So what happens when your partner isn’t a person—it’s an AI agent? As humans increasingly use AI systems to make economic and social decisions, this isn’t a hypothetical question. And the answer, based on our new research, may surprise you: people don’t simply refuse to cooperate with AI. Instead, the social norms that ordinarily guide cooperation lose their power.

The Game: How We Tested Human–AI Cooperation

We used a classic decision making task from behavioral economics called the Public Goods Game. It works like this: you and three partners each get $8. You decide how much to put into a shared pot. Whatever goes in gets doubled and split equally. If everyone contributes generously, everyone benefits. But there’s a clear incentive to free-ride—keep your money while others chip in. In fact, this is the best solution if you are pursuing pure self interest.

Before each round, participants predicted how much their partners would contribute. Then they made their own decision. Afterward, they saw what everyone actually did. This prediction step turned out to be crucial.

We randomly told some participants their partners were other humans. Others were told their partners were AI systems. Critically, the partners’ actual behavior was identical (we controlled it), so any differences in cooperation between human and AI partners were entirely driven by what participants believed about who they were playing with.

Norms Moved People Less When the Partner Was AI

In Study 1 (N = 794), we varied two things: whether participants played with AI or human partners, and whether those partners behaved generously (prosocial norm) or stingily (antisocial norm). In a previous paper we simply manipulated these norms and found that they doubled the rate of cooperation!

The crucial result was not that people refused to cooperate with AI. In fact, overall cooperation levels were similar regardless of partner type. There was no main effect of playing with an AI versus a human. What differed was how strongly social norms shaped behavior.

When partners were human and behaved generously, participants followed suit, contributing significantly more than when norms were generous. But when partners were AI, this prosocial boost was weaker (see Figure 2 below). The same generous behavior from partners produced less reciprocity when people believed those partners were machines. Put simply: norms that reliably drive cooperation between humans lost much of their force when AI was involved.

Norms Failed Because People Couldn’t Predict What AI Would Do

Why did norms lose their grip? We suspected it had to do with prediction. Social norms work by giving us a reliable map to navigate our social environment: “in this context, people will probably do X”. If that map is messy or hard to read, norms can’t do their job.

That’s exactly what we found. People were systematically worse at predicting AI behavior than human behavior—their prediction errors were significantly larger. They also showed a directional bias, tending to overestimate how much AI partners would contribute.

But this wasn’t a social learning problem. When people saw what their partners actually did and got feedback, they updated their expectations just as efficiently for AI partners as for humans. The error-driven learning machinery in their brain worked fine. The problem was in where they started: their initial mental model of AI behavior was miscalibrated from the outset.

This suggests that the cooperation deficit with AI isn’t about trust, algorithm aversion, or disliking machines. It’s about predictability. When people can’t form accurate expectations about a partner’s behavior, the social norms that sustain cooperation fail.

Describing AI as “Goal-Directed” Closes the Gap

If the problem is a miscalibrated mental model, can we may be able to fix it by giving people a better one? In Study 2 (N = 314), we used the same game but added a simple manipulation: we changed how we described the AI partner.

Some people were told the AI operated mechanistically (executing stochastic decision rules). Others were told the AI was intentional (capable of forming goals, making plans, and adapting its behavior). This is a psychological lever known as the “intentional stance”: the idea that humans naturally predict other people’s behavior by attributing goals and beliefs to them.

The results were striking. When AI was described mechanistically, the familiar cooperation gap appeared: people contributed less to AI partners than to human ones (a difference of about $0.47 per round). But when AI was described as intentional, the gap vanished entirely, contributions to AI and human partners were statistically indistinguishable (see Figure 4 below)

The same pattern held for prediction accuracy. Under mechanistic framing, people had larger prediction errors for AI—especially in prosocial contexts. Under intentional framing, the AI–human prediction gap disappeared.

The Underlying Process is Calibration, Not Learning

We dug deeper into exactly how intentionality framing worked. It didn’t make people learn faster or pay more attention to feedback. Once people saw a discrepancy between their prediction and reality, they corrected course equally well regardless of their partner. Instead, intentionality framing corrected the systematic bias in initial expectations. It gave people a better starting model of how AI would behave. The stochastic noise in their predictions was unaffected.

In other words, the bottleneck in human–AI cooperation is calibration, not learning. People don’t have a broken learning mechanism for AI. They have a broken starting map.

What This Means for the Future of Human–AI Collaboration

These findings flip the usual narrative about how to improve human–AI interaction. Much of the conversation focuses on making AI more trustworthy, more likable, or more human-like. Our work points to a different lever: making AI more predictable.

If people can form accurate expectations about what an AI system will do, social norms engage and cooperation follows naturally. This doesn’t require giving AI a face or a personality. It requires making AI behaviorally legible, helping people build a working mental model of how the system operates. In practice, this could mean designing AI systems that communicate their goals (“I’m prioritizing fairness in this allocation”), provide transparent behavioral histories, or adopt interaction patterns that invite mentalizing rather than mechanical reasoning.

However, there’s a tension here worth flagging. While intentionality framing improved cooperation in our studies, it could also lead to over-trust or the misattribution of human-like moral agency to systems that remain fundamentally algorithmic. The goal isn’t to trick people into thinking AI has a mind. It’s to give them the right cognitive tools to predict its behavior.

As AI takes on increasingly collaborative roles, from team-based work to shared economic decisions, the question isn’t just whether people will trust AI. It’s whether they can read it well enough for the norms that sustain collective action to do their work.
This post was drafted by Laura Globig with edits from Jay Van Bavel. Here is our Working paper — any feedback is welcome!

News and Updates

Jay was interviewed about our book on NPR’s Hidden Brain podcast this week. He responded to questions from listeners of the podcast about identity, groups, and intergroup relations:

Jay will be giving a talk on “Why Outrage Goes Viral: Influence, Identity, Polarization” live in Brooklyn, NY on August 6th at 6pm. This is part of the Lectures on Tap series where professors, experts, and storytellers give thought-provoking lectures in NYC bars. Get your ticket here and enjoy a conversation over a drink with Jay after the talk! Let him know if you are a newsletter subscriber.

Use our Mailbox to send us your questions

We also now have a mailbox where you can send us your questions and we will share the answers in a newsletter once a month.

Talk to the “Power of Us Chatbot”

We created a “Power of Us Chatbot” for subscribers. It is trained on our book and newsletter to answer any questions you have about our content and applying it to your own questions. The chatbot was created by Androw Ramy and Rémi Thériault in the Center for Conflict & Cooperation. It’s also an ethical chatbot (as you can see below, it blocked Dominic from attempting to blackmail Jay). Let us know if you like it or have any suggestions. You’ll need to use the password: _subscriber

Join us for “Ask Me Anything” sessions for summer! Premium subscribers can join us for our monthly live Q&A with Jay or Dom where you can ask us anything from chatting about research, career advice, professional dilemmas, to opinions and recommendations on pop culture happenings. At our July meeting, multiple people showed up and we designed a study together explore how people update their beliefs.

