Governments, companies, and scientists performing cyber security research need reference data sets, based on real systems and users, to test the validity and efficacy of the predictions of a given theory. However, various ethical and practical concerns complicate when and how proprietary operational data should be shared. In this paper, we discuss hypothetical and actual examples to illustrate the reasons for increasing the availability of data for legitimate research purposes. We also discuss the reasons, such as privacy and competition, to limit data sharing. We discuss the capabilities and limitations of several existing models of data sharing. We present an infrastructure specifically designed for making proprietary operational data available for cyber security research and experimentation. We conclude by discussing the ways in which a new infrastructure, WINE, balances the values of openness, sound experimentation, and privacy by enabling data sharing with privacy controls.
展开▼
