Paper Title: The echo chamber effect on social media

Author(s) and Year: Matteo Cinelli, Gianmarco De Francisci Morales, Alessandro Galeazzi, Walter Quattrociocchi, and Michele Starnini (2021)

Journal: PNAS (open access)

TL;DR: You might assume that giving people control over their social media feeds would trap them in an “echo chamber,” a space where they only hear opinions that match their own. However, a major study of over 100 million posts across four platforms revealed differences from platform to platform. When users can actively shape what they see, such as on Reddit, they are actually less isolated. It’s platforms with hidden, unchangeable algorithms, like Facebook and Twitter (editor’s note: now X), that lock us into segregated bubbles.

Why I chose this paper: When I browse social media, all the recommendations I receive are based on the content I follow. This creates a false impression that the whole world is paying attention to the same things as me and agrees with my views. This, of course, is completely absurd. I was particularly curious about the underlying differences between various platforms that contribute to this effect.

Special note: This bite was prepared as part of a group project for the Science Communication and Outreach course at KU Leuven.

When you’re scrolling through your social media feed, every comment and news item that pops up on the screen seems to perfectly resonate with your thoughts. You might feel a sense of comfort that the world agrees with you, but this comfort may be a carefully crafted trap. With the evolution of algorithms, social media may now be becoming an “echo chamber.”

The Background

Social media is currently the dominant force in news, with many information disseminations occurring through the internet. A recent study has shown that fake news spreads faster than real news. This outcome is driven by several underlying factors, including our limited attention spans and algorithms that restrict our information choices. In order to retain more users, platforms use algorithms​ to recommend content that users are likely to view, creating a sense of satisfaction and illusion among users that many people support them, including their contended ideas. While this customized environment might feel satisfying, it raises a critical question: When we discuss polarizing or sensitive topics online, are these underlying algorithms actually trapping us in isolated ideological bubbles?

Echo Chamber

The authors define this closed environment formed by repeated interactions within a group with similar attitudes as an “echo chamber.” The existence of echo chambers can lead to the continuous reinforcement of views within this space, ultimately leading to extremization and making it crucial to study the impact of social media on news consumption, particularly for polarized topics.

The Research Question

While this polarized network phenomenon seems increasingly prevalent, the study notes that there is still a lack of comprehensive scientific research across different platforms. Therefore, the authors primarily focused on two specific issues:

1. Despite different interactive environments (e.g., retweets on Twitter vs. group discussions on Facebook), is there a common phenomenon of opinion isolation within the same platform, where users of similar interests cluster together?
2. Does the degree of isolation among different groups vary on each platform?

This study is therefore a comparative research across four major platforms: Facebook, Reddit, Twitter, and Gab, an American social media platform known for its minimal content moderation. It attempted to determine whether echo chambers are universal phenomena or are influenced by the platforms’ algorithms.

The Methods

The researchers analyzed content surrounding specific controversial and polarizing topics, such as gun control, vaccination, and abortion. To measure where users stood on these issues, the authors relied on ratings from the independent fact-checking organization Media Bias/Fact Check (MBFC). MBFC scores the political bias of news outlets on a scale from “extreme left” to “extreme right” (assigned a value from -1 to +1). By averaging the bias scores of the content a user actively engaged with or posted, the researchers could quantify their ideological leaning. Furthermore, because user interaction looks different across these platforms, the authors tailored how they mapped the social network links:

1. Twitter: based on the “follow” relationship;
2. Facebook and Gab: if two users leave comments on the same post, it is considered as interaction between them; and
3. Reddit: based on users’ responses to others’ posts or comments.

After completing the preliminary preparations, the author employed a classic SIR model (Susceptible-Infectious-Recovered model) from epidemiology to simulate the diffusion of information. Simply put, it involves randomly selecting a seed user in the network and observing how information spreads among the population. If the information only circulates among people with the same stance, it indicates that the echo chamber is isolated.

The Results

Algorithms vs. Autonomy

By simulating how information goes viral, the researchers revealed exactly how these echo chambers operate. On Facebook and Twitter, the authors have found the bias in dissemination to be strong: your message is overwhelmingly likely to reach only those who already agree with you. The reason why this happens on these platforms, but not as much on Reddit, lies in the flexibility of the algorithms. On Facebook and Twitter, the information flow is dictated by rigid platform algorithms, leaving users with little control and exacerbating their isolation. In contrast, Reddit grants users the autonomy to curate their own communities and shape their feeds, which helps break down these rigid barriers.

A Different Kind of Bubble?

The styles of different platforms vary significantly. In the analysis of Facebook and Twitter, the authors observed a strong clustering phenomenon. On these two platforms, a user’s stance highly overlaps with the stance of their “circle of friends” they interact with on a daily basis. If you are against vaccines, your timeline is almost solely filled with voices opposing vaccines, and this high degree of stance consistency leads to severe social isolation.

The paper revealed that Reddit and Gab present a completely different ecosystem. Unlike Facebook and Twitter, users on these two platforms do not split into deeply isolated, opposing factions. However, this does not mean they are perfect hubs for diverse debates. Gab is an alternative platform known for its minimal content moderation, which has resulted in widespread hate speech and a user base that is overwhelmingly right-leaning (in fact, the authors declare left-leaning users are almost entirely absent). Consequently, users on Gab, and similarly left-leaning users on Reddit, tend to form a single, massive community. Information flows freely among them, but as the researchers point out, this might be because we are observing the dynamics inside one giant, platform-wide echo chamber.

The Impact 

This study tells us that the severe online polarization experienced today is not necessarily an inevitable consequence of our innate tendency to associate with like-minded individuals. Rather, the study suggests that the underlying algorithmic design of social media might be one of the key variables shaping how information flows on a platform. This research provides a crucial new perspective: to effectively study how content spreads, we must examine the specific environment of the platform in question. Ultimately, whether users can actively curate their own feeds directly determines whether they will be trapped in isolated echo chambers.

When it comes to sharing news and engaging in science communication, it is crucial that we tailor our strategies to the specific algorithmic environments of each platform. This adaptability is essential if we want to break through isolating barriers and reach a more diverse audience, rather than just preaching to our own echo chambers. Furthermore, this study can serve as a reminder to policymakers that prioritizing user engagement through rigid algorithmic recommendations might come at the cost of social consensus. However, if users are granted more autonomy over their information feeds, a healthier and more balanced public opinion environment can emerge.

Written by Chao ‘George’ Gao

Edited by Elien Van Clemen and Mykyta ‘Nik’ Kliapets

Featured image credit: Photo by camilo jimenez on Unsplash (Unsplash License)