In 2006, software engineer Michal Pospieszalski discovered a dangerous flaw in the US voting machine.
Hired by the Institute of Election Science, where he served as Chief Technology Officer, Pospieszalski flew to the headquarters of Election Vendor Election Systems & Software (ES&S) in Omaha, Nebraska. His job was to analyze the company’s Ivotronic voting system.
Over the course of more than a week, Pospieszalski has discovered a wide range of issues, including “bad code practices, backdoors, static passwords.”
“The biggest thing that wasn’t there was proof of end-to-end encryption,” Pospieszalski said. Decryption In an interview. “Even full external security can’t tell if a machine is legal to vote or if it counts twice, three times, 10 or 1,000 times.”
What’s missing from today’s voting machines
Pospieszalski, CEO of blockchain security and identity software company Matterfi, said the vulnerability is not hypothetical. It can be easily exploited by people with access to voting machines and voter registration systems.
“You can run the same vote in 10 times, but even today it’s still true – it counts as 10 votes,” he explained. “And the scanners don’t know much more, nor does the tabulators. The central district’s tabulators are like, “Oh, that was 10 votes.” ”
Pospieszalski said that the separation of voters’ records systems often makes settlements impossible without referring to the original paper records.
“There is no anonymous serialization of each vote, which allows the system to know that each serialized vote must be counted only once,” he said.
The solution, according to Pospieszalski, is based on encryption technology originally developed by David Chaum, a cryptographer who pioneered digital caches in the 1980s and introduced blind signatures, which included software rather than hardware.
Chaum later founded the early digital currency, Digicash, and proposed a cryptographic voting system that allowed public verification while maintaining anonymity. His work laid the key foundations for both secure electronic voting like Bitcoin and the latest cryptocurrencies.
“All you want is the last machine, the central count tabulator or election management system, to get the definition of votes, and every vote has a Kaauman blind serialization,” says Pospieszalski. “So, like in LA County, a printed output vote has a serial number. That serial number doesn’t identify the voter, but tells the central district tablater, “Hey, this is a unique vote.” ”
“I saw them, and someone was fooled,” he added. “Especially if 50 people see it.”
Pospieszalski’s proposed model includes three counts: paper voting, traditional digital aggregation, and third encryption.
“The way to see fraud is that you say you have 100 digital counts and you say you should only have 90 cryptographic counts,” says Pospieszalski. “Now you know someone injected 10 votes.”
Lessons from Antrim County
In 2020, Pospieszalski was hired to conduct forensic analysis in Antrim County, Michigan.
“In Antrim County, it was about 2,000 votes for Biden, and the next day it was 2,000 votes for Trump,” he recalled. “What really happened was that the definition of votes was misunderstood and the system made it seem that Trump’s vote was for Biden.”
He said “everything has returned to normal” when the vote was rerun in the revised definition file.
Pospieszalski emphasized that the optics of the situation gave public doubt while the error was technical.
“There were no major hostile attacks, but people want the answer, just as voters are being riled by the media, especially the right-wing media,” he said.
However, he found no evidence of remote hacking or software backdoors, but Pospieszalski said he encountered signs of possible votes during his analysis.
“If you have 42 choices and you can see 100 votes in your analysis where all 42 are filled out in the exact same way, you’ve seen: “That’s what I found some evidence in Antrim County.”
Asked why serialization of cryptographic voting has not been implemented, Pospieszalski pointed out that it would change the system and company’s unwillingness.
“They propose all sorts of really, really hard to use plans… if you’re a manufacturer of voting machines, this wouldn’t make sense,” he said.
Some technologies aim to improve election safety and confidence. In April, New York State Sen. Clyde Vannell introduced a bill that uses blockchain technology to secure voter records and election results. Blockchain is touted as a secure voting solution, but Pospieszalski argued that the core issues don’t require that level of complexity.
“All you’re trying to do is solve a simple problem, count accurate votes accurately,” he said. “There’s no need for extra complexity. Many people push blockchain because it’s popular, but it’s not really necessary.”
In contrast, Pospieszalski says his solution works on the current machine.
“I’m just saying, look, upgrade to an existing system, work with Dominion, work with ES&S.
Asked how adoption would occur, Pospieszalski proposed law or mandate from the jurisdiction overseeing the election.
“Voting makers and their customers (county) need a large precinct to drive change,” he explained. “We’re going to be in business if the law says that by 2028 or 2032 we need to include end-to-end cryptographic proofs in our voting system.”
According to him, the benefits will become clear in future elections, especially in fierce contests where trust is vulnerable.
