Why are we confident our voting machines were not hacked?
November 11, 2016 2:37 PM   Subscribe

I would like to understand why we should not be concerned that voting machines weren't hacked during the election.

I saw a Politico article a few months ago that discussed at length how vulnerable our voting machines were, especially in states like PA, where they are more computerized. I think I understand that a widespread hacking would be difficult, but not impossible, to orchestrate because the states systems are independent from each other. But one or two? On top of that, we have the widely accepted DNC hack. Even if the FBI looked into it, Comey hasn't exactly come across as Hillary's biggest fan. Here are some other articles I've seen that are making me wonder why people aren't talking about this more.

http://www.cnn.com/2016/10/12/politics/florida-election-hack/index.html

www.washingtonpost.com/amphtml/posteverything/wp/2016/07/27/by-november-russian-hackers-could-target-voting-machines/?client=safari

Can someone explain why we should be confident this didn't happen? How can it be verified without throwing everything into even more upheaval? Should we assume that of course it has been looked into?

I know how this sounds, but it seems like a fair question.
posted by anonymous to Law & Government (22 answers total) 10 users marked this as a favorite
 
Most machines don't have any kind of paper audit of votes cast on them, so you can't really be confident that no tampering took place. Read this Wired piece if you want to feel really bad.
posted by rachelpapers at 2:44 PM on November 11, 2016 [5 favorites]


We are not confident, and we should not be.
posted by erst at 2:45 PM on November 11, 2016 [9 favorites]


Tampering would produce discrepancies between exit polls and the official vote count. If those discrepancies correlate with a specific type of voting machine, that would be strong circumstantial evidence of tampering or malfunction (i.e., miscalibrated touch screens)
posted by qxntpqbbbqxl at 2:57 PM on November 11, 2016 [1 favorite]


Tampering would produce discrepancies between exit polls and the official vote count.

Is this true? My understanding is that exit polling is notoriously unreliable. Not that I think the election was hacked. I'd need to see some convincing evidence and none has been put forward.
posted by cnc at 2:59 PM on November 11, 2016 [7 favorites]


Terry Gross did an hour on Fresh Air a couple of weeks ago about voting security. It's really fascinating.
posted by BlahLaLa at 3:00 PM on November 11, 2016 [1 favorite]


Michigan, which was one of the biggest upsets in favor of Trump, is paper ballot only.
posted by theodolite at 3:03 PM on November 11, 2016 [17 favorites]


Because people on the Left aggressively claimed the voting system couldn't possibly be altered on a large scale, when the people on the Right claimed that would happen for the Right to lose. Now that the other team won, reversing would seem hypocritical and cause those people on the Left to lose face, so it's important to them to continue defending that claim.

Hopefully statistical analysis can offer more validation eventually.

Personally, I'm convinced that individual machines are not secure. Whether that could be leveraged into outcome-changing tampering, or whether anyone would ever try, I don't know. But secure, verifiable systems seem fundamentally important to me, regardless. It's too easy to go down conspiracy-theory holes, otherwise.
posted by solitary dancer at 3:06 PM on November 11, 2016 [5 favorites]


Is this true? My understanding is that exit polling is notoriously unreliable.

Exit polling a noisy signal, but it's not random. It has a relationship to the signal coming out of the voting machines, and that relationship will show statistical anomalies for tampered polling places when compared with un-tampered polling places.
posted by qxntpqbbbqxl at 3:10 PM on November 11, 2016 [1 favorite]


Of course we're concerned, but what can we do? Post in anguish on social media?

The issues's been 'on the radar' since 2003, when that big boss of Diebold (maker of voting machines) promised to supply the votes Bush needed to win re-election. That caused some states to go back to paper ballots.
posted by Rash at 3:11 PM on November 11, 2016 [4 favorites]


For me it's the 'little old lady' factor. Our elections are run a the precinct level by older patriotic women that have little to gain from breaking the system. The issue has been looked at before and never has there been a large scale corruption, and there is strong belief in keeping the system at a personal level any part of the country I've seen.
posted by sammyo at 3:15 PM on November 11, 2016 [8 favorites]


The unexpected results seen in this election have an understandable pattern (majority-white rustbelt areas went to Trump at higher rates than polls suggested, to oversimplify). If it was just Pennsylvania, or Wisconsin, or somewhere else, that showed an unexpected pattern, maybe it would be worth looking into further. But as you note, it's extraordinarily unlikely that someone was able to co-ordinate anything on a multi-state basis.
(This doesn't preclude the possibility that one state's system was hacked in the direction of the broader trends - but that wouldn't change the result, if that's what you're concerned about).
posted by une_heure_pleine at 3:23 PM on November 11, 2016 [8 favorites]


While hacking a voting machine isn't necessarily difficult from a technical standpoint, most that I've read about require some level of physical access to the machine. Note that in the linked politico article, Appel needed extended physical access to the machine, and the exploit itself was fairly involved. To successfully alter the election results, an adversary would need to physically access as well as defeat any tamper evident physical security. This is a feasible thing to do for a nation state actor, but it's orders of magnitude more work than the DNC hack, and much more risky.