Join us! Upgrade your subscription using the button below. Invites to RSVP will be sent via email from powerofusbook@gmail.com.

• August TBA @ EST with Dom

• September 10th @ 3:00pm EST with Jay

• October TBA @ EST with Dom

• November 5th @ 4:00pm EST with Jay

Catch up on the last one…

Last week, we shared results from a different experiment that used AI to evaluate 26 books with potential to bridge political divides. Two books were chosen for a human trial where researchers will see if reading nonfiction books can reduce polarization and other prosocial traits.

The Power of Us

Can AI Help Us Find the Books That Bridge Political Divides?

Last fall, we proposed a simple research question: can reading the right nonfiction book make people less polarized and more compassionate toward those who see the world differently…

Read more

23 days ago · 12 likes · 3 comments · Rémi Thériault and Dominic Packer & Jay Van Bavel

Learn more about THE POWER OF US

If you like our newsletter, we encourage you to check out our award-winning book “The Power of Us: Harnessing Our Shared Identities to Improve Performance, Increase Cooperation, and Promote Social Harmony”. You can learn more about the book or order it from the links on our website (here or scan the QR code below).

and recently discussed the impact of social technology during a Substack Live event. Jacki is a clinical psychologist at Brown and author of the fantastic newsletter “Techno Sapiens”. She takes an evidence-based approach to all the big debates around social media, smartphones, and AI and regularly dissects new studies and the big headlines on her newsletter. We highly recommend subscribing to her newsletter if you are interested in these topics.

On the live event, we discussed what the evidence really says about teen social media bans, detoxes, and the “scalpel” interventions that have surprisingly positive long-term effects. We also dug into why this debate about social media is so polarized on social media. You can watch the full video (above) or read our summary below.

Social media debates have ironically become a perfect example of the problem they’re trying to solve — debate that is polarized, moralized, and dominated by extreme voices. In this live conversation with we tried to do something unfashionable at the moment: be precise about what we know, what we don’t, and what we can do next.

We agreed that social media likely contributes to many harmful outcomes for individuals and society, but effects are often small or multifaceted and the public deserves nuance. But the debate on this topic is overheated because moralization turns complex evidence and real tradeoffs into good vs evil (read this summary of a new study on how AI debates are moralized). And people with the most extreme views end up dominating the discussion online, which crowds out the voice of people who have nuanced or ambivalent opinions.

We also discussed digital detoxes, smartphone bans, and other strategies to reduce the harms of these technologies. We agreed that people have big expectations for bans or detoxes that often don’t bear out, but there are real benefits — one of the biggest reliable gains may be sleep. However, dosage and compliance challenges with the research make any effects hard to detect. This is very challenging research and it likely underestimates any impact of social media.

Our biggest takeaways

1) Our views about the impact of social media have changed dramatically the more we have learned

The initial days of social media were extremely promising. From people connecting with old friends to the push for democracy in the Arab Spring, we thought that this technology would have a largely positive impact on humanity. Yet as we have collected more data and seen the impacts in our personal lives, we have updated our beliefs to have a more negative view of this technology. That said, we both use social media and think it has some positive uses (such as sharing research or connecting with people outside your social network).

2) Social media probably plays some role in mental health—but it’s not the only story

We went into this conversation without knowing what we each believed about the impact of smartphones, social media, and AI technology. But we immediately agreed its hard to argue that something teens (or anyone) do for several hours a day has no impact. And these impacts might be broad and far-reaching, from well being and attention to social connection and sleep. However, we agree that the effect sizes of social media are likely smaller than headlines suggest, and mental health is multi-causal (although social media isn’t the sole or even the primary cause of adolescent mental health).

3) Small effects can matter when billions of people are exposed

We also discussed how a narrow focus on the effect size might be shortsighted. Even if the effects of social media — or detoxes and bans — are modest, the scale of it is enormous, so tiny shifts can still have meaningful population-level consequences. We are approaching 6 billion social media users and people are on these platforms for roughly 2.5 hours a day (which is even higher for young people). The research also shows a number of negative effects across a wide variety of outcomes, whereas other apps and technologies do not. This means that even very small effects can have a very significant impact on humanity.

4) Status quo bias is baked into our debates about tech policy

We put phones into classrooms and platforms into lives without strong evidence of benefits — so requiring overwhelming proof to change policy entrenches status quo bias. It creates a powerful asymmetry between the burden of proof to insert technology into every aspect of our lives versus remove it from our lives. This bias is extremely favorable to tech companies. We should question this bias and perhaps a better approach is to start with a thought experiment: What would we ideally want to do if we were creating a policy from scratch? Would you allow kids to bring phones into the classroom or not? Then test this against the evidence — is there compelling evidence that would overturn this policy? We should be doing this right now with AI and chatbots.

5) People might be resistant to social media bans because of the way the issue is framed.

People have very strong attitudes based on how an issue is framed. The term “ban” might feel extreme since it evokes the notion of a nanny state of the loss of freedom. But most companies (like Instagram and Snapchat) already have policies disallowing kids under 13 from using their platforms. Therefore, the exact same policy could be framed as simply raising the age of use a couple of years. This seems far less extreme and is less of a challenge to the status quo since it builds on an existing policy that has widespread support (we’ve never heard an academic arguing to repeal these company policies—again revealing the power of the status quo and pro-tech implicit assumptions underlying how researchers think about these issues).

6) “We need perfect RCT evidence” is an unrealistic standard for policy

The other problem we discuss is the challenge of doing truly definitive research on the causal impact of technology, like smartphones or social media. Correlational studies in this space are revealing, but the causal direction between social media and well-being could go in either direction. Therefore, people often call for clear RCT (Randomized Control Trial) experiments. But these studies are extremely difficult to conduct and create clear results because they are extremely expensive, suffer from selection bias and drop-out problems, occur for a very short period of time (often a few weeks), and it’s hard to confirm that we have completely eliminated social media for people (e.g., even if we remove if from their phone, they can often still access it from a laptop or desktop computer).

7) Individual “detox” strategies are a weak test because social media is a group phenomenon

To date, most studies that have examined the impact of removing social media for individuals. But this is a highly limited approach. If you log off but your peers don’t, you can become socially excluded and this alone might reduce your mental health and increase anxiety. This is why school-wide or national policies may change behavior more effectively than isolated individual choices. By requiring an entire community to log off, no one faces the cost of social ostracism. The current bans are allowing scholars to finally analyze some of these impacts at the collective level.

8) The most efficient interventions may be “scalpels,” not sledgehammers

There is a great deal of debate about the efficacy of social media bans (like the teen bans in the UK, France, Greece, Turkey, Indonesia and Malaysia). One of the most promising findings we discussed: our research finds that unfollowing a small number of highly polarizing accounts (and following more uplifting/educational accounts) reduced partisan hostility and increased positive emotion — especially awe — with effects lasting up to 6 months. The reason these effects persisted is because many participants continued to unfollow these accounts after the study was over. Presumably they realized that removing these accounts had a positive impact on their information diet. This works because a tiny group of extreme users dominate the discussion and often post relentlessly. These folks might be ruining the experience for other users and unfollowing them permanently.

