The subject of this piece was an interview subject from a previous story that I reported for Los Angeles magazine. A recent texting session between us started off with what I assumed was the setup for a joke: “Well, I got kicked off Twitter again.” But there was no punchline. The suspension was real. Interviews and research soon followed. This person agreed to be interviewed under the condition that he remain anonymous.
While the Twitter war between Donald Trump and the company’s CEO Jack Dorsey rages on, another First Amendment battle is being fought in the social media trenches. Coverage is scant. Rachel Maddow has never aired a segment on it. The New York Times op-ed contributors don’t cover it either. The same goes for The Washington Post, BBC News, NPR, and every other news app that pings your bubble. Despite its obscurity, the conflict is real. On one side is Donald Trump, who routinely violates Twitter’s rules of engagement without fear of retribution. On the other side are those who post dissenting political views, violate the same rules of engagement, but are booted from the social media platform quicker than you can say “MAGA!” As the tweets fly, the casualties pile up.
One of those casualties was a comedy writer in Los Angeles. Like most soldiers, he never saw it coming.
The way he remembers it, February 8, 2020 was like any other day. It started off with strong drip coffee and a toasted bagel followed by a quick scan of the Los Angeles Times (hard copy) and the usual social media procrastination. He was scrolling through the president’s Twitter feed when this post caught his attention.
The writer, a cable TV veteran who analyzes the president’s Twitter account as closely as his agent’s emails, read between the lines and grimaced. In a single run-on tweet Trump had managed to soften the ground beneath Joe Manchin’s vulnerable Senate seat and mock Mitt Romney’s faith, suggesting that the former Latter-Day Saints minister, who wears traditional Mormon underwear beneath his starched Brooks Brothers shirts, was less than “highly religious” and not too bright. But the thing that really wound up the writer was that uppercase “Impeachment Hoax.” The two words popped on his iPhone screen like a devil emoji.
Alone in his Westside bungalow, the writer muttered an obscenity. A registered Democrat who voted for Hillary in 2016, he was still incredulous (and incensed) that Trump had so easily dismissed the Mueller Report’s findings and emerged from the impeachment process relatively unscathed. Scrolling past this POTUS tweet without responding was unthinkable.
Venting his outrage through a flurry of thumb strokes, the venom flowed. Unlike the sitcom scripts he crafts, there were no edits or rewrites, no notes to consider or meetings to attend. The words were spontaneous, almost reflexive, raw copy banged out straight from the heart: “I think that maybe the only reason he voted against you is because you’re a giant fucking piece of shit. Beyond that, Mr. President, I have no idea.
That blunt character assessment was quickly followed by another tweet that sounded like a cross between medieval curse and a talking point from the Antifa playbook: “You should just shut the fuck up. I’ve never wished this upon anyone before and it’s probably bad karma, but here’s hoping you have a stroke that affects your ability to speak & your ability to tweet. You’re Mussolini. You’re an authoritarian with sycophantic followers.”
Watching his counter punches land on the President’s Twitter feed never failed to unleash a dopamine hit, but this time the buzz was fleeting. Twenty-nine minutes later, the writer’s account was suspended when a hate speech detection algorithm flagged his “stroke” post for breaking one of the so-called Twitter Rules. This online code of conduct covers various bad behavior categories, ranging from “Spam” and “Non-Consensual Nudity” to “Graphic Violence” and “Terrorism.” Think of it as a moral firewall, the thing that discourages users from exploiting the popular social network to do things like spread malice, steal intellectual property, and sell illegal goods and services.
Here’s another Twitter Rule that bears mentioning: “Abuse/Harassment.” It turns out that wishing traumatic brain injury on anyone, especially the President of the United States, is a serious breach of company policy.
