Is it possible to reach a consensus with someone whose political opinion differs from your own? What if you both rely on social media like Facebook and Twitter for information about news and politics?
According to a new paper co-authored by Professor Janos Kertesz, head of CEU’s Department of Data and Network Science, the kinds of “echo chambers” created by online media make reaching such a consensus unlikely. Not only do people with vastly different political opinions resist change, even those with less different beliefs have trouble finding consensus between themselves.
“New media use algorithms to deliver information directly to a user’s personal account. The algorithms are not optimized to make users as best informed as possible; they are optimized for popularity. So, the news that users read is what they want to read, what they like to read—and that has consequences,” said Kertesz.
In Algorithmic bias amplifies opinion polarization: A bounded confidence model, Kertesz and his fellow researchers from the University of Pisa and the Consiglio Nazionale delle Ricerche use data modeling to simulate interactions between individuals with differing opinions. To mimic the effects of social media echo chambers on these interactions, an “algorithmic bias” is applied to the simulations, so that individuals with similar opinions are more likely to interact with each other. This algorithmic bias leads to a greater segregation among individuals’ opinions and, where reaching a consensus is possible, massively lengthens the time needed to reach it.
“This research is just a model, and not a direct description of what happens in a real-life situation, but it does qualitatively describe what we observe in reality,” said Kertesz. “Our results have much to do with our present-day experience of polarized political discourse and phenomena like the worldwide success of populism.”