9) AI chatbots may be introducing new problems into our social lives

There is evidence that chatbots are highly sycophantic and they get increasingly positive the longer we interact with them. This is damaging our relationships and we found that it is making us more extreme in our beliefs. We need to create chatbots that will make people feel seen without reinforcing their biases. In the short term, you should change the settings on your chatbot to avoid undue flattery.

Let us know if you want us to do other substack live events with other authors. Tell us your favorite authors and debate or discussion topics and we will try to set some of these events up for subscribers.

Thank you to everyone who tuned into our live video—join us for our next live video in the substack app.

Get more from Dominic Packer & Jay Van Bavel in the Substack app

Available for iOS and Android

Get the app

News and Updates

Jay will be speaking at the Human Advantage conference by Ipsos. The event will explore how we preserve democracy, harness AI responsibly, build thriving organizations, and drive meaningful change on October 1-2, 2026. It’s a free immersive online event with leaders in behavioral science and top practitioners in policy and industry. You can explore the program and register here.

Join us for “Ask Me Anything” sessions for summer! Premium subscribers can join us for our monthly live Q&A with Jay or Dom where you can ask us anything from chatting about research, career advice, professional dilemmas, to opinions and recommendations on pop culture happenings. At our July meeting, multiple people showed up and we designed a study to explore how people update their beliefs.

Join us! Upgrade your subscription using the button below. Invites to RSVP will be sent via email from powerofusbook@gmail.com

• August TBA @ EST with Dom

• September 10th @ 3:00pm EST with Jay

• October TBA @ EST with Dom

• November 5th @ 4:00pm EST with Jay

Use our Mailbox to send us your questions

We now have a mailbox where you can send us your questions and we will share the answers in a newsletter periodically.

Talk to the “Power of Us Chatbot”

We created a “Power of Us Chatbot” for subscribers. It is trained on our book and newsletter to answer any questions you have about our content and applying it to your own questions (password: _subscriber). The chatbot was created by Androw Ramy and Rémi Thériault in the Center for Conflict & Cooperation. Let us know if you like it or have any suggestions.

In case you missed our last Column

Since the U.S. turned 250, we explained the bright and dark side of national identity:

Learn more about THE POWER OF US

If you like our newsletter, we encourage you to check out our award-winning book “The Power of Us: Harnessing Our Shared Identities to Improve Performance, Increase Cooperation, and Promote Social Harmony”. You can learn more about the book or order it from the links on our website (here or scan the QR code below).

In Edward Bernays’ classic book Propaganda, he described the people who manipulated and ruled over the masses. He argued that a class of “invisible” people shaped public perceptions on behalf of governments and corporations by using intelligent manipulation strategies. According to DiResta, online influencers operate like modern propagandists by leveraging and engineering viral trends to shape public discourse and opinion. These invisible rulers turn misinformation and fringe views into popular narratives that create, echo chambers and bespoke realities. We fully agree with DiResta that understanding these online dynamics can help explain the rise of anti-vaxxers, QAnon, and GameStop stock, which has important implications for public health, democracy, and economics.

Much of Invisible Rulers focuses on the impact of influencers with large social media followings who have the power to manipulate public opinion. Influencers—from TikTok dance video star Charli D’Amelio to right-wing conspiracy theorist Candace Owens—are often dismissed as frivolous or inconsequential. But they are a new class of elites who possess the power to direct attention, evoke emotion, and mobilize crowds. Indeed, our own research suggests that a very small percentage of these influencers generate the vast majority of misinformation and toxic content online. For instance, 0.01% of Twitter/ X users were responsible for the spread of 80% of misinformation during the 2016 presidential election. There is good evidence to believe that influencers can distort social norms or foment political polarization.

Despite influencers’ ability to capture attention and build a connection with their followers, they do not seem to be representative of the general population. A small number of extreme users constantly post fringe opinions which creates the illusion that these views are far more widely held than they really are–which we call the “funhouse mirror” version of reality. Invisible rulers often work together to spread hostile narratives and conspiracy theories, and push more nuanced, knowledgeable, and reasonable voices out of the discourse. For instance, a small number of passionate, but coordinated, anti-vaxxers drive the online narrative, and following these influencers is linked to vaccine hesitancy. While most people in the real world support vaccinations, they are far less vocal online. This allows anti-vaxxers to amplify their message and mobilize the harassment of public health officials, scientists, and policy makers. We experienced this directly when Jay was attacked by a swarm of anti-vaxxers who were mobilized by a conspiratorial influencer with over a million followers. This dynamic can make fringe or activist groups look normative online because people who do not agree tend to disengage from the debate.

While online influencers can be as consequential as the propagandists of Bernays’ time, they operate differently. While traditional propagandists directly influence the public, invisible rulers exist in a more complex trinity with algorithms and crowds. The intense activity of these extreme voices games the algorithm in their favor, which amplifies their messages and incentivizes others to join their narrative. Algorithms shape the flow of information by feeding people content and recommending groups designed to keep people on the platform. In turn, crowds provide feedback to influencers in the form of likes and shares, creating a parasocial relationship between influencers and followers, building the stature of influencers, and giving them feedback about which views to continue sharing. Together, this trinity creates bespoke realities that shape how groups see the world.

These bespoke realities pose a serious challenge to our social, political and economic institutions, which were built in a world that looked more like Bernays’ than DiResta’s. On January 6^th, 2021, for instance, the U.S. faced an insurrection designed to overturn the results of a presidential election. This event was fueled by a conspiracy theory that was spread and coordinated on social media. How, then, should institutions respond to the rapid growth of the online world and the rise of these invisible rulers? DiResta reviews a number of possible solutions targeting our digital environments—solutions such as increasing transparency, redesigning algorithms, and adding friction. We have found that algorithmic transparency and providing more options to take control of one’s content-recommendation algorithm are particularly popular solutions across the political spectrum.

A growing challenge, however, is that the U.S. Government and social media platforms have become increasingly uncooperative with such efforts. For instance, Meta removed fact checkers from its platforms and Elon Musk dismantled the content moderation policies on Twitter/X while incentivizing influencers to spread incendiary content. The solution will therefore require other approaches. At the individual level, for instance, people can take control of their feeds by unfollowing accounts posting extreme, undesirable content. Institutionally, our legal system can step in to check the power of Big Tech companies, as in recent lawsuits ruling that Meta’s “addictive design” harms young users.

