• Keine Ergebnisse gefunden

The role of costs and benefits or future costs and future benefits (MacKenzie &

Wajcman, 1999) certainly exerts an important influence on the shape and success of an innovation in market economies. Costs of development, market introduction, and production, return on investment, expected sales price compared to older technologies and competing solutions are all economic categories which influence the decision of the involved groups in an innovation project. For instance, Law and Callon (1992) describe nicely how the high – and over the project duration

continuously increasing – expected total development and production costs of a new military aircraft, the TSR2, raised substantial opposition among different actors in British government. In the end, economic reasons like the high costs, failure of securing overseas markets, and the availability of a cheaper alternative contributed together with other arguments to the cancellation of the project.

The users of e-Infrastructures are often confronted with high learning and installation costs for new computer applications and unclear returns on making these investments; they have multiple needs in regard to computers and their use in their professional communication and cooperation and they have to deal with different communication situations; they work in different organisational settings and financial arrangements; they are subject to pressures and demands from peers (and students) on the channels to be used for communication, endorsed research practices and methods, acceptable data and information sources, etc. In addition, there is likely to be under-investment and under-valuation of the human capital aspects of investment in e-Infrastructure, both because not enough attention is paid to securing continuity of key personnel and because inadequate resources exist to fund the documentation of software and practices for their use that can be used to aid continuity. US and UK scientists have expressed substantial concern about sufficient numbers of trained individuals for the full exploitation and maintenance of e-social science investments.6 The e-IRG proposes to increase efforts in the training of scientists and computer support personnel on working with grid

6 Unpublished summary reports NSF/SBE cyberinfrastructure workshops Sept 18, 2004 and

environments (Leenaars, et al., 2005). Extensive thought needs to go into devising the most effective management for e-Infrastructure projects. A cadre of

paraprofessionals may be needed to supplement Ph.D. researchers. It was noted that the actual learning of the new technologies is not time consuming; rather, it is their adaptation for specific uses in the laboratory that requires great amounts of (expensive) principal investigator time.

The producers of e-Infrastructures are particularly affected by standardisation and resulting network economies. There are several examples in the history of

computing in which the development of an industry standard either in relation to hardware, e.g. personal computers, microprocessors, or software, e.g. operating systems, human computer interfaces, provided a decisive push in the diffusion (Williams, 1997). An industry standard triggers two attractive consequences for the technology producers: First, the existing users of a technology benefit from

additional users because of network externalities and the customer base for this technology grows. Second, a large customer base creates economies of scale and makes mass production possible. Then products and ideas diffuse via social networks through a domino effect. Early adopters ease the adoption for less innovative second movers. This again helps others and the innovation spreads gradually. At a certain point the process tips and the innovation spreads

explosively, turns into an “epidemic”. The introduction of mobile phones in the mid 90’s is a good example. More and more people needed to be reachable when away from a fixed line; it became fashionable to communicate through mobile phones;

they became the standard communication device in certain contexts.

We examine the role of technology diffusion through social networks, arguing that these reduce learning costs, enhance usability and sustainability and create a social incentive structure. The role of economic incentives in technology adoption has been clear since Griliches’ (1957) analysis of the adoption of hybrid corn in developing countries. However, sociologists have long argued that social networks provide important ways in which technology is diffused, and in the hybrid corn debate, Griliches acknowledged the importance of such networks: “If one broadens my ‘profitability’ approach to allow for differences in the amount of information available to different individuals, differences in risk preferences, and similar variables, one can bring it as close to the ‘sociological’ approach as one would want to.” (Griliches, 1962, p. 330, cited in Skinner & Staiger, 2006).

Standardisation could also solve a major issue which hampers adoption of new technologies, namely the concern by (potential) users about the sustainability of new tools and the resulting interoperability. This is, of course, a fundamental issue in e-science more broadly. In order for social scientists to invest time and energy in e-social science, they need to be convinced that the tools that they are using will not become rapidly obsolete. For example, in the United Kingdom the very successful initial Pilot Demonstrator Project, SAMD (http://www.sve.man.ac.uk/

Research/AtoZ/SAMD), which has been used as a flagship example of the value added of e-social science, is built on a platform that has since become obsolete.

The successor project has essentially had to start from scratch because the new platform is not compatible with the earlier one. A related issue, which has also been raised in the United States, is that the successful development of middleware requires a support infrastructure that is beyond that envisaged by initial grants. Of course, hardening and sustaining research products is difficult because products are heterogeneous, the process is costly, and researchers are trained to break new ground, rather than sustain existing projects.

to institutional and field routines, practices, and cultures. They also have to bridge them whenever knowledge of different types needs to be combined and they need to create cross-disciplinary exchange and understanding.

Matching between technical capacities and surrounding conditions. In a recent study Wouters and Beaulieu (2006) argue that e-Infrastructures are not (yet) conceived in a way that permits their integration into the particular culture, habits, customs and organisational setting of fields in the social sciences and humanities.

They speak of a “misalignment between the emerging e-science community and other scholarly communities” (ibid., p. 62) and exemplify this by comparing the research practices and social relations in a particular social science field (women’s studies) with the current offerings of e-science. They conclude that e-science initiatives are still too much driven by computational research and the production of infrastructures for largscale data and computation, and that a different turn to e-science should be considered: starting at the analysis of different research fields, with particular research practices, communication and collaboration relations, and a specific social organisation to find out how their differing needs can be supported by new ICTs.

