Kriston Capps is a staff writer for CityLab covering housing, architecture, and politics. He previously worked as a senior editor for Architect magazine.
The 2020 count will be the first census to go online, and it faces many threats, from cyberattacks and scam artists to security fears and undercount risks.
With the next census, for the first time ever, respondents will be able to fill out their questionnaires online. This marks a major transition for the count, which guides the apportionment of seats in Congress and the disbursement of hundreds of billions of dollars in federal funds. Giving Americans the option to fill out the 2020 census by laptop or smartphone means dragging Article 1, Section 2 of the U.S. Constitution into the 21st century. For better or for worse.
Worries over the looming census run beyond the typical concerns about underfunding and understaffing (although those are fraying nerves this time around also). Putting the census online opens a Pandora’s box of new risks, including meddling from hackers and scammers, and there’s evidence that vultures are already circling. While the first-ever online census introduces challenges for consumer protection and data security, the greatest threat to the census itself may be inequality—specifically, the digital divide.
“Asking people to fill out a form on their phone is quite different and complicated from asking people to use a social media app,” says Greta Byrum, co-director of the Digital Equity Laboratory at the New School.
Beware the census scams
First, the good news: An overwhelming majority of adults in America know about the census and plan to participate. The brand is strong, according to the Pew Research Center, despite the Trump administration’s failed effort to pin a divisive citizenship question onto the questionnaire. Yet its (quite literal) household-name status also makes it a high-value target for players intent on misleading people.
For example, in October, the Republican National Committee issued a mailer in Bozeman and other areas in Montana that represented itself as a “2019 congressional district census.” The document was really a disguised solicitation for President Donald Trump’s re-election campaign, leading officials in Montana to condemn the “imitation census” as misinformation.
Other census-lookalike forms are designed to lure people to sites where they might be asked for identifying personal information or financial records (even though the census doesn’t ask for these details). “We’ve already seen cases of fake mailers, where they ask people to go to some random URL,” Byrum says. She gives an example of a library patron in Canandaigua, New York, who brought a mimic mailer in to the local library to ask whether it was an official census form.
When the official 2020 census launches next April, the mailers that come to households will direct respondents to a web address and provide them with a unique identifying code. That opens a window for fraud: Bad actors might design convincing spoof sites that look like an official census portal, or they might zero in on (say) a wifi network created for census response by a neighborhood complete count committee. All the usual malware maladies that plague email could be tried against the census, and the same people who are vulnerable to those attacks—older people and those less familiar with online interactions—may be victimized. Other scammers pretend to be Census Bureau staffers and use analog methods of deceit to lure victims into handing over Social Security numbers and other identifying personal information over the phone or at the door. Organizations like AARP have been warning members how to better identify census fraud threats and imposters.
The U.S. Census Bureau’s 2020 data collection push itself could also be a target. When Australia launched its first online census in 2016, it was subject to a distributed denial of service attack that crashed the site, forcing authorities to take it down. Security experts have warned the Bureau that census data will be vulnerable both during transmission and at rest. Earlier this year, officials from the Government Accountability Office testified before the House that the Census Bureau had flagged more than 500 corrective actions to be taken during a cybersecurity risk assessment, nearly half of which were deemed high risk.
“The Census Bureau has been extremely guarded about how they’re building these systems,” Byrum says. “There was a long delay on procurement of these contracts because of the [federal government] shutdown [in 2018–19]. The Census Bureau is really far behind on building the IT systems.”
Delays, budget uncertainties, and lapses in leadership have loomed over the census. While three full trials were planned to test all 50-odd new IT systems for the 2020 Census, the bureau scaled back its preparations to a single dress rehearsal in Rhode Island’s Providence County due to funding shortfalls. “When we went into the end-to-end pilot in Rhode Island in 2018, several of the systems were not completed yet. We haven’t seen them. They haven’t been tested in the field. They’re not going to be tested.”
New technology, and stubborn gaps
Even the system for ensuring that the census reaches hard-to-count households is brand new. For the 2010 census, the bureau hired about 160,000 temporary workers known as “listers” to canvas nearly every block in the nation and generate the agency’s master address file (part of a much larger temporary workforce). As a cost-saving mechanism, the Census Bureau scaled back the door-to-door canvassing operation for 2020. The agency is splitting this task into “in-field” and “in-office” efforts. The latter involves sophisticated data analysis techniques, including machine learning and satellite imaging, to generate a profile for places that have added addresses.
