In the fall of 2016, Keegan Hankes, an analyst at the Southern Poverty Law Center, paid a visit to the neo-Nazi website the Daily Stormer. This was not unusual; part of Hankes’ job at the civil rights organization was to track white supremacists online, which meant reading their sites. But as Hankes loaded the page on his computer at SPLC’s headquarters in Montgomery, Alabama, something caught his eye: a pop-up window that announced “Checking your browser before accessing … Please allow up to 5 seconds.” In fine print, there was the cryptic phrase “DDoS protection by Cloudflare.” Hankes, who had worked at SPLC for three years, had no idea what Cloudflare was. But soon he noticed the pop-up appearing on other hate sites and started to poke around.
There’s a good chance that, like Hankes, you haven’t heard of Cloudflare, but it’s likely you’ve viewed something online that has passed through its system. Cloudflare is part of the backend of the internet. Nearly 10 percent of all requests for web pages go through its servers, which are housed in 118 cities around the world. These servers speed along the delivery of content, making it possible for clients’ web pages to load more quickly than they otherwise would. But Cloudflare’s main role is protection: Its technology acts as an invisible shield against distributed denial of service (DDoS) attacks—hacker campaigns that disable a website by overwhelming it with fake traffic. The company has more than 7 million customers, from individual bloggers who pay nothing for basic security services to Fortune 50 companies that pay up to a million dollars a year for guaranteed 24-hour support.
Hankes wanted to learn something about Cloudflare’s business, and what really interested him was finding out who Cloudflare was protecting. After a few months of research, he felt confident he’d uncovered something important, and on March 7, 2017, he penned a blog post that denounced Cloudflare for “optimizing the content of at least 48 hate websites.” Those sites included Stormfront, the grandfather of white-nationalist online message boards, and the Daily Stormer, at that time one of the most important hate sites on the internet. A virulently anti-Semitic publication, the Daily Stormer was founded in 2013 by a thuggishly enigmatic white supremacist named Andrew Anglin. (“Total Fascism” was the upbeat name of one of his earlier publications.)
Without Cloudflare’s protection, the Daily Stormer and those other sites might well have been taken down by vigilante hackers intent on eliminating Nazi and white-supremacist propaganda online. Hankes and the SPLC weren’t accusing Cloudflare of spouting racist ideology itself, of course. It was more that Cloudflare was acting like the muscle guarding the podium at a Nazi rally.
Matthew Prince, the 43-year-old CEO of Cloudflare, didn’t bother responding to the SPLC’s pointed accusation. In fact, he has only the haziest recollection of hearing about it. He might have seen a mention on Twitter. He’s not sure. Prince is a genial, Ivy League–educated Bay Area resident who once sat in on lectures by a law professor named Barack Obama—the type of person you would expect to have a vivid impression of being denounced by a prominent civil rights organization. But for Prince the criticism was nothing new. At Cloudflare, he was in the business of protecting all kinds of clients, including some whose views vaulted way outside the boundaries of acceptable discourse. He’d already been accused of helping copyright violators, sex workers, ISIS, and a litany of other deplorables. It was hardly a surprise to him that neo-Nazis would be added to the list. Come late summer, however, he would no longer be able to take that breezy attitude. Prince didn’t realize it at the time, but that SPLC blog post was the first indication of the trouble to come. Five months later, Prince would be forced to make a very public decision about the Daily Stormer, one made against his own best judgment and that presented some of the thorniest and most perplexing challenges to free speech since the ACLU defended neo-Nazis who planned to march in Skokie, Illinois, 40 years ago.
How did an internet infrastructure company get locked into a vital free-speech dispute with a bunch of Nazis? That is a story that begins, like so many great tales, in the cubicles of San Francisco and the brothels of Istanbul.
In 2010, when Cloudflare first started, long before it counted customers in the millions, Prince and his cofounders, Michelle Zatlyn and Lee Holloway, installed a bell in their cramped SoMa offices. Whenever someone signed up for Cloudflare’s services, the bell would ring and the 10 or so employees would all drop what they were doing to see who their new customer was.
One day in 2011, the bell rang and Prince went to see who had signed up. “It was the moment where I was like, ‘We need an employee handbook.’ ” The new customer was a Turkish escort service that needed cyber-protection for a promotional website. But it was only the first. Within two weeks, some 150 Turkish escort sites had signed up for Cloudflare’s services. The young outfit had somehow become a go-to service for the Istanbul sex trade.
