Twitter Disinformation
(Note: The focus of this essay is on how left-wing bias in Twitter moderation, which I think was real, was a sub-set of left-wing bias in many mainstream institutions, particularly media, and has had the result of lowering trust in those same institutions. Because of this focus, I’m not going into all the other Twitter Files stories, such as the Trump suspension and the connection between Twitter moderation and federal agencies like the FBI, DHS, and DNI. This is a big set of stories and I’m only hitting one angle.)
The Twitter Files—a series of tweet threads put out by Matt Taibbi, Bari Weiss, and Michael Shellenberger, based on Twitter information shared by Elon Musk—are making the news. You can find them here, here, here, here, and here. (So far. More are being added all the time.) A general thesis of their tweet threads—which have gotten a lot of pushback—is that pre-Musk Twitter (now being called “Twitter 1.0”) had a moderation team that lowered the visibility of and/or banned accounts based on political bias. While anyone’s tweets might be targeted and anyone’s account shadowbanned (its viral visibility reduced) allegedly the targets were more often right-wing or heterodox than left-wing. (The biggest target was President Trump, whose Twitter account was suspended on January 8, 2021.)
The threads paint a picture1 of Twitter putting its thumb on one side of the political scale. The more extreme accusation is that Twitter’s actions helped bring about Joe Biden’s 2020 victory. (After the Twitter Files hit the news, Donald Trump immediately suggested that either new elections should be scheduled or he should just be installed as president as the “rightful winner” because of course he did.)
Pushback against the threads falls into three camps:
1 - Twitter is a private company and has every right to limit tweets in any way it wishes.
2 - The tweets are a cherry-picked sample and don’t prove a pattern of bias. They’re a big “nothingburger.”
3 - The accounts Twitter was limiting were spreading harmful messages and disinformation and so limiting their reach was actually a good thing.
The first is obviously true. Twitter may sometimes behave like a public square but it is under no legal obligation to treat everyone’s tweets equally. It is a moneymaking business and it can make money as it pleases, within the confines of the law. The First Amendment applies to the government censoring people, not private companies running their businesses.
The second point could be true. I don’t know what information Musk released and what he held back. Whether or not the released information represents a pattern of bias (and I think it does), it cannot by itself prove it.
The third argument has quite a few problems. It may be reasonable to suppress bad ideas—I’m not eager to see Nazi imagery spreading across Twitter—but the line between “dangerous” and “I don’t like it” can be pretty fuzzy.
For instance, this tweet:
Dr. Bhattacharya was controversial because he had argued that lockdowns were not the best solution for dealing with the disease. Instead, he had suggested a targeted approach that would have kept older and more vulnerable people shielded from contact with the less vulnerable majority. This would have avoided the economic and societal hits caused by the lockdowns. He was, however, no Covid denier. He was also a firm believer in the efficacy of the Covid-19 vaccine.
Jay Bhattacharya: I mean, you think that for someone who's older, especially, the vaccine is incredibly important. COVID is very deadly disease as we talked about for people who's older. And the vaccine, while we haven't... It's only been in human use for 10 months, right? So we don't know all of the side effects, but we've seen enough to know it's pretty safe.
Interview at Hoover Institution, October 2021.
Was he wrong about Covid lockdowns? I don’t think there’s a conclusive answer but it should be clear to most that Covid lockdowns had both negative and positive effects. The negatives were especially strong for children. For Twitter to downplay Bhattacharya’s words because they went against the mainstream narrative that general lockdowns were necessary was an extreme overreach. Yes, Twitter had the right to do so, but the evidence that it was preventing harm was weak.
Then there is the example of Twitter freezing the New York Post’s account because they had tweeted this story:
Twitter called it “unsafe,” removed links to the article, and even stopped the story from being sent via direct message. All because the story was supposedly spreading dangerous election-damaging information. Except the story was true. The Hunter Biden laptop was real, not a Russian plant, and the emails on the device were legitimate. It seemed to critics that Twitter had suspended the Post’s account because the story might harm Joe Biden’s chances of being elected.
None of this means Biden isn’t the legitimate president (sorry, not sorry, Donald). It’s likely that there is nothing on that Biden laptop except further evidence that Hunter Biden was a messed up guy with sticky fingers and a drug problem. Unless you can prove that Joe Biden had his hand in the till as well—which seems unlikely—the laptop isn’t a smoking gun. Moreover, Twitter’s efforts might have done more to publicize the story than ignoring it would have.