Another important development since DiResta’s book is the proliferation of AI models. It is unclear how invisible rulers will operate in a post-AI world. Will influencers have less impact in a post-AI world, as people can instantly fact check their claims or as AI influencers take the place of human influencers? Or will AI make influencers become more powerful, using AI to generate ever more realistic and dynamic misinformation or propaganda campaigns? Will AI lead us to generate even more customized and extreme bespoke realities? Or will AI potentially nudge people toward more accurate—and less extreme—beliefs? The future in this area is difficult to predict: but one recommendation from Invisible Rulers that stands is that institutions need to adapt to this new world, and cannot be passive in the face of these new changes. As she notes, “there will be no return to a handful of media translating respectable institutional thinking for the masses.”

While it might feel convenient to dismiss or ignore the online dynamics of invisible rulers, this new information ecosystem is now fully integrated with traditional media outlets who pull stories from social media and journalists who actively engage in online discourse. Even the giants of traditional media–like the New York Times–are subject to the online incentive dynamics to ensure traffic, ad revenue, and public relevance. This means the power of the new class of invisible rulers has huge implications for nearly every sphere of society. Individuals, organizations, and institutions need to understand and adapt to this new reality. We are all in the grasp of this new tyrannical minority and the algorithms and crowds who amplify their messages.

This article was written by Jay, Steve Rathje, and Raunak Pillai. You can download the full preprint here: Van Bavel, J. J., Rathje, S., & Pillai, R. M. (2026). Tyranny of the minority: How social media influencers, algorithms, and crowds shape public opinion. Under review.

News and Updates

Jay will be having a live conversation on substack with Jacqueline Nesi on June 4th at 4pm EST on the big question: “Is a digital detox really worth it? What tech is doing to our relationships and mental health.” All newsletter subscribers are invited to attend live and join a Q&A session.

Jacki is a clinical psychologist and professor at Brown University who writes the popular newsletter TechnoSapiens. She studies how technology and social media impact mental health (especially for teens), and how parents can help their kids navigate it. Here is is an example of one of her recent posts:

Techno Sapiens

Are school phone bans working?

5 min read…

Read more

3 months ago · 35 likes · 10 comments · Jacqueline Nesi, PhD

Talk to Us

Use our Mailbox to send us your questions

We also now have a mailbox where you can send us your questions and we will share the answers in a newsletter once a month

Talk to the “Power of Us Chatbot”

We created a “Power of Us Chatbot” for subscribers. It is trained on our book and newsletter to answer any questions you have about our content and applying it to your own questions. The chatbot was created by Androw Ramy and Rémi Thériault in the Center for Conflict & Cooperation. It’s also an ethnical chatbot (as you can see below, it blocked Dominic from attempting to blackmail Jay). Let us know if you like it or have any suggestions. You’ll need to use the password: _subscriber

Join us for “Ask Me Anything” sessions for summer! Premium subscribers can join us for our monthly live Q&A with Jay or Dom where you can ask us anything from workshopping research questions, career advice to opinions and recommendations on pop culture happenings. Upgrade your subscription using the button below.

Invites to RSVP have been sent via email from powerofusbook@gmail.com

• June 4 @ 4:00pm EST with Dom

• July 6th @ 1:00pm EST with Jay

• August TBA @ EST with Dom

• September 10th @ 3:00pm EST with Jay

Catch up on the last one…

Last week we interviewed George Newman about how great ideas happen—and why creativity is a team sport…

Last week, juries in two different states delivered multimillion-dollar verdicts against Big Tech. A New Mexico jury handed down a $375 million verdict in a case brought by the state’s attorney general against Meta for enabling child sexual exploitation. The next day, a California jury awarded a young woman a combined $6 million in damages from Meta and YouTube for the allegedly addictive and mentally distressing properties of social media apps, including algorithmic curation and so-called infinite scroll, where the app continually provides you with new content as you scroll down the page.

These verdicts have catalyzed a "tidal cultural shift" toward viewing social media companies not just as content platforms, but as manufacturers of products that require regulation, similar to classic legal cases against "Big Tobacco". In fact, this parallel is exactly how some Meta-employees felt about their attempts to target kids:

The focus of the public has shifted from the content posted on platforms and content moderation policies to the deliberate design decisions made by these platforms—such as infinite scrolling, algorithmic recommendations, and push notifications—that are engineered to maximize engagement, even at the cost of individual or collective well-being.

This is hardly news to us. We have been studying and writing about the challenges posed by social media for the past decade. In this newsletter, we have written about the Facebook papers shared by a prominent whistleblower and why schools should ban smartphones.

This should also come as no surprise to anyone who has been following these issues. A majority of people believe that social media is harming children—and this has now been the case for several years. It is also a rare case of bipartisan consensus, with Democrats, Republicans, and Independents all on the same page.

In many cases, the research and safety teams are deeply aware of these problems. But the senior management in the companies is often unwilling to implement changes. The pursuit of profit is a difficult temptation and they have focused on harvesting the attention of teenagers, rather than ensuring their safety. For instance, Mark Zuckerberg wrote in a 2017 internal memo: “Teen time spent [should] be our top goal of 2017”, even after safety concerns were raised by employees.

Whistleblower Frances Haugen was one of the first employees to bring these concerns to the attention of the public. Her testimony and public interviews provided a better understanding about how Facebook engineers their platform and the consequences this has for spreading misinformation and fostering social conflict.

“Facebook is optimizing for content that gets engagement or reaction. Its own research is showing the content that is hateful, that is divisive, that is polarizing—it is easier to inspire people to anger than other emotions.” — Frances Haugen, on 60 Minutes.

A few days after the whistleblower went public, Facebook’s chief executive, Mark Zuckerberg, spoke with employees about the revelations. According to the New York Times, he claimed that Haugen’s assertions about how Facebook polarizes people were “pretty easy to debunk.”

Two court cases—and numerous studies later—the data is firmly on her side.

Debate about the potential harms caused by Facebook and other social media platforms is still up in the air on some issues, but findings on the spread of divisive and polarizing content (in the United States at least) seem clear. Before the recent interview with Haugen on 60 Minutes or her appearance before Congress, Jay published an analysis of nearly 3 million posts on Facebook and Twitter with his colleagues from Cambridge University, Steve Rathje and Sander van der Linden.

In our research, the single biggest predictor of social media “virality” was dunking on a political foe. Each word referring to the political out-group increased the odds of a social media post being shared by 67% (see figure below). If a post came from a Democrat, a word like “Republican” or “conservative” led to increased virality. And if a post came from a Republican, a word like “Joe Biden” led to the same. And most of these posts were clearly negative.

One example of a viral post that spread in Republican circles was from Breitbart, promoting a video entitled, “Every American needs to see Joe Biden’s latest brain freeze.” Meanwhile, a post that went viral from Democrats came from The Daily Beast, claiming that Mike Pence was blatantly lying about Covid-19.

When Steve, Sander, and Jay published a report about these data in The Washington Post, Facebook published a fierce rebuttal. They argued that “Extremism is bad for our business” and presented a limited set of studies that focused largely on the impact of Facebook in other countries.