Curious about this niche-business popularity, a Cloudflare employee contacted the webmaster at one of the escort sites. The webmaster had heard about Cloudflare from a friend who read about it on TechCrunch, and he explained why he sought the company’s protection: Orthodox Muslim hackers had decided to take the law into their own hands and wipe the escort sites off the web. They had largely succeeded, until Cloudflare entered the picture.
To understand why the Turkish webmasters flocked to Cloudflare, you have to understand a bit more about where the company interjects itself into the invisible and near-instantaneous flow of bits that travel between an ordinary user and the servers that deliver the information. When you type a URL into a browser and hit Return, that request first goes out to a domain name server, which translates the human-readable URL (call it www.turkishescort.com) into the numerical IP address of the web server that’s hosting the content. At that point, a packet of bits is dispatched from the domain name server over to the hosting server, and the content you’ve requested is delivered back to your browser.
The trouble is that “you” might not be you at all. Your computer might be infected with malware that has commandeered it to serve in an army of zombie machines—a botnet—that hackers use to execute DDoS attacks. Your seemingly idle laptop might be helping to swamp an innocent website with thousands of requests per second, overloading the target’s servers and making it impossible for legitimate requests to get through. That’s where Cloudflare comes in.
“Putting yourself in front of things that are controversial actually
makes the system smarter. It’s like letting your kids roll around in
Cloudflare protects against these attacks by inserting itself between the browser and host servers that contain the content. From the user’s perspective, the experience is frictionless: You hit the bookmark for, say, a local newspaper and within a split second your screen fills with high school sports scores and reports on the mayoral race. But behind the scenes, your request for information has been filtered through one of Cloudflare’s data centers.
“At that data center,” Prince explains, “we’ll make a series of determinations: Are you a good guy or a bad guy? Are you trying to harm the site? Or are you actually a legitimate customer? If we determine that you’re a bad guy, we stop you there. We act essentially as this force shield that covers and protects our customers.”
During a visit in September to Cloudflare’s headquarters—now in more expansive offices in SoMa—Prince took me to the company’s network operations center, where monitors line the walls, each filled with graphs and brightly colored blocks of text. These represented hundreds of different attacks being attempted in real time across the Cloudflare network. Cloudflare separates the good guys from the bad using pattern recognition. If it sees a familiar nefarious pattern breaking out, it will stop it, like a human immune system attacking a virus. The cyberattackers who went after the Turkish brothels exhibited a distinctive pattern; at Cloudflare, that fingerprint was dubbed the “TE attack,” as in Turkish Escorts.
About a year after the first Turkish escort site became a customer, Prince got a call from someone he calls a panicked “Dutch gentleman.” The caller was responsible for the website of the wildly popular Eurovision song contest. It was two days before the final showdown of the television talent show, and the site had been taken offline by a DDoS attack. When Cloudflare’s security team looked at the data, they saw the family resemblance immediately: It was the TE attack. The Eurovision contest that year was being held in Azerbaijan, a predominantly Muslim country, and the hackers had decided that Eurovision should be knocked offline. Having seen the attack before, Cloudflare was able to get the site up and running in less than 30 minutes, plenty of time before the final rounds. Fast-forward another six months. Prince was summoned to a big financial firm in New York to help analyze a recent attack on its servers. In the conference room, the finance team slid their log files across the table to Prince and his colleagues. As they scanned the logs, smiles of recognition passed across their faces. It was the same maneuver the Turkish Escort attackers had used.
The TE attacks didn’t just help Cloudflare impress Wall Street titans, they also taught the company something about the value of protecting objectionable content. A site that someone, somewhere, deeply despises is the type of site that is likely to be attacked. And when sites are attacked, Cloudflare gets better at what it does; its pattern recognition improves. “Putting yourself in front of things that are controversial actually makes the system smarter,” Prince says. “It’s like letting your kids roll around in the dirt.” This is one of the reasons it makes sense for Cloudflare to offer a free self-service platform: By widening the pool of potential invasive agents, it makes the immune system more responsive. “It’s not obvious that a bunch of escorts that aren’t paying you anything are good customers. It’s not obvious that having people who get attacked all the time—including neo-Nazi sites—that you would by default want them to be on your network. But we’ve always thought the more things we see, the better we’re able to protect everybody else.”
