Scientific progress and lessons for institutional design: comments on "a toy model of scientific progress" by Susanne Lohmann

American Journal of Economics and Sociology, The, Jan, 2004 by Susan K. Snyder

The early years of public choice are closely associated with two universities: the University of Virginia and Virginia Tech (VPI). Scholars at these schools made founding contributions in public choice to help make it what it is today: an inherently interdisciplinary research field, respected both by economists and political scientists.

All agree that extraordinary personalities contributed to the success of public choice. Institutional factors likely played a role also, and we may be able to identify particular structural features of the universities that facilitated the growth and development of public choice. This should be of interest to those who would like to see the success of public choice repeated for other interdisciplinary fields. One cannot replicate academic entrepreneurs on demand, but perhaps the institutional setting in which academic entrepreneurs are most likely to flourish can be recreated to some extent.

How do we uncover the key institutional features that enabled public choice to thrive? One way to proceed is to create a model of scientific progress, and use this model to make inferences about the optimal organizational structure of a university. Lohmann's paper provides some groundwork for this endeavor: she outlines a framework for thinking about scientific enquiry and makes inferences about how university structure will inhibit or encourage scientific progress.

In these comments I will describe her model of scientific progress and propose some directions in which it could be further developed in order to create a better understanding of how universities should and will be designed.

The model starts with a true state of Nature. If the true state of Nature can be accurately predicted, then society can shield itself from disasters. As is true in our world, the true state of Nature is too complex for any one individual to predict; additionally, the true state of Nature can change over time, so prediction methods that were successful in the past may not be successful in the future.

The actors in the model are scientists who try to ascertain this true state of Nature. A key component of the model is that scientists have distributed information; different scientists analyze different data sets, using different methods. The underlying assumption is that there are exogenous limits on how much information each scientist can process.

There is collaboration in the model; scientists can read other scientists' work and make modifications on this basis. How work is judged is not detailed, though the underlying assumption seems to be one of strict positivism. Scientists cannot read every other scientist's work, however, again because there are limits on how much information each scientist can process. We can think of scientists gathering into small groups, or "clusters," in which they study similar problems with similar methods, share results, and make modifications.

Thus progress in science depends on scientists concentrating within their clusters--nothing could be accomplished if they did not somehow limit their scholarly contacts. Incremental progress is made within these clusters, but the danger is that ossification will set in and scientists will keep using the same old methods and the same old data even after there are no new conclusions to be reached. It is assumed that new methods or new data will come from scientists outside the cluster. If new entrants can make better predictions, science will advance in this new direction. This is a period of vibrance or revolution.

Then one can think of a university as providing a structure of defined clusters. A scientist chooses to locate in an existing cluster within the university, and is expected to converse with the other scholars in that cluster (and is given incentives to do so). The clusters are within a hierarchy that allows for entrants from other clusters to infuse new life into ossified fields--a framework that allows for creative new endeavors that would not evolve naturally within the existing clusters. An example of such an endeavor might be public choice in the 1960s and 1970s.

The framework outlined in Lohmann's paper provides a starting point for thinking about how universities should be designed. One area that requires more detailed modeling is the concept of research clusters and their organization. It is clear that the cognitive limitations scientists face necessitates clusters of research; what is not clear is how we should create the clusters. If we think of academic research as being analogous to a policy space in the public choice literature, there are at least two dimensions of importance: the kind of questions one looks at and the kind of methods one uses. That is, there is an underlying structure of the research itself, apart from the structure that an academic institution imposes. Thus one could imagine a university's research clusters being formed according to either of these dimensions. The social sciences could be organized by broad categories of research questions: economics, political science, and sociology, for example. Or the social sciences could be organized by research methodologies, such as rational choice modeling, experimental methods, and case studies.