The Internet Could Be So Good. Really.

Many people have put forth theories about why, exactly, the internet is bad. The arguments go something like this: Social platforms encourage cruelty, snap reactions, and the spreading of disinformation, and they allow for all of this to take place without accountability, instantaneously and at scale.

Clearly, we must upgrade our communication technology and habits to meet the demands of pluralistic democracies in a networked age. But we need not abandon the social web, or even avoid scalability, to do so. At MIT, where I am a professor and the director of the MIT Center for Constructive Communication (based at the MIT Media Lab), my colleagues and I have thought deeply about how to make the internet a better, more productive place. What I’ve come to learn is that new kinds of social networks can be designed for constructive communication—for listening, dialogue, deliberation, and mediation—and they can actually work.

To understand what we ought to build, you must first consider how social media went sideways. In the early days of Facebook and Twitter, we called them “social networks.” But when you look at how these sites are run now, their primary goal has not been social connection for some time. Once these platforms introduced advertising, their primary purpose shifted to keeping people engaged with content for as long as possible so they could be served as many ads as possible. Now powerful AI algorithms deliver personally tailored content and ads most likely to keep people consuming and clicking, leading to these platforms becoming highly addictive.

The unfortunate consequence of this model is that the best content for keeping eyes glued to screens is often the most emotionally provocative and polarizing content, regardless of quality or accuracy. Quieter voices get drowned out. Most people quickly come to understand that silence is safest and shift into a mode of passive consumption and emotion-driven sharing of content. Peer-to-peer communication is largely reduced to inconsequential chatter, given the risks of cancellation and trolling, which suppress meaningful conversation. Harms are most acute for youth, who feel social pressure to be on social media yet refrain from meaningful self-expression because of possible ostracism and bullying.

Read: We finally have proof that the internet is worse

The threats to democracy in an environment like this are clear. Social media distorts our understanding of others, amplifying false and harmful stereotypes that lead to dehumanization and violence. Moreover, the foundational truth-seeking function of open dialogue and debate is nearly impossible.

One might think that by now we would have learned to naturally self-regulate our use of social media, but the power of these platforms to capture attention and provoke reaction is profound. In 1985, the media critic Neil Postman famously wrote about his fears that a TV-centric culture meant we were “amusing ourselves to death”—the entertaining nature of the television medium had, he argued, subsumed its more serious uses for education and journalism. Postman, who died in 2003, surely would have been horrified by our current state of affairs.

As many reasonable people retreat to the private sphere, or at least away from the social web, the internet is left to the loudest, most polarizing voices. There are concrete steps in regulation that could and should be taken, yet regulation is not enough. We must also create alternatives that meet the needs of democracies in the digital age. These new networks do not need to compete with legacy social media such as Facebook, X (formerly Twitter), and TikTok in terms of entertainment value, but they must provide scalable spaces designed for high-quality and authentic public discourse. And in creating such spaces, we must develop new communication practices that effectively weave these spaces into our civic and democratic affairs.

I’ve been working on these problems with a team of colleagues at MIT and Cortico (a nonprofit I created with Russell Stevens and Eugene Yi in tandem with our MIT lab to translate research into scaled deployment) for many years now. In the lead-up to the 2016 presidential election, our research team analyzed Twitter and mainstream media coverage. We concluded that Twitter could not provide deep insight into why people voted as they did—despite the outsize attention that journalists placed on the platform as a lens through which to see the world. The platform was dominated by the loudest, most polarizing voices, making it difficult to understand the complexities and nuances of the thoughts and opinions of people across the country. In my view, by the end of 2016, Twitter had become corrupted as a social listening channel. (Obviously, it is even worse now.)

To gain better insight, some members of our research team embarked on a listening tour in 2017. The goal was to step outside our bubble in Cambridge and talk with people who had different life experiences and perspectives, particularly in rural and conservative regions of the country. We convened small groups of community leaders and asked open-ended questions about their hopes and concerns for their communities. What we found may seem obvious, but was also profound: Through these in-person dialogues, we witnessed the power of face-to-face conversations in dispelling myths and creating a nuanced understanding of others. We discovered that people were willing to engage and share their perspectives, provided we could find trusted intermediaries to bring us together. We also realized just how challenging it can be to truly escape our own bubble.

This experience led us to wonder how we might design communication spaces online that capture some of the magic we had experienced in person.

Read: Facebook is a doomsday machine

We started with a simple experiment. We designed a “listening box,” a tabletop kiosk that could display prompts similar to the ones we had used in person, record spoken responses, and play back responses from others. We installed the listening box in a public library in New York City, where people could record their stories in their own voice and listen to the stories of others. The premise was that a local newspaper would write articles using these voices as input. We learned—again, perhaps not surprisingly—that people shared far richer stories if one of our team members was present and listening.

As this was happening, I began collaborating with Kathy Cramer, a remarkable political scientist from the University of Wisconsin at Madison, who had ditched the standard tools of public-opinion research—surveys and focus groups—in pursuit of something better. She had spent years driving around rural Wisconsin, inviting herself to coffee klatches in the wild, and got permission to record open-ended conversations about civic matters with small groups of locals. She spent several more years making sense of those conversations. Her book The Politics of Resentment describes what she calls “rural consciousness”—a totally different worldview that was new to her even though she lived just a few hundred miles away. Kathy’s work inspired us to shift from listening to one person at a time to using facilitated small-group conversations as a basis for listening. Her approach to listening with curiosity and systematically making sense of others’ worldviews helped shape our vision for the kinds of properties a social network could foster in people.