Cloudflare has now logged millions of different kinds of attacks, each, like TE, with its own recognizable signature. This growing database of malice ultimately brought Cloudflare to its central, if largely invisible, position as an internet gatekeeper. The day before I visited Cloudflare’s offices, 22,000 new customers signed up for its services. Needless to say, there is no longer a bell ringing for each one that signs up.
Matthew Prince grew up in Park City, Utah. His father started out as a journalist and later became a drive-time radio host, and Prince has memories of “sitting around the dinner table, talking about the importance of the First Amendment and freedom of speech.” As an undergraduate at Trinity College in Hartford, Connecticut, Prince briefly considered majoring in computer science before deciding on English literature. He also founded a digital-only magazine. He went on to study law at the University of Chicago, where he attended those lectures by Professor Obama, before going to Harvard Business School, where he met Zatlyn.
Prince’s eclectic background gave him the confidence to grapple with Cloudflare’s speech dilemma at all its various layers: As a trained lawyer, he understood the legal implications of corporations policing speech acts; as the founder of a tech company, he was familiar with the technical abilities as well as business imperatives of dealing with customers; and as a liberal-arts-son-of-a-journalist, he thought a lot about what kind of rhetoric is acceptable online. Prince felt strongly that the invisible infrastructure layer of the internet, where Cloudflare operated, should not be the place to limit or adjudicate speech. In Prince’s governing metaphor, it would be like AT&T listening in on your phone conversations and saying, “Hey, we don’t like your political views. We’re kicking you off our network.”
In the years after the launch of Cloudflare, he argued, public-intellectual style, for the importance of preserving free speech online and the neutrality of the infrastructure layer of the internet. It was partly that history that allowed Prince and his colleagues to dismiss the initial investigation by the SPLC. “We’re always having controversies about things,” Zatlyn says.
“Employees were struggling. There are some really awful things
currently on the web, and it’s because of us that they’re up there.”
But Prince’s law school certitude would soon be challenged by another, even uglier, twist involving the Daily Stormer. In the process of standing guard outside its clients’ websites, Cloudflare’s filters sometimes trap legitimate complaints against these sites, the majority of which involve copyright infringement. Someone uploads a catchy song to a website without permission from the artist. Eventually, the songwriter takes notice, but her lawyer can’t present a cease-and-desist notice because the copyright violator is behind the Cloudflare shield. And so, over time, Cloudflare had developed a policy of passing along any complaint to its customers and letting them deal with the requests.
But a system designed to address copyright infringement proved to be less adept at dealing with Nazis. Ordinary people disturbed by the hate speech on the Daily Stormer would seek to register their complaints about the site to Cloudflare, the host. But instead of directly addressing the complaint, Cloudflare, following its usual policy, would pass those complaints, with the senders’ contact information, along to the Daily Stormer.
In early May, another story came out—one that Cloudflare could not ignore. The article, by ProPublica, revealed that people who had complained to Cloudflare about the Daily Stormer were getting harassing and threatening calls and emails, including one that told the recipient to “fuck off and die.” The ProPublica piece quoted a blog post under Anglin’s name: “We need to make it clear to all of these people that there are consequences for messing with us. We are not a bunch of babies to be kicked around. We will take revenge. And we will do it now.” It looked as if Cloudflare had ratted out decent people to an army of fascist trolls.
Recognizing that it had a legitimate problem on its hands that couldn’t be erased by invoking free speech, Cloudflare quickly altered its abuse policy, giving users the option of not forwarding their identity and contact information. ProPublica also reported Anglin saying that the hate site paid $200 a month for its Cloudflare protection, a point Cloudflare would not comment on. Despite Cloudflare’s pride in protecting any site, no matter how heinous, Prince says he was caught off guard by the Daily Stormer’s attacks on the people who complained. “What we didn’t anticipate,” Prince told me, ruefully, “was that there are just truly awful human beings in the world.”
A few months later, on Friday, August 11, a group of torch-wielding white supremacists marched in the streets of Charlottesville, Virginia; the next day a counterprotester named Heather Heyer was run over in what appeared to be an act of political violence. That afternoon, without mentioning Heyer’s death, Donald Trump blamed the violence in Charlottesville on “many sides,” and the whole country was suddenly engulfed by the question of what we were willing to do to stand up to Nazis. The Daily Stormer posted a repulsive piece under Anglin’s byline with the headline “woman killed in road rage incident was a fat, childless 32-year-old slut.” It only got worse from there.