This Washington Post article—“No, limiting the Hunter Biden laptop story didn’t cost Trump the election”—argues that the evidence for Twitter hurting Trump’s chances is weak. Searches for “Hunter Biden” when up after Twitter and Facebook imposed their limits. This was my own experience. Everyone was talking about Twitter shutting down the New York Post so I went and read the article (something I normally would not have done).
So no, Twitter’s behavior probably didn’t cost Trump the election. The Twitter files do not reveal the crime of the century, as right-wingers like Pedro Gonzalez (a senior writer at Chronicles, a conservative magazine) are claiming in silly tweets like this:
But the Twitter Files seem to show Twitter putting its thumb on the scale to favor Democrats. Even if the thumb didn’t change anything, the attempt (if attempt it was) still matters.
Monoculture
The Twitter Files are also further evidence of a trend in mainstream American institutions to favor left-wing views over others and to frame counter-narratives as dangerous misinformation. A recent Wall Street Journal editorial made the argument that it’s not so much that liberals live in a silo as they’ve become so institutionally dominant that they’ve created a world where contrary points of view can only seem bizarre and dangerous.
The left-liberal outlook has triumphed across American culture—in corporate boardrooms, in government agencies, in sports and entertainment institutions, in K-12 education bureaucracies, in universities and in media organizations. But that is precisely what has robbed progressives, especially those in the political class and in the media, of any ability to criticize themselves or doubt their own righteousness. They don’t engage with serious arguments advanced by the other side. They live in a world in which it is possible to pass through a month without encountering much in the way of serious conservative opinion. When they do encounter a conservative view, it is precategorized as “fringe” or “extreme” by the calm, omniscient NPR voice that relates its content.
Recently, Matthiew Yglesias made a related point on his Bad Takes podcast that hit home for me.
I feel people started to feel increasing pressure not to like pursue the truth vigorously but actually to suppress dissonant factual information in pursuit of a kind of moral clarity…
I do think it's become a serious problem in the world…
You want the press to seem credible to as many people as possible which means you want to be able to say that you call it like it is and you pursue the truth without prejudice or favor and giving up on that ideal is quite damaging…
A lot of people and organizations took stances because they believed they were more correct than they really are because there was a kind of echo chamber around a lot of these topics.
Yglesias is solidly left-liberal and he lives in the liberal media ecosystem (he’s a co-founder of Vox) so I think he knows what he’s talking about. His words also match my own experience. Over the last half dozen years, I’ve read stories in places like Vox and the New York Times with increasing skepticism. Too often they seemed to prioritize agenda-pushing rather than informing the public.
Yglesias is not talking about a massive conspiracy and neither am I. Instead, what’s happened is that liberals who dominate both big media companies and tech companies share the groupthink that the Wall Street Journal described. So much so that they’ve come to believe that contrary information should be downplayed or ignored, even if obviously true. You can’t live in lefty circles and not realize, for example, that talking about crime is generally frowned upon while talking about police brutality is generally applauded. Articles emphasizing the latter get far more play in left-wing spaces. The result is that liberal readers tend to see the police as far more dangerous than do moderates and conservatives (and vice versa for crime).
Given this skewed emphasis, how effective is the media at conveying the truth of police brutality? In 2020, Skeptic magazine asked Americans how many unarmed black men they thought had been killed by the police in 2019.
Most “very liberal” Americans thought the number of unarmed black men killed by the police was at least 1,000 (and 8% thought it was over 10,000). Merely “liberal” Americans thought the number was at least 100 (and 39% agreed with their very liberal friends that it was 1,000 or more). The actual number in 2019 was 12 unarmed black victims. Another 26 unarmed white victims were killed by the police that same year. (Data is from the Washington Post police shootings database.) These numbers are typical. In 2021, there were 11 unarmed black victims of police gunfire and 19 unarmed white victims.
So in this area, conservatives have a far more accurate view of the world than liberals.
Getting these numbers wrong matters. Police shootings are a serious problem but so are homicides, which kill far more people of all races. According to the CDC, 24,576 homicides occurred in 2020.2 This is a crisis for America (our homicide rate is about six times that of other rich western countries) and yet in a recent conversation, a friend told me that the police killing black men was a bigger problem because she truly believed that more unarmed black men were killed by the police than were murdered. The actual ratio is about 1000 to 1 in the other direction.
No wonder liberal crowds shouted, “defund the police” in 2020. They believed that the police were the main threat to black people’s lives. As Yglesias said, “A lot of people and organizations took stances because they believed they were more correct than they really are.”
