Echo Chambers and a Healthy Network
“Measuring magnetism: how social media creates echo chambers” (https://www.nature.com/articles/d43978-021-00019-4) by Michele Travierso gives a brief introduction to how different mechanisms of different social media platforms are enhancing echo chambers. Echo chambers refer to “situations in which beliefs are amplified or reinforced by communication and repetition inside a closed system and insulated from rebuttal.” [1] The research discussed in the article indicates that social networks and the associated software applications are reinforcing echo chambers among their users and creating segregations among different user groups. The researchers quantify the extent of echo chambering effects by user’s leaning on a given topic and the interactions between users and gathered data from Facebook, Twitter, Gab, and Reddit. The results show that the recommendation algorithms and the way of information spreading on different social media largely affect the scale of echo chambering effects.
Given that social media follows a network structure, we know that echo chambers are formed naturally: in a balanced network either all nodes are friends or nodes are polarized into two conflicting groups and under either cases, nodes are restricted in an echo chamber with similar ideas and opinions. Although we also know that the balanced network theory with negative and positive relationships oversimplifies the reality because there can be mixed relationships in the real world, the research shows that the power of such a model in describing the polarization of social networks on the internet.
However, as the article has shed light on, there are ways to prevent the formation of echo chambers and preserve the diversity of opinions on social media by reconstructing the networks. Theoretically speaking, a balanced network means a network in which nodes will not change their relationships with each other over a long period of time. In terms of social media, if we frame the different relationships between nodes as different opinions among users, we can clearly see that users’ opinions are not unchangeable in a stable social network. Actually, people change their thoughts constantly and it does not necessarily disrupt the stability of a social network. Therefore, social media platforms should keep their eyes a little bit away from the recommendation accuracy of their feed algorithms. Particularly, they should feed users every once in a while with some new topics which they have newly searched for some related keywords because they might be just thinking about some new ideas. As stated above, this is not harmful to the stability of the platform. From my personal perspective, this is even desired by users because I feel bored after being recommended content on the same “favorite” topics for months. The research results mentioned in the article also confirm this idea: platforms with more customizability of recommendation contents, including Reddit and Gab, have less degree of echo-chamber effects.
This also gives insights into the objectives of many state-of-the-art software applications of which the recommendation algorithm is representative. In both academia and the industry, the main objective of recommendation algorithm research has been accuracy for many years and many computer scientists have been working to increase the accuracy measure by tiny points. However, results from research papers similar to the one discussed in the article have investigated the ethical and cultural implications associated with the desire for high accuracy. Another example is that in the field of computer vision and face recognition, the pursue of high accuracy has raised public concerns on personal privacy and civic security. Therefore, it might be the right time for computer scientists to reflect on the myth of SoTA (state-of-the-art performance) and look more into what we should care more about when designing computer technologies.
Works Cited
[1] “echo-chamber noun – Definition, pictures, pronunciation and usage notes | Oxford Advanced Learner’s Dictionary at OxfordLearnersDictionaries.com”. www.oxfordlearnersdictionaries.com. Retrieved 2020-04-25.