On August 7, the joint police-prosecution investigation headquarters searched nine election commission offices.
The allegation is that the number of voters by time slot was entered differently from the facts.
But how are turnout figures made in the first place?
3-line summary
1. Turnout figures are statistics entered by people.
2. Nine election commission offices were searched on August 7.
3. Turnout and vote totals are different data.
What happened
It began with the ballot shortage incident in the June 3 local elections. A special committee for a parliamentary investigation was formed in the National Assembly, and the joint police-prosecution investigation headquarters is handling the investigation.
The nine locations searched on August 7 were the National Election Commission in Gwacheon, Gyeonggi Province; the election commissions of Seoul, Gyeonggi Province, and North Chungcheong Province; and the election commissions of Gangnam and Seocho districts in Seoul, Gimpo and Uiwang cities in Gyeonggi Province, and Jincheon County in North Chungcheong Province.
The alleged offenses listed in the warrant were reportedly falsification of electronic public records and obstruction of official duties.
Reports also described circumstances that drew investigators’ attention: an email the National Election Commission sent to provincial and metropolitan election commissions on the morning of the election. It said that the principle was not to correct voter-count errors, and that if there was no major error, the figures should be adjusted and reported at the next reporting time. One employee who sent this email was additionally booked as a suspect.
Searches and booking are investigative procedures for securing evidence and identifying subjects of investigation.
Where do those figures come from?
On election day, figures such as “Turnout at 11 a.m.: 23.5%” appear. Many people think these are figures automatically counted by machines—as if a counter rises each time a ballot enters the ballot box.
That is not the case.
At polling places, voters receive ballots after identity verification. The polling place counts how many people were given ballots and reports that number at designated times. Those reports move up through cities, counties, and districts, then provincial and metropolitan levels, and finally the central level; the resulting figure shown on screen is the turnout rate.
In other words, turnout is not a vote-counting result but a progress report. People count it, and people enter it.
Two things follow. First, errors can occur. Second, they can be corrected. These are precisely the two points at issue in this case.
From 968 people to 1,580
One reported circumstance shows how the structure works.
At Polling Place No. 5 in Yeoksam 2-dong, Gangnam District, Seoul, the cumulative voter count rose from 968 at 10 a.m. to 1,580 at 11 a.m.—an increase of 612 people in one hour.
The method investigators suspect is this: At some point, a figure higher than the actual number is entered, and until the actual voter count catches up, the same value as the previous time slot is repeatedly entered to conceal the error. Then, during some interval, the increase may register as zero people for an entire hour. Reports also described circumstances in which there was no increase for six hours.
Looking only at the cumulative figure, the numbers line up in the end. Only the intermediate process looks unusual. Because it is a method of adjusting the middle figures to make the total match, it also means the final turnout rate could remain unchanged.
Turnout and vote totals are different data
This is the core of the current political dispute. One side says, “If turnout is manipulated, manipulation of the vote share naturally follows.” The other says, “Calling false turnout entries manipulation of vote totals is a logical leap.”
Rather than taking either side, it is enough to look at how the two figures differ.
| Turnout | Vote total | |
|---|---|---|
| When | While voting is in progress | After voting ends |
| What is counted | Number of people who received ballots | Votes marked on ballots |
| Is there physical evidence? | Reporting records | Physical ballots |
The biggest difference is in the last row. Physical paper ballots remain behind vote totals. Vote counting takes place in the presence of observers, and if a dispute arises, a court can recount the ballots.
There is no such physical evidence for turnout. There is no way to retrace how many people came at what time other than through the records.
That is why “the turnout figure was wrong” does not automatically lead to “the election outcome changed.” For that connection to exist, separate evidence would be needed that the ballots themselves were tampered with. Conversely, because it is a statistic without physical evidence, the fact that it was wrong is itself a problem.
Why isn’t it counted automatically?
One question remains. With today’s technology, couldn’t it be counted automatically?
Identity verification at polling places is computerized. But what comes after is done by people: issuing a ballot, going to the voting booth, and placing it in the box. To count this automatically, electronic devices would have to be attached to the ballot box or ballots.
It is probably not because the technology does not exist. But the paper-and-person system involves a tradeoff. It is slower, but anyone can verify it with their own eyes. Digitization makes it faster, but verification moves into the realm of specialists.
The cost is that interim statistics pass through human hands. This case is a moment when that cost became visible.
Two steps for distinguishing progress statistics from vote counting
First, look at what “turnout manipulation” refers to. Whether it means statistics during voting or vote-counting results makes it a completely different matter. What was at issue this time was the former.
Distinguish the investigation stage from a final determination. A search is a procedure for securing evidence, and booking means someone has become a subject of investigation. Neither is a conclusion.
What remains on paper and what does not
Turnout is not a figure counted by a machine; it is a report that people enter into fields and submit. That is why it can be wrong, and why it can be corrected.
Physical ballots remain behind vote totals, but there is no such physical evidence for turnout by time slot. If the two figures are grouped together as one, the current dispute cannot be understood.
To trace the path of errors in turnout by time slot, the reporting times and reasons for corrections must remain in the records.
References
- Dong-A Ilbo report dated August 7, 2026 — nine locations searched; Yeoksam 2-dong case
- Yonhap News TV report on booking of the sender of the guidance email
- Yonhap News report on remarks at the parliamentary investigation special committee meeting