Michelle R. Smith/AP

Analysis: It’s Time to Worry About the Census

From cybersecurity issues to administrative problems to a legal drama over a possible citizenship question, the decennial headcount is in trouble.

Nobody will write songs about the census. Among the fabled pillars underpinning the country’s democracy, the great American headcount is often relegated to a dusty corner. In the nine interstitial years between each tally, analysis and development of a more perfect instrument take place mostly hidden from public view. There have been only 22 U.S. censuses—Presidents Reagan, Kennedy, Lincoln, and Jefferson never administered one—but the rarity of the event has not assigned it a special blue-moon-like significance among the public. For most people, the census is a vague, decennial annoyance, nothing more.

But the census is vital to the country’s functioning. It’s not just a count of all households or a measure of American characteristics. It’s also an augur of political, economic, and cultural forces—a predictor and an allocator of power. In times of social upheaval—between political parties, whites and nonwhites, urban and rural areas, economic elites and the working class—the census can function almost like an umpire. And today, when each of these intertwined conflicts is escalating, the incentive and ambition for working the ref are greater than they’ve ever been.

That’s why the 2020 census is in real danger, though it’s still more than a year away. There has been a cascade of bad news recently for the census itself and the agency that administers it, the Census Bureau. The major drama still enveloping the bureau is the Trump administration’s proposed addition of a citizenship question to the next survey. Given the anti-immigrant rhetoric and policies of this administration, the proposal has generated controversy: Some fear it will be used to intimidate immigrant respondents or punish districts with high noncitizen populations.

That controversy has spilled over to the courts, too. Last Thursday, in a ruling allowing 18 states to continue with a lawsuit against the Commerce Department, U.S. District Judge Jesse Furman of Manhattan essentially confirmed the worries of the proposal’s opponents: He doubted Secretary Wilbur Ross’s rationale that the citizenship question is necessary for enforcement of the Voting Rights Act, and he concluded that there was “evidence suggesting that Secretary Ross’s stated rationale for adding the question is pretextual.”

The citizenship question is only the start of the Census Bureau’s issues. It has faced constant budget reductions and shortfalls, hasn’t been able to field the kind of end-to-end testing this year that it needs to, has missed its marks on some quality-control processes, and has been embroiled in smaller controversies over a rumored proposal to ban noncitizens from working as census-takers. All of these administrative issues have arisen as the Senate gears up to begin confirmation hearings for a new Census Bureau director. Though the post is not usually high on partisans’ collective radar, Trump’s appointmentof former Bureau of Justice Statistics Director Steven Dillingham has become a hot-button issue, one that could further complicate the morass of problems for the bureau to sort out before 2020.

If all these complications weren’t enough to deal with, the 2020 census also faces a new and profound security threat. The 2016 election saw the major vulnerabilities of American voting infrastructure on full display. Russian hackers managed to infiltrate registration systems and voting machines, and experts uncovered other holes across a broad range of systems that nefarious actors could potentially exploit. And the American people proved as eminently hackable as the machines on which they rely: Much more so than the attacks on U.S. hardware and software, Russia-sponsored actors found success in manipulating the fears and behavior of voters themselves. Both these sets of exploits could be predecessors to a strategy to undermine the 2020 census.

As security experts Joshua Geltzer and Matthew Olsen wrote in The Washington Post last week, the census and its respondents are prime targets in the lead-up to 2020. The upcoming survey is slated to be the first to which people can respond online. That new format introduces a ton of new statistical and methodological issues to work through, and it also presents a target for infiltrators. The census database could be used for identity theft or foreign microtargeting purposes, and the data themselves could be prone to tampering, Geltzer and Olsen argue.

What’s more, even if the database and its underlying information prove secure, foreign actors could try to “undermine Americans’ trust in democratic institutions” by using Facebook campaigns and social engineering to convince people to refrain from taking the census or to mistrust its findings. It’s possible they could do that by merely exacerbating real controversies swirling around the census and the citizenship question, Geltzer and Olsen write.

All of these potential pitfalls with the 2020 census can be broken down into three main dangers: poor quality in the underlying data, data tampering, and the use of data—good or bad—for anti-democratic aims. These deficiencies could each be used for political purposes by domestic groups or foreign actors with interests in American elections. They could also be used purely to destabilize the country.

As the ongoing furor over the citizenship question shows, these three dangers can play off each other. For example, the main objection from immigration activists on the citizenship question—that noncitizens would refuse to take the census with the question included for fear of reprisal—is in part a data-quality concern: An undercount of noncitizens will bias the census data and make it unreliable. But it’s also a concern based on the history of racist leaders wielding administrative data against immigrants, and on the potential for the GOP to use the census to punish “sanctuary cities” or deport people. Additionally, data that undercount the total number of noncitizens—whether through non-response bias or tampering—also hamper districts with large numbers of them through the improper allocation of congressional representation and funds.

So far, the most prominent census-related debate has been the growing conversation about gerrymandering. The enumerated purpose of the census in the Constitution is to use the data gleaned to redraw congressional districts and reallocate representation to fit the population. Over the past few decades, as politics and demographics have become increasingly intertwined, the institutional incentives to game that system have mounted. Especially as Republicans have embraced their status as a party for white men, they’ve become more and more adept at using the census, especially the redistricting process, in order to maintain partisan advantage. But that advantage has been built mostly with the census data already in hand—influencing the data itself, perhaps by biasing it against noncitizens, would add another potent weapon to an arsenal that’s already proven its dominance over American electoral politics. And that is to say nothing of the possibility that between foreign and domestic interventions in data quality, the census could be rendered unusable or unreliable for those political processes altogether.

Without a census, there is no American democracy. Conversely, the census, like all other institutions upholding that democracy, can be warped and weakened to promote partisan aims or weaken the promise of fair representation. In a country where the stalwart institutions of democracy are fraying, where white-nationalist agitation against the demographic strength of people of color is only growing, and where anti-immigrant rhetoric is chained tightly to political fortunes, now is the prime time for warping and weakening. The fear among many voting-rights and immigration activists is that the 2020 census will be trouble—but 2020 might only be the start.