Norms as a Basis for Governing Sociotechnical Systems: Extended Abstract

We understand a sociotechnical system as a microsociety in which autonomous parties interact with and about technical objects. We define governance as the administration of such a system by its participants. We develop an approach for governance based on a computational representation of norms. Our approach has the benefit of capturing stakeholder needs precisely while yielding adaptive resource allocation in the face of changes both in stakeholder needs and the environment. We are currently extending this approach to address the problem of secure collaboration and to contribute to the emerging science of cybersecurity. 1 Challenges in Sociotechnical Systems We define a sociotechnical system (STS) as a microsociety in which autonomous stakeholders interact. How can these stakeholders collaborate effectively even though their interests may only be imperfectly aligned? Complexity and change in interactions among stakeholders and in their environment make administering STSs nontrivial. We address the challenge of enabling stakeholders to administer or (self)govern such systems in a manner that respects their autonomy. A particular benefit are the resulting gains in adaptability in accommodating any exceptions and opportunities. The Ocean Observatories Initiative (OOI) [Arrott et al., 2009], which facilitates scientists and research institutions in acquiring, storing, analyzing, and sharing information from the world’s oceans, is a paradigmatic STS. Its stakeholders include oceanographers, educators, and members of the public as well as research laboratories and universities. The OOI has three key features. One, autonomy: the stakeholders own and share resources such as Underwater Autonomous Vehicles (UAVs), buoys, ocean sensors, and research databases. Thus, the OOI would support collaborations in which it would not own all resources involved. Two, lifetime: a requirement for the OOI was to sustain operation for decades. Thus we must accommodate changes in stakeholder needs without relying ∗ c © ACM. Reproduced with permission. This paper is an extended abstract of [Singh, 2013], which appears in the ACM Transactions on Intelligent Systems and Technology. upon any specific technology to be available throughout the lifetime of the system. Three, scale: an STS could have thousands of stakeholders, tens of thousands of physical resources such as ocean gliders, and potentially millions of cyber resources such as datasets. At those scales, automation and adaptation are essential for administering resources according to the preferences of the stakeholders. How can we accommodate continually changing stakeholder needs? How can multiple stakeholders collaborate in a sustainable, efficient manner? How can individual ownership and control be respected as autonomous parties interoperate? How can resources be added or dropped dynamically at runtime? How can coalitions be constructed and enacted to dynamically share resources while entertaining challenges such the stakeholders’ needs changing unexpectedly, as in an emergency? How may we accomplish all of the above adaptations over a wide range of resource granularities and timescales? 1.1 Governance: Norms and Organizations We term dealing with the above challenges governance. Governance contrasts with traditional management, which presumes authority (superior to subordinate) relationships. In STSs, the collaborating parties are autonomous peers. Today, governance is carried out “by phone call”—by ad hoc negotiations among humans. However, modern STSs require decision making at fast timescales, so manual negotiations would simply not be feasible. From the perspective of governance, the stakeholders of an STS are themselves participants. Recognizing their autonomy, we observe that we cannot prescribe a decision-making strategy for each participant. Instead, each STS can prescribe its rules of encounter via a set of norms. Informally, a norm characterizes sound or “normal” interactions among members of a social group, reflecting their mutual expectations. We emphasize interactions: behavior that has no effect on others is irrelevant for our purposes. Two examples of norms in a scientific setting are putting an instrument in power-save mode at the end of an experiment and closing unneeded data streams from sensors. Norms may arise through top-down legislation or bottom-up conventions emerging from norms implicit in participant strategies [Savarimuthu et al., 2009]. We restrict ourselves to norms that carry sufficiently strong expectations (as in contracts), so that their satisfaction or violation is significant to at least one of the interacting parties. Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence (IJCAI 2015)