On Wednesday, the House is expected to pass the American Rescue Plan Act and send it to President Biden to sign. The robust relief package includes a broad set of measures to strengthen the faltering social safety net: a remarkable child tax credit, enhanced unemployment insurance, billions to reopen schools and the marquee policy, a $1,400 check sent to most U.S. residents making less than $75,000 a year.

Politicians across the political spectrum have traditionally feared this kind of broad policy approach. In 2021, it was primarily Republicans like Sen. Susan Collins (R-Maine), who expressed strong concern that the checks are not targeted enough. What if they go to the “wrong” families? Who is poor enough to need support from the welfare state? These moralized questions are based on the assumption that the government can sort people effectively; that it has the technological capacity to know the population well enough to target relief with incredible precision.

Yet, despite the promise of new tools to track citizens, the U.S. government’s ability to administer individualized social services has always been remarkably faulty. People remain hard to know. We move, have children, fall sick, get promotions, take in relatives, elope, look for side jobs and make a messy digital paper trail. Trying to match each citizen with the “right” amount of support is a technical nightmare marked by dangerous errors. Technology has never been a silver bullet; it misses all the time. Design oversights and bugs cause glitches, but just as often with government tech, glitches are rooted in policy failures.

Since the New Deal, policymakers have maintained faith in the federal government’s technical ability to sort out who is and is not deserving of public support. President Franklin D. Roosevelt’s secretary of labor, Frances Perkins, remembered her excitement when her team learned in 1937 that IBM punch card machines could be used to track individualized payroll data. “I’ll never forget how startled I was when I saw the first IBM machine,” she remarked years later. She added, “It’s an amazing convenience!” By the 1950s, the Social Security Administration had its own general-purpose computer. Historians have paid close attention to the pride with which Americans made themselves machine-readable to take part in government services (including Social Security number tattoos).

Despite growing fears of federal surveillance, the postwar U.S. government remained the primary consumer of computing resources and computers changed how the government saw the population. By 1980, the government-eye view of the citizen resembled a string of databases that formed what historian Dan Bouk has called “data doubles,” digital representations of each citizen powered by unified data sets. Your data double determines if you are eligible for Social Security benefits, welfare checks, food stamps, Pell Grants or a home loan from the federal government. The policy choice to depend on data doubles has created a host of problems.

Seven years ago, Americans saw up-close what happened when the government overburdened a technical system to sort citizens for individualized health-care plans. After the passage of the Affordable Care Act in 2010, the government built the website HealthCare.gov to take in personal information and match a verified legal resident with the right government subsidies and relevant private insurers. This was all done to create a set of opt-in individualized plans from which users could choose.

Yet, when Americans tried to log on to the site to craft an insurance package, the system crashed. Engineers, parachuted in to fix the site, have described the difficulty constructing data doubles. How would the consumer credit agency Experian database match with the government subsidy requirements? Would tax data integrate with log-in information? The enrollment platform broke under these demands. The task of tech experts, who eventually patched the website, was not simply to solve the technical challenge of interoperability across different databases. They were asked to solve inherent political problems built into the policies of individualized public services.

A broken Web interface was bad, but, as President Barack Obama himself argued, failed tech undermined a central argument of his administration that government itself worked. But what if Obamacare asked technology to do the wrong thing? Systems like HealthCare.gov falter because they encode a specific version of what it means for government to function well, that is to individualize targeted services to help only those who need help and no others. If what it means for government to work well shifts away from a fixation on sorting the deserving and undeserving, the technology on which the work of government relies could be contracted to ensure the fast and direct receipt of support.

The pandemic has demanded just such a shift toward a broad-based approach. The stimulus checks are still capped, phased out above certain income thresholds, but direct financial relief issued with broad strokes stands out as a notably rare form of policy in the United States. The law strips out the usual opt-in, form-heavy, targeting apparatuses (and the reliance on online enrollment) that have come to define U.S. social policy. Treasury Secretary Janet Yellen, no stranger to navigating technocracy, knows the amount of work that goes into distributing checks. She warned against over-targeting relief: “The truth is there are pockets of pain that go beyond what can be reached in those highly targeted ways.” Yellen knows the government both cannot and should not achieve the type of perfect targeting imagined by those who don’t actually have to make the system work.

Technology can help make American Rescue Act’s mandate of widely accessible public support a reality. Just because computerization has enabled a highly individualized means-testing approach to policy in the past doesn’t mean it has to in the future. The package includes billions for technological modernization at the state and federal level. Updating the federal government’s technological infrastructure can enable a broad approach to public goods. We will know that these technical systems work not when public services are perfectly targeted but when they widely distributed and quickly received. When necessary financial support arrives in your bank account through direct transfer or is correctly routed by the U.S. Postal Service, then the machinery of government will be working.