What Psychological Factors Drive Echo Chambers in the YouTube Comment Section?
Did you watch a YouTube video (be perfectly honest)?
I also bet that you lurked around in the comments section. You might even have left a few spicy comments yourself! 🌶
If my previous posts have done any good, it’s to have convinced you to start your own YouTube channel. 📺
For many viewers, the comment section is a place to voluntarily enter an echo chamber. This is neither a good or bad thing on the face of it. For instance, extreme fandom, upon which many of the most successful YouTube channels are built, is a form of echo chamber. 🎉 🎉 🎉
So how exactly can this echo chamberness be accurately measured? Here are the 3 high level concepts we need to track and hopefully measure:
📜 Homogeneity of commenters’ opinions: This is primarily influenced by 2 major psychological phenomena. The first one, selective exposure applied to individuals, means that people going into a comment section are more likely to do so in order to absorb info that is consistent with their pre-existing beliefs and to reject info that isn’t. It’s really the same as confirmation bias in my opinion. The second factor is social contagion. It is both the mechanism via which commenters spread information among each other and mimic each other. Using these 2 phenomena, it is easy to see why in an online world where many like-minded people meet echo chambers can form.
💪 Constant reinforcement: As explained in the above point, the dominant dynamic online is homophilic groups. As a result, people, in addition to seeking people and opinions similar to theirs, get their own opinions constantly reinforced! This simply means repeated interactions with similar people holding similar views.
☃ Polarisation: People tend to be in denial about how much people agree with their points of view. If we are honest with ourselves, there are always nay-sayers and good points of contentions. The comment section is no different. A healthy discussion usually involves many diverse points of view interacting with each other (i.e. low polarisation). Another discussion where view points which disagree don’t interact much if at all with one another is a high polarisation situation.
Both comments homogeneity and polarisation can be measured using the TF-IDF measure combined with cosine similarity. As for constant reinforcement, it can be measured using TF-IDF/cosine similarity in tandem with network analysis applied to the comment-reply network!
My current company is working on an AI platform that allows you to better understand and interact with your YouTube audience via the comment section and conversational AI! Get in touch with me if interested 😉
(Source: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9739846/)
(Source: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7936330/)
