What Do Twitter’s Users Actually Want?

Eighteen takes on what should—and should not—be forbidden on the platform.

person shouting through their hands
Charles Phelps Cushing / ClassicStock / Getty; The Atlantic

This is an edition of Up for Debate, a newsletter by Conor Friedersdorf. On Wednesdays, he rounds up timely conversations and solicits reader responses to one thought-provoking question. Soon after, he publishes some thoughtful replies. Sign up for the newsletter here.

Last week, I asked, “What should be forbidden on Twitter?” You responded with many recommendations for the social-media platform as Elon Musk attempts to purchase it and take it private.x

Michael sympathizes with the status quo:

I am married to someone who reviewed graphic / violent posts at Twitter. Our discussions changed my mind on the topic. Social media is not analogous to regular speech. If Musk wants to make a genuine town hall, with no moderation, he should do away with retweets, likes, and other elements that create mob effects. Twitter has spent the last seven-ish years working on content moderation, a major focus for a lot of bright people, and the outcome is highly specific. My suggestion: run with the existing rules, but—I say this to satisfy Musk—break ties in the direction of leaving posts up.

R.C. has mixed feelings:

If you are asking me as a customer what they should ban: I go to Twitter for entertainment and information, but also to be able to interact with others. If there were content I found odious and grotesque and couldn’t block I would stop using the site. Whereas I as a citizen of this country would tolerate and defend people’s rights to use hate speech or even speech that advocates violent revolution, unmolested by any arm of the state. But Twitter just isn’t on that level for me. So, with those caveats, here’s what I would consider a dealbreaker for my Twitter habit: Child Pornography. Snuff videos. Rape videos. Nazi / White Supremacist / Neo-Nazi / Skinhead propaganda videos or messages. Videos or messages that explicitly call on people to commit violent crimes.

M. argues for a First Amendment standard:

My impression is that people arguing for content moderation (AKA censorship) aren’t making a case about why Twitter, specifically, warrants its own unique standard of speech. Instead the arguments I see are (sometimes) veiled arguments that the free speech standard of our society as a whole should be altered. For example, the statement you provided from the NCAA, doesn’t identify anything unique about Twitter that warrants different rules—you could swap “Twitter” out of their statement and replace it with any other platform and the argument would be the same.

If anyone believes that Twitter should have its own unique standard that is different from the First Amendment, the onus is on them to explain why Twitter warrants a unique solution. Twitter certainly has some distinguishing characteristics such as the number of people it reaches, the ease of searching its content, and the speed that a message can spread. But being the most (or one of the most) popular and accessible forums for speech is a poor reason to restrict speech.  

Twitter does have at least one somewhat unique characteristic: an algorithm(s) that amplifies certain messages over others. I do believe those algorithms need to be carefully considered so as not to reward hucksters and trolls. Focusing the discussion there—on what that algorithm should be (if there should be amplifying at all)—is more appropriate than debating a new standard of free speech.  

That’s a long way of saying that I support the First Amendment standard of free speech for Twitter. This would mean all sorts of abhorrent content (racial slurs, Holocaust denial, etc.) and/or misinformation (masks might suffocate you, Russia is the aggrieved party, etc.) would be allowed. While I think those ideas are horrible, allowing people to express them is a good thing. I want to know who holds racist views so that I won’t vote for them or patronize their businesses. I gain confidence in good ideas—such as covid vaccines or the qualifications of Ketanji Jackson—when I know I can access the arguments against them and see that those arguments are nuts. I also believe censoring ideas often makes them edgier and more appealing. If [the ideas are] left to fester in the open, honorable people will expose how rotten those ideas are with their own free speech. Finally, given who is buying Twitter, I think it’s important to say that the First Amendment standard would also protect criticism of Elon Musk, his companies, or the markets he wants to sell his products in (China). People should be free to criticize Musk or promote competitors.

D. wants to give people a choice of how to experience the site.

This debate brings to mind Henry Ford’s famous quote about the Model T, when he said, “any customer can have a car painted any color that he wants so long as it is black.” How can we possibly have one choice in content moderation that works for everyone?

Let’s allow each user to decide what content moderation they want. Allow competing standards with their own AIs, and force the user to pick one when joining (including a free speech absolutist standard). Of course, allow them to change their selection at any time. Give the user a user-friendly interface allowing them to choose, from the outset, whether only people they are following can reply to their tweets. This would fix a lot of problems. Also give users an option for an old-fashioned chronological feed from those they follow.

Martin agrees. “The goal is not to stop dumb people from saying dumb things, but to make sure they are not heard by broad swaths of people who have no interest in listening,” he writes.

