Skip to Content
MIT Technology Review

The Census is a target for disinformation—here’s how it could be protected

It’s an all-hands-on-deck situation to stop the vital count from being compromised. But there are still concerns that Big Tech isn’t being transparent enough.

November 6, 2019
social mediasocial media
social media

The 2020 US Census is only five months away. This nationwide population count, which takes place every 10 years, determines everything from Social Security payouts to how the government spends money on schools and hospitals to where businesses open new stores—and everyone is worried about attempts to skew the count. 

Some groups, like young children, have always been hard to count. But the post-2016 era has provided something new to worry about: disinformation campaigns—run by internet trolls, or foreign powers that want to see congressional representation change—that encourage people to avoid the Census entirely. That would lead to undercounting, which means less money, resources, and power for certain communities. An analysis from the Urban Institute found that Latino and black populations in particular are likely to be undercounted next year, in part because the Trump administration’s attempt to add a citizenship question to the Census has made immigrant communities afraid of answering the survey and then being deported. 

As a result, politicians are becoming increasingly concerned that not enough is being done. Earlier this week, Democratic lawmakers called for Twitter to reveal how it is going to fight against Census disinformation, saying it is “vitally important” that the company follow through on its promises to do better. Similarly, members of the Congressional Asian Pacific American Caucus, a group of lawmakers interested in Asian-American issues,  are planning to ask Facebook, too, for updates to its Census plan.

So far, most of the rumors going around are caused by innocent confusion or people not knowing what the Census does, says Stephen Buckner, assistant director for communications at the Census Bureau. But more targeted approaches will almost certainly come as we get closer to the Census, he says. To prepare, the bureau has been working with Big Tech and community groups on coordinating a way to fight back. 

Here’s what we know so far about how people are trying to protect this important survey. 

What the Census Bureau is doing

First, the Census Bureau runs the rumors@census.gov tip line, which people can email with questions or examples of false information. It updates an FAQ webpage debunking the most common misconceptions. 

But an email account and a list of frequently asked questions is hardly enough to get the word out. The bureau also has a team of about a dozen social-media analysts monitoring the big platforms for rumors, both in English and in other popular languages like Spanish. (They plan to hire dozens more analysts in the next few months.) For instance, a few months ago the bureau learned of a rumor that robbers were posing as Census officials to break into houses. At this point, the bureau updated its FAQ to debunk the claim, and heavily optimized the SEO so that it would be the first page to show up on Google when people searched for the rumor. 

Most important, the bureau reached out to its partners. It plans to partner with hundreds of thousands of organizations—over 300,000 in total, from Fortune 500 companies to philanthropic organizations to neighborhood associations to local businesses. These include 13 agencies focusing on different ethnic groups that might not speak English. The role of these partners is to help reach the people who would never email a Census hotline or navigate to a government FAQ page. Members of these partner organizations are on closed Facebook groups and WeChat, putting them far closer to what’s happening on the ground. That means they can get information to people the Census Bureau would never reach.

In the case of the robbery rumor, the Census started coordinating with these local and national groups to tell everyone it was a hoax. As a result, the number of mentions that the rumor was getting on social media dwindled, and emails about it to the rumors account went to zero. 

What Big Tech is doing

Ultimately, it’s the tech companies that hold a lot of the power to curb Census disinformation. All the big platforms—Facebook, Google, Twitter—have been working with the Census Bureau, but some have been more transparent than others about what, exactly, they are doing. 

So far, Facebook has done the most. The company has already said that it will ban disinformation about the Census. It has an internal team that, like the Census Bureau team, is monitoring content related to this issue using a combination of artificial intelligence and human judgment. The company will release its policy on Census disinformation this fall. 

Google has said it will it will set up a team specifically to counter misinformation. It will also be releasing its plan for the 2020 Census soon, says Buckner. As for Twitter, last year it launched an archive of all potential efforts at foreign interference (of any kind) on the platform. It also plans to set up a team dedicated to protecting the Census. That’s not enough for the Democratic lawmakers who want the company to ban Census disinformation and improve its moderation of suspected bot accounts. “We believe that social media platforms must do more and do better than was done in 2016 to deter and mitigate interference in our upcoming count,” this week’s letter reads. Similarly, Senator Brian Schatz has been hounding tech companies to be more transparent since July. 

It does appear that each one of these platforms is doing more than it was in 2016, but we still don’t know what Twitter and Google’s exact policies toward Census disinformation will be. (Twitter and Google did not respond to questions about when this information would be released.) Nor do we know how exactly their internal teams work, how they’ll detect bots, or what they’re doing to research and preempt new disinfo techniques. 

For academics and researchers who study disinfo, the lack of transparency from these platforms is an old problem. “We need researchers to have access to study what went wrong,” Dipayan Ghosh, a platforms expert and fellow at the Shorenstein Center at Harvard, told the New York Times. “Silicon Valley has a moral obligation to do all it can to protect the American political process.” The more opaque these companies are, the harder it will be for experts to know whether these efforts will be enough. ”Again, it really comes down to two things with the Census: it’s about power and money,” says Buckner. “You get all the political representation based on Census counts, billions of dollars. We only do it once a decade.”