Yet their own internal research, shared by this whistle-blower, contradicts their rebuttal. According to a disturbing new analysis of the Facebook Papers by The Atlantic, employees at Facebook have long known that the social media giant amplifies extremism, spears misinformation, and encourages political polarization. And they are growing increasingly angry with the way that company leadership has refused to incorporate these insights.

“Again and again, the Facebook Papers show staffers sounding alarms about the dangers posed by the platform—how Facebook amplifies extremism and misinformation, how it incites violence, how it encourages radicalization and political polarization. Again and again, staffers reckon with the ways in which Facebook’s decisions stoke these harms, and they plead with leadership to do more.

And again and again, staffers say, Facebook’s leaders ignore them.”

It also turns out that people don’t like this type of content. We asked a diverse sample of Americans how they feel about the content that spreads on social media. As you can see below, people believe that divisive content, moral outrage, negative content, high-arousal content, and misinformation are all likely to go viral online. However, they wish this type of content would not go viral on social media. Instead, they want to see far more positive content spreading online, such as accurate, nuanced, uplifting, and educational content.

The aversion to social media extends well beyond specific topics. About half of Gen Z wishes TikTok (47%) and X (50%) didn’t exist. That’s despite—spending four hours a day on social media, as more than half of respondents to a new survey say is normal.

Scholars have been sounding the alarm about these problems for several years. Indeed, we devoted an entire chapter to toxicity online in our book “The Power of Us“. In the book, we describe a clever experiment that offers causal evidence that Facebook increases polarization and reduces well being. A group of economists led by Hunt Allcott decided to find out what would happen if they paid people to leave Facebook for four weeks in advance of the 2018 US midterm elections. They paid more than a thousand volunteers to log off the social network for a few weeks and compared them to a control group of volunteers who continued using Facebook.

This experimental design allowed the researchers to detect the influence of using Facebook on polarization and other potential harms…or, more accurately, whether ceasing to use the platform reduces these problems.

Not only did the volunteers who logged off Facebook report an increase in their psychological well-being, but they became less politically polarized (across a variety of measures). Remarkably, the reduction in polarization from logging off Facebook for a single month was equivalent to nearly half the amount polarization has increased in the US since the mid-1990s!

In defense of Facebook, they are certainly not the only platform exacerbating social conflict by amplifying polarizing content and misinformation. We recently published a review of the research on the topic and found that Twitter/X is even more polarized than Facebook or WhatsApp in some countries. And the effects are mixed—one study in Bosnia, for example, suggests that Facebook might help expose people to different perspectives if their real life social networks are homogenous.

But when this social media technology was introduced to the Marubo tribe deep in the Amazon via access to Elon Musk’s Starlink it revealed the same basic pattern we have seen over and over again in many studies. Members of this small community quickly became obsessed with social media to the extend that the community had to shut down access to the internet each day to ensure they would hunt and gather enough food for their own survival.

More strikingly, it also spurred the kind polarization we saw in our American samples. According to the NYT “Elon Musk’s Starlink has connected an isolated tribe to the outside world — and divided it from within.” I described this natural experiment a few months ago on Andrew Yang’s podcast, which you can watch below:

In the United States, where antipathy toward out-group partisans is at a 40-year high and often seems like a sectarian conflict, social media amplifies existing disagreements, facilitates the spread of conspiracy theories, and allows for the organization of anti-democratic activities (like the January 6th insurrection). But the Amazon example also illustrates that this technology can have a massive disruptive influence even in a community without these political dynamics.

This technology poses a serious problem for public health, as well. As Biden’s Press Secretary noted in July, twelve people produced 65% of anti-vaccine misinformation on social media platforms. At the time, all of them remained active on Facebook despite some being banned on other platforms (including ones that Facebook owns!). Indeed, news exposure from Facebook was one of the single biggest predictors of vaccine hesitancy in the United States—with people who only got their news from Facebook expressing greater vaccine hesitancy than Fox News viewers.

New research suggests that there may be some alternatives. A group of researchers from Stanford developed a novel method for reranking social media feeds. They found that increasing exposure to antidemocratic attitudes and partisan animosity posts in the newsfeed increases affective polarization and negative emotions, whereas decreasing exposure reduces them (see figure below). These changes were comparable in size to 3 years of change in United States affective polarization.

As political polarization and societal division become increasingly linked to social media activity, these findings provide a potential pathway for platforms to address these challenges through algorithmic interventions. These interventions may result in algorithms that not only reduce partisan animosity but also promote greater social trust and healthier democratic discourse across party lines.

It’s unlikely that everyone would be willing to take a four week vacation from Facebook. But this research suggests that changes to algorithmic feeds can mitigate some of the harms of social media. There is a growing awareness that people need to take this research more seriously and think about the potential costs to society given the impact of Facebook, X, WhatsApp, and other social media platforms.

The problem with social media platforms is that many of their design features and algorithms are optimized for engagement, rather than accuracy, cooperation, or other civic values—including the values that promote a healthy democracy. It is increasingly clear that denying the body of internal and external research about these issues is simply no longer an option for Meta or the rest of us.

Catch up on the last one…

If you missed our live event last month, read on to hear more about our advance screening to the AI Doc: Or How I Became an Apocaloptimist!

The Power of Us

Reflections on The AI Doc: Or How I Became an Apocaloptimist

“This is the last mistake we’ll ever get to make” — Aza Raskin…

Read more

4 months ago · 13 likes · 6 comments · Center for Conflict + Coop., Jay Van Bavel, and Anni Sternisko

“This is the last mistake we’ll ever get to make” — Aza Raskin

This week, we curated a diverse audience of over 150 experts—ranging from academics, journalists, and tech—to join us for a special screening of The AI Doc: Or How I Became an Apocaloptimist (which is now out in theaters!). We also included several newsletter subscribers who joined us in person.

After the screening, Anni hosted a panel discussion with Jay, the author of The Anxious Generation Jonathan Haidt, and Academy Award-winning producer Jonathan Wang. The room was buzzing afterwards and we decided to share a recap of the discussion to share with the rest of the world

While much of today’s conversation focuses on how AI (Artificial Intelligence) will shape the future. We took a step back and asked a different question: How can we guide AI toward a future we want, and how do we keep that future human?

“The rare feeling of being in a room assembled with real purpose, where the aim was not simply to generate more noise around AI, but to think more carefully about what this moment demands of us.” — Event attendee, Paul Gimenez

Our Key Take-aways:

1) Watch the documentary, but don’t watch it alone

Is it a super accessible primer on the promise and perils of AI that your family should see? Yes. — Event attendee

We have been studying the psychology of technology for over a decade, but the documentary rattled many of us in a new way. Watching together, we were on a roller coaster of despair, excitement, and hope. Jay said his first reaction was existential terror. His second reaction was hope—especially when people spoke about collective action.

This is a huge problem that will require a lot of collective action (…) hearing everyone react to it made it feel like we were all (…) reaching a common ground. — Event attendee

2) AI will supercharge the good and the bad

AI can potentially help us solve climate change, find a cure for cancer, and offer children around the globe access to individualized, high-quality tutoring. Yet, the same powerful technology can advance bioweapons, feed children sexualized content, and erode our shared understanding of truth. The two outcomes are intrinsically intertwined.

