America's new voting machines bring new fears of election tampering

By design, tens of millions of votes are cast across America on machines that cannot be audited, where the votes cannot be verified, and there is no meaningful paper trail to catch problems – such as a major error or a hack.

For almost 17 years, states and counties around the country have conducted elections on machines that have been repeatedly shown to be vulnerable to hacking, errors, breakdowns, and that leave behind no proof that the votes counted actually match the votes that were cast.

Now, in a climate of fear and suspicion over attacks to America’s voting system sparked by Russia’s attacks on the 2016 elections, states and counties across the country are working to replace these outdated machines with new ones. The goal is to make the 2020 elections secure.

“There’s a lot of work to do before 2020 but I think there’s definitely opportunities to make sure that the reported outcomes are correct in 2020,” said Marian Schneider, president of the election integrity watchdog Verified Voting. “I think that people are focusing on it in a way that has never happened before. It’s thanks to the Russians.”

The purchases replace machines from the turn of the century that raise serious security concerns. But the same companies that made and sold those machines are behind the new generation of technology, and a history of distrust between election security advocates and voting machine vendors has led to a bitter debate over the viability of the new voting equipment – leaving some campaigners wondering if America’s election system in 2020 might still be just as vulnerable to attack.

The draw of the new machines, called ballot-marking devices (BMD), is the promise of a paper ballot. The voter will use a touchscreen to vote, then the BMD will print out the votes. Theoretically, the voter will look over that ballot, make sure it is correct, and insert it into an optical scanner that will quickly count it and save the paper in a secure lockbox. This paper trail solves the problem of the previous system by allowing election officials to audit the election.

“In computer security, systems need to be resilient,” Schneider said. “That means you’re able to monitor, detect, respond and recover from any event, whether it is a bug or whether it is malicious interference. Having a paper ballot that allows you to do that auditing allows you to have a resilient system.”

But there are concerns with the integrity of the paper trail a BMD would create at every stage.

At least two companies that sell the machines, Dominion and Election Systems & Software, combined a BMD, which prints a filled-out ballot, with a scanner, which counts the votes. Large election jurisdictions such as Delaware, Georgia, New York and Philadelphia are purchasing these “hybrid” systems, which some observers say creates two problems.

First, the printer and the scanner share the same paper path. If a voter leaves any races blank – a common practice called undervoting – the machine could in theory autofill those races. Neither the voter or election administrators would be able to detect the change. Second, the hybrid machines have a feature critics are calling “permission to cheat”. Voters can opt to not to review their ballots, meaning that the BMD prints the ballot straight to the scanner and into the lockbox. In such cases, there would be no way to confirm that what the voter intended to vote was actually what was printed and counted.

Due to these problems, the New York state board of elections is reviewing the state certification of the voting equipment. But other jurisdictions are not.

In many ways, getting a verified paper trail is a question of human behavior. If voters verify the ballots, then BMDs should be fine, according to Ron Rivest, a professor at MIT and a renowned cryptographer and computer scientist.

“If it’s the case that almost nobody looks at the paper records produced by ballot marking devices, then we have a problem,” Rivest said.

This question has not been thoroughly researched. Mike Byrne, a psychologist who studies elections, was astounded when he researched usability of voting systems in the mid-2000s. No one had done it before, he said, and $3bn had already been allocated to states to purchase new systems. Now, once again, the US is introducing a new voting technology has not been thoroughly studied before it is purchased.

“I think it’s premature to claim that BMDs either are the right answer or aren’t the right answer. We don’t yet know what the right answer is and nobody wants to admit that,” Byrne said.

Anti-BMD activists have frequently referenced a study that indicates that voters either forget some choices – and therefore cannot verify that they are the right ones – or forego checking their ballots at all. It appears to be the only study on this issue to date, though it was not peer-reviewed and should not be considered definitive, according to Byrne.

But there is another potential weakness, too. Many BMD models on the market print a sort of two-in-one ballot with one section to be read by machines and another to be read by humans.

Barcodes – or QR codes – that represent a voter’s choices are printed on the ballot along with plain text showing, presumably, the same information in a way people can understand. When the ballot is scanned, it is the barcode that is scanned and counted, not the text that voters can read. If a barcode is printed that represents a different choice, or the scanners were hacked, voters would not know the difference.

Hanging over all of this is a question that scares many: what if voters did find problems with their ballots? And what would election officials do if it was found that ballots were cast on hacked machines? There may not be a better way to sow chaos in American democracy than to force dozens of jurisdictions across the country to redo elections.

More likely, complaints about the machines would be written off as user error and manipulation would go undetected. After all, thousands of complaints were made about the old machines in Texas and Georgia in the 2018 midterms. Neither the results nor the machines in either election have been investigated.

Some solutions are being trialled. Both Los Angeles and Travis county, Texas – home to Austin – have developed innovative voting tools aimed at serving all voters and preserving a paper trail that can be accurately audited. But they are rarities – and Travis’ tool will not be ready for the 2020 elections. As to the rest of the states and counties looking to upgrade their machines, they will either have to move to paper and pen or buy the machines currently on the market – with all the potential flaws and insecurities.

Some states simply do not have the money to upgrade, others messed up the procurement process and are stuck in litigation, and others are simply frozen by politics. But in the end, there are likely to only be three states – Louisiana, South Carolina and most of New Jersey – and dozens of counties across the country with paperless voting machines in 2020. The problem is: no really knows if that has solved any problems or just simply opened a new set of vulnerabilities, masked by trust in a new system.

(THE GUARDIAN)