This work is in the tradition of earlier work that stresses the influence of the cultural particularities of a field on how the internet is used. The importance of differing work organizations and social structures as well as the external relations determine in Walsh and Bayma’s (1996) study how the internet is used by mathematicians, chemists, experimental biologists and physicists. Kling and McKim (2000) show that field-specific constructions of trust and of legitimate communication influence whether and in what particular way e-publishing has become part of the

communicative forum: Whereas high-energy physicists quickly adopted the arxiv.org e-print server as a central communication channel, some fields in computer science have established pure electronic journals, and molecular biologists rely on digital databases and shared digital libraries (like PubMed Central). Taking the case of a humanity field, namely corpus-based linguistics, Fry (2004) has highlighted that cultural elements exert a strong influence on the uptake and use of ICTs.

Political considerations. Winner (1999) provides several examples for technical solutions that are not primarily shaped by a technological paradigm or efficiency considerations, but consciously by political goals or subconsciously by a lack of consideration or awareness: the low height of bridges in New York intended to keep public transport and thus poor people and minorities out of certain areas;

inefficient moulding machines that had the only advantage that they could be run by unskilled labour were used to destroy the union influence in a firm; the 1970s movement of handicapped people made society aware of the design deficiencies of many technologies for handicapped people and the resulting social exclusions.7 Political shaping in this sense means that arrangements and considerations of authority and power influence the form that a technology takes. Concerning e-Infrastructures in particular political considerations of different players in science policy like universities, sponsors of research and research infrastructure (like science foundations, research ministries and the European Commission), publishers, scholarly societies and others should be taken into account. Past initiatives at national level on promoting e-Infrastructures in the social sciences and humanities in the US and the UK certainly contribute to the fact that both countries are currently at the forefront of the discussion on e-Infrastructures. Publicly

sponsored actions, like the building of demonstrators and prototyping of ways of

7 However, Winner (1999) makes clear that technologies can also induce certain social conditions; for instance, nuclear power plants require a hierarchical management and control

making e-Infrastructures more usable and deployable contributes to their spread in the UK. Because the evaluation and maintenance of data tools and products are both costly and not a natural part of research culture, they are unlikely to happen without a coordinated strategy to develop a critical mass of resources and appropriate incentive systems. The principal near-term opportunity is to survey existing mechanisms of hardening and sustaining e-Infrastructure at various levels and test the most promising approaches. Possible examples of successful

hardening might be found in the US Digital Government program (see Burton &

Lane, 2005, http://www.nsf.gov/funding/pgm_summ.jsp?pims_id=5459).

Academic production function. In terms of the basic unit of analysis, we think of each individual academic as operating like a firm that tries to maximize individual academic profits. This is a classic constrained optimization problem in which output is a function of inputs such as time, labour resources, and computational

resources, and the cost constraint is described by prices of each input. In this very oversimplified framework, academic ability is the capacity to convert these inputs into outputs, where again, in simplified terms, output is the standard academic currency: the quality of refereed academic publications (as measured by a variety of factors, including citations) and academic grants. However, the broader societal problem faced by the European Commission is that each individual researcher is operating within a local, but not a global, optimum. The new capacities of

e-Infrastructure offer the potential to fundamentally advance academic knowledge by generating new knowledge by means of creating new data, and new methods to analyse data. The challenge for the European Commission is to identify the factors that are needed to catalyse the adoption and use of e-science, broadly defined, and to change, in some ways, the nature of the academic production function. In this framework, we particularly focus on identifying the benefits to individuals, narrowly defined, and society, broadly defined, associated with social scientists adopting and using e-Infrastructure.

Assigning scientific credit and ownership rights. In addition, an important social aspect is that there is inadequate scientific credit for dissemination of existing research datasets or code, and this results in disincentives to sharing both. Barriers to wide data sharing result from their character as research resource: the

production of empirical databases is costly; ownership and access to databases constitutes an important resource and input to empirical research. Hence, scientists might be unwilling to share these resources as long as they haven’t drawn all the benefits from them. Or they might not want or be able to provide sufficient information for other scientists to use the available data with confidence. As Woolgar and Coopmans (2006) argue, the sharing of raw data might not be fully realised and hindered by practices that are not in line with the idealistic and mostly discarded Mertonian norm of communalism. Until issues of intellectual property rights are worked out, individual scientists and private firms may be reluctant to participate in shared developments. In other words, there is substantial

misalignment both in assignment of ownership rights and in how academic credit is granted. Ownership rights in data generated in a collaborative project are difficult to assign, yet the data themselves may have substantial financial value. Likewise, some social science communities and departments do not have a tradition of granting academic credit to tool builders or researchers who share their data widely.

Cross-disciplinary communication and collaboration. A final issue relates to the problems of reaching an accommodation of research agendas where computer scientists and researchers from the user disciplines are collaborating in

e-accruing to developers of collaborative tools and disseminators of data; and compared to other sciences, there are no established protocols for allocating credit among, e.g., researchers in the social sciences and tool developers (perhaps from other disciplines) (Burton & Lane, 2005).

3. Stock-taking of e-Infrastructures in the social sciences and