All of which is to say, there are reasons why a nation state wouldn't, plus the expectation that there would be more evidence of tampering if a nation state did. Neither of which rule the scenario out explicitly, but they do push it much more into the territory of conspiracy theories.
posted by phack at 3:32 PM on November 11, 2016 [2 favorites]


On Election Day, both my son and I checked online to ensure we were registered to vote.

The state's website said neither one of us was registered.

We went to the poll anyway and of course we were registered.

But I keep circling back to that. How many people also got erroneous information that they couldn't vote and then decided to stay home?
posted by yes I said yes I will Yes at 3:34 PM on November 11, 2016 [14 favorites]


Can someone explain why we should be confident this didn't happen?

There are 3,143 counties and county-equivalents in the United States. Each county contains multiple precincts which each report separately. Assuming that the person in charge of reporting the precinct count to the secretary of state's office is going to double check that the count that the SoS got was the same as the count the machines show, you'd need to rig the machines to produce phony votes.

To rig an election via voting machines, you'd need to effect enough machines to spread the phony votes around so as not to raise eyebrows. At least in Virginia, machines can vary on a county-by-county basis. So, you'd need to come up with hacks for each potential type of machine, figure out how to rig the votes so they seem plausible, precinct by precinct, then distribute the malware.

This is a big job, which would likely need to involve individual actors across the country (or at least across targeted swing states -- Florida has ~67 counties, I can't find details on the number of precincts). If you could do it all via the internet, I'm sure an advanced persistent threat style actor (Russia/China) could manage it. But if you need to put "voting machine repairmen" or the like in place to hit all the machines, there's no way a criminal conspiracy that big stays secret, and I don't think post-cold war spycraft is up to it either.

How can it be verified without throwing everything into even more upheaval?

Taking Florida again, which as a mix of paper and electronic w/o paper trail voting systems, I'd look at the error in the exit polls between paper precincts (presumed accurate) and pure electronic (possibly hinky). Correcting for [appropriate factors known to people who understand polling science], the ranges should look the same. That data might actually even be public.

Should we assume that of course it has been looked into?

Nope. The distributed nature of US voting practices is both a blessing and a curse. You're not going to get any county election official with a tight budget to do a deep dive into whether the machines that the vendor promises are "good" are actually what they say they are.

Which is silly, because electronic machines with voter verifiable paper trails are known technologies, it's only money stopping all counties from buying those and then doing spot recounts of some statistically significant number of votes after every election. But it's hard to justify the money when people are convinced that it isn't a problem.

You may be interested in the second season of Scandal.
posted by sparklemotion at 4:01 PM on November 11, 2016 [2 favorites]


So, I've actually been curious about this. Not really because I think the election is hacked, but because I think there are good learnings possible about voter behavior.

It's weird to me that turn-out was lower than 2012. It's basically unheard of that an open election would garner fewer votes than an incumbent election. Especially with such close polls. Tight races historically garner higher turnout.

I think the news channel I was watching specifically called out a suburb of Detroit have significantly fewer votes than 2012. And I wondered if it experienced depopulation, higher apathy, less affinity with 2 white candidates, greater difficulty getting to the polls due to some structural or social scenario, or just had fewer votes for the top position.

Because that last scenario is how you could hack an election, and get around the exit polls. You just suppress vote counts in urban areas. And it would be easier with paper ballots than electronic ballots, because you're only affecting the machine that scans the ballots. The likelihood that people push for a recount is pretty low, as long as you kept the variance between voter signatures and presidential votes low enough it mimicked actual voter behavior for choosing to not pick a presidential candidate.

The data is going to be analyzed to death, because it's a big deal that the polls got this wrong, and presidential elections are always fascinating to graduate students. So it will be noticed eventually. But our ability to really have something definitive before the inauguration is pretty unlikely.