Within seconds of receiving this notification, he filed an appeal. But the human moderator who reviewed his case sided with the A.I. and upheld the suspension. That decision was based on the offender’s Twitter rap sheet, which includes multiple Abuse/Harassment citations for tweets aimed at Trump and several of his media surrogates. Last winter, for instance, the writer fired a broadside of profane tweets targeting two Fox News personalities. First up was Geraldo Rivera, who was labeled a “deuche [sic] bag” and an “ass” (January 26, 2020). Then it was Lou Dobbs’ turn. He started by calling the arch-conservative “dumb and evil,” and closed with gangster swagger: “I hope I run into [Dobbs] one of these days” (February 7, 2020). He’s also had Twitter skirmishes with Trump’s base, specifically the Breitbart and QAnon crowd fixated on the Deep State and Wuhan lab conspiracies.
But it’s the “freaky orange bastard” that the writer saves his best material for, like this rimshot retweeted by some of his followers: “Hey, Mr. President. I hear there’s a big rally in your honor near the book depository. Come on down!”
Depending on the moderator, a menacing tweet like that could qualify as either a “Violence” or “Hateful Conduct” transgression under the Twitter Rules. Oddly, the post evaded detection. The phrase “book depository,” and all the grainy slo-mo carnage that it conjures within the collective psyche, didn’t trigger any of the code stuffed inside Twitter’s digital minesweepers.
The writer was also fortunate that his tweet didn’t surface on a screen at the NSA. Secret Service agents have been known to knock on doors for far less (see: U.S. Code Title 18, section 871). In April, a Texas man was sentenced to 18 months in prison for threatening the President’s life. According to one court document, the defendant admitted to issuing this statement on YouTube: “I am waiting for Trump to visit Dallas before I attempt to assassinate him.” The writer didn’t realize it, but his shtick was wearing thin.
Encouraged by the feedback, he recycled the “book depository” tweet and worked it into his standup act, a common sideline for sitcom writers. The bit either killed or bombed. Dead President jokes are tricky that way. But the “stroke” post was different. Drop all the F-bombs and Dealey Plaza references you like, but don’t ever joke about Donald Trump (84.3 million followers) suffering an illness that renders him tweetless.
It may have helped if this was the writer’s first run-in with the Twitter police, but it wasn’t. Over the years, he has served several suspensions of varying lengths. Just three weeks prior to this incident, his platform privileges were revoked for spitting out a filthy riposte to one of Trump’s many gaslight tweets about the “Ukraine Hoax.” He caught a break that time. During the appeal process, the moderator proposed a deal: remove the offensive post, and the seven-day suspension would be rescinded. He considered the offer, but only briefly. He just couldn’t bring himself to tap the “Delete” button. Suspensions are easy; comedy is hard.
This time, though, there was no deal on the table. All the anti-Trump rhetoric had finally caught up with him. A moderator studied the evidence, uncovered a disturbing pattern of Twitter Rules infractions and launched the nuclear option: permanent suspension, better known among Alt-Rights and professional provocateurs as the “perma-ban.” Twitter’s Enforcement Options page is blunt and explicit: “Permanently suspending an account will remove it from global view, and the violator will not be allowed to create new accounts.”
Officially banished from the platform, the writer joined an infamous club that includes Martin “Pharma Bro” Shkreli (sexual harassment), InfoWars’ Alex Jones (abusive behavior), Roger Stone (threatening journalists), Milo Yiannopoulos (staging a racist campaign against Saturday Night Live regular and Ghostbusters star Leslie Jones), the American Nazi Party (promoting racial violence), and ISIL member Jalbib Al-Jazrawi (death threat).
Since the suspension, which is binding and final, the writer has had time to reflect on his social media sins and ponder a life without Twitter. As is often the case when punishment is doled out, there is remorse. “I wish I hadn’t used that language, I was a jackass,” he says. “Those tweets were extremely offensive and vitriolic.” But the mea culpa, tinged with what seems like genuine contrition, ends abruptly. The switch from regret to anger is stunning. This is the effect Trump has on people. “It isn’t fair! I didn’t say, ‘I wish the President was dead!’” His voice rises in volume as he pleads his case, like a locomotive gaining speed as it swallows a steep grade. “What I posted wasn’t that bad. Trump’s tweets are far worse, it’s not even close!”
