The Price of Big Science
The quality of the expanding output is a significant question that has not gone without comment , even by Derek Price himself , as well as philosophers of science , Robert Merton , Thomas Kuhn , and Karl Popper . Indeed , Popper ( 1976 ) viewed the growth of scientific output as a deficiency of science , since , in his view , the “ growth of knowledge … is not a repetitive or cumulative process , but one of error elimination …” ( p . 144 ) and therefore should not increase in volume , but decrease as errors are eliminated from the understanding of the natural world . Price also wrestled with the quality question , noting that perhaps it would be better to take a different approach to science : “ One may study the growth of only important discoveries , inventions , and scientific laws , rather than all such things …” ( 1961 , p . 32 ), thus reducing the overall output of science .
Indeed , several philosophers of science have suggested that science should be conducted by an elite group of highly talented researchers , rather than a broad group . Bradford ’ s Law ( Garfield 1971 ) informed the earliest efforts by Francis Narin , Eugene Garfield , and Henry Small to identify high quality science and are based upon the premise that a small percentage of output would represent the highest quality work ; thus a limited dataset would constitute the bulk of material worth tracking ( Garfield 1976 ). ( This is justification for the limited number of journals indexed in the Web of Science .) 4
The question of how quality is measured and what is included in different databases is beyond the scope of this paper . We have the more modest goal of understanding scientific knowledge growth as part of a project to study why scientific collaboration ( as measured by co-authorships ) continues to grow . To address the question of expectations about the growth of science and why it has exceeded hypothetical expectations , we developed a systems model as to whether scientific publishing operates a complex system ( Katz 2006 ). Showing that scientific publishing operates as a complex system ( rather than simply as an aggregation of lists that continues to grow exponentially ) would provide us with a set of conditions upon which we could test , understand , and further study the system for additional research . The following discussion presents the model and initial findings .
Scientific publishing has many features in common with complex systems 5 which lead us to make this measure . The features in common include the dynamic growth of output , the emergent nature of that output , the openness of the system in accepting inputs , and the self-organization of researchers into disciplines and collaborative teams . Complex systems are dynamic and have a propensity to exhibit scaling properties along a power law form . These scaling properties appear to be a feature of scientific publication , at least in co-authorship ( Wagner and Leydesdorff 2005 ) and in citation behavior ( Katz 2000 ). Table 2 compares features of complex systems with features of scientific knowledge production whose statistical features have been shown by Lotka ( 1929 ), Price ( 1963 ), and others ( Katz 2000 ) to display similarities with other complex systems .
To further test this systems hypothesis and to explain why Price ’ s theory of growth cannot be upheld , we constructed
4
Estimates have been made of the extent of scientific publication within and outside of SCIE . Bjork , Roos and Lauri estimate 2008 and Wagner and Wong ’ s ( 2011 ) calculation suggest that the extent of scientific publication outside SCIE may be considerably larger when the publications of developing countries are fully
counted .
5
( Baranger , n . d .).
113