The Murmur of the Flock

Let’s take a moment to consider the model underlying the researchers’ work. The video is 2 minutes long, but watching even for a short while will give you the model’s best visualization.

You are watching the murmuration of starlings. Those amazing patterns that form are driven by simple rules; do not collide with the birds around you; follow those around you. It requires no leader, allowing them to travel and avoid predators. (You can read more about the starlings and murmurations here.) The researchers thought that these murmurations were echoed in our use of social media and might give us new insights into what the national or global “we” might be thinking.

The researchers made use of Tweets from 2018 and 2019. They began by identifying “flocks” of Twitter users, networks of individuals sharing social media accounts based upon likes and retweets. They assumed that in the social media world, information moved from the “elites” to average accounts – that we move towards opinion influencers and that the networks around them behaved, like those starlings, responding as a group to their environment’s advantages and dangers.

They began by using “political opinion leaders” as the central node of a flock and looked for those who were followers. They identified 193,000 accounts with 1,310,000 followers; too large a dataset. They reduced the accounts to 100, reflecting the biggest with roughly 13,000 followers each. Their analysis considered those 100 accounts and the 1,000 most active followers. Each flock was given a name based on a profile analysis of their membership, e.g., Bernie Bros., the Trump Train, cultural elites, etc. (Do not be offended by those choices. Nielsen segmentation includes “Fast-truck families,” “Big Fish, Small Pond,” and “Conservative Media.”)

They first determined that they had birds of a feather, homogenous, interactive, stable networks or flocks. They found that within their groupings

• The accounts were more homogenous, sharing more followers within the community than outside their grouping
• The flock was interactive, retweeting an average of 44% of the time within the network
• The flock was stable; 60% of the accounts identified in 2018 remained identified within the community in 2019
• Flocks predictably used the same hashtags, i.e., expressed similar opinions

This is really a more technical way to point out that we live in social media echo chambers. And to be honest, it doesn’t break new ground. Nielsen’s segmentations have been used for years, fueling many advertising campaigns. The company itself was purchased in 2021 for $2.7 billion. What the researchers brought to the table, unlike Nielsen’s static views, was the flock in motion. Specifically, how 10 of these flocks responded to three news events; the Mueller investigation, passage of anti-abortion measures, and the death of Jamal Khashoggi.

In each instance, the flocks generated different opinions, of varying intensity, over different time intervals.

• The Mueller investigation – liberal and conservative flocks were equally engaged (based on the number of Tweets). They were “relatively in sync,” tweeting similar numbers when events unfolded. Of course, the groups used very different words. Conservative flocks used terms like maga or witch hunt; liberals used obstruction of justice and impeachment.
• Anti-abortion laws – conservative flocks were the most vocal, generating 60% of the tweets. They were also the most consistent tweeting throughout the “news cycle.” Liberal communities only entered the fray when “the Alabama governor signed the most extreme abortion ban.” And again, word choice was quite different. Conservative flocks used pro-life words, heartbeat, babies, infanticide; liberal flocks chose terms like access, rights, and bans.
• Death of Jamal Khashoggi – Tweets spread first from the “Middle East Correspondents” flock to “National Political Journalists” to liberal flocks. Conservative communities were late to the party. Interestingly the early movement among journalists was “focused on the event itself,” only when the liberal and conservative flocks got involved did it become politicized. The liberals emphasize President Trump's connection to the Saudis and the conservative community, Khashoggi’s ties to the Muslim Brotherhood.

“By identifying different flocks and examining the intensity, temporal pattern, and content of their expression, we can gain deeper insights far beyond where liberals and conservatives stand on a certain issue. This is because these flocks are segments of the population, defined not by demographic variables of questionable salience (e.g., white women aged 18–29 years), but by their online connections and response to events.”

The method is meant to supplement public opinion polls which, by their nature, are static. We might follow the flocks to gain insights into how these groups experience and respond to actual world events in real-time. Consider how that knowledge might have altered our experience of the COVID vaccine rollout.

Source: Social Media Public Opinion as Flocks in a Murmuration: Conceptualizing and Measuring Opinion Expression on Social Media Journal of Computer-Mediated Communication DOI: 10.1093/jcmc/zmab021