When he complains to friends, they say the moderator made the right call. He’s stunned and disappointed by this reaction. The way he sees it, justice was swift but far from righteous: “I’m annoyed because of the inequity of the ruling.” What would be fair? “Maybe Twitter should block my feed from Trump’s account, but banning me? That’s absurd.” He isn’t a lawyer, but he’s written a few courtroom scenes and knows in his heart what’s right. “If you pull my Twitter account,” he argues, “you have to pull Trump’s account too.”
Jillian York, director of international freedom of expression at the Electronic Frontier Foundation (EFF), a non-profit digital rights group based in San Francisco, embraces this tit-for-tat social justice theory: “If you’re going to have these [Twitter] policies, they should apply to all users equally, whether it’s a ban or rules that are more inclined toward free expression.” As for attacking a public figure via tweet, she’s seen it all before. In the world of “digital free speech,” it’s as common as Botox in Beverly Hills. “People go down all the time for that kind of thing,” says York. “I know someone who was banned from Twitter for saying ‘I hope you die!’ to Thomas Friedman. Which is funny because Thomas Friedman doesn’t really use Twitter.”
The other side of this legal issue, holding public figures accountable for their tweets and handing out suspensions if necessary, is gaining favor. Last year, The New York Times ran an op-ed titled “Trump Is Too Dangerous for Twitter.” In the piece, contributing writer Kara Swisher highlighted a Trump retweet that warned of civil war if the Dems were successful in removing him from office. California Senator Kamala Harris has also lobbied to blow up the @realDonaldTrump. During the fourth Democratic presidential debate, she attacked Trump for trying to sway the House impeachment hearings with a barrage of angry tweets: “Here we have Donald Trump…using that platform as the President of the United States to openly intimidate witnesses, to threaten witnesses, to obstruct justice, and he and his account should be taken down.”
More recently, during the May 20 airing of Morning Joe, Mika Brzezinski called out Twitter for allowing Trump to spread a debunked conspiracy theory that suggests her MSNBC co-anchor (and husband), Joe Scarborough, murdered a female office intern in 2001 when he was serving as a Republican congressman. “You will be hearing from me on this, because this is B.S.,” Brzezinski railed during the broadcast. Then she tried to rule-shame Twitter CEO Jack Dorsey into banning Trump: “@jack At what point is @Twitter a part of this? TAKE DOWN TRUMP’S ACCOUNT–the world world [sic] be safer. Retweet if you agree.”
All his life, the writer has fought to have his name appear below the title where an audience can see it. But now, he insists on being anonymous. He claims that this sudden aversion to the spotlight isn’t just about staying employable. It’s about staying alive. “Even though the people who hire me tend to be liberal, I could be blackballed for being the foul-mouthed guy with anger management issues,” he explains. “It’s also not unreasonable to think that I run the risk of being shot by a crazy person wearing a MAGA hat if I go public.”
Anonymous or not, the writer maintains that his constitutional rights have been ripped asunder. “This is about freedom of speech,” he says like a Sorkin character in a West Wing reboot. “Twitter’s position is that Trump can say whatever he wants because he’s the President, but that regular citizens like me have to watch what they say.” In addition to censorship, he accuses Twitter of blatant political bias. On this point, he is unequivocal: “Twitter is either run by conservatives or they’re afraid of retribution from the Trump administration.” In other words, sinister forces lurk beneath the fluttering wings of that cheerful blue logo. Trump and his base, of course, insist the opposite is true. They claim that their conservative voices are being stifled by Twitter’s inherent liberal bias.
The more the writer talks, the more agitated he gets: “Twitter makes totalitarian decisions without giving you a chance to speak and defend yourself, and your appeal can only be as long as a tweet.” That’s the rule, but you try mounting a decent Abuse/Harassment defense in 280 characters or less. Not so easy. It’s like climbing Everest without supplemental oxygen. Some have pulled it off, but many more have failed trying.
“I don’t like [Trump,]” he says in the kind of tone that makes you think maybe this guy should be on a Secret Service watch list. He adds that this dislike is well deserved: “[Trump’s] a horrible despot, he’s a demagogue, he’s a fascist. And, as an American, I have the right to tweet things that might upset his base.”
