters/257/
[3] See Raymond, E.S., 1999. The Magic Couldron,
at http://www.tuxedo.org/~esr/writings/magic-cauldron/
and particularly http://www.catb.org/~esr/writings/cathedral-bazaar/magic-cauldron/ar01s15.html
[4] See Barber, D. 2001. The Open Source Development
Model: is it applicable to other industries?, Mojolin.com, at http://mojolin.com/articles/open_source_model.php?session=vTVi4tc1GfTb
and D. Stark and G. Neff, 2003. Permanently Beta: Responsive Organization in
the Internet Era, in P.E.N. Howard and S. Jones (Ed.) The Internet and American
Life, Thousand Oaks, CA: Sage., at www.coi.columbia.edu/pdf/neff_stark_pb.pdf
[5] For an elaborate discussion of technology
fusion, see Kodama, F. 1992. Technology Fusion and The New R&D. Harvard
Business Review, July-August.
[6] See http://sohodojo.com/techsig-sponsors.html
and http://organis.org
About the author
George N. Dafermos is an independent researcher and author based in Crete, Greece. He can be contacted via e-mail at dafermosATdatahostDOTgr .