Scientific information policy

The goal of information policy for science is to ensure that the society acquires empirical knowledge of all phenomena as rapidly as possible. As discovery is the basis for invention and innovation, the scientific information policy aims to promote a rapid rate of invention and innovation. For most scientific activities leading to invention there is no need for an explicit information policy, since there is no conflict between scientific curiosity and other social concerns. But for many scientific activities leading to innovation there is a fundamental conflict between scientific curiosity and the desire for privacy and trade secrets.  

These conflicts  between curiosity and privacy currently restrict the advance of empirical science concerning all types of human behavior. In some areas of social sciences such as conditioned learning experiments in psychology and agricultural economics controlled experiments are the rule. But in most areas of the social sciences and business disciplines the empirical research must discriminate between alternative hypotheses on the basis of observations of actual political economic behavior. In social systems the principle impediment to observing behavior is that the observed subjects vigorously oppose being observed. What this means is that empiricists in testing an hypothesis must make ingenious use of the available observations rather than collect a data sample best suited to test the hypothesis. Numerous social and political economic phenomenon are simply not observable in any systematic fashion.

The difficulty of obtaining observations has created a theoretical bias in many disciplines of social science and business administration. Currently these sciences, such as economics, have developed mathematically rigorous theories of behavior, but at the same time can not accurately explain the behavior they are ostensibly studying on a moment by moment basis, that is real time. For example, there are no simulation models which will simulate the behavior of a firm in real time to an accuracy of even one significant figure.

The academic incentives of publishing in prestigious journals to some extent impede rather than promote scientific advance, inasmuch as the criterion for publishing in the respected academic journals is, unfortunately, mathematical rigor and sophistication rather than the ability to explain the phenomena under study in real time. To be sure, the bias towards the abstract mathematical manipulation of models is useful as a devise for ranking the mathematical ability of academics.

But in fact these academic incentives tend to create mathocracies rather than sciences explaining natural phenomenon. This is not to degrade the efforts of the many conscientious and able scientists who labor to extend the empirical contents of their respected disciplines. Rather, the problem is that a major conflict exists between generating observations which are efficient at discriminating between alternative theories and concepts of privacy and property rights.

To ensure a rapid rate of discovery for innovation, the right to learn  needs to be defined to provide observation rights to all phenomenon. The minimum scientific observation right is a representative sample of observations on any phenomenon. Because the study of behavioral relationships does not require a knowledge of the identity of the subjects in the sample, the compromise with privacy is to delete labels in scientific datafiles. Observations involving legitimate trade secrets like military secrets would be released with a time delay.

To alleviate fears of abuse of data and to promote accuracy in observations, data collected for scientific purposes must not  be used for administrative purposes. This means that scientific records should not be subpoenaed to reveal individual information, although the methodology of obtaining and using such records could be challenged. In this regard the incentives to promote accuracy are more important than the occasional discovery of illegal activity. Scientific records would, however, be used for policy purposes such as performing studies which might indicate the need for future changes in laws or procedures. To preserve the separation of scientific data from administrative data, agencies with no administrative functions should collect scientific data.

Private agencies collecting scientific data would likewise be regulated to ensure due care in preserving privacy. The use of scientific data for private operational decisions would also be prohibited. Scientific data could only be used privately to construct better decision functions, and if the new decision functions were superior to the old, they might create an operational demand for new inputs from all participants in a decision process. Until this happened, however, even if decision makers knew the identity of subjects in a scientific sample, they could not use the data for decision purposes. For screening functions in the public domain this separation of scientific and administrative data would be enforceable.

Given the proposed safeguards, scientific information policy can be used to greatly increase the empirical content of sciences and applied disciplines concerning political economic phenomenon. First consider the collection, use and release of data collected by government. Currently government at all levels collects information on economic agents for administrative purposes. Since the computerization of government records in the 60s, administrative agencies and researchers outside government have made increasing use of administrative data for statistical studies. This data has defects for studying behavior as the purpose of the data collection is for compliance with administrative law without consideration of issues of statistical methodology. Also inconsistencies in data collection and the Privacy Act and other legislation inhibit the creation of behavioral data samples from many administrative datafiles. Finally, researchers outside government can rarely obtain disaggregated business data which the government has collected.

Because of proposed safeguards, the collection and dissemination of administrative data could better serve scientific purposes. With the proposed distinction between administrative data and scientific data, the combining of data from many sources for statistical purposes should be encouraged. The quality of such composite data can be improved by greater attention to statistical considerations in the collection of administrative data. Also as informational society advances, government agencies, to fulfill their legislative mandates, will increasingly collect data based on good statistical methodology to understand the behavior under their own particular mandate.

Moreover, information collected by the government can be released to researchers outside government either immediately or with a time delay. In a constantly innovating society, the purpose of protecting trade secrets is to create incentives for constant innovation in order to keep ahead of one's rivals. The value to a firm of data collected by the government so that the firm may examine the behavior of its rivals is rapidly discounted. Over time the files of data collected about business behavior would be released in a disaggregated form.

This means that for administrative data series such as the SIC code aggregated data would over time be available at the level of the individual firm and product, and eventually all the internal information of all public and private institutions would become public. Administrative data concerning individuals would be released in the form of unlabeled representative samples. The major change this represents from current policy is the increase in collection of representative samples of behavior.

While data collected by government agencies could better serve scientific purposes, more steps need to be taken to promote discovery of political economic behavior. Currently much empirical research asks what hypotheses can be tested by the available data, not what the most important hypotheses are that should be tested. Scientific information policy should provide empirical scientists representative data samples in order to test hypotheses of any behavior and the compliance in supplying these data would be compulsory. The concept that the right to learn guarantees the right to obtain a representative sample will appear very threatening to many. And, indeed, given the current political design, the fear of abuse of power is well grounded in numerous empirical examples. With a more open information policy firms would fear their trade secrets would be stolen, and that adverse publicity might inspire the public to demand new government regulation.

The scientific observation right is intended to cover all phenomena; however certain practical considerations place some limits on how this right is to be exercised. First, the observer pays for the cost of observing. Second, this right is a right of the scientifically competent, since funding for observations will be provided by such agencies as the National Science Foundation. In competing for funding, the proposals for observations will be subject to peer group review. Other government agencies will have an interest in resolving hypotheses important for innovations or policy, while private foundations would fund more controversial studies.

While professionals and professional societies could use the legal system to obtain the desired observations, in most cases they would not have to do so. In an industry-wide study, say of incentives versus performance of managers a professional society would, in order to obtain voluntary compliance, work with the trade association to soothe any fears of loss of trade secrets. Professional societies observing community behavior would have a similar interest in smoothing the situation with local officials, and much of the burden of data collection would be minimized by machine talking to machine.