Beware the Never-Ending Disinformation Emergency

Credit to Author: Gilad Edelman| Date: Fri, 11 Mar 2022 13:00:00 +0000

To revist this article, visit My Profile, then View saved stories.

To revist this article, visit My Profile, then View saved stories.

“If you put up this whole interview,” Donald Trump said during a podcast livestream on Wednesday afternoon, “let’s see what happens when Instagram and Facebook and Twitter and all of them take it down.”

Trump named the wrong platforms; the podcast, Full Send, a mildly Rogan-esque bro-fest, was streaming on YouTube. But otherwise his prediction made sense, because during the interview he reiterated his claim that he, not Joe Biden, was the rightful winner of the 2020 election. “The election fraud was massive,” he said during one of several riffs on the theme. “I call it ‘the crime of the century.’ We’re doing a book on it.”

YouTube has a strict policy against claims that the 2020 election was stolen. Yet the video stayed up for more than 24 hours, drawing more than 5 million views. YouTube took it down Thursday evening, a few hours after WIRED inquired about it. It's the latest example of how platforms can struggle to enforce strict misinformation policies—and it raises the question of whether this kind of content ban makes sense in the first place.

Consider what happened to the Hill.

Last week, YouTube suspended the Hill, a political publication in Washington, DC, for seven days after its YouTube channel aired clips of Trump claiming election fraud. One came from his recent speech at the Conservative Political Action Conference. The second was a snippet from a Trump interview on Fox News, which was broadcast on the Hill’s daily commentary show, Rising.

The latter clip wasn’t even primarily about the election. In it, Trump gives his less-than-statesmanlike analysis of the Russian invasion of Ukraine, which the Rising hosts proceeded to mock. But right at the end of the clip, Trump says, “And it all happened because of a rigged election.”

This was enough to trigger YouTube’s election integrity policy, which prohibits “false claims that widespread fraud, errors, or glitches changed the outcome” of past presidential elections. Under the policy, you can only include those claims if you explicitly debunk or condemn them. That’s where the Hill went wrong. “Upon review, we determined that the content removed from this channel contained footage claiming the 2020 US presidential election was rigged (which violates our election integrity policy) without sufficient context,” said YouTube spokesperson Ivy Choi in an email. One “strike” gets you a warning, two gets you a weeklong suspension, and a third gets you kicked off the platform.

With all the attention paid to online misinformation, it’s easy to forget that the big platforms generally refused to remove false content purely because it was false until 2020. It was Covid-19, and then the election, that got them past their squeamishness about weighing in on factual disputes. Two years into the pandemic and more than a year after January 6, however, it’s worth asking: What’s the endgame for policies adopted during an emergency?

“If we’re actually thinking about how speech works, surely context should be relevant.”

Evelyn Douek, Harvard University

It’s important to remember that platforms have very good reasons for not wanting to be the “arbiters of truth,” in Mark Zuckerberg’s famous words. As Trump seems to understand, it feeds people’s sense that there are ideas that powerful entities are afraid of discussing. “If we talk about the election fraud, they will not cover it,” Trump said on the podcast, referring to the “corrupt” media. He challenged the hosts to stand up to the censorious social media overlords. “Let’s see what happens when they threaten you,” he said. “It’s a test.” And, of course, platforms will inevitably restrict perfectly legitimate content while letting bad stuff slip past, because no one can do perfect enforcement. In addition to the  podcast interview, Trump's full CPAC speech—showing a clip of which helped get the Hill suspended—was still available, from CPAC's YouTube channel, 11 days after it first went up. YouTube also took that video down only after WIRED inquired.

In the Hill’s case, YouTube’s election integrity policy seems to rest on particularly questionable assumptions. Notice that when I quoted Trump’s comments from the podcast, I didn’t add that his claims were false. Were you therefore at risk of believing them, if you didn’t already? The unstated premise of a policy like YouTube’s is that, in the year 2022, there are a meaningful number of people out there who would have been.

This set of assumptions can put journalists who rely on YouTube to reach their audience in a strange position.

“You kind of feel like an idiot telling your audience that, in fact, Hugo Chavez did not overturn the 2020 election,” says Ryan Grim, one of the cohosts of Rising, referring to the conspiracy theory about the late Venezuelan president’s posthumous involvement in rigging US voting machines. Grim’s main job is DC bureau chief for the Intercept, a left-leaning publication. It bemuses him that anyone might think a clip of Trump aired on Rising amounts to an endorsement. “It makes you feel like a dancing clown, juggling for the YouTube executives,” he says.

On the other hand, Jeff Allen, a former Facebook integrity researcher and cofounder of the Integrity Institute think tank, sees some logic in YouTube’s policy. When YouTube changed its rules to allow mentions of Nazi ideology for educational purposes, he says, actual white supremacists tried to game the system by smuggling in racism under the guise of education. Requiring heavy-handed disclaimers of election fraud claims could help ward against similar mischief. “There are probably lots of YouTube channels that are exploring the bounds of YouTube's anti-election misinformation policy, for the purposes of actually advocating for election conspiracy theories,” he said in an email.

Still, by asking news hosts to explicitly denounce any mention of election fraud, YouTube isn’t just making its own content decisions; it’s injecting itself into the editorial processes of actual media outlets.

“It absolutely puts YouTube in the editor’s meeting, in a really bizarre way,” says Grim. A better approach, he argues, would be to let legitimate media organizations make their own decisions about how to present controversial material. This would raise tricky issues about who qualifies as legitimate, but platforms already make these distinctions in many contexts. And a system that treats a child in their bedroom and a reporter in a newsroom exactly the same would be absurd. “I think they should leave newsrooms alone,” Grim says. “I know that makes a lot of independent content creators angry, but as a starting point, if people are running professional news operations, it’s hard to see why you also need YouTube executives in the room for that.”

