Can Israel’s election count be tampered with? An official explains the process
Amid technical glitches, inconsistencies in the Central Election Committee’s published tallies, and even allegations of fraud, here’s how the ballots were processed
Simona Weinglass is an investigative reporter at The Times of Israel.
Last Thursday, two days after the elections, New Right party leader Naftali Bennett learned that his party was about 1,380 votes shy of earning any seats in the Knesset and demanded a recount, hinting at possible foul play. Sources in his party went so far as to allege that the elections were being “stolen” via a corrupted count.
On Sunday, the Central Elections Committee granted Bennett access to the original “double-envelope” ballots — the “extra” votes from soldiers and diplomats on whose votes New Right had pinned its hopes of making it into the Knesset — so that he could confirm for himself that the count was honest. At the same time, the committee chastised his party for its insinuations of wrongdoing.
In addition to New Right, several parties, including United Torah Judaism and Meretz, were in touch with the committee in the days immediately after the election over what they believed were mishandled ballot boxes.
With the votes finalized on Tuesday — and UTJ bumped up a seat, Likud down a seat, and New Right still outside the Knesset — Bennett’s party remained insistent that it was the victim of fraud in the vote-counting process, asserting discrepancies in 8% of ballot boxes. The Central Elections Committee dismissed the claim as unfounded.
Ahead of the formal, official submission of the certified vote count to President Reuven Rivlin on Wednesday, The Times of Israel sat down with Giora Pordes, spokesman of the Central Elections Committee, to understand precisely how the vote count process unfolded, and whether there was any potential for fraud. We also spoke to three other sources familiar with the process, including one who worked entering data at a vote counting center, who asked not to be identified.
Pordes explained, step by step, what happened from the moment the first voter entered a polling station on April 9 until the final vote tally was counted and checked, and detailed the safeguards put into place to prevent vote fraud. He was thoroughly confident, he said, that the final election results Rivlin will receive on Wednesday are indeed an accurate accounting of the people’s choices.
‘Keeping each other honest’
Pordes began by noting that every polling station is required to have at least three officials present, and often has four. (There were 11,500 polling stations nationwide on Tuesday, said Pordes, plus another 300 “double envelope” stations for soldiers, diplomats overseas and other absentee voters.) The first is the secretary of the polling committee, a person who has declared that he or she is not affiliated with any party and who is a paid temporary employee of the Central Elections Committee. The other two-three individuals are the chair of the polling committee, the deputy chair and (though this is not mandatory) an observer. The latter two or three officials are affiliated with different political parties (though they can come from the same part of the political spectrum). All the officials except for the observer must be present for any voting and any counting of votes to take place. The idea, said Pordes, is that each official watches the others and keeps them honest. (ToI was told by a source that, in fact, not all polling stations had three officials present; some had just a secretary and chairman. ToI could not confirm this claim.)
In order to become secretary of a polling station, one must apply online, take a course, and pass a background check, two written tests and a face-to-face interview. “We have nine regional election committees,” Pordes said. “Each committee decides on the lowest test score required to pass. It’s usually 80 and above, but in Jerusalem the minimum test score was 86.”
Each polling station has 300-800 registered voters. As voters arrive at the polling station on election day, the officials vet their identification, check off their names in a ledger, and give them an envelope to vote. The voter stands behind a divider where he or she selects a paper ballot for the party of their choosing and places it in the envelope, and then comes out from behind the partition and drops it into the ballot box.
Asked about rumors that in some ultra-Orthodox communities, ostensibly due to modesty concerns, men were able to vote on behalf of their wives by bringing their wives’ ID cards, Pordes said that this scenario was highly improbable because of the three or four officials monitoring everything. “It can’t be; the officials watch one another.”
At the end of the day, when the polling station closes, the three or four officials present tally up the vote: they open the ballot box, take out the envelopes, remove the voting slips from the envelope and check them, and record the number of votes for each party. The observer doesn’t have to be there but if the three others — the secretary, chairman and deputy chairman are not all present, even if one of them just goes to the restroom — the ballots cannot be counted. “When the poll closes, at least three people have to be present for the vote count: the secretary, the chairperson and the deputy chairperson,” Pordes said.
The polling committee members are required to count the ballots twice. Then, in pen, they write down the number of votes each party received, as well as the total number of people who voted, on a paper form called a protocol. “The number of votes for each party must add up to the total number of votes. If it doesn’t add up, they need to check the ballots again.”
Once the protocol is complete, the voting slips are put back into the ballot box. The secretary of the polling station then personally delivers the protocol to one of 25 vote counting centers nationwide. He or she “can take it by bike” if they like, he said.
This part of the process is unsupervised, Pordes acknowledged, so it would appear there is nothing to prevent the secretary tampering with the protocol en route. (The ballot box is also delivered to the vote counting center.)
At the vote counting centers, employees hired through temp agencies check the protocols to make sure all the correct lines are filled out, The Times of Israel was told by a staffer who worked at one of the centers. They then photograph them and scan them into a computer as an independent record. These employees then pass the protocols to a second team at the center — the Central Elections Committee’s data entry team — which enters the data into the Central Election Committee’s software system. This software, known as the Democratia computer system, was created by Malam Team, a private computer company, especially for the CEC.
Once the data is entered into the CEC system, sources told ToI, it is rechecked and then appears on the CEC’s website, which provides ongoing updates of the count as it proceeds, including party totals, percentages of votes counted, and district by district breakdowns.
Pordes noted that he received complaints and cries of foul play when the tally entered into the CEC system for a polling station in the West Bank settlement of Bat Ayin — and then published on the CEC website — showed 87 votes going to the Arab sector party Ra’am-Balad. (Similarly bizarre vote counts were recorded for several other polling stations, and initially published on the Central Elections Committee website: cases in which Arab parties did well at settlements, or Labor scored high and United Torah Judaism very low in ultra-Orthodox areas, or recorded turnout at a polling station exceeded 100%.)
