Is social media strengthening or weakening public discourse?

The Town Square Is on Fire—And We Handed Out the Matches

Social media promised us a more connected world. What we got instead was something far more complicated — and far more human.

I remember the exact moment I realized something had fundamentally shifted in how we talk to each other. It was 2019. A friend of mine—someone I'd known for fifteen years, someone who'd helped me move apartments twice—posted something political. Within forty-eight hours, the comments had devolved. Not between strangers. Between people who'd shared meals together. I watched it happen in real time, refreshing the page like I was waiting for an accident to stop being an accident.

That's the question I've been sitting with ever since: Is social media actually making us better at public discourse, or is it engineering our worst conversational instincts at scale?

The answer, I'll tell you now, is neither clean nor comfortable. And that's exactly why it's worth taking seriously.

What we thought we were building

When platforms like Twitter (now X), Facebook, and YouTube expanded in the late 2000s and early 2010s, the dominant narrative was democratization. Finally, anyone with a phone and a perspective could reach the world. The Arab Spring seemed to vindicate this. Grassroots movements organized, marginalized voices found audiences, journalists bypassed editorial gatekeepers. There was something genuinely thrilling about it — a distributed public square where ideas competed on merit, not on access to printing presses or broadcast towers.

"Public discourse" has always been a contested space. What's new isn't the conflict. It's the speed, the scale, and the business model underneath it all.

That narrative wasn't entirely wrong. It just wasn't the whole story. And the part we missed was sitting in the algorithm.

The architecture of outrage

Here's what research has steadily revealed over the past decade: engagement-maximizing algorithms don't optimize for truth, nuance, or productive disagreement. They optimize for emotional activation. And the emotion that activates most reliably, most profitably, is moral outrage.

A 2021 study from the NYU Centre for Social Media and Politics found that on Twitter, tweets containing "moral-emotional" language—words like "destroy," "shame," "corrupt"—received significantly more retweets than neutral-toned versions of the same information. The platform wasn't just a mirror. It was a megaphone with a preference for amplifying the loudest, most agitated voices in the room.

This isn't a conspiracy. It's an incentive structure. The platforms discovered, through billions of data points, that anger keeps eyes on screens. And eyes on screens translate to advertising revenue. So they built products that, not incidentally but functionally, reward the performance of outrage over the practice of reason.

What the data actually says

A 2023 Pew Research study found 64% of Americans believe social media has a mostly negative effect on how things are going in the country—up from 55% in 2018. Yet the same survey found 70% still use at least one platform daily. That gap between what we believe and what we do is not irrationality. It's a product feature.

But let's be honest about what was broken before

Here's where I want to complicate the nostalgia. The idea that public discourse was ever healthy—before social media—is a myth worth puncturing.

Yellow journalism predates the internet by a century. Talk radio in the 1990s was incendiary. Cable news discovered the ratings value of conflict long before Twitter existed. The town square of the 18th century was exclusionary by design—women, enslaved people, and the poor had no seat at it. Gatekeeping wasn't just editorial wisdom; it was also power protecting itself.

Social media did something genuinely new: it gave voice to people who had been systematically excluded from public conversation. Disability activists, queer communities in rural areas, survivors of systemic injustice—many found platforms, found each other, built movements that changed policy. That is real. That is not nothing. Any honest accounting of social media's effect on public discourse has to hold that alongside the polarization data.

The new architecture of public thinking

What I've come to believe—after years of watching this, reading the research, and frankly, experiencing it firsthand in my own feeds—is that social media has fundamentally changed not just what we say in public, but how we think before we say it.

There's a concept researchers call "context collapse": on a platform like Instagram or X, you're speaking simultaneously to your grandmother, your college roommate, your professional network, and strangers who found you through a hashtag. Ordinary people, suddenly, face a version of the same challenge that presidents and CEOs used to navigate with communications teams. Most of us aren't equipped for it. So we self-censor, or we perform, or we retreat into communities where everyone already agrees with us — what Eli Pariser famously called the "filter bubble."

The result is a public square that looks more crowded and diverse than ever, but often functions as a series of parallel echo chambers, each convinced of their own righteousness, rarely actually encountering the friction that genuine disagreement requires to produce insight.

Where I think the real leverage is

I've thought about this for a long time, and I want to offer something more useful than a diagnosis. Because the problem isn't social media per se — it's the specific design choices embedded in current platforms. And design choices can be changed.

First: friction as a feature, not a bug. Twitter's own internal research found that adding a prompt asking users "Would you like to read this article before sharing it?" reduced retweet rates of misleading content by 33%. That's a tiny design intervention with measurable impact. Slowing down the share reflex is not censorship; it's what thoughtful discourse has always required. We just forgot to build it in.

Second: rewarding depth over reaction. Platforms that prioritize long-form engagement, comment quality, and sustained conversation threads produce demonstrably different discourse than those built around the hot take. Reddit's best communities (and Reddit's worst, too) illustrate this — when norms and moderation are thoughtful, remarkable conversations happen. The infrastructure matters.

Third: media literacy as civic infrastructure. Finland has integrated media literacy education into its national curriculum for over a decade. Finnish students learn to identify manipulation techniques, evaluate sources, and recognize emotional appeals masquerading as evidence. The result: Finland consistently ranks among the most resilient countries to misinformation in European studies. This is not a technology problem with a technology solution. It is a civic education problem requiring civic investment.

And then there's AI — a tool, not a replacement for your thinking

I'd be remiss not to address the newest variable in this equation: artificial intelligence. In the last three years, AI-generated content has flooded social media at a scale we're only beginning to measure. Synthetic personas, AI-written op-eds, algorithmically generated news summaries — the volume of machine-produced text in our feeds is accelerating.

The response I hear most often is either panic or dismissal. Neither is useful. The more honest and productive frame is this: AI is a tool, not a replacement for human judgment. And the quality of AI's contribution to public discourse depends entirely on the quality of human intention behind it.

A researcher using AI to synthesize hundreds of studies and communicate findings accessibly is not the same as a political actor using AI to manufacture credibility at scale. The technology is neutral; the design and deployment are not. What we need are clearer norms, better labeling standards, and—crucially—humans who understand enough about how these tools work to use them thoughtfully and spot when others are using them manipulatively.

The most AI-literate person in the room isn't the one who's most impressed by what AI can do. It's the one who understands what it can't: verify its own claims, hold genuine convictions, take moral responsibility, or replace the irreplaceable texture of human experience in argument. Learn the tool. Know its limits. Don't outsource your thinking to it.

Where I land

Social media is neither saving public discourse nor simply destroying it. It is amplifying what was already in us—our tribalism and our solidarity, our capacity for outrage and our capacity for empathy, our willingness to caricature the other side, and, sometimes, our willingness to genuinely listen.

The platforms are not neutral pipes. They are environments with architectures that shape behavior. Changing those architectures is possible — but it requires pressure from users, regulators, and researchers who refuse to accept "engagement" as the only metric that matters.

And it requires each of us to do something that social media actively works against: slow down. Read the whole thing. Ask yourself why you feel what you feel before you share it. Seek out one voice you disagree with and try, genuinely try, to understand how someone reasonable could hold that view.

That's not a platform feature. That's a practice. And it's the only one I know of that has ever actually worked.