When the media puts their collective thumb on the scales in the name of being virtuous their readers are left poorly informed.
So Twitter’s old moderation policy fit in with a general tendency on the left to prioritize advancing certain narratives over allowing free debate. This meant ignoring facts that went against the prevailing worldview on the left. This seemingly led Twitter moderators to push ideologically motivated agendas. Or as David French put it in The Dispatch:
The picture that emerges is of a company that simply could not create and maintain clear, coherent, and consistent standards to restrict or manage allegedly harmful speech on its platform. Moreover, it’s plain that Twitter’s moderation czars existed within an ideological monoculture that made them far more tolerant toward the excesses of their own allies.
If you have opaque moderation policies and an ideological monoculture that assures you that you are always right it is hardly surprising that you will often favor one side over the other. Inevitably this will lead to a decline in trust, especially on the right.
Twitter’s moderation policies were in sync with mainstream media’s editorial policies. A smug belief that you are the righteous ones and that your readers should be protected from dangerous ideas will lead you down some deeply distorting paths. If you are so sure that you are correct that you silence or muffle dissenting voices, you are almost guaranteed to end up very wrong about some issues. As in science, media narratives require constant challenges to stay reasonably close to the truth.
(Important clarification: Despite my criticisms, I don’t think mainstream media is hopelessly compromised, as some right-wingers argue. There is still no paper with the breadth and depth of coverage of the New York Times. Other journals like the Wall Street Journal and The Economist also do an excellent job. The bias I've mentioned is there but it has only affected certain kinds of stories. I also think there has been a clear shift in all these papers back towards better editorial standards. I trust 2022’s Times more than 2020’s.)
Free Debate
None of this means that Twitter has to publish everything or that the mainstream media is part of some vast conspiracy against the truth. Instead, I’m preaching the importance of listening to contrary voices, of not assuming that one political view is the only correct and righteous one. Deboosting or shadowbanning or whatever it is that Twitter was doing had the result of suppressing contrary voices, and that was bad. In many ways, it was worse for liberals than it was for conservatives.
This doesn’t make Elon Musk the good guy and Twitter’s Trust and Safety department the baddies. I think Musk is doing good by releasing this information to Taibbi et. al., but he’s a childish troll, too in love with sending out nasty tweets. I worry that he is bent on a kind of vengeance spree against the left and that would be very harmful, for conservatives as much as for liberals. (Think about how much Trump’s trollishness benefited liberals in 2018 and 2020.)
I also think the Trust and Safety folks, led by Yoel Roth, were doing their best with a tough job. Their own liberal biases, reinforced by the echo chamber Yglesias spoke of, convinced them that they had to suppress dangerous voices to protect America and its democracy. They were not evil, just wrong.
If you don’t see the groupthink that I see in left-dominated institutions, more anecdotes won’t convince you. I’ll just end with the obvious point that confidence in the media has steadily dropped over the past few decades. I’m sure that pushing narratives and campaigning against “disinformation” isn’t the only cause of this drop in confidence, probably not even the most important cause, but I can’t see how it’s not involved to some degree. People don’t like being lied to. Perhaps even more important, banishing ideas that liberals deem harmful may end up blinding liberals to reality. This will have negative consequences for public policy and for electoral success.3
Speaking of pictures, the picture thumbnail for this essay is of a giant sculpture called “Echo,” by the artist Jaume Plensa (although New Yorkers just called it “that big head thing”). It was set up in Madison Square Park in Manhattan in 2011. If it looks weird in a photo, trust me that it looked even weirder in person.
The FBI reports a lower number because not all law enforcement agencies submit homicide reports to the FBI.
Although given the poor candidates Republicans chose in 2022, the Democrats may be able to make many mistakes in campaign messaging and still win quite a few elections. It often seems a race to see which party can shoot itself in the foot more often.
For decades, conservatives have been complaining about a "liberal media," and I always dismissed it. Only in the past couple of years have I started to concede that they may have a point.
Doesn't mean I'm going to start voting GOP. I know where my values lie. But man, the echo chamber effect really messed up all of us across the spectrum.
Twitter doesn't really have the right to selectively apply political bias in how they curate content as long as they enjoy the privilege of being shielded from the kinds of liabilities that go along with being a content publisher rather than a platform.
And the whole point of the Hunter Biden laptop was the 10% "for the big guy", of course Biden himself was implicated.