The idea that tweets are a protected form of free speech under the United States Constitution isn’t a widely held legal opinion. Twitter is a private company, and it can decide who and what it censors. People also agree to certain restrictions when they click on that little box at the bottom of the User Agreement page. The “Twitter Rules and Policies” form the bedrock of this social media contract. Or, as the Santa Clara University law professor Eric Goldman succinctly put it, “Nobody has a right to be on Twitter.”
Armchair legal scholars may scoff at that position, but Goldman is here to torpedo their flimsy arguments. “There’s a sense of entitlement here that dominates the public discourse,” he says, “but it’s not justified at all.” In the world of Internet law, this is as open-and-shut as it gets: “We just assume that these social media companies owe it to us to publish our every thought. They don’t owe us anything.” So the writer has no case? “I’ve legaled at least a dozen lawsuits involving people being suspended or kicked off platforms—the plaintiff always loses,” he says.
Digital law expert and First Amendment lawyer James Rosenfeld concurs: “Since Twitter is a private actor, not a state actor, they have leeway to do what they want.” That’s a very important distinction. State actors, like police officers and court officials, must honor a citizen’s constitutional rights while private actors, like corporations and media organizations, get a free pass. That’s why cops have to read the Miranda Warning before making an arrest. It’s also why Twitter can suspend a comedy writer for wishing a stroke on Trump.
“There’s not even a legal obligation to administer the rules evenly,” says Rosenfeld. “And there are legitimate public interest loopholes that allow Twitter to not shut down Trump for comments that other users would be banned for. I’m sympathetic to the writer, but there’s not a lot he can do about this suspension.”
That’s always been Jack Dorsey’s excuse for why he doesn’t trash Trump’s Twitter account: public interest. “We hold all accounts to the same Rules, and consider a number of factors when assessing whether Tweets violate our Rules,” reads a company statement from 2017. “Among the considerations is ‘newsworthiness’ and whether a Tweet is of public interest.” That was released right after the president committed a “Violent Threat” by reminding the world that his “Nuclear Button” is “much bigger & more powerful” than Kim Jong Un’s nuclear button (“and my Button works!”).
Since then, Trump has broken enough Twitter Rules to suspend a small village. Among other things, he has tweeted that Alexandria Ocasio-Cortez and The Squad (aka Ilhan Omar of Minnesota, Ayanna Pressley of Massachusetts, and Rashida Tlaib of Michigan) should “go back” to where they came from, threatened to destroy cultural landmarks in Iran if the country retaliated for the American killing of its top general (“HIT VERY FAST AND VERY HARD”), promoted two unproven COVID drugs as “one of the biggest game changers in the history of medicine,” and floated the tinfoil-hat theory that the 75-year-old man assaulted by police at a BLM protest in Buffalo, New York, was secretly an “ANTIFA provocateur.”
Twitter has been less tolerant of this kind of thing lately. In May, it affixed a “potentially misleading” fact-check label to two Trump tweets that falsely claimed mail-in ballots would lead to pervasive voter fraud. In retaliation, Trump signed an executive order that seeks to weaken the ability of platforms to censor user content. “In a country that has long cherished the freedom of expression, we cannot allow a limited number of online platforms to hand-pick the speech that Americans may access and convey on the Internet,” the order says. “This practice is fundamentally un-American and anti-democratic. When Large, powerful social media companies censor opinions with which they disagree, they exercise a dangerous power.”
Slapping a “warning label” on Trump’s tweets in a new development in this conflict. The first one came on May 29. If he had just called the George Floyd protesters “THUGS” and left it at that, the post wouldn’t have resonated. Even threatening to send the “Military” to Minnesota didn’t register much of a blip. It was a quote, tacked on at the end, that pushed the tweet over the edge: “when the looting starts, the shooting starts.” Those words, used by Miami’s police chief in 1967 to terrorize the city’s black neighborhoods, are as familiar to African Americans as MLK’s birthday. The cited violation is at the very top of the Twitter Rules page: Safety (“violent threat and glorification of violence”). In response, Trump said this about Twitter at a White House briefing with journalists: “I think we shut [Twitter] down, as far as I’m concerned. But I’d have to go through a legal process. If it were able to be legally shut down, I would do it.”