And so we began experimenting with what we called a “digital hearth,” which had two functions: first, to record a small-group conversation, and then to play back excerpts from previous conversations at the command of the facilitator. This second feature was important; it let the facilitator bring in perspectives different from those in the group. We also built software to store speech recordings from the hearth, manage data privacy, and provide tools for finding patterns within and across conversations.

Using our new tool, we led small, facilitated group conversations in which people would speak and listen to one another about local issues, with the digital hearth present to record the conversation and bring in voices from prior conversations. Facilitation experts helped us develop the open-ended conversation prompts, which made it easy for anyone to participate. We incorporated well-known dialogue practices to encourage the sharing of personal experiences rather than opinions—a method that has proved to increase mutual respect in ways that sharing facts and opinions does not.

Through all of this work, we came to a big realization about what our conversations had that many online conversations do not: clear roles for key participants. We had conversation organizers, facilitators, curators, prompt designers, and analyzers to find and make sense of patterns of experience that emerged across conversations. Inviting individuals to play these roles in a social dialogue network is an opportunity to participate in civic and democratic processes. Some online spaces already have moderators or admins, but we began to wonder what it would look like if we built a platform that resembled the conversations we had around our digital hearth. And that is the path our teams at MIT and Cortico are now on.

Obviously, such an approach requires participants to have shared values and commitments. Many of today’s trolls and fly-by meme makers aren’t really looking for civic discourse. And even for those who do want to play an active role in building a better internet, designing new kinds of social platforms is not enough by itself. People must form communication practices to weave constructive digital communication spaces into civic affairs. This is a major challenge when so much of government and civic digital and bureaucratic infrastructure lags woefully behind the capabilities of the modern web. (In-person public meetings with open mics tend to have the same problems as social media—the loudest, most extreme voices take over).

In the book Small Is Beautiful, the economist E. F. Schumacher suggests the idea of creating small-scale technology “with a human face” in which the effects of people’s actions are limited in scope so that feedback cycles can help regulate social behavior. In this spirit, we should create small-scale networks in which people with a stake in the game can see that their participation makes a difference. Community-scale social networks can foster learning, listening across divides, mediation between groups in conflict, deliberation, and decision making grounded in listening. Imagine, for example, community-powered networks that connect with city councils, or youth-led networks that connect with school administrators. Community networks could be connected into networks of networks that span state, national, and international scale. Such laterally interconnected systems would be more resilient to top-down political manipulation. Or, to put a finer point on it, we could build networks that reflect the will of the participants, rather than the capriciousness of a single owner or profit-motivated board.

Read: The age of social media is ending

Should we automate the jobs to be done to operate social networks for more efficient scaling? With the stunning advances in generative AI, it is tempting to automate these roles. In echoes of the Lippmann-Dewey debates from a century ago, it is easy to see an argument for, perhaps even the inevitability of, leaving the work of democracy to technocrats and AI. But people should not give up their right to democratic participation so easily—that’s a bad strategy if our goal is to strengthen our public sphere and build healthy pluralistic democracies. Full automation of the functions needed to create and operate social networks in the civic sphere is a path to autocracy in which humans are taken out of the loop, consolidating power in the hands of whoever controls the technology. Instead, we should harness the power of AI to design assistive tools for people who organize, facilitate, analyze, mediate, deliberate, and decide. In other words, we must assiduously defend and empower decentralized human agency.

As we have learned from our own work, scalable digital social networks can be designed to create roles that many of us can play in our local networks and shape how we are governed. Technology creates opportunities to form new habits at scale. In an era when trust has eroded in virtually all institutions, including those foundational to democracy—government, media, higher education—there is a clear opportunity to create a new communication infrastructure for people to understand and shape their communities from the ground up.

The question that many people ask when someone proposes a social platform that is kind and productive, for a change, is some variation of Sure, but do people actually want that? And it’s true that some people—perhaps even a majority—participate in social media because of the chaos, not in spite of it. But competing on the bases of speed, scale, and attention is not necessary to create societal value. LinkedIn has a global-scale impact on people’s professional life, yet this social network is boring compared with Instagram or TikTok. Because LinkedIn’s purpose is not entertainment, being seen as boring or buttoned-up does not limit its value or use. (The fact that LinkedIn is a place where people use their real names to look for jobs helps disincentivize bad behavior.)

More generally, many alternative business models can enable scalable social networks that don’t need addictive consumption to thrive. Twitter’s revenue had long been 90 percent from advertising (until Elon Musk’s recent decimation of the ad business). Historically, the other 10 percent—about $600 million in 2021—came in part from data licenses that allow companies, news organizations, and others to listen to people’s chatter.

We’ve seen glimmers of an alternative path. Mastodon is a nonprofit organization with more than 1 million users who support servers and software development largely via small donations. And telephone networks, the original type of global social network, are paid services that optimize for quality of voice connection, not addiction. (Of course, telephone conversations usually take place among only a few—usually two—participants, and don’t have paths to create larger scale networks.)

The bottom line is this: We need to see and hear the humanity in others for democracy to function. We can and should create social networks designed for public discourse that prioritize inclusion, where underheard voices and perspectives can flourish, and where people take and offer disagreement in good faith. Ad-supported social media is not the only game in town, nor should it be. Scalable, sustainable business models, including not-for-profit models and publicly funded models, are not just possible, but desperately needed.

When you buy a book using a link on this page, we receive a commission. Thank you for supporting The Atlantic.