Our Approach

A landmark intersection of engineering and social sciences at MIT, IDSS takes a holistic and data-driven approach to analyzing complex, high-impact systems in society.

IDSS research is rooted in three core analytical disciplines and aims to address broad overarching challenges inherent to highly interconnected and data-rich systems. IDSS researchers work in a broad range of applications, including five key research domains of expertise.

Analytical Disciplines

Statistics and Data Science
While mechanistic models may exist for some physical systems, they often overlook how users of these systems actually make decisions. Furthermore, the emergence of new technologies enabling the Internet of Things (IoT), mobile devices, and big data presents us with new opportunities to understand both physical systems and social behavior. The key issue is how to generate reliable and actionable algorithms and models from diverse data with only partial information about the underlying processes. New and powerful techniques in modern statistics as well as machine learning algorithms are needed to support these applications. A powerful use for such models is prediction. Given the complexity of the underlying applications, prediction methods will continue to be limited by the ‘curse of dimensionality’ and will demand new innovations in experimental design, computation, and model reduction.

Information and Decision Theory
Information and decision theory is a broad discipline that includes research in: systems and controls, optimization and game theory, networks, and inference and statistical data processing. Systems and control theory addresses the challenges associated with designing, modeling and controlling complex, distributed systems. Optimization and games are powerful techniques for arriving at synthesis, and addressing algorithms for single or multiple decision makers. Communications and networks research focuses on issues of performance (discovering both fundamental limitations, and close-to-optimal methods), as well as scalability (i.e., maintaining  information and algorithmic efficiency as network size increases). Inference and statistical data processing looks at estimation and learning in dynamical systems, for example: estimating the state of a dynamical system or identifying a dynamic model for such a system.

Human and Institutional Behavior
At IDSS, research into the behaviors of humans and institutions looks at nontraditional aspects of modeling these behaviors as they relate to particular societal issues. This includes the behavior of individuals, groups, and institutions (such as markets, regulators, and governments) as it impacts upon the operation and behavior of systems. Aspects of this field are anchored in the social sciences, e.g., economics, sociology, psychology, political science, management and policy. The availability of large amounts of data on individual and institutional behavior will enable empirical models to be derived that are consistent with fundamental theory, and actionable from an analysis and design perspective. Topics include: rational decision theory, behavioral decision theory, organizational behavior, and public policy design.

Overarching Challenges

There are many challenges in developing and modeling data-driven systems, regardless of domain or discipline. IDSS research is unified in its efforts to address these challenges.

Resilience and Systemic Risk
Interconnections within a networked system can propagate shocks, amplifying their effects and making the system more prone to disruption or failure. Systemic risk arises when shocks to one part of a system threaten to create, or cause, cascading failures. IDSS aims to build a foundational science that measures and minimizes these types of risk, resulting in more resilient systems.

System Design and Architecture
Good architecture is easy to recognize in retrospect, but harder to predict or design. As we transform the nation’s power grid, develop smart and autonomous transportation systems, and enable real-time data exchange in financial markets, secure network architecture is essential. IDSS incorporates foundational theory, practical algorithms, and concrete applications to develop a framework for robust and efficient system design.

Sustainability and Policy
Improving societal well-being across the domains of ecology, economics, politics, and culture requires systematic evaluations of public and scientific innovations. Through quantitative analysis and design, IDSS researchers are working to derive models that account for society’s complexities, an effort that is critical to well-informed public policy.

Identifying models for decision-making is a hallmark of decision theory. However, the magnitude, diversity, and structure of modern data sets creates new challenges around how to use them. IDSS research addresses issues of technical management, security, privacy, and data integrity, as new tools are developed to explore data sets and models and to make better decisions.

Research Domains

IDSS researchers are working in a broad range of applications. Five key research domains of expertise include:

MIT Institute for Data, Systems, and Society
Massachusetts Institute of Technology
77 Massachusetts Avenue
Cambridge, MA 02139-4307