To cut down on suspensions, Twitter recently tested a new feature that asks users if they’d like to “revise” a reply that contains “harmful language” before it’s posted. Introduced in May, the company believes this screen prompt will encourage polite conversation by slowing down the tweeting process, a mental exercise that’s more often impulsive than thoughtful. If it proves to be popular, this abort option will be baked into the Twitter interface.
All these chess moves Dorsey has made in the past few months to enforce the Twitter Rules and level the playing field are part of an initiative to improve “platform health.”
This technology arrives too late to save the writer from his own tweets, which only intensifies the pain. He says the Revise button was designed for everyone but the Stable Genius in the Oval Office: “Trump can tweet whatever he wants, and I can’t. He’s never had a single one of his noxious tweets removed. That tells you everything you need to know.” But aren’t the fact-checks a form of punishment? He laughs and calls it “window dressing,” a toothless gesture to placate the media critics who accuse Jack Dorsey of playing favorites, particularly when it comes to Donald Trump. “There are two sets of Twitter Rules,” says the writer, “one for the president, and the other for everyone else.”
Trump doesn’t always get his way. In July 2019, a three-judge panel on the United States Court of Appeals for the Second Circuit, in New York, ruled unanimously that Trump had been violating the Constitution by blocking political adversaries from his Twitter account. The rationale for this decision was simple: Since the President uses Twitter to conduct government business, he cannot selectively block people from accessing his account and posting comments. Simply put, Trump uses Twitter like a private actor when he’s really a state actor. Judge Barrington D. Parker summed up the ruling nicely: “The decision is unusual only in that it involves Twitter, a relatively new form of public, interactive communication, and the president. However, the opinion is consistent with every precedent of this court, and the dissent does not demonstrate otherwise.”
Katie Fallow, a senior staff attorney at Columbia University’s Knight First Amendment Institute, was on the team that represented the plaintiffs. She praised the ruling, but stressed that Trump’s voice is still worth hearing: “I don’t think Twitter should ban Donald Trump. Because for all the negatives this is still a way for us to learn about what our president is thinking. All of his tweets encourage more speech rather than less.”
Knight v. Trump was hailed as a triumph for the First Amendment in the highly contentious and rapidly evolving arena of cyberlaw. Trump may pride himself on being a “fighter,” but he shouldn’t get his hopes up. To win on appeal, the White House lawyers will have to prove that their client’s Twitter account isn’t a public forum or, as the Knight lawsuit so eloquently argued, a “digital town hall.” That’s a heavy lift, made even heavier after the same New York appeals court announced in March that it wouldn’t be revisiting the case.
There’s no small measure of irony here. The scales of justice may have prevented Donald J. Trump from blocking critics from his Twitter account, but that hasn’t stopped Twitter from blocking those very people.
Like many celebrated tech brands, Twitter doesn’t court the media. They tolerate it. Barely. The San Francisco-based company is a virtual black box where executives speak through press releases and the rank and file leak to journalists at their own peril. Although Twitter declined an interview request for this story, a company spokesperson did agree to field some questions via email. There was one catch, and it wasn’t insignificant: discussing the writer’s suspension was off the table, even with his permission. That ground rule also applied to Twitter suspensions in general, particularly those involving rabid political posts.
There’s no way of knowing how many other users have been suspended from Twitter for trolling Donald Trump. Are there hundreds of other cases like the writer’s? Thousands? More? Pressed for a “ballpark figure,” the Twitter spokesperson wouldn’t budge. It was the kind of stonewall response, equal parts polite and curt, that PR people spin like cotton candy: “Afraid we do not have that data, in ballpark, or otherwise.”
