Grant: Discovery Grants Program - Individual

Our social interactions and daily activities are increasingly mediated by technological networks, producing complex socio-technical systems sometimes referred to as social machines. The increasing role of connected devices (smartphones, wearables...) and data-driven algorithms has made these social machines powerful forces for disruptive change, shaping the way we connect, the way we work, and the way we vote. If we are to anticipate and accommodate these changes, and design new social machines to bring about a better future for our society, it is crucial that we acquire a better understanding of existing social machines, of the role of their automated components and of the interplay between these components and social processes. We propose to develop new analysis and development techniques for social machines, to be integrated into a coherent engineering methodology for social machines of the future. The first objective of this research is to develop a modelling formalism to analyze a social machine in both its social and technological dimensions. Inspired by formalisms from the multi-agent research community (e.g. electronic institutions and protocol languages), we will adapt these formalisms to better account for algorithmic components. This will support our second objective, which is to evaluate Fairness, Accountability and Transparency (FAT) properties of social machines. There is a growing awareness that the large-scale deployment of algorithms and automated agents in online systems, intended to solve simple problems at an individual level - e.g. show more relevant ads to people -can have insidious and unpredictable side effects at the global level - e.g. reduced employment opportunities for women or ethnic minorities. However, these effects often result from the complex interplay of social and algorithmic processes. Being able to disentangle these two aspects would be an important breakthrough in FAT-related research, and could produce useful verification and validation methods to be used in the early stages of engineering new social machines. Finally, we will further advance the research in engineering methods for social machines by developing a series of tools to support the rapid prototyping of social machines, in a Model-Driven Engineering approach using the above modelling formalism. A key innovation in our approach will be to leverage existing Web applications (e.g. Twitter, Slack) to support social processes, instead of generating raw code. This will produce a very lightweight approach which would help democratize the creation of social machines, allowing communities ignored by the mainstream tech market to create useful systems to fulfill their needs and aspirations.