Jay and Jonathan Haidt have written a lot about how tech companies are incentivized to maximize attention because this keeps users on their platforms and engage more. But this can have a number of downstream consequences, including: polarization, distorted norms, and declining adolescent mental health. Meta just lost two trials this week: A $370 million lawsuit over child sexploitation and another $4.2 million for creating addictive design features that harm users.

I expect it will become essential viewing for anyone trying to think seriously about artificial intelligence and its human consequences.— Event attendee

AI may perpetuate this problem: In a set of studies, led by Steven Rathje, Jay finds that people enjoy interacting more with sycophantic chatbots (chatbots that validate beliefs indiscriminately) vs. more balanced chatbots which led to more extreme political beliefs. And a new study came out this week showing that AI chatbots are 50% more sycophantic than humans. The authors concluded “that seemingly innocuous design and engineering choices can result in consequential harms”. This is a recipe for disaster.

3) Misaligned incentives and the psychology of AI arms races

The documentary explains that there is no good AI vs. bad AI. The promise and peril are inextricably linked and it is therefore the underlying incentive structures that we have to understand and guide. And these incentives are currently steering us in the wrong direction, creating a race similar to the nuclear arms race during the cold war; companies and countries are incentivized to cut corners in AI safety to avoid losing the race.

What future should we thrive for and how do we get there?

The documentary challenged us to reflect on what future we want and what it means to be human.

As we have seen with social media, it is difficult to predict the impact of new technology. Social media played a key role in democratic movements such as the Arab Spring, but it was also linked to the Rohingya genocide. We need to learn from that and design the technology to avoid these risks, rather than dealing with the damage to individuals and society in hindsight.

Jonathan Haidt urged us to create resilient societies with a strong immune system to buffer unpredicted stressors. Societies that have institutions that can react fast and civilians that have moral maturity and wisdom. This is a tall order for our current political leadership so the burden will fall on each of us to advocate for better solutions.

Psychological research can help us get there. For instance, the study by Jay and Steve also found that engaging with sycophantic chatbots did not increase belief extremity among participants who scored high in intellectual humility—these individuals were more receptive to chatbots who gave them critical feedback and a well rounded view of an issue.

The documentary producer, Jonathan Wang, spoke about the need to focus on human beauty and creativity. In a world that gets more and more complex and noisy from the abundance of content, people are craving meaning, beauty, and genuine social connections—uniquely human experiences. Human-made art and story telling rather than synthetic and hyperpersonalized content may be one way to get there.

4) The psychology of collective action

Changing incentives and institutionalizing alternatives requires coordination and collaboration at an unprecedented scale. As the world is entering an era of democratic instability, intergroup conflict, geopolitical isolationism, and distrust this may seem impossible. But psychological research and practical experience can help us come together.

We have studied intergroup conflict for several decades and found that a shared sense of identity and superordinate goals can encourage people to collaborate instead of compete. Research also finds that changing social norms may be a powerful tool to mitigate selfish behavior. For instance, a study conducted in Finland found that Finish students were less susceptible to selfish behavior, likely because Finland scores high on social norms of cooperation.

One of the biggest contribution of the movie is that it gives us a common language and knowledge around AI and how it affects us personally and society at large. Common knowledge is essential for breaking pluralistic ignorance and coordinating action. Jay noted that our own research finds that the overwhelming majority of people want transparent algorithms and they want control over these algorithms (see figure below).

If we are largely on the same page, you may ask how we can possibly rally people? The PhD student in our lab—Danielle Goldwert—is studying this question for her dissertation. She finds that a feeling of collective efficacy is critical to mobilize people. This sentiment is echoed in historic events like the civil rights movement:

Agency and hope

Claiming our agency was one of the biggest themes of the movie and the panel discussion. Hopelessness is the enemy of change. Indeed, one study finds that sense of agency predicted collective action, but only among groups that also scored high in hope.

How do we find hope?

A recent poll finds that 80% of U.S. adults believe the government should maintain rules for AI safety and data security, even if it means developing AI capabilities more slowly. Recently, the U.S. Congress voted 99-to-1 to strike down a bill proposal that would limit state-level regulation of AI In such a deeply polarized country this should give us hope. We can agree on things, we just need to find the same language and collaboration to get things done.

Our panelist Jonathan Haidt said that it is the most exciting time to be a social psychologist. We couldn’t agree more. Our field has all the tools to study these issues. Understanding these dynamics may prove very helpful for taking action. You can build your own understanding by watching the movie, talking about it with your friends and family, and becoming a part of the conversation.

The documentary will be released nationwide in theaters March 27. Catch it in your local theater and find out where you are on the Apocaloptimist Scale.

We thank the NYU Arts and Science Office of Research and Focus Features for their partnership in making this event possible.

Notes of the week

This screening was the first in a broader series of events we’ll be opening up to newsletter subscribers, including online courses for premium subscribers led by Jay and Dom, so keep your eyes peeled for more to come.

In the meantime, Jay was interviewed on NPR's Hidden Brain about the power of social identity. We often think of beliefs and decisions as individual choices. But in many cases, they are shaped by our social identities. When people join groups, they don’t just coordinate behavior—they begin to see the world through a shared lens.

This has profound implications for: leadership & conformity, organizational culture, intergroup polarization, and collective action. The themes we discussed in this very newsletter about AI. But social identity also helps explain why changing minds can be so difficult: you’re often engaging with identity, not just information.

If you’re interested, you can listen to the full discussion here.

Catch up on the last one…

After the conservative activist Charlie Kirk was fatally shot at a university rally in Utah last week, Spencer Cox, the state’s Republican governor, called social media companies a “cancer.”

Senator Chris Coons, a Democrat of Delaware, blamed the internet for “driving extremism in our country.”

President Trump, who helped found the Truth Social platform, also pointed fingers at social media on Monday and said the accused gunman had become “radicalized on the internet.”

The aftermath of Mr. Kirk’s killing reveals how inflammatory and hateful online content has become. Countless people have been fired for their comments on social media, misinformation and conspiracy theories are running rampant, and the companies themselves are no longer even promising to solve it. Mentions of “civil war” have surged on X and Elon Musk seems committed to fanning the flames of conflict himself.

F…

Read more

While working on their undergraduate degrees, Roy and Neel Shanmugam invented Interview Coder: an AI-powered, undetectable app that helps interviewees pass coding interviews for computer science-related jobs. Their plan was to “use it themselves, get offers from top companies, film everything, and ride the shock factor”.

Roy focused more on distribution and virality than the product, so he and Neel built it in 10 days using open-source AI and launched it on LinkedIn. He then filmed and posted his entire Amazon recruitment process where he used InterviewCoder from online assessment to final offer to show how it works. After an Amazon employee saw the video, they reported it to Columbia which resulted in Roy’s dismissal from the university. Now, they are building Cluely, a tool that aims to help people cheat on everything. Eventually, they allegedly want to embed Cluely via micro-chip form into human brains.