For that matter, why should YouTube police election fraud claims at all, long after the election is over?

Social media companies are private entities that aren’t bound by the First Amendment. But their power over communication is so significant that their policies are nearly as important as law itself. In a new book, Cheap Speech, the election law expert Rick Hasen argues that platforms should ban political speech “only upon a clear showing that the speech actually threatens to undermine, rather than support, democratic governance.” One clear example of that kind of speech is encouraging violence against political opponents. A second is spreading false information about when, where, or how to vote in order to deprive people, through trickery, of their constitutional right to vote.

YouTube, Twitter, and Meta all have policies forbidding that kind of content. But YouTube goes a step further by disallowing claims that any past presidential election was rigged—without explicitly saying why it has that policy.

Perhaps YouTube takes Hasen’s position that rigged-election talk undermines public confidence in elections.

“An election requires not only that you hold a fair election but that people believe in that election that it was done fairly,” he tells WIRED. “When you undermine that, you risk undermining our entire democracy. If you believe that the last election was stolen, you’re going to be more likely to be willing to take steps to steal the next one.”

The tricky thing is that it’s hard to find the limits of that logic. In 2020, quite a few people were alarmed over the prospect that Louis DeJoy, Trump’s postmaster general, would use mail delivery slowdowns to prevent Democratic ballots from being counted. And to this day, millions of Democrats believe that the 2000 election was wrongly awarded to George W. Bush based on an inaccurate vote total in Florida. Should they not be allowed to make that argument? (Technically, YouTube’s election integrity policy forbids this claim, too.) Blocking people from questioning election results seems likely to backfire, if the goal is to increase public confidence.

Another reason to ban Trump’s big lie might be that it has been linked to violence. Certainly, this rationale applied during the period around the election and January 6. The January assault on the Capitol was the direct consequence of Trump egging on his supporters to “stop the steal” and prevent Congress from certifying the election results. In that context, it made perfect sense to treat Trump’s claims of election fraud as too dangerous to tolerate.

More than a year on, however—and with claims of electoral fraud continuing to saturate Republican politics, notwithstanding social media platform policies—the calculus has changed. That appears to be the view that Twitter has taken. In January, CNN’s Daniel Dale reported that Twitter had, in March 2021, quietly stopped enforcing its policy against claiming the election was rigged. “When enforcing our civic integrity policy, we prioritize content that can cause direct, immediate, real-world harm,” Twitter spokesperson Elizabeth Busby said in an email to WIRED. “With the US 2020 presidential election certified and implemented, and President Biden seated in office, these harms and risks look different than they did more than a year ago.”

Even though nobody had noticed the policy shift for nine months, Dale’s article was met with quite a bit of outrage. In a recent op-ed, Hasen called it “a step in the wrong direction.” The prominent press critic Jay Rosen tweeted that Twitter’s decision “makes no sense,” because Trump’s “lies still have political valence.”

The backlash reveals something about content moderation rules: It’s easier to put a new restriction in place than it is to take one away.

“It’s mostly been a one-way ratchet,” says Evelyn Douek, a doctoral candidate at Harvard Law School who studies content moderation. “It rarely goes back the other way; it always tightens and tightens. To date, we haven’t seen loosening at the end of periods of risk.”

Early in the pandemic, Douek warned about this exact issue in an essay for the Atlantic. Pointing to the unprecedented steps major platforms were taking to police pandemic-related misinformation, Douek analogized the situation to an “emergency constitution,” when a government claims special powers and temporarily restricts civil liberties. This, she argued, made perfect sense. In a situation like the early days of an exponentially spreading pandemic, access to accurate information could be the difference between life and death. The question, she wrote, was how long these exceptional measures would last.

Today, Douek argues that platforms need to adjust their tolerance of “bad” speech to fit the circumstances of the moment. Blunt content bans might make sense during the height of an unfamiliar pandemic, or during a violent assault on the transfer of power. But when those dangers subside, the same restrictions on speech may not justify the burdens they place on communication. These judgment calls are difficult. But a first step would be for platforms to at least commit to making them.

“If we’re actually thinking about how speech works, surely context should be relevant,” she says. “The social context is completely different now than it was around January 6, when there really was a clear and present danger that this kind of information might influence action.”

The election isn’t the only domain where this will come up. Platforms continue to impose all sorts of restrictions on pandemic-related misinformation. This comes from a well-intentioned desire to keep people safe during a public health crisis. But as the pandemic drags on for years, it gets harder for social media to justify keeping its emergency measures in place. Now is a good time to remember, too, that platforms can accomplish much more by changing their algorithms to reward credible material than they can through Whac-a-Mole content moderation.

It might seem strange to worry about platforms cracking down too hard on Trumpian claims of election fraud, especially as Republican election officials around the country are sending ominous signals about what they’re prepared to do in the 2024 election. Rigged election talk is harmful. But banning content can be harmful, too, in ways that are harder to measure. As Trump demonstrated in Wednesday’s podcast interview, it feeds people’s sense of unfairness and martyrdom. And even if you agree with the platforms’ policies on election and Covid misinfo, you might not want to keep giving them ever-more power to rule certain ideas out of bounds.

The question isn’t whether lies are dangerous. They can be. But a lot of speech is dangerous. The question is whether the benefits of banning it outweigh the costs. They say a lie gets halfway around the world while the truth is still putting on its shoes. But eventually the truth gets its shoes on. At some point, you may have to let it run.

https://www.wired.com/category/security/feed/