In a business where information is hard currency and even the most trivial data points are analyzed and securely cached, that’s highly doubtful. Given today’s polarized politics and the fact that Twitter is Trump’s digital bullhorn of choice, it’s reasonable to assume that the writer’s case isn’t unique. For proof of this consider an academic paper that only the Trump era could inspire: “Characterizing Twitter Users Who Engage in Adversarial Interactions Against Political Candidates.” This 2020 study, conducted by Cornell University researchers, focused on a dataset of 400,000 Twitter users’ 1.2 million replies to 756 candidates for the U.S. House of Representatives in the two months leading up to the 2018 midterm elections.
The findings won’t surprise anyone who’s ever thumbed through one of Trump’s 3 a.m. Twitter tantrums after it’s gone viral. To quote the study: “First, highly adversarial users reply more in general to candidates from the party the user opposes, suggesting that they are actively seeking out these opportunities for cross-party communications. Second, highly adversarial users are significantly more likely to have profiles that include partisan attacks, suggesting an adversarial disposition independent of any specific conversation.” Translation: Liberal Twitter account holders tend to attack conservative politicians and vice versa; and both loyalist camps tend to have user histories indicating a penchant for political axe grinding.
There’s an acronym that crops up frequently in this 13-page document: AFR (adversarial follower ratio). Forget about the math. Here’s all you need to know about this metric: users who post mean tweets about a politician are more likely to follow other users who post mean tweets about that politician; and more mean tweets attract more like-minded followers, which ramps up the AFR and increases the odds of suspension.
The line charts and bar graphs sprinkled throughout this bone-dry text blend into one another, forming a single image that glazes the eyes like glaucoma. One of them, though, is so simple and devoid of calculus hieroglyphs that it grabs your attention: the AFR graph. More than being intelligible, this diagram offers a persuasive argument that the comedy writer in Los Angeles is part of a large and fast-growing cohort.
In the study’s dataset of about 1,000 “highly adversarial” users, the suspension rate ranged from a baseline of 2 percent (all collected accounts) to 20 percent (accounts with an AFR of 55 percent). “Yes, we definitely saw some banning happening,” confirms Mor Naaman, one of the paper’s authors.” In empirical terms, the clinically vague phrase “some banning” translates into 150 suspensions after just a few months.
Like all peer-reviewed scientists, Professor Naaman is mindful when interpreting numbers while speaking to journalists. In addition to the standard correlation does not imply causation and more research needs to be conducted caveats, he adds this cautionary note to the AFR discussion: “Of course we don’t have the data on why the accounts were suspended, and certainly not whether it was justified.” Balancing that proviso is this hypothesis, which isn’t verified but smacks of the truth: “Simply replying to Trump may make your tweets more likely to be reported by others, and therefore more likely to be reviewed and removed.”
Even if only a small portion of those 150 “highly adversarial” users were scrubbed from the platform for tweeting wicked things about Trump, the implications of the AFR graph loom large. Remember: These were House races, midterms that typically generate modest enthusiasm. Imagine how much more heated the responses would be if it was the runup to the general election, and Trump was doing the tweeting. Moreover, the experiment only included 400,000 users. The Twittersphere is a big place: 330 million monthly active users, 48.35 million in the United States alone. It’s impossible to extrapolate the data from that graph to get the big picture on Twitter users suspended for mixing it up with Trump (Naaman says the “settings” were “very specific”), but it’s reasonable to assume the number stretches well into six figures.
To silence the “highly adversarial” demographic, Trump and his base have floated the idea of rolling back a law that most people had never heard of until recently: Section 230. It’s part of a larger piece of legislation known on Capitol Hill as the “CDA” (Communication Decency Act). Passed in 1996, the CDA was originally an attempt by Congress to shield minors from the evils of Internet porn sites. First Amendment proponents, however, argued that the CDA restricted online free speech. Lawsuits followed. The Supreme Court eventually struck down all of the CDA’s anti-free-speech provisions, but Section 230 survived intact.