Says Cole:

I am a 23 year-old Twitter addict who is three months off the site, with no plans to return. I cannot recommend leaving strongly enough. There is a big gap in the free speech crowd. On one side are the 4chan freaks who want to post cartoons of graphic sexual deviancy combined with swastikas and other truly bone-chilling racist language. The other side just wants to spout off socially-conservative takes from 1972 without anyone bothering them.

My proposed path forward would involve users getting a pop-up screen, similar to the sliders in a video game, asking them how “free” they would like the speech they take in to be. Anyone should be able to say anything that is legal, but the scope of what people choose to see will vary. It would function a lot like muting words on Twitter, but as preset options, with capabilities to customize further. For example, I don’t want to see violence or slurs or sexual content, but political disinformation does not bother me. I expect the majority of people will choose settings similar to what’s already the code of conduct; the people who run Twitter are not morons, and Jack Dorsey was pretty good on free speech. Under this approach, many will realize they aren’t the free speech purists they thought they were. Everyone talks a big game about their takes on Democrats being censored, but I suspect not many of them want to see compilation videos of people getting hit by trains or ugly men getting naked and shouting the n-word.

Whereas Bob would, I suspect, shift the locus of social-media polarization from ideology to pet preference: “It’s pretty simple,” he writes. “The content moderation rule for Twitter should be: the only posts allowed are pictures of dogs.” Careful, sir, cat people are a powerful faction online.

Isaac argues that fighting over content moderation misses the point because the costs that social media imposes on society are mostly due to the deliberately addictive designs of the platforms:

Any algorithmic policy that attempts to truly eliminate all problematic speech will doubtless end up banning tons of accepted speech (however this is decided). Any algorithm that is permissive will let all kinds of toxicity through. I don’t think it is actually possible to make a policy that is totally acceptable to everyone. What deeply saddens me is that these companies are incentivized to promote social media addiction. People destroy their lives gambling because casino companies have figured out how to hack the human brain’s reward system. It doesn’t work on everyone, but some people are absolutely sucked in by the flashing lights and the chance to win big. Social media companies have used the exact same strategies to create addiction, it’s just that instead of using money as the reward they use social validation. Humans are a deeply social species and I think many people fail to appreciate what a powerful motivator this can be.

From the perspective of society, less engagement means less social media addiction, less online toxicity, less misinformation, and a social order that has a chance of getting things done. Instead we get a race to the bottom of the brain stem. It has been fairly disheartening to see this media cycle so focused on pointless content moderation policies, and to see so many well intentioned people miss the forest for the trees so badly. We need to stop focusing on these granular policies and start focusing on the design of these platforms in general. The upshot of all of this is that we could probably fix the design without instituting some capricious regime of censorship by mega corporations.

Chris wants to protect the ability to lie but opposes ease of spreading lies:

I’ve grown up with the American faith that free speech is a universal force for good, because it allows people to respond to bad ideas with good ideas, and the best ideas will rise to the top. But why should we think that “rising to the top” will always happen, regardless of the particulars of evolving media? If we want a forum where discussion of ideas actually helps us find consensus and discover truth, we have to pay attention to the rules of the medium. This is why, for example, the legislatures use systems like Robert’s Rules of Order: collective intelligence arises from not just the intelligence of the individual members of a group, but from the way that the members interact with one another.

I don’t like banning lies; but maybe if Twitter changes the underlying rules, the lies won’t be very harmful. I think the apparent need to ban some speech on Twitter stems from a flaw in the way Twitter works that causes engagingly irritating ideas to spread more widely and faster than good ideas, due to what the user interface makes easy and what it makes hard: e.g. it’s easy to make an anonymous account and reply with dumb one-line retorts to lots of people, but hard to say something longer and more thoughtful with an account you can prove is your own. With different rules in place, liars, trolls and assholes would not have nearly as much influence, and might not need to be banned.

Among Perry’s picks for what to prohibit: “No bomb-building instructions or poison gas recipes. And anyone who posts about causing harm to someone should go to jail if it happens.”

Sebastian takes a page from the gun-control debate:

What if Twitter et al. institute a “cooling-off” period? You submit your tweet as usual but it would not be immediately published. Rather, an hour later (or even just 5-10 minutes!), Twitter sends you a notification asking if you would like to publish the tweet. I suspect it would materially cut down on inflammatory speech. The added friction would simply require that we abide by the age-old wisdom: “take a deep breath and count to ten before you say something you might regret.” It might render Twitter more humane, measured, and productive.

Matt would offer the least protection to possessors of the blue check that Twitter bestows on “verified” users like me:

There should be a line drawn between “Blue Check” people and normal users. It should be forbidden to doxx a private person for some social infraction, but to speak all manner of profane things about an elected official or celebrity should be allowed. I only draw the line at threats of violence or incitement for the blue check world. What about misinformation? For the truly nefarious actors, we can be aggressive. My own parents were scammed after buying survival supplies from an “Anti-Biden” ad on Facebook. Would I gladly ban those actors, absolutely! But should my mother get banished for posting that Kamala Harris is a member of the Illuminati? Probably not. If anything, Twitter is a wild west library and is better for it. You can find facts, entertainments, and fictions. Caveat emptor.

