Quantum Voting Study Tests When Noise Changes the Winner

What if there was an election where ballots were not static? Each voter’s preference would exist as a spectrum of possibilities until measured. In some scenarios, voters are quantum-linked, causing a single measurement to solidify a group’s collective choice. Now, apply this to a real, noisy quantum chip and ask a critical question: Does the correct candidate still win?

Mostly, yes. Until it doesn’t.

This is the conclusion of a study published in Scientific Reports by a team from Bar-Ilan University, the European Institute of Science in Management, and Chapman University. They subjected “quantum majority rule” (QMR) to simulations and actual IBM superconducting processors to observe how physical errors influence a collective decision. Though the concept suggests futuristic democracy, the reality is a stress test for quantum error correction, presented as an election.

Understanding quantum majority rule

QMR, a concept developed by Zhengfeng Bao and Nicole Yunger Halpern, encodes each voter’s ranking as a quantum state. It then builds a group preference distribution through a combination of quantum and classical processes.

The underlying motivation addresses a long-standing challenge in economics. Decades ago, economist Kenneth Arrow demonstrated that no voting system can convert individual preferences into a group decision while satisfying all reasonable fairness criteria simultaneously. This can lead to voting cycles, where voters collectively prefer A over B, B over C, and C over A, resulting in no clear winner. QMR redefines these mathematical conditions using quantum analogues, effectively circumventing a quantum version of Arrow’s theorem, an elegant theoretical solution.

This particular study didn’t aim to re-prove Arrow’s theorem. Instead, researchers investigated what happens when this precise mathematical framework encounters imperfect hardware. Their setup was intentionally small: five voters, three candidates, and six possible strict rankings. The algorithm assessed pairwise preferences, constructed a directed graph illustrating who beats whom, and used Tarjan’s algorithm, which is a standard graph-processing method, to detect cycles. A Condorcet winner (the candidate who wins every head-to-head matchup) served as the benchmark for evaluating whether noise altered the outcome.

The winner persisted as the underlying data collapsed

A surprising finding emerged: The team examined a case where candidate C won by narrow margins and then significantly increased readout error, a specific glitch where a quantum computer misinterprets a qubit during measurement.

The full ranking distribution progressively deviated from its ideal form. Yet, candidate C continued to win. This held true even with a readout-error probability of 0.4, which is remarkably high. Only at 0.5, essentially a coin flip for every measurement, did the result unravel, with agreement to the classical winner plummeting to approximately 2 percent. Runs on an IBM device model and actual IBM hardware remained within this stable zone.

This indicates that the winner is not a reliable indicator of system health. A voting rule can consistently name the same victor as its underlying foundation quietly deteriorates. For this reason, the team meticulously tracked three separate metrics: how often the QMR winner matched the classical benchmark, how frequently the winner changed across repeated runs, and the extent to which the entire distribution had shifted.

The researchers then moved beyond hand-picked, favorable examples. When generating random electorates, some configurations failed at a readout error of just 0.01. One profile showed only 43 percent winner agreement. The difference wasn’t the chip; it was the structure of the electorate. A voting profile already close to a majority-cycle boundary is inherently unstable, and even a small error can disrupt it. On IBM’s 156-qubit ibm_marrakesh processor, an almost-cyclic case maintained 100% agreement, while an exact cycle dropped to 70%. Quantum error correction is a central bottleneck in the field, and this study reveals the same fragility through a unique lens.

Entanglement’s ephemeral impact

The team also conducted a supplementary experiment to assess whether entanglement influences voting behavior. They used a simplified system, inspired by a different proposal rather than the complete QMR.

Groups of voters were placed into GHZ states, which create strong correlations such that measuring one voter locks the entire block into a single choice or its opposite. Across 10,000 small voting rounds, the entangled groups eliminated ties under ideal conditions, as the block resolved collectively instead of fracturing. Individual voters appeared locally identical in both setups. The correlations, however, altered the group statistics.

The researchers candidly described the subsequent outcome:

“The entanglement effect largely disappeared when the researchers considered large populations in which only limited groups of voters could be entangled”

Introducing local bit-flip noise caused the GHZ behavior to revert towards ordinary randomness. At a bit-flip probability of 0.5, it vanished completely.

Scope and limitations of the study

The team is transparent about the study’s limitations. Five voters and three candidates constitute a small scale, and scaling up drastically increases qubit requirements, making a real election on current noisy machines impractical. They primarily modeled readout errors, omitting crosstalk, leakage, and deep circuits. The entanglement experiments were simplified demonstrations, not the full QMR framework, and therefore don’t inherit its Arrow-beating properties.

Crucially, there’s no quantum advantage demonstrated here. The core QMR distribution was computed using classical machines. The quantum processors served as sampling devices to measure how hardware errors distort a known outcome. Furthermore, this isn’t a secure voting protocol, lacking ballot secrecy, voter authentication, and resistance to coercion.

The true value of this work lies in the bridge it constructs. It connects abstract quantum social-choice theory to physical hardware and suggests encoding voting registers into error-correcting codes to test the resilience of protected logical qubits. This is a practical and modest contribution. The election framing is merely a metaphor, and the underlying substance is an experiment in noise.