Digital rights lawyers praise 47 U.S.C. § 230 as one of the most valuable legal instruments for protecting innovation and freedom of expression on the internet. It’s the only thing that prevents men like Jack Dorsey and Mark Zuckerberg from being held responsible for publishing the content that their users post. Section 230 allows Twitter or Facebook to delete any post, for practically any reason, if they claim the censorship is for the good of the platform or their user’s well-being. It’s a Get Out of Jail Free card for social media moguls. The law is broad and inclusive: “No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”
Jeff Kosseff, a cybersecurity law professor at the United States Naval Academy, calls that sentence “the 26 words that created the internet.” It’s only a slight exaggeration. In fact, those same twenty-six words constitute the unwieldy title of Koseff’s book on Section 230, which was published last year.
Nobody takes the President’s assault on Section 230 seriously. After all, if Jack Dorsey wasn’t protected from all liability for the tweets that his users post, it’s almost certain that Trump’s account would be the first to go. Instead, the White House has encouraged its GOP base to monitor Twitter for them. Last year, the administration launched a Typeform website that permitted individuals to report any instance of online “political bias” that suppressed their conservative voices. “SOCIAL MEDIA PLATFORMS should advance FREEDOM OF SPEECH,” reads the sparse website copy. “Yet too many Americans have seen their accounts suspended, banned, or fraudulently reported for unclear ‘violations’ of user policies.”
Any Trumper who clicks on that link today will be disappointed to learn that the site is no longer accepting complaints: “The White House received thousands of responses—thank you for lending your voice!” Not that it matters. This is pure stagecraft. Trump and Twitter share a symbiotic bond that neither wants to break.
Thomas Davidson, a PhD candidate in Cornell University’s sociology department, knows that many of Trump’s supporters have a legitimate gripe. As the lead author of “Racial Bias in Hate Speech and Abusive Language Datasets,” a study that analyzed 130,000 tweets for discriminatory content, he’s well acquainted with the A.I. programs that social media companies use to patrol their platforms and ferret out bad actors.
“My research indicates hate speech algorithms are prone to error,” Davidson says with sterile dispassion. ”They must be more precise; language contains a lot of context and nuance. You don’t just pick out a word and say, ‘OK, that’s hate speech.’” And the algorithms are far from perfect. “Even racial slurs and homophobic terms can be misconstrued. It’s often the victims of hate speech who are flagged for bans after sharing something that was sent to them.” Here’s another factor that skews racial bias statistics: “Some of the more hateful things our algorithm caught were rap song lyrics. There were a lot of false positives there.”
If a legal team is ever tasked with disproving the writer’s accusation that Twitter suspends users based on an inherent conservative bias, one of the exhibits they would mark into evidence is a video that actor Michael Rapaport uploaded to his Twitter account on March 25. Posted in response to a Trump tweet that encouraged “real people” to “get back to work ASAP” in the midst of the COVID pandemic, the scathing rebuke runs 1:28 and is so downright nasty it would make Martin Scorsese blush. During the 24 F-bomb tirade, Rapaport singles out Trump and his family members for a heavy dose of “Abuse/Harassment.” Even “little [redacted] Baron” isn’t spared.
After seeing the video, the writer utters only one word: “Unreal.”
When the Rapaport video is brought to the attention of the Twitter spokesperson, his reply is boilerplate evasive: “Please feel free to report the account and if it’s found to be violative, we will take action, pursuant to our policies.”
That was four month ago. Since then, not only has Rapaport’s Twitter account not been suspended, the video has gone viral, racking up 3.2 million views and 74.2K likes. Did the actor’s celebrity and respectable following (513.8K) influence Twitter’s decision? Not according to the company spokesperson. “We enforce our policies evenly and equally for all,” reads his email, “political ideology is never part of our evaluation.”
While the scientists collect more AFR data and the engineers tweak their algorithms, the war between Donald Trump and Jack Dorsey escalates. On June 23, Twitter attached another warning label to one of the president’s tweets. The company stated that the post, which targeted the protesters in Washington, D.C., violated the platform’s “Abuse/Harassment” rule. “There will never be an ‘Autonomous Zone’ in Washington, D.C., as long as I’m your President,” reads the tweet. “If they try they will be met with serious force!” Dorsey defended his decision not to delete the tweet by noting ‘its relevance to ongoing public conversation.” He did, however, block all users from engaging with the tweet except to post a comment.
