G. Michael Purdy
Division of Ocean Sciences, National Science Foundation
Any discussion of the merits of "large" versus "small" science programs (as alternative mechanisms for the organization and funding of basic research) must begin with a description of the factors that govern progress in research.
The core of basic research in the natural sciences is the generation of new ideas that explain natural phenomena in useful ways. Therefore, one essential goal of any organizational structure designed to support basic research must be the creation of new ideas. This is not a simple matter! Ideas are created by individuals. They are not arrived at by consensus, they are not directly the result of any formal process, and the best ideas cannot be produced according to any predetermined schedule. It is not always possible to predict which area of science will produce the best new ideas or, indeed in what direction these new concepts will lead. Ideas require stimulation beyond simply the curiosity of a bright mind, and the source of this stimulation can vary widely. Unexpected observations, new theoretical approaches, other investigators' ideas, or even the discovery of an error or an oversight in some previous work—all can play the catalytic role that converts a long period of unsatisfying bewilderment into a joyful flash of insight and understanding.
It is wrong, however, to represent basic research as nothing but idea generation. Progress in research depends on many other less abstract factors. If models and hypotheses are to be verified, appropriate data and observations are needed. If complex data sets are to be understood and made useful, data analysis tools are essential. If new fields are to be explored, the necessary measurement technologies must be developed. The design of any structure to support basic research must take into account these factors and many others.
Large science programs, such as those described later in this volume, involve many investigators in their planning and implementation and necessarily depend on the development of a consensus among the participating researchers concerning investigative strategies and plans. The process of developing this consensus allows participants to share ideas and opinions, and produces, most often, the optimal set of compromises required to match objectives with available capabilities and resources. The successes of this planning process for large programs are well documented in the articles cited later in this paper. A range of models for the management of large research programs have been developed and implemented, and their strengths are clear. They have achieved their objectives of developing global strategies for coordinated data collection, of building new cross disciplinary connections in the community, and of efficiently directing substantial resources toward focused research problems of particular significance to society. The strengths of community consensus-based planning for large research programs are well established. The shortcomings are not so obvious, but they are precisely the strengths of the individual-investigator, small science approach to the support of research and are most effectively described in these terms.
By far the single most important attribute of the individual-investigator, small science approach to the support of basic research is its superior ability to recognize, select, and support the best new ideas, new approaches, new investigators, who often challenge existing dogma and take the research in unpredicted directions. Small science projects are built around the single most important resource: the individual investigator. As emphasized earlier, ideas come from individuals and ideas are the foundation for all research progress. An effective system for the support of research must be as open as possible to all investigators and all ideas, so there is the richest possible field of opportunities from which to select when the harsh reality of prioritization and resource allocation is faced.
The Division of Ocean Sciences at the National Science Foundation (NSF) supports five core disciplinary programs in biological, chemical, and physical oceanography; oceanographic technology; and marine geology and geophysics. These five programs are the engines that generate the ideas that drive ocean discovery. The rich diversity of the topics that these programs support makes it impossible to summa