The algocracy. Understanding and explaining how public organizations are shaped by algorithmic systems

Abstract

This thesis develops the algocracy, a new ideal type of rational-legal authority, that helps to understand and explain how algorithmic systems shape public organizations. The construction process entails theoretical and empirical exploration. First, the characteristics of the algocracy are derived from scientific literature and compared to related ideal types: the machine and professional bureaucracy as well as the infocracy. Second, based on the developed characteristics the case of predictive policing at the Berlin police is explored in order to refine the ideal type. The algocracy draws on automated advice that is based on the algorithmic analysis of data to coordinate social action. Uncertainty that is inherent in expertise-based decision-making is reduced and quantified by algorithmic systems. In this way, non-routine work is standardized in algocratic organizations. This allows a more centralized organizational structure and emphasizes the hierarchical order. Thus, higher levels of control and obedience are achieved in organizations that use automated advice instead of standardization of skills (e.g. by training) as their prime coordinating mechanism. It means that the algocracy is a further rationalized organizational configuration of the professional bureaucracy. This has implications for the performance of professional labor which is subjected to more vertical control and requires more analytical and reflective skills in the algocracy. These characteristics of the algocracy are illustrated by the case of predictive policing. The Berlin police uses a self-developed system called ‘KrimPro’ which predicts the risk of domestic burglaries on a spatiotemporal level. KrimPro is operated on a central organizational position. The predictions standardize and formalize the work of the police professionals who plan and decide on operative measures also because the use of the automated advice is promoted by rewards and hierarchical mechanisms. The control of the performance of non-routine work increases and even exceeds organizational boundaries. Thus, the case of predictive policing successfully demonstrates characteristics of the newly developed ideal type; the algocracy.