You (Yes, You!) Are Trapped In An Audience
We need to be more aware of the physics that govern the spaces where we argue, protest, think, speak, and connect
Welcome to Galaxy Brain — a newsletter from Charlie Warzel about technology and culture. You can read what this is all about here. If you like what you see, consider forwarding it to a friend or two. You can also click the button below to subscribe. And if you’ve been reading, consider going to the paid version.
Ben Smith’s Times media column this week chronicled a Facebook group of former Jeopardy! contestants who conjured up a conspiracy theory about a contestant flashing a white power symbol and, despite proof to the contrary, refuse to back down. The example is striking, in part, because the conspiracy is coming from an unusual bunch:
I should stress again that these are smart people, who were in general more polite than the journalists who reluctantly take my calls most weeks. And that, I think, is the point here. The contestants’ investigations of Mr. Donohue had all the signal traits of a normal social media hunt gone awry — largely, that you assume your conclusion and go looking for evidence. And they followed the deep partisan grooves of contemporary politics, in which liberals believed the absolute worst of a Trump supporter. But they also contained a thread of real conspiracy thinking — not just that racism is a source of Trumpian politics, but that apparently ordinary people are communicating through secret signals. It reflects a depth of alienation among Americans, in which our warring tribes squint through the fog at one another for mysterious and abstruse signs of malice.
The Donohue incident reminds me of a piece of advice that researcher Erin Gallagher relayed to me near the end of our long interview on disinformation and information networks:
Ask yourself, ‘what audience are you in? What is the content you’re seeing everyday and sharing? Who makes it? Why are they making it?’ That’s a big ask, but it seems critical.
Average internet dwellers tend to look extremely critically at outside audiences — but rarely apply the same lens to their own networks. You’ve probably seen someone suggest that social media accounts expressing views counter to their own are ‘bots’ or ‘foreign influence operations’ or the result of toxic filter bubbles and echo chambers. There’s a sense that, in someone else’s information networks, conversations are being distorted by nefarious online dynamics and/or manipulated by grifters or by a motivated minority. This is…frequently true! And it’s likely also happening in some fashion in your own information networks — in the audiences that you are a part of.
I don’t mean to produce some false equivalence between dangerous conspiratorial groups/movements and, say, your knitting group on Facebook. I only mean to suggest that most online audiences are shaped by similar forces. Take this study of Hungarian college students on Facebook from 2017. Here’s a bit from the abstract:
The results indicate that only a politically interested minority of university students post or share political content on Facebook. However, posting is shaped by dissatisfaction with the way democracy functions, and accordingly, obtaining regular information about politics through Facebook leads to more negative perceptions about the way democracy works. Based on these findings, it may be assumed that the negative evaluation of democracy by students who are informed about politics through Facebook results from the fact that on this platform information and opinions are mostly provided by their discontented peers.
This is an example of the not-so-subtle ways that the dynamics of a social network influence conversations. The loudest, most opinionated voices can set the tone of the conversation (again, not always a bad thing!) But you can see the dynamic happening in the Jeopardy! incident (emphasis mine):
Several members of the group who thought the reaction to Mr. Donohue’s hand was “unhinged,” as a 2020 contestant, Shawn Buell, told me, stayed silent. Mr. Buell said he assumed he would be shouted down. (He said he had initially considered membership in the group “an honor,” but had learned to stay silent this January after group members bitterly condemned the “Jeopardy!” icon Ken Jennings
I don’t know whether the former contestants are generally good or bad people, and it basically doesn’t matter (Smith suggests they’re smart and polite). Regardless, the norms of the group and its posting cadence and tone are likely influenced by the most outspoken, active members. That might be because their posts are the most polarizing and the most engaging and enticing algorithmically. Or that might be because many group members can’t be bothered to or don’t enjoy posting and would rather read/lurk, leaving the content creation up to the loudest. These dynamics aren’t exclusive to high-functioning game show contestant groups — they are everywhere.
A particularly grim example of this phenomenon comes from a recent internal Facebook report on the ‘Stop The Steal’ groups that helped fuel the events of January 6th. According to the report, revealed by BuzzFeed News, the movement was powered by a semi-organic process of ‘super inviters’:
The biggest Stop the Steal groups had 137 super-inviters, who invited 67% of the groups’ members, according to the report. These accounts were each responsible for inviting more than 500 people to groups. Facebook’s analysis found the super-inviters worked in coordination, lied about their locations, and used private groups and chats to coordinate activity.
What an online audience sees is the result of an information environment interacting — in potentially unexpected ways — with its audience and participants. And each environment behaves differently: a private Facebook group will produce a different conversation than, say, a public Twitter thread, even if most of the participants are the same.
I suppose this is all common sense, but most of us are conditioned not to see these dynamics in our own online experiences. I fall victim to this shallow thinking all the time, especially on a platform like Twitter. I often take my (algorithmically filtered) timeline for granted as some kind of standard Twitter experience, despite the fact that it is curated by, uh, me. Sure, I think a number of people are having somewhat similar experiences to me on a day-to-day basis (because of overlap in followers or topics of conversation) but there’s also a lot more people who are experiencing a completely different internet. The “hellsite” I complain about is usually a prison of my own design.