What happened in the Bat Ayin case (and presumably in the similar cases too), said Pordes, was that either the person who filled in the numbers in the protocol at the polling station erroneously wrote down the tally for the Union of Right-Wing Parties in the square next to the Ra’am-Balad party, or the data entry typist erred when entering the numbers in the CEC computers. He did not know which of the two was the case. “It was after a long day and they were tired. Either the polling station officials or the typist entered the number on the wrong line.” The error, he said, was subsequently corrected.
At the vote counting center, any irregularity spotted by the software — for instance, if the number of votes for each party does not add up the total number of people who voted at a polling station — is flagged. The problem is then brought to the attention of a district court judge stationed at each of the centers, who stays up all night to adjudicate such cases.
The judge tries to figure out the source of the irregularity and rules on how it should be resolved, and the count is amended as appropriate. Most often in such cases, the tally is off because the officials at the polling station made a good faith arithmetic mistake, the Times of Israel was told by a data entry person who worked at a counting center in the Sharon region. Dozens of such minor irregularities were brought to the attention of the judge at his vote counting station all night long, this person said.
While it is conceivable that a data entry person could try to deliberately enter the wrong data, this source told The Times of Israel, he is required to print out his work and attach it to the handwritten protocol, and Central Election Committee officials made the rounds to compare the figures typed into the system with the handwritten protocols. (The protocols are later delivered to the Central Elections Committee’s HQ at the Knesset.)
Vote counters and data entry personnel are not allowed to bring their phones into the protocol processing center. USB drives on the computers used for data entry are taped shut, the employee who did data entry on the night of April 9 told The Times of Israel.
Once the numbers are typed in, the Malam Team system tallies the vote count. Asked if a hacker could attack Malam Team’s software and change the numbers, Pordes said he thought it unlikely since it is a closed system that is not connected to the internet.
He said that if someone were going to sabotage the vote count, it would have to have been before the elections, and the bug would have to have gone unnoticed by the company’s employees.
“Malam Team has safeguards in place against this,” he said. Asked what they were, he said that was an issue for Malam Team.
A software bug
There was one major computer bug on election day, Pordes acknowledged, though not in the software that actually counts the votes.
For most of the day on Thursday, April 11, when the “double envelope” votes had been counted and the entire election vote count was completed but was still being rechecked, the Central Elections Committee’s website showed the New Right Party just above the 3.25 percent threshold to get into the Knesset, with 138,101 votes or 3.26 percent of the vote. But at mid-morning that day, an official from the Central Central Elections Committee had told reporters that this figure was not correct and that the New Right had not in fact passed the threshold. It had, rather, scored 3.22%, and was 1,380 votes beneath the threshold.
Pordes explained that this confusing situation was the result of a software bug — not in the counting software itself, but in the interface between the the program that counted the “double-envelope” ballots and the Central Elections Committee website.
This “double envelope” count was not carried out by the Democratia computer software, but, rather was done using a different computer system, he said.
“We had 11,500 polling stations throughout the country, as well as an additional 300 ‘double-envelope’ polling stations,” he said.
There were 234,000 votes from those 300 extra polling stations, Pordes went on, recorded in a separate system, and there was a bug in the software that transferred the double-envelope count to the CEC website. “It took us eight hours to fix the bug; that’s why we didn’t publish the correct results until Thursday at midnight. The problem was not in the counting, it was in the transfer of the data. I don’t know if you’re familiar with the Israeli song, ‘Everything because of a small nail.'”
In fact, the discrepancy lasted for longer than eight hours. It was acknowledged at around 10 a.m. and was not corrected until almost midnight. Pordes did not say why a note warning that the figures were inaccurate was not placed on the website.
The inaccuracies on the site prompted a flood of complaints on Thursday, with New Right leader Naftali Bennett, as of this writing, still hoping the count would be amended in his party’s favor.
Camera clamor
On election day itself, meanwhile, there was a furor throughout the day over the deployment by activists paid by Likud of a reported 1,200 hidden cameras at polling stations in Arab areas. This operation was ruled illicit by Hanan Melcer, the Supreme Court justice who oversaw the elections, and the camera equipment was ordered removed. The camera deployment was decried by an Arab legislator as “political terrorism” intended to deter Arab voters, whose turnout was significantly lower in these elections than in 2015. The hidden camera operation continues to make waves, with an Arab MK on Monday castigating Melcer for failing to confiscate the equipment and the films.
Asked about the cameras, Pordes replied mildly, “it is normal for one of the officials at the polling station to be a Likud member. After all, the system is designed so that officials from different parties watch over each other. Some of the Likud officials started taking videos. We got complaints. Judge Hanan Melcer decided that photographing inside the polling station is not permitted, but that if they see rioting they can film that, and they can also film the counting process.”
In fact, while some of the cameras were indeed brought into the polling stations by registered Likud observers, others were found hidden on the bodies of activists who had no official role at the polling stations. A PR firm proudly took responsibility for the operation, and said it had been hired by Likud.
The way he saw it, Pordes said, the election day voting process itself passed almost without incident. “There was no need to call police, no unruly behavior, no theft of ballot boxes,” he said, enumerating incidents that had occurred in previous years.
The sources with whom ToI spoke saw a potential vulnerability in the vote count if the officials at a polling station colluded at the expense of parties whose representatives were not present. The data entry source, based on his personal experience, said he considered the vote count process itself to be securely safeguarded. Assuming Israel’s district court judges to be honest, he said, then the vote count is honest.
The Times of Israel Community.