Introduction
For more than a century, the democratic world has organised itself around a simple principle: one adult, one vote. It is elegant, easy to explain and hard won. But elegance is not the same as effectiveness, and it is fair to ask whether the rule we inherited is the best rule we can design.
This article proposes a modest refinement: every adult still votes, but each vote is weighted by the voter’s result on a standardised, recognised cognitive ability test taken at registration. Nobody loses their voice. The weights simply lean slightly towards those who have demonstrated stronger reasoning ability.
The idea is not new. John Stuart Mill argued for plural voting in 1861, and philosopher Jason Brennan has revived the debate in recent years. What is new is our ability to test fairly, cheaply and at scale, and our growing evidence about how voters and politicians are actually selected.
The problem: we ask voters to judge what many cannot assess
Governments today manage trillion-dollar budgets, monetary policy, energy transitions and national security. Yet decades of survey research show that many voters lack even basic knowledge of how government works.
- Legal scholar Ilya Somin notes that only about 34% of Americans can name the three branches of the federal government (Manhattan Institute).
- Voters routinely overestimate foreign aid spending by five to ten times, when it is about 1% of the US federal budget (The Regulatory Review).
- Public political knowledge has changed very little in recent decades, despite the internet making accurate information easier to find than ever.
Somin argues this is not stupidity but “rational ignorance”: because a single vote almost never decides an election, most people have little incentive to study policy. The result is the same either way. Complex decisions about economics and governance are delegated to an electorate that, on average, cannot evaluate them.
Brennan’s book Against Democracy (2016) builds on this evidence. He argues that political power is power over other people, and that citizens have a right not to have that power exercised incompetently (Springer review).
The proposal: one adult, one weighted vote
Every adult citizen keeps the right to vote. At registration, each voter completes a standardised, recognised cognitive ability test. Their vote then carries a weight equal to their score divided by 100.
Because IQ scores are normed to an average of 100 with a standard deviation of 15, roughly two thirds of voters would carry a weight between 0.85 and 1.15. This is a gentle tilt, not a revolution. No one is silenced, no one gets double votes, and the majority of the electorate sits within 15% of the average.
That modesty is the point. The goal is not to hand power to an elite, but to nudge outcomes, and the incentives around them, slightly towards reasoned judgement.
Testing: accurate, simple and ready for scale
The obvious objection is logistics. The traditional gold standard, the WAIS-IV, takes 60 to 90 minutes for its core subtests and is administered one on one by a licensed psychologist (Pearson). In the US, a full assessment typically costs US$1,200 to US$2,500. Testing a whole electorate that way would be impossible.
But that is yesterday’s model. Psychometrics has already moved on.
Computerised adaptive testing (CAT). Built on item response theory, an adaptive test selects each question in real time, offering harder items after correct answers and easier ones after mistakes. Research shows this cuts testing time and improves accuracy for the same number of questions, while large item banks improve security (PLOS ONE). A study of adaptive cognitive tests found validity broadly comparable to traditional paper forms (Studia Psychologica).
Proven short formats. Even a three-minute speeded reasoning test has shown excellent reliability (0.96 to 0.99) and a substantial correlation with full test batteries (Intelligence, 2023). Too short for a voting weight on its own, but proof of how far efficiency can go.
Where AI adds value. Modern AI can help generate and calibrate large, fresh item banks, detect irregular response patterns, deliver tests in many languages and formats, and adapt accessibility features for people with disabilities. Combined with secure, supervised testing centres, the same infrastructure many countries already use for driver licence knowledge tests, a 30 to 45 minute adaptive assessment is realistic.
A workable model would look like this:
- Register to vote and book a supervised test slot, online or at a government service centre.
- Complete an adaptive test drawn from a large, independently audited item bank, in your preferred language.
- Receive your weight, valid for a fixed period (for example, 10 years), with a free right to retest.
- Results are held by the electoral commission, encrypted, and never shared with employers, insurers or parties.
Once built, the per-person cost would be a small fraction of a clinical assessment, and the system would sit inside existing electoral administration.
Learning from history: why literacy tests failed
Any proposal to test voters must confront the United States’ experience with literacy tests. Supporters claimed they ensured an informed electorate. In practice they excluded immigrants and the poor, and across the South they were used to stop African Americans registering (Smithsonian). The Voting Rights Act of 1965 suspended them in many states, and a 1970 amendment banned them nationwide (Britannica).
That history is a warning, not a verdict. The tests failed for identifiable reasons, and each one points to a design safeguard.
Free testing, free retests, and investment in education alongside the reform
Two further protections would make the system harder to capture:
- A floor and a ceiling. Capping weights between 0.7 and 1.3 (equivalent to IQ scores of 70 and 130) means no vote can count for less than 70% or more than 130% of the average. It guards against manipulation at the extremes and makes clear that no voter’s influence can be reduced to a token.