After reading the post, an anti-fascist vigilante hacker known as the Jester tweeted, “Nice site, Andrew. Be a shame if something ‘happened’ to it.” But the threat was empty as long as Cloudflare continued to offer its protection. “That night I was at home and I get this DM on Twitter from the Jester,” Prince says. “And he’s saying, ‘Hey, these guys are jerks. I want to DDoS them off the internet. Will you get out of the way?’ ” Prince says he responded with a link to a speech he’d given at an internet security conference defending free speech principles. (The Jester did not respond to a request for comment.) Meanwhile, the wrath against Cloudflare was rising. “All of a sudden, a ton of people were yelling at us on Twitter,” Prince recalls. The online service GoDaddy, which maintained the Daily Stormer’s domain, announced it was canceling this arrangement. The Daily Stormer tried to move its domain registration to Google Domains but was denied. Cloudflare seemed to be the last major player willing to do business with the neo-Nazi site, appearing once again to go out of its way to protect hate speech.
On Monday afternoon, Prince and his management team gathered to address the growing controversy. The backlash weighed heavily on the minds of the rank and file at Cloudflare. “There was definitely water-cooler talk,” recalls Janet Van Huysse, who oversees employees and human resources. “We were all over the news. People were struggling. There were a lot of people who were like, ‘I came to this company because I wanted to help build a better internet, and we believe fiercely in a free and open internet. But there are some really awful things currently on the web, and it’s because of us that they’re up there.’ ” A range of feelings would emerge during a town-hall–style meeting for employees conducted later in the week. One attendee said to Prince, “I don’t have a good answer for what we should say going forward as a proud Cloudflare employee. What should I say?” Another asked why they would consider kicking neo-Nazi sites off the platform, but not alleged ISIS sites.
On Tuesday night, Prince was hosting a dinner for Cloudflare interns at his home in San Francisco. At one point during the event, Cloudflare’s general counsel, Doug Kramer, pulled Prince aside and said, “It seems like this keeps ratcheting up.” Checking his phone surreptitiously during the meal, Prince noticed that fellow technologist Paul Berry, founder and CEO of a social media service called RebelMouse, had taken to Twitter to denounce Cloudflare for hosting “Nazi hate content that even @GoDaddy took down.”
After the interns left his apartment, Prince and Tatiana Lingos-Webb, his fiancée, cleaned up and did the dishes. (The two have since married.) Stung by Berry’s tweet, Prince started bemoaning the ease with which people seemed willing to abandon the basic ideals of free speech. “Maybe there is something different about Nazi content,” Lingos-Webb ventured. “And I looked at her and said, ‘You too?’ ” Prince recalls.
“I went to bed angry,” Prince says, “and woke up in the morning still angry.” Checking Twitter, he discovered that someone on the Daily Stormer site—in full frog-and-scorpion mode—had decided to antagonize the one service left supporting it. An anonymous comment about the site’s technical challenges noted the moves by GoDaddy and Google to oust the Daily Stormer: “They succeeded in everything except Cloudflare, whom I hear are secretly our /ourpeople/ at the upper echelons.” Overnight, Prince and his colleagues had been welcomed into the ranks of practicing white supremacists.
Prince called Berry to walk through his reasoning for continuing to protect the Daily Stormer. Prince had known the RebelMouse CEO for years through the technology conference circuit and respected his opinions. It was a tense call. Berry told him he understood the predicament. “But when you work that fucking hard to build something that’s that successful, you get to choose who uses it,” Berry recalls telling Prince. “And you get to set a code of conduct that leaves it clear for people—a code of conduct that says we will not support white supremacy, racism, hate.”
“I’d try to make my argument,” Prince says now of the conversation, “and Paul would say, ‘It doesn’t matter, Nazis.’ And I’d say, well, it kind of does, because if the phone company is listening in on my phone calls and then decides that they don’t like what I’m talking about and starts pulling the plug, that seems creepy to me.”