As a Silicon Valley resident, it’s hard not to hear about the actions and ambitions of local start-ups. I actually have some mutuals with the exec team at Cluely and their “frat house” office is nearby. But still, their reach beyond tech without a polished product is impressive. Earlier this year, they recieved a $15 million seed from A16z just months after their initial fundraising round of $5.3 million. The translation from online clout to cash is real.

Most people have an intuition about why a 21-year-old CEO and his start-up’s unique culture is so notorious. But here’s what social psychological factors play into it:

How Roy Lee became a “king of distribution and virality”

1. Activating morality: Research has found that content with moral underpinnings is more likely to go viral on social media. Specifically, posts that use moral-emotional language to talk about social transgressions tends to spread quickly on social media. Words that can activate both morality and emotion are “abuse”, “honor”, “spite” and “greed”. Thus, Cluely’s tagline, “Cheat on everything” is highly effective in capturing attention, controversy, and clicks.

2. Negative content: Roy’s understanding of Twitter/X users is that “They love controversy, drama, and attention; they love to either dog on people or watch people get dogged on.” In an interview, he explained that if half the people in the world wouldn’t feel very negatively about a post, it probably won’t go viral — most viral content requires a controversial twist. He is correct. Content that is divisive, negative, and controversial tends to get more engagement compared to positive or educational content.

   Notably, Cluely’s ads have depicted scenes of violence and verbal aggression to game the attention economy. High-arousal, negative content tends to attract attention. Arousal refers to the intensity of an emotion, regardless of its valence (negativity or positivity). Anger and excitement are high-arousal emotions, sadness and calm are low-arousal emotions.

   

   Being keen to high-arousal, negative information likely has evolutionary origins. It can be valuable for people to pay attention to these types of information because it can signal threats to wellbeing. However, social media environments are an example of evolutionary mismatch, because many people pay too much attention to negative information online that doesn’t help in real life.

3. Acknowledge a shared sense of reality: Hiring in tech has slowed down in recent months/years. Computer science and engineering majors are now among the top 10 majors with the highest unemployment rate. Interview Coder’s popularity likely stemmed from a shared frustration among job seekers who struggle in coding interviews. While the questions are meant to evaluate problem solving and critical thinking, success often relies on familiarity with the problem and the ability to memorize (sort of like studying for the SAT). These interviews are an obstacle for fresh grads, and Roy’s marketing created a sense of shared reality in a tough job market.

   

   Content that rallies people around a solution to a collective problem is more likely to be shared because people see it as potentially beneficial to their communities.

4. Being unhinged: Unhinged marketing has worked for Cluely because it increases brand awareness, which ultimately increases conversions in the long-run. In an interview, Roy explained, “If I say extremely crazy shit online, it will make more people interested in me and the company and drive more downloads and conversions for Cluely. If anything, I need to become crazier online so that people will keep funneling attention towards the core product.” In another article, he shared his plans to throw a $1.5 million dollar rave in September which would promote Cluely.

   

Overall, the strategies that Roy and his marketing team have employed are prime examples of types of content that use human cognition to go viral on social media. Social media algorithms are designed to promote this content because they too, understand social psychology. All this to say, be mindful of your social media consumption and take notice when trolls like Roy Lee are ragebaiting for clicks.

News and Updates

Today is the last day to enter our book giveaway! Check out our interview with about his new book, Misguided to leave a comment and enter the giveaway. Paid subscribers are automatically entered. Good Luck!

*Closes Aug 5th, 11:59PM PST

The Power of Us

How misguided beliefs are formed: Interview + book giveaway with Matthew Facciani

Since 2020, the number of people who regularly get their news from social media has risen—particularly for TikTok and Instagram. While these platforms offer convenience and rapid, bite-sized updates, they also breed rampant misinformation that’s hard to detect and even harder to stop…

Read more

a year ago · 5 likes · 7 comments · Matthew Facciani and Dominic Packer & Jay Van Bavel

Catch up on the last one…

More on the funhouse of social media in last week’s newsletter. Ideas on social media do not reflect people’s beliefs in real life.

The Power of Us

How social media distorts our perceptions

The experience of interacting with others on social media is becoming more and more distant from reality. Jay recently wrote an article for the Guardian exploring this disconnect. He explains how social media creates a deeply distorted picture of society, amplifying outrage and division far beyond what most people experience in their daily lives…

Read more

a year ago · 5 likes · Center for Conflict + Coop.

The gap between the inflammatory tone of online discourse and the relative calm of everyday life has grown more pronounced over time. A recent article from our lab by Claire Robertson, Kareena del Rosario and Jay explains how social media functions less like a neutral reflection of public opinion and more like a funhouse mirror—creating a distorted reflection of reality.

A small, hyperactive group of users—just 10%—produces about 97% of political content on platforms like X (formerly Twitter). Even more striking, only 0.1% of users are responsible for 80% of fake news. This small group of "super-users" skews public perception by flooding social…

Read more

We decided to share a column from Annie in which she talks about the real world challenges of moral decision making in the modern age of artificial intelligence. In particular, she describes fascinating research about the moral programming people want in driverless cars. This turns out to be a really critical issue because no one wants to buy a driverless car that will make a moral decision they wouldn’t make (e.g., smashing them into a pole to save a stray cat). Annie explains how deeply human moral intuitions inherited from the selective pressure…

Read more

I was recently invited to discuss my research on the profound impacts of social media on our psychological makeup and social behaviors on the podcast “Plain English with Derek Thompson”. He distilled my research into a "devil's playbook" of online engagement and we discussed four dark laws on online engagement that are shaping our social relationships.

The fundamental aspects of group psychology—our ability to join and identify with groups—is among the most unique human traits. Indeed, we are the only primate who will engage with prosocial behavior with in-group mem…

Read more

I’m not normally prone to such sweeping statements, I prefer to understand modern challenges and social problems through a much larger cultural and evolutionary lens. But I feel like I stand I solid ground on this one. Check your phone right now and see how many times you’ve picked it up this week (click settings/screen time/see all app & website activity on an iPhone).

I average 101 pick ups per day (with a peak of 130 on Sunday, when I should have been paying more attention to my family). This is particularly problematic for someone whose livelihood hinges on thinking deeply about topics. I often feel like Sugriva, the poor chimpanzee in this video, scrolling mindlessly through my apps.

Unfortunately, we are now held captive to an onslaught of content engineered to appeal to our social instincts. We respond to the various notifications because for many reasons, but chief among them is the desire to stay connected with our friends, family, colleagues, and communities. This is what makes this technology uniquely difficult to escape: if everyone we know is using it, disengaging feels like a form of social suicide.

This is why social media is a collective trap. Users feel compelled to stay engaged, even if they hate it. A recent paper from a group of economists found that the majority of people would prefer to live in a world without Tik Tok or Instagram (which currently have 3.5 billion users). Indeed, nearly 60% of Instagram users wish the platform wasn’t invented.