(note: I don't think the election was stolen. But I do think that this data is a great data set to study a lot of things we think we know about voter behavior. But with all those unknowns, fraud could potentially be one of those)
posted by politikitty at 4:41 PM on November 11, 2016 [2 favorites]


The complexity of managing an attack involving critical exploits on multiple platforms at a huge number of locations using only remote access to generally non-networked devices seamlessly in a very short timeframe with who knows what other random factors (version mismatches, unexpected configuration settings to name a few) complicating matters is pretty daunting. Even if you only target vote aggregators & not tabulators, you have to make it work flawlessly the first time. You need an advance team conducting information gathering to know which locations have which platforms & find & test the network paths to each one, without exposing yourself anywhere along the way. And then there's the payload, you need to know in advance which votes or vote tallies you're going to alter in which direction & how far. Which precincts, counties, states will you fix? How many do you dare to hack? Too many & you risk exposure. Too few & you can't guarantee a win. I'm good & I know some of the best but I would walk away from that gig & so would they, even given an unlimited budget & a very large team of guys who really know what they're doing. It just doesn't scale.
posted by scalefree at 5:20 PM on November 11, 2016 [2 favorites]


All this seems to assume that an attempt would happen right at the time of the election event, on a town or state-wide basis.

The more likely vector, I'd think, would be by the programmer at the time of creation of the software powering the machines. Many programmers skew conservative and cynical, and it would be easy to have come up with an algorithm that adjusted outcomes subtly, based on the pattern of inputs, with some random variation.

Plus:

Because people on the Left aggressively claimed the voting system couldn't possibly be altered on a large scale, when the people on the Right claimed that would happen for the Right to lose.

So, the right claiming vocally and emphatically that the election was "rigged" essentially forced the left to defend the system, thus putting them into a corner if they later came to doubt that.

It's time to get people more focused on voting rights and embracing the complexity of the modern system.
posted by amtho at 6:45 PM on November 11, 2016


I used to work for a voting systems company. It was a while ago, but the systems we deployed then are still in use in a lot of places. People might have assumed it was a conservative company based on where it was located, but I'm pretty far to the left, and anybody doing a token amount of background research would have known that about me. I had coworkers on the left and right, but management always struck me as apolitical (and one boss in particular is still a guy I'd cite as one of the two best people I've ever worked for).

We worked in red and blue counties alike, and occasionally in uneasy partnerships between cities and counties who were getting the same systems, where one was blue and the other was red, and each side was watching the other pretty closely.

If I was managing your election, I'd program and test your ballot myself, and then I'd make as many copies as were necessary for the machines we were deploying for you. The tabulation software was independent of the ballot creation process and was the same for every county that used our systems. The ballot didn't really know what Republican or Democratic was except as a field that ran alongside the candidate's name; we'd run tests on new machines with ridiculous ballots and no real parties on them.

scalefree's and sparklemotion's comment above rings very true to me. From my experience -- and admittedly, there are a couple other companies and their systems may be different -- the failure point you need to watch for is the aforementioned little old ladies doing those unglamorous jobs in the county clerk's office, out of sight. (Not that I ever saw any dishonesty myself.)
posted by chimpsonfilm at 10:11 PM on November 11, 2016 [4 favorites]


Not everyone is. I don't necessarily believe the things in the article but I don't know what anyone could do to prove any of it, either, and it doesn't look like it's entirely speculating that the machines themselves were the problem.
posted by dilettante at 9:16 AM on November 12, 2016


"Hacking" in this context appears to require that the target be available online. In most locations, even when the vote count is computerized, it is offline.

The only episode of mis- or malfunction that I recall being reported on election day involved people who wanted to vote for Trump having the machine report it as a vote for Hillary instead.
posted by yclipse at 9:26 AM on November 12, 2016


The only episode of mis- or malfunction that I recall being reported on election day involved people who wanted to vote for Trump having the machine report it as a vote for Hillary instead.

Which largely turned out to be misaligned touch screens that registered your touch below the spot you touched.
posted by scalefree at 4:07 PM on November 12, 2016


On Election Day, both my son and I checked online to ensure we were registered to vote.

The state's website said neither one of us was registered.


Update: I spoke to my Town Clerk who said she had the same problem. Apparently, *somehow* the zip codes for people in my town were wrong in the state of Massachusetts database. So when people typed in their name and accurate zip code, the site told them they weren't registered to vote.

This seems like something that should be investigated further.
posted by yes I said yes I will Yes at 3:04 AM on November 15, 2016


« Older GoFundMe for Faithless Electors's Fines/Fees   |   Oh, Columbus? Newer »
This thread is closed to new comments.