Hanging up with Berry, Prince got in the shower. He had barely slept; his friends seemed to be taking the wrong side of the free-speech argument. But he could see where the controversy was heading: If he kept protecting the Daily Stormer, the inevitable next step was a customer boycott of Cloudflare, with real business consequences. But if he kicked them off, “I thought through the parade of horribles that would follow,” he recalls. Suddenly every controversial site on the Cloudflare network would be subject for review, and Cloudflare would have helped establish a precedent for deep infrastructure services regulating speech. As he stood in the shower, all these thoughts swirled around his head. “It was literally one of those lean-your-head-against-the-wall moments—like, what the hell are we going to do?”
“But then I had a thought: Maybe we can kick them off, and then talk about why that’s so dangerous. Maybe that can change the conversation.” Prince would betray his principles and then make the betrayal into an argument for why those principles matter.
“Matthew called me at around 10 that morning, and said, ‘We’re kicking them off,’ ” Zatlyn recalls. She’d gone to bed feeling Cloudflare was not the right place to censor and assuming they would stick to the company policy. “I was speechless, a little stunned. ‘I’m surprised to hear you say this. I was not expecting that. But OK.’ ”
By late morning, the company’s trust and safety team had completed the procedures to remove the Daily Stormer from the Cloudflare network. And Prince drafted a blog post. It began in a just-the-facts mode: “Earlier today, Cloudflare terminated the account of the Daily Stormer.”
“Our terms of service reserve the right for us to terminate users of our network at our sole discretion.” The tipping point for Prince was the suggestion on the Daily Stormer site that top managers at Cloudflare “were secretly supporters of their ideology.”
But then Prince took a rhetorical twist: “Now, having made that decision, let me explain why it’s so dangerous.”
Prince spoke about the peril posed by DDoS attacks. We might all agree, Prince argued, that content like the Daily Stormer shouldn’t be online, but the mechanism for silencing those voices should not be vigilante hackers.
His bigger argument was about the danger of private companies like Cloudflare (not to mention Google or Amazon Web Services) determining what constituted acceptable speech. “Without a clear framework as a guide for content regulation,” Prince explained, “a small number of companies will largely determine what can and cannot be online.” Perhaps his most striking point came in a separate memo he wrote to his staff. “Literally, I woke up in a bad mood and decided someone shouldn’t be allowed on the internet. No one should have that power.”
Prince’s dilemma over the Daily Stormer has been present in net culture from the early days of online communities. But where debates about what forms of speech should be forbidden often seemed academic and remote, today they are at the center of social discourse. White-supremacist movements that once were deemed beyond the pale are more vocal, their ideas spreading openly into the mainstream, with political leaders not always willing to condemn them. Hankes, of the SPLC, says that even fringe hate sites “can have a tremendous impact” because of social media’s ability to amplify extreme ideas. “Our position has been that pretty much everyone south of the internet service providers”—in other words, anyone hosting or protecting online content—“has the responsibility to take a stance on these issues,” Hankes says, “or be ready to answer for the consequences of people who are taking advantage of their services.”
The original free-speech ethos that shaped the internet has also grown shakier: Back then, strong First Amendment values were one of the few areas of agreement among the libertarians and progressives who shaped the early culture. Today, that alliance is less stable. Aggressive anti-hate-speech movements on college campuses have aroused ire among libertarians, and among progressives there is a growing sense that Big Tech has become a breeding ground for bile. Every other week, it seems, there’s another flare-up over Twitter’s terms of service and the rampant harassment and abuse that plagues that platform.
“Honestly, I am so sad,” Berry told me. “I grew up in the Valley; I’ve been writing code since I was 10, and I believed in technology.” But now, Berry says, he sees money to be made as a platform company triumphing over civic decency. “Right now we have a tension between financial success and actually being human.”
“Literally, I woke up in a bad mood and decided someone shouldn’t be
allowed on the Internet. No one should have that power.”
The immense size of those gatekeepers—like Google, Facebook, Twitter and, in its own way, Cloudflare—has also challenged the older vision of cyberspace as a realm of unchecked speech. There have been dark wells of hate online since the Usenet era, but back then, misanthropy was distributed across thousands of different platforms. Even if you felt some speech was objectionable enough to silence, it was a practical impossibility to get rid of it all. No single entity could silence an idea. But in a world where Facebook and Google count their audiences in the billions, a decision by one of those big players could, essentially, quiet an unpopular voice. In December, in fact, Twitter started enforcing new rules to suspend accounts of people who use multiple slurs or racist or sexist tropes in their profile information.
