Independent site. It does not belong to the government or to the TSE (Brazil’s Superior Electoral Court). Based on studies and behaviors of official public data. No insinuation, only bringing these identified patterns to the public.
Elections 2026 Cases from the voting machines’ internal logs

← The “compasso” case

How the study began

The Santa Rosa case

It all began with a voting machine in Acre, a very distinctive one. This is the path we took to understand its pattern.

Part 1 · The voting machine

Flávio ahead in the 1st round, Lula with almost 4 times more votes in the 2nd round

Vote for president in each voting machine in the city, from the 1st to the 2nd round, from the poll tape (boletim de urna). “—”: the polling section did not exist in 2022.

Part 2 · How we got to it

A comparison, voting machine by voting machine

Part 3 · The official files

The same machine in both rounds

Each voting machine leaves three files on the TSE website: the poll tape (the result it prints when it closes), the vote by vote (each vote recorded, without saying who voted) and the log (everything the voting machine did, second by second). We counted this voting machine’s vote by vote: it adds up to exactly the poll tape’s result, in both rounds.

The log comes compressed in 7-Zip (a .jez file). Inside it, the logd.dat file is text: one line for each thing the voting machine did.

Part 4 · The log, line by line

Two consecutive votes with the same time

Reading this voting machine’s log line by line, what later became the “compasso” case showed up: consecutive votes recorded with exactly the same time between one and the next. The time of each vote runs from the previous recorded vote up to it.

    The lines are as in the official file: the time, what the voting machine did and the check code that the voting machine itself calculates for each line. The number on the left is the line’s number in the file. The log says that the voter ID was typed in, but does not include the voter ID number.

    See this voting machine’s profile, with both rounds →

    Part 5 · The next question

    Does this happen in the rest of Brazil?

    From voting machine 95, the same reading of the log was done on the voting machines of the city, then in other cities and, finally, on all the voting machines in the country, in both rounds. We gave this continuation the name “compasso” case.

      The voting machine’s log says when each voter voted, not for whom.