Strum, Wesley, Joel Genuth, and Ivan Chompalov. Structures of Scientific Collaboration. Cambridge, Mass.: MIT Press, 2007. Pp. xi + 280. Hardcover, $38.00.

A study of the organization of fifty three “big science” physics projects, each a collaboration involving at least three independent organizations. The specific areas were Geophysics, Ground-based Astronomy, Material Sciences, Medical Physics, Space Science, and Particle Physics. The data was collected in the 1990s by a group from the American Institute of Physics.  The collaborations fit into four clusters: Bureaucratic, Leaderless, Non-Specialized, and Participatory.  The particle physics projects were in the last cluster and were basically the only ones in that cluster, probably because of the technology involved and the corresponding acculturation of graduate students.  Except for particle physics, the clusters did not coincide with scientific specialties.  Several reasons for collaborating exist, but the important one appears to be access to instrumentation—“the structure of collaboration is best viewed in terms of the practices through which data is acquired.”  Besides the issue of who gets to work with the equipment, sharing of data can be problem in a collaboration. The organization of projects must satisfy the needs of the engineers and scientists involved but also the requirements of the participants’ home institutions, which usually have constraints on acceptable structures. The process of starting a collaborative project influences a project throughout its life—success is determined more by the beginning than the end.  Larger projects, not surprisingly, had more administrative structure. On the other hand, when the scope of the collaboration is wider, decision-making tends to be more participatory and local.  If the scope is wider, probably less micro-management is done.  Leadership style and technology are related: technologies that make participants interdependent lead to decentralization of leadership.  Trust between participants does not seem to be a determinant of success, although a lack of trust leads to conflict.  One determinant of success, though, is funding difficulties at the beginning—projects having difficulties raising funds, but ultimately in doing so succeeding, had more positive final outcomes.  Much of the discussion involves “bureaucracy,” used here to mean formal organization, with a hierarchy.  The authors point out that the presence of bureaucratic procedures may very well aid in getting the science done, and may give more local autonomy to scientists and engineers in doing their work.  It seems evident that collaborations will be more necessary in the future as the scope of significant problems enlarges.  Knowing how to plan and manage such projects will be useful, and much insight can be gained from this book.  Some of the exposition was slow going for me; perhaps a person trained in the social sciences will have an easier time.