Collaboration, Self-Similarity and Scale-Independent Policy

by J. Sylvan Katz

There are probably as many, if not more, reasons for researchers to collaborate as there are reasons for individuals to marry. Some marriages are arranged, some are entered into as a matter of convenience, and others are an expression of a deep emotional bond. Similarly, some research collaborations emerge from political memoranda of understanding between nations, some collaborations result from a formal requirement of funding agencies for partnerships and still others evolve from the growth of professional respect and trust. The factors that drive scientific collaboration are numerous and difficult to quantify.

A modern science system has many interacting political, economic, social and cultural processes. It is difficult, if not impossible, to determine exactly how each of these influence the co-operative and competitive actions that naturally emerge from the intrinsic curiosity of scientists working alone or in groups. In fact, our current understanding of systems that exhibit complicate internal dynamics suggest that their properties can not be characterised by traditional quantitative measures such as averages. These complex systems tend to exhibit scale-independent or self-similar properties that are commonly characterised by power law distributions.

There is strong evidence to indicate that a science system has many self-similar characteristics. For example, Lotka showed that the publishing productivity of scientists follows a power law distribution. Recently it was demonstrated that the amount of recognition (citations) received by a scientific community increases in a power law relationship with publishing size of the community. These non-linear effects can be so large that they bring into question traditional bibliometric measures such as impact (citations/paper) that have not have be adjusted to compensate for this fact.

Does collaborative activity exhibit scale-independent properties? A bibliometric analysis of UK collaboration shows that is a distinct non-linear relationship between institutional publishing size and the amount of institutional collaboration. On average, institutional collaboration showed a power law relationship with the publishing size of the institutions. A greater proportion of publications from smaller institutions than from larger institutions involved domestic, intra-sectoral, inter-sectoral and industrial collaboration. On the other hand a greater proportion of the papers from larger institutions than from smaller institutions involved international or intra-institutional collaborations. Industry collaboration showed a mixture of the two non-linear effects.

This presentation will explore the implications of the power law relationship between publishing size and various types of collaboration. It will also discuss the potential for developing scale-independent policies to support and nurture the inherent processes in a science system that encourage collaboration.

Back to Papers ListBack to Papers List