- Independent governance. The test should be run by the electoral commission, not the government of the day, with regular bias audits checking that no question performs differently across groups of equal ability.
There is also an uncomfortable truth to face openly: average test scores differ between groups, largely tracking differences in education and economic opportunity. A fair test can still produce unequal outcomes. That is why this reform should be paired with, and ideally preceded by, serious investment in education, so that the system rewards reasoning rather than privilege.
The real prize: changing who is drawn to politics
Weighting votes does not directly test candidates, and it does not measure emotional intelligence. Its larger effect is indirect: it changes what it takes to win, and so changes who decides to run.
Economists have long worried about this. Standard models suggest that free-riding and lower opportunity costs give less competent people a comparative advantage in entering politics (Quarterly Journal of Economics abstract). If capable people see politics as a contest won by slogans, many will stay away.
But the evidence also shows selection can be shaped. The landmark study Who Becomes a Politician? (Dal Bó and colleagues, 2017) used military conscription test data for Swedish men to show that politicians there are significantly smarter and better leaders than the population they represent. Elected politicians from every social background scored higher than those who ran and lost (Berkeley Haas). Similar positive selection has since been found in Norway, Denmark and Finland.
The lesson is that political selection responds to what voters reward. Change the reward, and you change who applies. A weighted vote works on both ends of the talent pool:
- It draws capable people in. When a larger share of the electoral weight sits with voters who scrutinise policy, expertise and competence become assets on the campaign trail. Capable people who currently dismiss politics as a contest of slogans see a real path to winning.
- It pushes incompetent candidates out. Candidates who rely on emotional provocation, empty promises or personality alone find their message loses ground at the margin. Fewer of them win, and fewer see politics as an easy career.
- It changes party behaviour. Parties preselect whoever they think can win. If competence wins more often, parties have a stronger reason to put forward competent candidates, and a weaker reason to back unqualified ones.
Over several election cycles, these small shifts compound. The pool of people seeking office changes, and so does the quality of those who reach it.
Mill made a version of this argument himself: a system that gives weight to demonstrated competence tells society that intellectual effort matters (Mind). Nobody has tested this at national scale, so the size of the effect is unknown. But the direction of the incentive is clear.
The strongest objections
A serious proposal should meet its best critics head on. These are the arguments any reader, and any legislator, will raise.
- Equality is the point, not a bug. Many democratic theorists argue that equal votes express equal moral standing, and that this is valuable in itself, not just a route to good outcomes. Critics of Brennan make exactly this case (Public Seminar). Response: every citizen keeps a vote; the proposal adjusts weight, not membership.
- Voting is about interests and values, not only competence. A person’s vote expresses what matters to them, their wages, their health, their community. Intelligence does not make anyone a better judge of another person’s interests. Response: the weighting is deliberately small, so every group’s interests remain strongly represented.
- Political wisdom is widely spread. Critics of Mill noted that the knowledge relevant to governing is dispersed across society, not concentrated in one measurable trait (PolSci Institute). IQ measures reasoning ability, not political knowledge, judgement or character.
- Capture risk. Whoever controls the test controls the weights. History shows voting rules are often bent by those in power. Response: independent administration, open methodology, audits and caps.
- Participation may fall. People told their vote counts for less may disengage, and scholars have argued that Mill’s scheme would have harmed participation (Ethic@).
- Legal barriers. Equal suffrage is embedded in law in most democracies. In the US, testing as a condition of voting has been banned since 1970, so this reform would require major legislative or constitutional change almost anywhere.
These objections are substantial, and some are fundamental rather than technical. Any honest advocate should expect to win the argument slowly, if at all, and should welcome the debate.
Conclusion: a small change with a long reach
Many modern elections are decided by razor-thin margins, where a shift of one or two percentage points changes who governs. In that world, small design choices matter. A gentle weighting of votes, most sitting between 0.85 and 1.15, would not overturn democracy. It would add a quiet, persistent pressure towards reasoned choices, and over time towards more capable candidates.
The case rests on three claims: voters are often poorly equipped to judge complex policy, modern adaptive testing makes fair assessment feasible at scale, and political selection responds to what voters reward. The first two are well supported by evidence. The third is plausible but untested.
That suggests a sensible path. Rather than a sudden national overhaul, the idea could be trialled where the stakes are lower: in advisory ballots, citizens’ assemblies or internal party preselections. If it improves decisions without eroding trust or participation, the case for going further grows. If it does not, we will have learned something important about why one person, one vote has endured.
Democracy has never been static. Who votes, and how, has been redesigned many times. It is worth asking, openly and with evidence, whether the next redesign should reward not just participation, but reasoning.