And Benjamin would kick me off Twitter entirely.

“I would ban all journalists,” he writes. “They enjoy it too much and it is disconcerting to find out that the people who are delivering the news to the American people are really dumb.” #Notalljournalists, Benjamin! “I would ban me,” Benjamin goes on, “because I am afraid that one day in a drunken stupor, I might use the damn thing.” You see, he is not a fan of Twitter:

Free speech is not intelligent speech. Choosing who can say what on Twitter is like picking a poison to murder someone with. Any poison will do. The world’s self-destruct button has already been pressed. So, I would censor nothing. Twitter is a garbage dump, a toxic wasteland. It is a place where the self-righteous thunder self-righteously and pat each other on the back for being good. It is a place where people go to say things they can’t say in any other social setting without starting a brawl. It is a place where using words, people take the equivalent of a bowel movement. Twitter is America, the freaks, phonies and fools, hypocrites, pinheads, and puke brained imbeciles. Allowing a free for all of free speech on Twitter won’t make it any worse than it already is. It will make it sillier and more fun. The unintelligentsia will have more to get hysterical over.

So let it fly, let it hit the fan and let’s see what sticks. Most people, the ones who never use Twitter and never listen to the journalists who love it so much, know it doesn’t matter.

Twitter is us, and we scare the hell out of each other. So, what.

Strong Holden Caulfield vibes with this one!

Starla believes we should focus on amplification:

What we have now is a problem with everyone believing their message should have access to the same amplification as the President, despite the fact that they may be simply an uncredentialed dude with an ax to grind. Anyone should be free to say anything within the confines of their space—but how and when that is shared to other networks and spaces is the real question. Solving the problems related to amplification of misinformation requires the business models of social media companies to be drastically re-ordered. A useful model would provide safer, smaller spaces, like living rooms or soap boxes, for each person’s right to free speech. Amplification or sharing out to the public sphere, however, must require some kind of credentialed gate-keeping if we intend to live in a factual world with some sense of shared reality and values.

I’ve yet to live through a presidential administration that didn’t lie to me, reaching me through media gatekeepers to do so. Did the credentialed press of the War on Terror period, to cite one example, result in the dissemination of reliable facts and a shared sense of reality and values? And that’s to say nothing of the era when the president was Donald Trump, a man who spread far more disinformation and misinformation than the typical “uncredentialed dude.”

Luciano wants a diptych-based social-media platform:

For me, the underlying problem with the unfettered Twitter (like the unfettered internet, or unfettered democracy, for that matter) is that the debate that is supposed to happen never does. Instead, the shouts become louder until the voices of many are really just the loudest voices of a few. Instead of creating an open popularity contest, where clicks and likes drive the debate, have the platform built to foster debate, where a tweet is always shown with an equally popular or compelling counter-idea SIDE BY SIDE. This way, instead of seeing just a Trump or AOC tweet, the user will always see a counterargument. The more popular or influential the user, the more the algorithm would find plausible and direct counterargument tweets. You’d never see just a single argument.

Peter wants to criminalize group libel:

Against slander, false accusations and libel the only remedy that is available is civil law. Going that route can be a very long and costly process that minorities and poor people can’t afford. Members of the majority can raise the same accusations over and over again, making it impossible for the minorities to keep arguing against the repeated false accusations. These false accusations can give rise to real violent and harmful consequences as in the case of Antisemitic statements and Holocaust denials, to give one common example. That is the reason why certain false accusations have to be addressed by criminal law, rather than by civil law which puts the burden of proof on the injured party.

Jack takes his argument in a direction I did not expect:

Twitter is not the government. It is more powerful than the government when it comes to facilitating or denying ordinary voices and opinions. To me, this suggests it has the duty to be as protective of free expression as the comparatively inconsequential government is.

And Jonathan says we should sort onto different platforms by ideology:

Any notion of content moderation imports a system of values that is to be forcefully imposed by some authority upon subject peoples. As a society, we collectively lack anything approaching a consensus on either the things we value or what constitutes a valid authority. Consequently, it is impossible for anyone to impose a universal program of content moderation that will come anywhere close to appeasing an overwhelming majority of people. There can be no Universal Twitter except to the extent that people can either accept being deeply dissatisfied with the way things work or converge on an accepted system of values. As neither of these options seems particularly likely, the best path forward is for different groups with sufficiently aligned values to agree upon some system of authority that will moderate content on its own platform.

For my part, I would urge tolerance and forbearance, this moment’s most underrated virtues, so we can stay on the same platform together. Thank you all for your various perspectives, and I’ll see you Wednesday.