Is it really just because of our phones? Researchers from the University of Luxembourg, the German Institute for Economic Research, and the Barcelona School of Economics have embarked on an ambitious project to better predict and address the rise of online polarisation.
Parsing through the thorny relationship between social media, polarisation, misinformation, and populism is a difficult undertaking, but one worth examining for the future health of our democracies, according to Christos Koulovatianos, professor in finance at the University of Luxembourg.
With a background in economics and political economy, the professor has always been interested in the interplay between economics and politics, noting that “it is well empirically documented that a bad economy is a trigger for losing elections for the incumbent party”.
His personal experiences have made him all the more uniquely interested in the interplay of the two.
“As a Greek person, I have experienced major fiscal crises and the state going bankrupt. Once in 1989 and once in 2010, the latter being the one most people are familiar with and which is still ongoing,” he said, adding, “Populism was at least partly at the root of what happened in both cases, which triggered my interest in political economy. It’s this question: why do people tend to sabotage themselves in a democracy through populism?”
Studying populism
The current project he is working on, which is in its first year, takes an empirical approach to society’s current woes – the interaction between social media, polarisation, misinformation and populism.
“So far, we do not understand this interaction, and that’s the issue. We feel that there is something going on there; our job is to make sense of it, and this is a very difficult thing to do,” Koulovatianos said.
Honing in on populism made the professor and his team realise that it is an ill-defined concept. A decision was made to focus on two aspects of populism in their research.
“Number one is that it creates a division in people; usually, someone is perceived as the elite and another group as the opposite of that elite. It can consist of any two opposing groups, such as fascists and anti-fascists. We want to track this polarisation,” he said.
The second aspect is the denial of factual information linked to such thinking. “This is fantasy, a departure from reality, the idea that if you somehow remove the elite, all problems will solve themselves magically.”
Trying to elucidate people’s detachment from reality is key to the study, as all crises have an easily observable material component but are also accompanied by how people, who can be prone to panic, think about them. The professor gives the example of the stock market, where certain disasters are self-fulfilling.
The researchers mainly use a combination of Network Theory and Decision Theory in their analysis. First, they empirically sweep data from social media sites to observe how ideas and people cluster together.
For example, they scrape websites for hate speech in order to analyse the tone of words used, which can be done with the help of AI. “We tried to see how people who write comments, for example, on YouTube or LinkedIn, cluster with others and whether this clustering follows hate speech,” the researcher said.
Their theories observe a double way causality, where one feeds the other, “our models try to study how this interaction might make people depart from reality and start doing foolish things”.
“For example, starting a war is a very stupid idea. It doesn’t end well, but it starts from a kind of populist rhetoric. We’re not there yet, but in the future we would like to understand how people who become fanatic about something start spreading fake news and to poison the internet, which is a very complex topic,” he said.
Then, they apply Decision Theory to study the trends in people’s thinking. “Decision Theory is the same thing your GPS uses in the car: it tracks where you are, but it also has a plan on where to go. So every time you miss a turn, for example, it always recalculates,” he said.
“That machine does something that looks simple, but it’s an extremely complicated calculation done through what is called dynamic programming,” Koulovatianos explained.
This method is used extensively in studying economics and asset prices, the professor says, and is an example where finance and decision theory come together to better understand what people are thinking about.
In politics, this can be used to observe decisions around voting, said Koulovatianos. “I anticipate a candidate to do good or bad things. Therefore, I do or don’t vote for them.”
Even if some of the reasoning may appear simplistic, creating abstractions is important for the scientific method to be able to analyse different aspects individually.
Networking interacts with decisions in many ways and is a crucial element in understanding trends in how people think.
A perfect example is the Covid-19 pandemic, when “so many, even educated people, just did not want to read the studies and statistics about vaccines. The element we look at is that when people cluster into groups, they reassure each other about wild ideas such as the flat earth theory, it starts becoming more believable”.
“In fact, there is a statistical way of showing that independent robots would reconfirm each other about certain directions, they would basically do the same thing,” Koulovatianos added.
Better predicting political crises
The first year of the project has focused on laying the groundwork.
“One of the goals is to see why anti-intellectualism is rising through social media”, Koulovatianos said, “this first year we are observing data, we’ll hopefully get insights that nobody else has had before, that’s the goal.”
In the second year, the researchers will add theory on top of the collected data, “and the aim of this will be to say something about how to regulate social media.”
The project is very ambitious in its scope and has many objectives, but Koulovatianos summarises them as “an attempt to incorporate new elements in a holistic view of populism.”
Today, experts can predict an oncoming financial crisis in a central bank through data analysis. “We can detect a crisis early on and even prevent it to a good extent. This is missing from understanding political crises, like populism and so on, and that’s what we hope to achieve. We want to simulate, according to some yardstick, the evolution of the value system of people,” said the University of Luxembourg professor.
This would look like trained scientists who could counsel governments, he said, “they could warn, ‘hey, this thing is going too far. We might have mass misinformation, and you need to debunk it.’ We need to find a constitutional and legal manner to do this, which is yet to be invented, but if we understand something, experts will be able to better legislate on this.”
Does the professor believe all the issues the study concerns itself with are solely caused by social media?
“Definitely not,” he said, “they’re human vices that always existed, but operating through social media can put those vices on steroids by encouraging peer-induced behaviours because people want to be accepted and likeable, so people try to imitate each other. We try to at least detect this.”