On October 14, the Quaest institute published its first poll ahead of Brazil’s presidential runoff, scheduled for October 25: Flávio Bolsonaro — 46%, Lula — 44%, a statistical tie. The poll’s most interesting number lives in the fine print: nine percent of voters say they will either spoil their ballot or not show up at all. In a country where voting is compulsory, and skipping it can leave you without a passport, a paycheck, or university enrollment, nine percent is not a margin of error. It is a movement.
What those nine percent choose looks like a logical mistake. A spoiled ballot goes to no one: it takes no vote away from Lula, Bolsonaro, or the candidate for federal deputy in your district. It vanishes from the calculation altogether. A rational voter either votes for someone or stays home. A Brazilian gets up on Sunday morning, goes to the polling place, stands in line, and votes against everyone—knowing in advance that this system has no “against everyone” box.
It is worth figuring out what a spoiled ballot actually does for at least one reason: in 2022, null votes were nearly twice the winner’s margin. The system no one has managed to break reliably records dissent—just not where it is sent.
The electronic ballot box, the urna eletrônica, offers voters a short script. Candidate photos appear at the bottom of the screen, a keypad above them. You can enter a candidate’s number and press “confirma.” You can press the “branco” key—and the vote will be counted as blank. Or you can enter a number that does not exist: 00, for instance, or any combination that matches no registered candidate, and confirm your choice. The machine will politely warn you that the number was not found, but it will record the vote as null, nulo.
The difference between branco and nulo is the difference between silence and a line delivered into the void. A blank vote says, “I abstain from choosing”: no declaration, no gesture, a simple surrender of position. A null vote is an active gesture: to cast one, you have to know that the system will reject an invented number, and enter a nonexistent one deliberately. In its glossary, the TSE explicitly distinguishes the terms: a null vote is “recorded exclusively for election statistics.” In other words, the only meaning the system is prepared to assign this act is the record of dissent itself.
In the designers’ ideal scheme, it is a neat ledger of discontent: the system registers dissent itself—and nothing else.
Compulsory voting seeped into Brazil long ago, in several stages. As early as 1846, the law imposed fines on those who missed voters’ assemblies; in 1873, deputy João Alfredo Correia de Oliveira proposed extending the requirement to all elections—the parliamentary commission replied at the time that the principle was “not new to our legislation” and already existed in municipal practice, but the bill did not pass. The 1932 Electoral Code finally made voter registration and voting compulsory, fully and definitively. In the TSE glossary, it is definitivo—forever.
The consequences of skipping an election are laid out in the same Code with accounting-grade dryness. A fine—from 3 to 10% of the regional minimum wage. Fail to pay it and fail to provide an excuse, and you lose the ability to obtain a passport and identity card, apply for or take up a public-sector post, receive a paycheck from a government institution beginning in the second month after the election, participate in public procurement, renew enrollment at public educational institutions, or obtain documents at diplomatic missions. Miss three elections in a row and your voter registration is canceled—and each round counts as a separate election: you can physically “miss” an electoral cycle by skipping one single runoff. The excuse can be submitted through e-Título, the electoral justice app, or on paper within 60 days after the round.
Against that backdrop, nine percent promising to spoil their ballots are not acting like capricious people, but like people who honestly make it to the ballot box and refuse the final step. They are required to show up. They are not required to choose. A null vote is that “no.”
Why a spoiled ballot “does nothing” can be explained by a single formula. Article 211 of the Electoral Code: a candidate is elected if they receive an absolute majority of votes, “excluding blank and null votes from the count.” Null votes are thrown out not of turnout, but of the denominator. The denominator is then rebuilt from valid votes alone, and the more nulls there are around it, the smaller the share of the population that actually decides the outcome.
Hence the paradox that usually slips past notice: a null is not a vote “against everyone.” It is a proportional gift to everyone who cast a valid ballot. If, out of one hundred people, twenty spoil their ballots and eighty split sixty-twenty, the winner gets not sixty percent of the country, but seventy-five percent of the denominator. Your null does not shout “no”—it silently turns up the volume on someone else’s “yes.”
The second layer of protection is Article 224. Only the Electoral Justice can void the result and call a new election, and only if a court invalidates more than half of all votes cast. The key word is “invalidated,” anulado, not nulo. The TSE spells this out every election season: null votes “are not transferred to the winner and do not cancel elections”; they are “merely a record of voters’ dissatisfaction.” A null is a voter’s decision; invalidation is a judge’s verdict. The system digests the first in silence; the second happens in cases of outright fraud—and since 1996, when the electronic ballot box came into use, there has not been a single presidential-scale rerun for that reason.