Dorsey has made an example of some other high-profile tweeters. It’s as if he’s warning the president that anybody can crash and burn on Twitter. Even world leaders. On March 25, two of Venezuelan president Nicolás Maduro’s tweets were removed for violating a new company policy that forbids users to spread COVID-19 disinformation through Twitter accounts. The following day, Brazil president Jair Bolsonaro had one of his COVID-19 propaganda tweets removed. And the very next week, Trump’s personal lawyer, Rudy Giuliani, suffered the same fate.
One of Trump’s tweets was finally censored on June 18, when Dorsey’s A.I. dragnet spotted a doctored video on the @realDonaldTrump feed. The banned content, featuring a bogus CNN “Breaking News” graphic and titled “Terrified Todler [sic] Runs From Racist Baby,” proclaims that “America isn’t the problem, fake news is.” Twitter’s warning icon, placed below the video and linked to the company’s “Manipulated Media” policy page, wasn’t much of a deterrent. Less than two hours after the president tweeted the 60-second clip, it generated millions of clicks. When the parents of one of the children in the video accused Trump of violating copyright law, Dorsey removed the video.
And just this week, Donald Trump Jr. had his Twitter account suspended for sharing a post that contained “misleading and potentially harmful” information about COVID-19. The video he posted on July 28 stated that people “don’t need masks” to prevent the coronavirus from spreading, a position that’s contradicted by overwhelming clinical evidence both here and abroad. It wasn’t much of a punishment. The @DonaldJTrumpJr account was only down for 12 hours, but the mini-suspension sent a message: having “Trump” in your Twitter handle no longer confers special treatment. That same day, Dorsey deleted a tweet posted by Trump himself. The retweet, from the @stella_immanuel” account said: “Covid has cure. America wake up.” After removing the bogus message, Twitter replaced it with a gray box that reads, “This tweet is no longer available.”
Relations between Donald Trump and Jack Dorsey have devolved from there. In June, Trump’s then-campaign manager, Brad Parscale, fired this warning shot: “[email protected], your days are numbered.” Parscale linked his tweet to the upstart social media site Parler, which has attracted staunch Trump supporters like at Matt Gaetz, Lindsey Graham, and Ted Cruz. Jack Dorsey wasn’t spooked by the high-profile defections. Several days later, Twitter added another public interest notice to a Trump tweet for violating its Abuse/Harassment rule.
Knowing he’s not the only Twitter casualty is of little consolation to the writer. With COVID still spreading, record unemployment, more Trump scandals, and the November election looming, he has plenty to tweet about but no account to log on to. Like anyone in Hollywood who writes for a living, he knows the pain that comes with being shut down. He’s been fired from TV shows, endured WGA strikes, and suffered bouts of writer’s block. Those were occupational hazards, but this is different. This wound was self-inflicted.
Sensing his frustration, I ask the cruel question: If you knew a perma-ban was the price to pay for the “stroke” tweet, would you still post it? The pause is just long enough to suggest ambivalence, but I know better. The silence on the other end of the line isn’t soul searching. It’s bad acting. “No, I probably wouldn’t,” he finally admits.
The writer misses Twitter. He misses everything about it: the idiot-proof interface; the bite-sized, real-time content; the silly hashtags; all the Likes and Retweets; even that chirpy sound effect. But what does he miss the most? There’s no hesitation this time. “I miss sharing my thoughts and getting a few laughs,” he says. The sense of loss is palpable. If the perma-ban has five stages of grief, acceptance is still a long way off. So he makes do with Facebook. But it’s a poor Twitter substitute, like switching from Châteaux Margaux to Two-Buck Chuck. He won’t admit it, but he misses Trump. Because despite all their political differences, he still appreciates the president’s material: “For comedy writers, Trump’s tweets are required reading. Some of that stuff is really hilarious.”
Stay on top of the latest in L.A. food and culture. Sign up for our newsletters today.