Political volatility

Using techniques such as natural language processing to quantify trends and changes in the volatility of public opinion before and after widespread use of social media

Project status



It often seems that political life is faster-moving, more unpredictable, and more unstable than ever before, leading to electoral shocks, policy surprises, and even regime changes. Yet while modern politics seems more volatile, there has been no systematic investigation into the nature of this volatility, or its potential impact on the policy-making landscape. This project seeks to address this gap in knowledge by quantifying the level of volatility in the ‘issue attention’ economy.

Explaining the science

Social media platforms have been shown to inject instability and uncertainty into social life, public opinion, civil society, and the policy-making environment. They exert social influence on users by showing in real time what other people are doing (social information), as well as reducing the costs of becoming visible, and so increasing the likelihood of acting. Individual acts of participation in turn create further social information, which may influence someone else’s decision about whether to act, leading to feedback cycles and chain reactions.

These feedback influences at work on social media platforms have been shown to introduce instability into cultural markets, and have now penetrated the ‘issue attention’ economy and the policy-making landscape, creating political turbulence. This disorganised environment, in which fluid and overlapping groups of individuals mobilise around common (often temporary) social issues and goals, manifests itself, for example, in a highly unequal distribution of participants; most YouTube or Facebook videos have very few views while a small number are watched millions of times, accumulating these views quickly and dramatically.

Project aims

In this project, we are testing this theoretical model described in 'Explaining the science', with fifty years’ worth of public opinion polling and media coverage data from the UK and Germany. We are using natural language processing and information theoretic approaches to quantify the trends and changes in the volatility of public opinion before and after social media became widely used.

Understanding the system properties of this volatility in civil society and the role of social media platforms will shed light on undercurrents of public opinion that have, in recent years, burst to the surface in highly unpredictable ways, and can inform the creation of early warning signals that may aid in prediction and explanation.


Contact info

[email protected]