Information inundation and polarization

We are just two weeks after the election and while people focus on different questions, from why did the polls misestimate voting patterns to questions on concessions and legal battles, I would like to focus on another question, which may seem somewhat unrelated at first. Still, I find it as one of the root causes of these two issues. In one of my most recent research project, my co-authors and I try to answer a simple question: why is it that we become more polarized even though we have more information sources than ever. Why is it that we "can't understand how anyone would vote for the other candidate"? How is it possible to have multiple narratives that seem to be completely conflicting on questions that seem factual in nature? While it is easy to dismiss these as fake news issues, I do not think it is all that simple.

In a recent research paper, which I co-authored with Kimon Drakopoulos and Vahideh Manshadi, we are studying the process of learning and opinion formation. The paper is motivated by the relatively recent changes in how people access and consume news. Specifically, more and more people access news through social platforms such as Facebook, Reddit or Twitter (or even Substack). The immediate implication of this fact is that the cost of distribution has decreased. One does not need to write for the NY Times to have a distribution channel to mass consumers, allowing independent journalists to access their audience directly. The other implication is that we see more sources than ever that have access to consumers. While in the past, it was clear that some outlets are more rigorous in their journalistic processes (for example, double fact check every source), and other outlets are more frivolous in what they publish, it is nearly impossible to track this information at this day and age. Even within the NY Times, some journalists are more careful when posting news and others are less so. Again, I am not talking about fake news and biased reporting. Still, even when trying to estimate the impact of covid-19 and the appropriate strategies, many sources of information may sound reliable, like the World Health Organization, while providing "noisy" estimates about the future. In our opinion, the result is that even for simple questions such as whether masks are useful to reduce the effect of Covid-19, we see polarization. The reality is that for each one of these questions and facts, there is an immense amount of information and information sources, and the question is whether the high number of sources makes us more informed and if not (which I am sure your intuition tells you the answer is “no”), why.

Let's start with the naive approach: well, as you can see here from the Op-ed of David Chavern at the Wall Street Journal,  

"The rapid growth of digital connectivity has pushed demand for information to unprecedented heights. Never in history have so many people consumed so much news. This is a boon for democracy. Although reporting is often an irritant to those in power, high-quality news and analysis are essential to any political system that depends on giving citizens the facts so they can draw their own conclusions."

But as I wrote earlier, unless you are a very discerning audience, it's tough to keep track of each journalist's credibility and journalistic standards. In fact, the same reporter can be a very credible source when it comes to, let's say, healthcare news might not have the same credibility and ability to analyze the topic when it comes to climate policy.  

As I mentioned earlier, a lot of this discussion recently is about fake news, but this is not what our study here is about, not in the way people think about fake news. We study the impact of these phenomena by developing a model in which we focus on how we, as citizens or users, learn information about facts. I use the term facts to describe the truth about a news story. For example, how many people are going to remain uninsured after repealing the Affordable Care Act? We then focus on how people consume information that is provided in the manner described above. Given the amount of information and the fact that most of it appears over time, the process we study is of a sequential updating of our opinion (or beliefs) on a specific fact. The process has two steps: first, the user decides which "post" or "news story" to consume (a stage we call "screening"), and then the user reads the full article or post and uses the information to update their opinions (a stage we call "consumption"). We show that when screening which information to consume, because users have not only to find what information is most useful, but they also need to decide which information is most credible: from the user's vantage point, the most credible information is the one that is closest to him or her, in terms of their initial (or current) opinion.

In this paper, we assume that people are almost rational (I will not get to the technical discussion here) and are not easily biased due to psychological tricks or emotions. Yet, we show that even if people are good “statisticians” in the sense that they try to find the information that will reduce their ignorance by the most significant amount, they will choose to consume information that will slow their ability to learn. In fact, we show that precisely this screening stage will result in polarization: everyone is sort of "stuck" in their side of the political map. For those familiar with the notion of confirmation bias, we are essentially illustrating a mechanism that generates confirmation bias even for a seemingly rational reader.

Again: things can be worse if some of the sources are deliberately misleading. Or if the platform is trying to show you, on purpose, news you are more likely to “like”. But we offer that even in the absence of these, we will be less educated and more polarized about things that should be factual in nature. Depressing.

What can platforms do? It is clear that some of the tools currently used are not very helpful. For example, for a while, Facebook was trying to avoid claiming whether something is truthful or not, but merely showing you other relevant posts. It is clear that this policy only worsens things. FiveThirtyEight's strategy of classifying pollsters by their credibility can be a good starting point if applied more broadly to other instances and not only polls.

A good summary of this short article is this single Tweet by Garry Kasparov

"The point of modern propaganda isn't only to misinform or push an agenda. It is to exhaust your critical thinking, to annihilate truth."

I will just say that it's not only propaganda. Social media platforms are already doing it.