As a result, the Census Bureau is only physically canvassing a quarter of the blocks that the agency covered for the last census. During the single (and only) end-to-end trial conducted of the census, the in-office (digital) canvassing results differed from the in-field (analog) canvassing results for 61 percent of the blocks tested, according to a final internal report on the trial.
“If there’s an over-representation of folks who have internet at home, we don’t know that the nonresponse follow-up systems as it exists is going to be able to identify who has not been counted,” Byrum says. “We’re not sure there’s any corrective mechanism to identify or measure an undercount.”
There won’t be another dress rehearsal before Census Day (April 1, 2020). The 2018 practice run in Providence County did not exactly inspire confidence, according to James Diossa, the mayor of Central Falls, Rhode Island. Outreach was nonexistent. Worse still, Commerce Secretary Wilbur Ross announced the citizenship question in March 2018, midway through the test, adding to the confusion. “There was no information, no advertising, no discussions happening from the Census Bureau around this test trial run,” Diossa told CityLab earlier this year.
Yet outreach is an enormous obstacle for the 2020 census, thanks to the deep divides in the ways that American reach and use the internet. In New York City, for example, more than 917,000 households lack access to broadband at home—29 percent of the city, per a July report on the census from the Office of the New York City Comptroller. This digital divide tracks neatly with existing borders that define marginalized populations, including race, class, and ethnicity. Nearly half of the homes in Borough Park, Kensington, and Ocean Parkway in Brooklyn lack broadband access at home, while on the Upper East Side that figure is just 15 percent.
Broadband access isn’t the only measure of the digital divide. Sticking with New York, about 38 percent of households without internet access at home pay for data on a mobile device. Smartphones may be ubiquitous among communities of color, particularly in low-income communities, but that isn’t a closing of the digital divide, says Maya Wiley, professor at the New School and founder and co-director of the Digital Equity Laboratory. “Try doing your homework on a mobile phone,” she says.
Counting on trust
Black and Hispanic adults, who are more likely to have unreliable access to the internet in the first place, also harbor greater doubts about the census, according to the research from Pew. And no wonder: The Trump administration took great pains to introduce a citizenship question as a way to give an edge to Republicans and non-Hispanic whites. While the effort to add the citizenship question failed, the distrust lingers, and putting the census online raises a whole new category of objection.
“Folks would rather not transmit their data through systems that they neither understand nor trust,” says Melva M. Miller, executive vice president for the Association for a Better New York, a nonprofit that has identified 2020 census outreach as a priority.
Maximizing New York City’s self-response rate is one of her association’s goals going into a census that could see the state as a whole lose billions of dollars in federal funds as well as one or more seats in Congress. Developing messaging to reach hard-to-count communities means coming up with the strategy that’s most likely to reach a trusted figure within a particular demographic, whether that’s a maternal head-of-household, religious leader, or social media platform. And the answer changes wherever you go.
“I was in a conference and sitting on a panel with a woman who is organizing in the state of Arkansas, and she mentioned that there’s been some hesitation among the minority community specifically in Arkansas around filling out the form online. Their preference was to complete the form over the phone,” Miller says. “In our focus groups [in New York], we saw the absolute opposite. Filling out the census over the phone was the least favorite option, even after enumerators knocking on individual doors.”
Public libraries are likely to be the front line in census outreach: That’s where many people who don’t have home access to the internet go to get online. And as trusted arbiters of information across many different communities, librarians have been preparing for the 2020 census for at least two years, according to Larra Clark, deputy director for policy at the Public Library Association (part of the American Library Association). In fact, librarians are already doing some heavy lifting for the 2020 count: They’re helping library users apply for and train for jobs with the Census Bureau, processes that have migrated online with this census.
“Every time we see a government activity move online, whether it’s only online or partly online, every single time we see an impact on our public libraries,” Clark says. “So much about the census is about what public libraries do every day ensuring people have a safe and effective online experience.”
Librarians, faith leaders, and other standard bearers have their work cut out for them. For the 2020 census to succeed, they’ll have to help communities across the country bridge the gulfs of digital illiteracy and lack of accessibility. Success assumes that the government’s untested census technologies hold up to attacks from pirates, hackers, and foreign governments. And if everything works—well, we’ll never know, really. The Census Bureau isn’t conducting a control trial to see how the online census measures up to past efforts.
“If we have a census where a large percentage of the population don’t have faith in the results,”Byrum says, “then we’re in a very poor position when it comes to how we make those decisions or how we litigate going forward regarding these very important issues.”