In fact, these social media users were willing to pay $28 and $10 to have others, including themselves, deactivate Tik Tok and Instagram, respectively. Yes, they would pay their own money to delete these apps from existence.

Yet they stay online because they feel socially compelled to use social media despite how awful it makes them feel. This is what the authors refer to as a collective trap, where people are stuck in an inefficient situation due to a collective coordination failure. They are driven by a sense of anxiety about being social excluded or a fear of missing out.

In collective traps, it’s often impossible to escape with individual solutions. This explains why all my attempts to minimize screen use—from deleting my social media apps to removing my iPhone from my room—ultimately fail. It is not a lack of willpower, but an issue of collective action. Like the participants in the study, I would probably be better off if the apps were simply removed from the world.

There is currently a fierce discussion about the potential harms of these technologies and how to regulate them. For instance, Jon Haidt’s new best-selling book “The Anxious Generation” has sparked a heated discussion about how harmful this is to kids and what might be done about it. I bought a copy but haven’t read his book yet, so I don’t want to weigh in on his specific claims. But his writing on the topic—and the response from many scholars in the field—has led to a critical discussion about these issues.

(FWIW I’m largely convinced that social media has a number of harms to society, but I’m still not sure how much they contribute to several mental health problems. The data I’ve seen offers a clear correlation and some causal evidence that social media reduces well-being—but the effect size tends to be pretty small. Of course, this doesn’t seem to account for social network effects which means it might be an underestimate. To get a better handle on this topic, my lab is about to launch a global experiment on the impact of social media so stay tuned for the results).

The collective trap posed by social media and technology is something I face in my own classroom—where I teach Introduction to Psychology to over 300 undergraduate students. Over the years, it’s become increasingly difficult to hold their attention long enough for me to deliver a lecture. Of course, my problems pale in comparison to high school teachers who have to hold the attention of middle schoolers all day, every day.

The internet is now full of stories of teachers who have painstakingly counted the number of disruptions in class. For instance, a math teacher in Hillsborough County asked her class of 23 students to put their phones on the corner of their desks, and every time they received a text or a notification to mark it on the board. Can you guess how many times they were disrupted in 30 minutes?

This is what the board looked like after half an hour:

For those doing the math at home, there were:

• 58 texts

• 182 social media notifications

• 28 other notifications

That's 268 disruptions in a half an hour! Can you imagine trying to teach or run a meeting with that many distractions? Can you imagine how hard it is for the kids to pay attention? This is the new normal. Instead of learning, we are either peaking at our phones or feeling a sense of FOMO because no one is pinging us.

Even setting aside the potential costs for our mental health or social connection with the people around us, this has to be extremely damaging for learning and memory. There is extensive evidence that divided attention, task switching, and cognitive load are associated with inferior cognitive performance. At a minimum, we are distracting ourselves into stupidity.

This is the type of problem that requires collective solutions. As it turns out, my kids were recently part of a pilot experiment at their local school that provided one clever solution.

My kids go to a public middle school in NYC where they lock up their phones for the day. The kids need their phones to commute through NYC (including the subway system) and coordinate social events with their friends. But once they arrive at school, the phones instantly become a distraction.

To address this problem, last year the school gave each kid a Yondr Pouch. Every morning, all the kids lock up their phones in the pouch and the teachers unlock the pouches at the end of the school day. Naturally, the kids hated this policy. A few managed to break into the pouches, but they were forced to pay a small fee for every broken pouch. Eventually they got used to it.

After a year, the school finally emailed parents the results of the pilot project last month. This is what the school observed:

“Overall, the program has been a massive success. We are happy to share that we continue to see the benefits of using Yondr, with increased student engagement in the classroom, less time spent in the bathrooms and hallways, more genuine connections within the community and a decrease in reports of cyberbullying.”

These pouches provided a collective solution: by making all the kids put their phone in the pouch, they were forced to interact with one another which fostered genuine social connections. No kid was forced to be the first mover or face social isolation if they didn’t own a phone.

Of course, I haven’t seen the raw data so I don’t know how the school tested these effects. I would also love to see a large experiment across several schools with a proper control condition to look at the long term educational and social impact of this policy. But given everything I’ve read, this seems like a great step in the right direction and I’d love to see more schools and organizations employ (and study!) similar solutions.

The bottom line is that we should start to think of more issues in terms of collective problems—and solutions. Many solutions fail because we have a tendency to understand these issues through a narrow individualistic lens. But many problems are fundamentally social and require coordinated solutions. As our society grapples with the impact of social media, smart phones, and other technology, we should bear this in mind. Otherwise we are doomed to a form of Sisyphean futility—failing over and over again until the end of eternity.

News and Updates

The Guardian recently recommended our book, The Power of Us! Read “The big idea: do our political opponents really hate us?” here.

Last week, John Miles emailed us to let us know that Jay’s episode on his podcast, The Passion Struck Podcast was the most downloaded episode in the last year! It’s also available on YouTube if you want to watch the video instead of listen to the podcast.

Catch up on the last one…

Last week, we shared summaries for every chapter in The Power of Us! This post is for paid subscribers only, but free subscribers can see the preview.

Why is it that this content constantly goes viral if no one we know actually likes it? Jay wrote an op-ed this week on this paradox of virality with his Postdoc Steve Rathje for the Boston Globe. You can read the full article here, but we will summarize some of the key points in this newsletter.

Lots of research suggests that negativity is more likely to go viral than positive posts. And we have found that this is especially true when people post negative stories about their political opponents: For each individual term referring to the political out-group was used, it increased the odds of a social media post being shared by 67%!

It might seem like people take a bit too much pleasure in the suffering of others—especially their enemies. But a new study we published in the journal Perspectives on P…

Read more

In our experience, it feels like much of the world rewards bullshit and other forms of misinformation. For instance, there are nearly 5 billion people on social media and the primary incentive structure is about capturing attention–regardless of accuracy. This is why it is often called “the attention economy” rather than “the accuracy economy” or “the trust economy”. In the words of Facebook whistleblower, Frances Haugen:

"Facebook is optimizing for content that gets engagement or reaction. Its own research is showing the content that is hateful, that is divisive, that is polarizing—it is easier to inspire people to anger than other emotions."

It turns out that people are sick of this incentive stru…

Read more

"Facebook is optimizing for content that gets engagement or reaction. Its own research is showing the content that is hateful, that is divisive, that is polarizing—it is easier to inspire people to anger than other emotions."

— Frances Haugen, on 60 Minutes.

Read more

“The past few months have brought a hurricane of horror stories about social media sites gone haywire. Facebook and Instagram hijacked by Russian trolls. YouTube struggling to fight an army of conspiracy theorists. Twitter coming to terms with its own fake news problem. Even Pinterest isn’t safe from misuse by bad actors.

They put together this short video on some of our research, along with an interview with Jay:

Read more