Allow me to drag myself even more: The simple fact that I frequently complain about Twitter and call it a “hellsite” is the result of my being trapped inside specific information environments. Over the last 12 years, my incessant use of Twitter has been shaped by the people that I choose to follow or who are thrown into my timeline (including a number of whom love to complain about Twitter as a bit of in-group signaling). Sometimes, this can be useful: I’ve learned quite a bit about how to present my reporting in informal and compelling ways and how to evolve my thinking in real-time on the platform. But I’ve also certainly mimicked more toxic behaviors from master dunkers. Given how much time I’ve spent on the platform, the collective experience has affected my politics, my understanding of pop culture, how I evaluate some of the media I consume and likely the way that I think and speak publicly.
This might sound damning to you — what a lemming! — but it shouldn’t. This is happening to you, too. This is what happens when you consume information. Some of the individual bits of data you process have a profound effect on you immediately; others don’t. Over time, though, the ecosystem that feeds you the information also shapes your thinking and opinions and even your ability to break out or influence the ecosystem that feeds you the information.
Consider, once again, Shawn Buell, the Jeopardy! group member who thought the white supremacy conspiracy was “unhinged.” Even though his opinions weren’t shaped by the group, his actions were and he chose to remain silent. It seems likely his objection would not have changed the group’s dynamics, but you never know — it theoretically could have been the crack in the dam that might have caused the group to disagree and splinter. Instead, the norms of the group perpetuated themselves and away we go!
Near the end of the column, Smith cites an excellent Zeynep Tufekci piece about the problem of polarization and the proclivity of some on the left to only “focus on misinformation over there.” Indeed, Smith’s column was tweeted by plenty of folks as a good example that the left also has its share of tin foil hats (though not in equal proportion).
Not to get too meta here, but even these tweets are an example of an information ecosystem influencing behaviors! It is obviously a good thing to be open-minded and honest about conspiratorial behavior happening among all political persuasions/cultures. But among journalists on Twitter, it’s also advantageous to project an air of open-mindedness whenever possible. It builds trust among certain followers and also provides receipts that one can put forth in the event of accusations of bias. Now, I’m not picking on any of these tweets or even suggesting that this kind of cold, calculation was made consciously. It’s merely a norm that is learned over time spent in a particular community (in this case, media and politics Twitter).
Again, in a vacuum, none of these group/ecosystem dynamics are “good” or “bad.” They are existing forces, with their own physics. You wouldn’t call gravity de facto ‘bad’ because planes crash. It simply exists. What seems most important, then, is that we become aware we’re surrounded and influenced by these forces — whether we like it or not.
In the context of the last decade of social media, the Jeopardy! affair is another exhausting example of our information systems amplifying and enabling our worst behaviors. A natural response is to want us all to disconnect from the machine.
But the more I write about these online phenomena, the clearer it becomes that we’re not going to log off — at least not en masse. Many of the information dynamics I described above are what make the internet exciting and chaotic and, at times, glorious. Being part of an audience means belonging, in some sense. People want to belong. And it is far easier to simply assume that the audience you belong to is a bit more savvy and a bit more self-aware and a bit more resistant to these dynamics.
(Aside: I worry that people will read this as some kind of moral relativism. I don’t mean it that way. Some online groups and movements are obviously far worse than others. Some individuals are more susceptible to information operations than others. The networked propaganda of the MAGA media ecosystem is not comparable with perceived mainstream media bias. Similarly, not all information ecosystems are run by charlatans or with the intent to manipulate or deceive. Most people are probably selling something, though.)
But it is precisely because we won’t log off that we need to be more aware of the physics that govern the spaces where we argue, protest, think, speak, and connect. So many of the challenges posed by the modern, platform-based internet are sprawling and structural — and cannot and should not be fixed by shifting responsibility onto individuals. But there is a power in knowing what audiences you’re a part of, and how those audiences shape you and how you shape those audiences in return. As Gallagher told me, “it’s a big ask, but it seems critical.”
I left Twitter last Sept because I became exhausted by my liberal 'progressive' bubble policing my comments & questions. It was the *best* behaviour modification for my mental wellbeing, perhaps in my entire life so far. I wrote about it, if anyone wants to look up 'Sex is great, but have you tried quitting Twitter' by Mike Hind.
Thank you for your beautifully en pointe piece on this.
With my social platform of choice (Instagram) I have found tremendous value in something shy of the “log off” - the hard reset. To manage my own usage behavior I’ll regularly go on a follow/unfollow spree, delete the IG app and switch to web (a far inferior experience but also one that’s less user targeted), or take long-ish breaks entirely. This apparently has the effect of confusing the algorithms powering the explore/recommendation engine — I’m not sure how this works technically, but the effect is that when I reengage, I’m reset back to the default recommended content for engagement (cute puppies, feel-good stories, the occasional celebrity) versus whatever “more like this” pattern I was seeing in the past. I guess I like to think I’m somehow outsmarting the rabbit hole-ing phenomenon and keeping my ecosystem from germinating too deeply in any one direction. But I wouldn’t have put it into these words without being prompted by this essay.