Governance of Data and Learning Machines through the lens of a Rugged Ecosystem

Digital technologies are increasingly being adopted throughout the economy, and the use of IoT and AI is becoming more and more pervasive. Yet, the digital arena is largely unregulated and dominated by few players, while key questions on the governance of data value chains and the control over learning machines are mostly unanswered. The aim of this paper is to develop an evolutionary, landscape model of digital ecosystems to study the effects of different data and digital capital governance on industry dynamics, concentration, and productivity. Building upon a growing body of empirical evidence on the adoption of digital technologies, the model can account for the most relevant specificities at the core of data-driven technologies and their functioning. As a result, the simulations mark a clear distinction between standard - non-digital - capital, and digital one, challenging the most conventional prescriptions on control and governance arising from economic literature on property rights. Moreover, primary role is given to the governance over data, whose role as both main input for learning machines and codification of knowledge embedded in production processes makes them pivotal. To conclude, the paper elaborates upon the management of existing digital ecosystems, and proposes alternatives.