Game theory will help predict social conflicts

Fake news, post-truth politics and trolls are fostering populism and social conflicts. Game theory can fight back


Social media have deeply modified interactions among individuals. Photo by rawpixel on Unsplash.

In the last decade, social media and digital discussion forums have been changing the way individuals and groups interact. But how can researchers keep one step ahead of such a rapidly changing social space? Game theory uses mathematical models to study conflict and cooperation between entities and is used across economics, politics, psychology and computer sciences to simulate everything from predicting election results to mapping the spread of information. Now social scientists in particular are looking to game theory and text analysis as a way to understand new social dynamics and for early detection of rising conflicts.


Anyone who uses social media regularly can see how debates about news – from women’s rights to terrorism to Trump and Russia– can quickly escalate to threats, including violence. The 2016 Oxford Dictionary Word of the Year was ‘Post-truth’ and in the 2017 Collins Dictionary it was ‘Fake news’.


But what are the precise factors that affect the dynamics of usual opinion exchange? And how do cultural and cognitive differences (frames) instigate or amplify conflicts? The EU-funded Future and Emerging Technologies (FET) project ODYCCEUS is developing a platform named PENELOPE to analyse the enormous quantity of information circulating on digital and social media platforms to better understand how meaning and representation are shaping the dynamics of conflict.


Inputs exchange happens at micro, meso and macro societal levels


In this context, we are all part of a multiscale game: at the micro level there are individuals; at the meso level lie groups and communities; and at the macro level are regions and nations, all exchanging inputs, ideas and feeding back (or not) with each other. Beyond this complex structure, high connectivity among agents (individuals, groups or macro-communities) and the accelerated dynamics of social interactions make it hard to capture a clear picture of this process. Hence, a rigorous approach to analyse the mass of data including game theory, opinion dynamics, language analysis and text mining is necessary.


This theoretical background will be made available through two innovative participatory apps: the Opinion Observatory and the Opinion Facilitator. These apps can set up several tools to create workflows of elaborated data coming from selected sources and could be linked via API (application programming interface) to pull out patterns and trends in the data to find the key drivers of the system. Four types of users are targeted: social scientists, media researchers, the public at large, and developers who want to contribute with new tools.


Moreover, the Opinion Observatory will be able to take a global view of ongoing debates, while the Opinion Facilitator will make the conceptual maps used by participants visible to lead the negotiation towards more precise terms, thus avoiding miscommunication. Indeed, a fundamental assumption of this project is that conflict is not just an effect of diverging interests, like a natural consequence of negotiation or discussion. Conflict emerges also from a disparity in cultural background and produces different mental models, leading individuals to divergent interpretations of the same situation.


In that flow of data, the apps do not store any data. Only custom workflows and conceptual maps are stored. However, a dedicated ODYCCEUS team is dealing with the privacy policies to assure the compliance with EU law and in particular with GDPR.


ODYCCEUS has been selected as “Success case” in the integration of Social Sciences and Humanities (SSH) within H2020 by the network Net4Society, and a factsheet is set to be published by end June 2018. The €6M ODYCCEUS project involves eight partners and is coordinated by the Max Planck Institute For Mathematics In The Sciences in Leipzig, Germany.


Banner image: via