Now the figures from 2022: the runoff, official TSE tally. Lula — 60 345 999 votes (50.90% of valid votes), Bolsonaro — 58 206 354 (49.10%). Difference: 2 139 645. Null votes: 3 930 765. Blank votes: 1 769 678. Add the last two figures and you get 5.7 million people who came to the country’s biggest election and declined to choose. Invalid votes were nearly twice the winner’s margin. The 2022 victory was already smaller than the volume of the protest null—only the denominator says nothing about it.
The null vote in Brazil has never been monolithic. In 1989, in the first direct presidential election after the dictatorship, the runoff between Collor and Lula produced 4 094 692 invalid votes—5.83%. At the time, it was the gesture of people offered a choice between two candidates neither of whom was their choice: Collor embodied young business populism, Lula the labor movement. That voter was not voting between candidates, but against the entire vertical order on offer.
Thirty-three years later, in 2022, the share of invalid votes (null + blank) in the runoff was more modest—around 4.6% of all votes cast—but the absolute number rose to 5.7 million. Campaigns for voto nulo surfaced in both camps during that cycle—and each time, too late to change anything.
The current Quaest poll adds the freshest line to the story. Nine percent of the current electorate say: I will spoil my ballot or I will not show up. And it matters to read the figure precisely: this is not “9% will spoil their ballots”—it is the combined category branco/nulo/não vai votar, which also includes potential no-shows. The transfer of votes from this group, according to the same poll: 43% of these voters say they will vote for Lula in the runoff, 38% for Bolsonaro, 17% will again take no part. The protest null is not a third force, but noise with a tint: it sounds slightly different in both directions, and at full strength in neither.
There is a separate layer—the trend this poll line completes. 2026 has already set a first-round record: 33.4 million people did not turn out, 21.08%—the highest figure since 1998. Another 2.3 million pressed “branco,” 3.7 million spoiled their ballots. TSE president Cássio Nunes Marques announced that the 20% absenteeism threshold is above what the Electoral Justice considers healthy. With every cycle, the compulsory-turnout system captures more and more people who formally exist within it while participating in it less and less.
A spoiled ballot has a third quality, and it is the most revealing. Because a null’s real weight is zero, it is an ideal target for myth: no one can test “what if they actually decide things”—but you can fantasize endlessly in an empty space. Brazilian fact-checker Aos Fatos analyzed WhatsApp chats from the first half of 2022: of 405 messages about null votes, blank votes, and absenteeism, 143—35%—contained false or distorted claims. Of those, 51 chains were about “erased millions”: supposedly, additional counts show that blank and null votes were actually valid, but electronic machines “erased” them. Even the old 2018 fake—that very “7.2 million invalidated”—is alive and still circulating through chats, although official data show that these votes are invalid by the voters’ own choice.
The outline of these myths is always the same: the people’s true will is hidden in invalid ballots, and if they were counted honestly, someone else would have won. The truth is that the count is honest—it simply ruthlessly excludes empty ballots from the calculation, exactly as the 1965 Code prescribes. Disinformation does not latch onto the technology’s dark corners, but onto the one place where the system candidly admits: we register your gesture, but we do not count it. An admission of design weakness becomes fuel for theories.
Let us return to the opening scene. Sunday, October 25, a line before the ballot box. A person for whom the fine is real money reaches the machine. They cannot not come: passport, job, university. They do not want to choose: they dislike both candidates, and there are another 33 million like them between the two. So they perform the only act the system has left them: enter a number that does not exist and press “confirma.”
Compulsory turnout faithfully transcribes what this person thinks into a line marked “nulo” in the evening statistics—and there its power runs out. What follows is what the lawmakers of 1932 did not foresee: dissent that is denied a channel does not disappear. It accumulates—in the 21% absenteeism and the 9% “I’ll spoil my ballot” of the current season. Compulsory voting has traveled from a tool for consolidating the young democracy of 1932 to a machine that reliably produces its own stock of dissent: every citizen is first forced to the ballot box, then everyone acts surprised that they arrived with no intention of choosing.
A spoiled ballot decides nothing—there is little arguing with that. But that is exactly why it is honest: showing up and entering a nonexistent number is the only form of participation the system cannot get wrong, distort, or appropriate. The choice “against everyone” is not provided for in Brazil, and so it is expressed literally: as a number that does not exist.