Skip to content

Tech and Democracy / Elections

Abstract

Digital technology was supposed to be good for democracy, more information, more participation, more accountability. Instead it became one of democracy’s central battlegrounds. Two distinct threats converged: the manipulation of voters through data-driven microtargeting and disinformation, dramatized by the Cambridge Analytica scandal; and the integrity of the vote itself, contested through hackable voting machines, foreign interference, and (most corrosively) false claims of fraud that erode trust in legitimate results. This article traces how the optimistic “e-democracy” of the 1990s gave way to the anxious politics of the algorithmic age, where the contest is no longer just for votes but for the shared reality in which voting makes sense.

Computers on Election Night

Computers entered elections through television. On election night, 4 November 1952, the CBS presenter Charles Collingwood asked a flashing console in the New York studio, “Can you say something, UNIVAC?” The real UNIVAC in Philadelphia had in fact already predicted Dwight Eisenhower’s landslide, but the team there did not dare pass on the forecast; two hours later NBC’s smaller Monrobot announced it first. Britain followed: the BBC rented an English Electric DEUCE for the general election of 26 May 1955, used apparently only by radio, and on 8 October 1959 put an Elliott 402F in the television studio while ITV used a DEUCE. West Germany’s first computer projection came at the federal election of 19 September 1965, when an IBM 1620 run by the infas institute for WDR named Ludwig Erhard the winner two and a half hours after the polls closed. From 26 to 28 February 1971 WDR also experimented with computer-tallied viewer voting in ORAKEL, three evenings of live discussion on the environment during which callers rated statements from 1 to 5 (527 called in for the first set of questions, more than a thousand on the second evening); its designer Helmut Krauch published Computer-Demokratie in 1972, hoping for a direct electronic democracy.

From E-Democracy Optimism to Algorithmic Anxiety

The 1990s and 2000s brimmed with optimism that the internet would strengthen democracy: open government data, online deliberation, lower barriers to organizing, and an informed citizenry no longer dependent on gatekeeping media. Howard Dean’s 2004 and especially Barack Obama’s 2008 campaign showcased the internet as a tool for grassroots mobilization and small-dollar fundraising, technology as democratic empowerment. The Arab Spring (2011) briefly seemed to confirm that connectivity favored people over power.

The mood inverted around 2016. The same tools that mobilized voters could manipulate them; the same platforms that informed could misinform; and the personal data that powered convenient services could profile and target citizens with unprecedented precision. Democracy’s relationship with tech shifted from hope to alarm.

Cambridge Analytica: Data as a Political Weapon

The emblematic scandal was Cambridge Analytica (2018). The firm had obtained data on up to 87 million Facebook users, harvested via a personality-quiz app that, under Facebook’s permissive pre-2015 API, also scraped the data of quiz-takers’ friends without their consent. Cambridge Analytica claimed to use this for “psychographic” microtargeting (tailoring political ads to individuals’ psychological profiles) in the 2016 Trump campaign and the Brexit referendum.

Two truths sit in tension. The scandal was real and consequential: it exposed how loosely platforms guarded user data, triggered Facebook’s largest crisis, led to a record $5 billion FTC fine (2019), and accelerated privacy law. Yet the efficacy of psychographic manipulation was likely overstated, much of Cambridge Analytica’s “mind-control” claim was self-promotional salesmanship, and rigorous evidence that psychographic targeting actually swung votes is thin. The lasting lesson was less “they hacked our minds” than the structural revelation that the surveillance-capitalist data infrastructure built for advertising is natively a political-manipulation infrastructure, available to any campaign or foreign actor who wants it.

Microtargeting and the Fragmented Public Sphere

Beyond any single firm, political microtargeting changed the structure of campaigning. Broadcast-era politicians delivered one message to a mass audience, visible to all and checkable by opponents and press. Targeted digital advertising lets campaigns deliver different, even contradictory messages to different micro-audiences, invisibly: “dark” ads seen only by their targets. This corrodes the shared public square that democratic accountability depends on: you cannot fact-check or rebut an argument you never see. It also enables precise voter suppression messaging aimed at discouraging specific groups from voting. The EU and others responded with rules forcing disclosure of political-ad funding and targeting (the EU’s 2024 Political Advertising Regulation), and several platforms restricted or banned political microtargeting.

Election Integrity and the Machinery of Voting

The second front is the integrity of the vote itself. Mechanical voting machines are old: lever machines from the U.S. Standard Voting Machine Company (founded 1900) were used in a sixth of US districts in the 1930s presidential elections, and punched-card Votomatic systems followed in the 1960s. Germany’s electoral law of 1956 already allowed approved counting devices, but it took until 1999 for Dutch Nedap voting computers to be used, first in Cologne; in 2005 about 1,800 of them served some two million voters in the federal election. On 3 March 2009 the Federal Constitutional Court ruled their use unconstitutional because voters could not check how their votes were recorded and counted, and the Netherlands had already dropped electronic voting in 2007. Computerized voting raised genuine security concerns elsewhere too: after the 2000 US Florida “hanging chads” debacle pushed states toward electronic machines, security researchers repeatedly demonstrated that direct-recording electronic (DRE) voting machines could be hacked or could fail silently, with no way to audit a paperless result. This produced a hard-won technical consensus: voter-verified paper trails and risk-limiting audits are essential, software-only results are untrustworthy because they cannot be independently verified. (Notably, this is also why serious experts overwhelmingly reject internet/blockchain voting: it sacrifices the auditability and secret-ballot guarantees that paper provides.)

Foreign interference added another layer: the 2016 US election saw both an influence operation (the IRA troll farm) and intrusions into voter-registration systems and a major party’s email, establishing election infrastructure as a national-security concern (the US designated it “critical infrastructure” in 2017).

Dead End: The “Stolen Election” and the Collapse of Shared Reality

The most dangerous failure mode is not a technical vulnerability but a political-epistemic one: the weaponization of distrust in elections, culminating in the “Big Lie”, the false claim that the 2020 US presidential election was stolen through fraud, which fueled the January 6, 2021 attack on the Capitol.

Here the threads converge into democracy’s deepest tech-enabled crisis. Ironically, the 2020 US election was, by expert consensus and exhaustive audits, among the most secure in history, yet a coordinated disinformation campaign, amplified by engagement algorithms and partisan media, convinced tens of millions that it was fraudulent on the basis of no credible evidence. This is the failed end-state of the whole story: when a society loses a shared standard of evidence, election security becomes irrelevant, because perceived legitimacy decouples entirely from actual integrity. A perfectly run, fully auditable election can still be rejected by those who have decided in advance not to believe it.

The lesson is that democracy’s vulnerability to technology is not mainly about hacked machines (those are auditable and fixable) but about the erosion of the common informational ground on which democratic consent rests. You can secure the ballot box completely and still lose the republic if citizens no longer share a reality in which the count means anything. Defending elections turned out to require defending not just infrastructure but truth itself, a far harder and more political task than any cybersecurity patch.

📚 Sources