Prince is aware of that power, but he also carefully titrates the various elements in the internet concoction. He argues that there is a fundamental difference between sites like Facebook or Twitter, which provide content, and deep infrastructure like hosting or security services. For Prince, the relative invisibility of Cloudflare to ordinary consumers makes it the wrong place to address speech. “I think that gives us a framework to say infrastructure isn’t the right place to be regulating content,” he says. “Facebook and YouTube still may be—and it’s an easier question for them, because they’re advertising-supported companies. If you’re Procter & Gamble, you don’t want your ad next to terrorist content, and so the business model and the policy line up.”
If this sounds like passing the buck, Prince’s argument does get philosophical support from civil liberties groups. The Electronic Frontier Foundation, which has taken a stand that what it calls “intermediaries”—services like Cloudflare and GoDaddy that do not generate the content themselves—should not be adjudicating what speech is acceptable. The EFF has a strong presumption that most speech, even vile speech, should be allowed, but when illegal activity, like inciting violence or defamation, occurs, the proper channel to deal with it is the legal system. “It seems to me that the last thing we should be doing is having intermediaries deputizing themselves to make decisions about what’s OK,” says Corynne McSherry, legal director of the EFF. “What law enforcement will tell you is that it’s better for them to be able to keep track of potentially dangerous groups if they’re not pushed down into the dark web.” She adds: “I want my Nazis where I can see them.”
In the months following the Charlottesville weekend, the Daily Stormer bounced around a series of websites, briefly appearing on Russian and then Albanian domains with new URLs. Prince himself has grown more certain that his company should not be in the speech-regulation business. Since ejecting the Daily Stormer, Cloudflare has received more than 7,000 complaints about sites in its network. “The weirdest was a totally nonpartisan cooking blog,” Prince says. “We’ve considered trying to make some of the recipes, to see if they’re just really terrible.” Though Prince’s blog post vowed to establish a framework for managing objectionable sites on its network, little has changed. “We’re still having the debate, but I think the likely outcome is that as an infrastructure company, we’re going to err on the side of being neutral and not do what we did to the Daily Stormer again,” Prince says now.
Cloudflare can legitimately embrace free speech tradition in the defense of its policy. But it is also protecting its business interests. Network software and algorithms have allowed Big Tech to organize and distribute (and in Cloudflare’s case, protect) staggering amounts of information. Looking for patterns in DDoS attacks, detecting the signatures of the Turkish Escort attackers—these are the kinds of problems that can be solved at scale with code. But evaluating 7,000 websites for, say, potential incitements to violence is not something that lends itself to a final determination by software alone; it invariably requires human judgment. Facebook and Google have confronted this issue in the past year with the infiltration of Russian ads and fake news into their feeds and screens. But humans are expensive. Only after public outcry did Facebook and YouTube pledge to hire thousands of human moderators to deal with suspicious ads and with videos that are inappropriate for children. Prince may be right that a service like Cloudflare’s is the wrong place to make those assessments, but it’s also convenient: Opting out of that obligation makes his business much easier to run.
One still-unresolved debate at Cloudflare is about how the company should memorialize the decision to eject the Daily Stormer. “We do a transparency report twice a year, and one of the things that we have is a list of ‘things we have never done.’ ” It’s a short list, and one of its key statements is, “We have never terminated a customer or taken down content due to political pressure.” That is no longer true. “So we’re having this conversation now internally,” Prince says, “about whether we have to remove that.”
As of December, Prince says, the company was leaning toward keeping the statement but adding an asterisk that links to a full account of the Daily Stormer affair. “So when the next controversy comes along, we’ll be able to point to that and say, ‘This was the one time we did that, and here are the dangers it creates.’ ” The Daily Stormer, however, has not been invited back.
- Tech, Turmoil, and the New Censorship: Zeynep Tufekci explores how technology is upending everything we thought we knew about free speech.
- Everything You Say Can and Will Be Used Against You: Doug Bock Clark profiles Antifa’s secret weapon against far-right extremists.
- Please, Silence Your Speech: Alice Gregory visits a startup that wants to neutralize your smartphone—and un-change the world.
- The Best Hope for Civil Discourse on the Internet … Is on Reddit: Virginia Heffernan submits to Change My View.
- 6 Tales of Censorship: What it’s like to be suspended by Facebook, blocked by Trump, and more, in the subjects’ own words.
This article appears in the February issue. Subscribe now.