Election security: What’s being done locally

The machines on which most of Brown County’s residents will vote sit in the basement of the jail, in a cage secured by three different locks.

One key is in the hands of the Republican Party, one is in the hands of the Democratic Party, and the other is held by Brown County Clerk Brenda Woods.

Kyle Conrad called it “the most secure voting equipment in the state of Indiana.” He works for Governmental Business Systems, the vendor that’s been providing Brown County’s vote-gathering and vote-counting machines for the past 10 years or so.

In the wake of hacking allegations in the 2016 presidential election, the security of elections across the county has been under debate and scrutiny.

[sc:text-divider text-divider-title=”Story continues below gallery” ]

Before every local election, the Brown County Election Board conducts a public test of the voting machines to be used.

About 10 people attended the test on Sept. 28, mock-voting for candidates for town, county and state offices for three precincts on touch-screen machines. Then, they checked that the vote totals that printed out for their machine matched the votes they had put in.

Tests also were run on the optical scan machine, which counts the bubbles filled in on paper ballots.

Nobody participating in the tests reported irregularities.

A handful of them took the opportunity to ask questions about how the process works.

In the past three months, two organizations have written position papers stating their belief that electronic voting machines should be replaced with paper ballots in order to increase security and voter confidence in the outcome of elections.

The National Academies of Sciences, Engineering and Medicine advised that “all local, state and federal elections should be conducted using human-readable paper ballots by the 2020 presidential election” and “every effort should be made to use paper ballots in the 2018 federal election.”

The League of Women Voters of Indiana wrote that “auditable election results are essential to maintaining voter confidence in the integrity of the process.”

Brown County does not have vote centers — a place where anyone from any precinct could go to vote on Election Day — but if we ever did, we could not use paper ballots there, Woods said. As early as next year “I think that might be something the (election board) will look towards,” she said about vote centers.

One of the main reasons we aren’t all using paper ballots now is to lessen human error, she said. In her opinion and experience, electronic voting machines are less subject to those problems.

What we have

In Brown County, paper ballots only go to voters who apply to receive an absentee ballot by mail.

Since the 2016 general election, all other voters use an AccuVote TSx touch-screen machine. These used to be the machines that only voters with an impairment used, but when the state legislature passed a law regarding how “overvotes” were to be handled — ballots on which a voter picked too many candidates for a race — replacing paper ballots with electronic voting machines was a way to comply, Woods said. Federal funding through the Help America Vote Act allowed for their expanded use.

Anyone who votes on Election Day, or absentee in person before then, also will use an electronic poll book — a tablet-like device — to sign in and get an electronic ballot.

Electronic poll books replaced paper poll books in 2016. Previously, when voters checked in with their ID at their polling place, poll workers would find their name on a paper book, verify their name and address, and the voter would need to sign next to their name on that book.

Sometimes, voters would sign on the wrong line, Woods said. The electronic poll book only gives them one place to sign.

The electronic book also will keep track of people who requested an absentee ballot by mail, and won’t allow those people to vote in person if that ballot hasn’t been surrendered, Woods said.

Previously, some people would try to “beat their ballot” in the mail and attempt to vote in person, Woods said. Why? “I don’t know; people do all kinds of strange things, just because they can,” she said.

During the primary, a couple precincts reported problems with their electronic poll books. The base unit that powers them stopped working, preventing people from signing in to receive a ballot and causing lines to grow. Also, in two precincts, the poll book lists were switched, showing the wrong names, Woods said. Problems were resolved that day so that people could continue to vote; in the meantime, emergency paper ballots were issued.

Woods said the power problem happened because some of the poll book stands were wired incorrectly with replacement parts. Since then, all of them have been refurbished and inspected, and the company has now sent battery backups for each poll book stand at no cost to the county.

Also, during this Election Day, each precinct will be equipped with a paper list of voters to use if needed, and “encoders” to prepare cards for the voting machines if there’s any problem with the electronic poll books.

“All of the issues we had in spring, we kind of tried to head them off for the fall,” Woods said.

How it works.

After checking in with the poll workers at the electronic poll book, voters are given a card to insert into a voting machine when it is their turn to vote. That card will pull up the blank ballot, Conrad explained.

To vote, voters touch the box on which the candidates’ names appear. If the race has multiple candidates and you’re only allowed to vote for one, the machine will not allow the voter to overvote, or choose more candidates than they’re supposed to.

After going through all the races on their ballots, voters are shown a screen with their choices. If they skipped a race and didn’t vote for anyone, if their vote didn’t record correctly for the candidate they intended to choose, or if they changed their minds and want to pick different candidates, they can return to those screens and make changes.

After confirming that their votes recorded as they should, the voter ejects the card from the machine and returns it to the poll workers.

The votes are not recorded on that card; they are recorded on the voting machine’s memory, Conrad explained. Cards are recycled to different voters all day.

Once polls close on Election Day, poll workers print out a “tape” from each machine that shows the total votes recorded for each candidate. Adding up all the “tapes” at a polling place gives the total votes cast in that precinct on that day.

Copies of those tapes are given to runners who have permission to observe the process at the polls, and they take copies back to their parties’ headquarters. The Brown County Democrat also sends runners to collect tapes so that results can be reported online.

Then, the machines get packed up and transported back to the courthouse by precinct officials. The security seal that’s been covering the memory card port on the machine since the completion of the public test is taken off at that time, and a bipartisan team pops the cards out of each machine and takes them to a “receiving team” at the courthouse, Woods said.

At the end of the process that evening, Woods’ office prints a tally of the unofficial totals in each race.

The results aren’t “certified” until the election board reviews provisional ballots a few days later to see which of them are valid. Provisional ballots can be given to voters if there’s some problem or question about whether or not they are qualified to vote.

The “certification” is simply Woods signing off on the results, she said.

Hacking concerns

AccuVote machines are one of the most widely-used electronic voting platforms, according to a study by Princeton University researchers published in 2006.

They obtained AccuVote TS machines — a different model than what Brown County uses, but “closely related” to the TSx model we do use — ran experiments on the hardware and software, and “considered whether real election practices would leave it suitably secure.”

They found there were ways to alter vote totals through the use of malicious code.

Greg Bowes, a candidate for Brown County prosecutor, was familiar with those studies and related theories, and questioned Conrad about the machines while trying one at the Sept. 28 public test. He asked him about how and whether the code that programs the machines could be manipulated, such as showing a different result than the votes that were actually inputted.

“This chip could have a whole lot of code on it,” Bowes said.

“It doesn’t,” Conrad said.

The Indiana Secretary of State’s office tests all of the election equipment used in Indiana as an added layer of security, through the Voting System Technical Oversight Program at Ball State University.

That code is not available for just anyone to manipulate, Conrad said. In addition, the machines themselves are physically secured in the basement of the jail after the public has had a chance to see and test them, he said.

“Part of the trouble is, how do we restore confidence?” Bowes said while participating in the public test. “And this is one step, doing this right now.”

Another theory is that machines can be hacked over the internet.

Brown County’s machines are not connected to the internet; the only way they could be is to plug in a cable to a wall, Woods said.

Ballots are not transmitted over the internet except when the clerk’s office corresponds with absentee voters overseas, she said. Not many voters receive or return their ballots that way; as of last week, she had sent five and received two back.

Human errors

With the urging of the League of Women Voters of Indiana, Monroe and Montgomery counties replaced their voting machines with paper ballots, the League report says.

Indiana Secretary of State Connie Lawson isn’t convinced that’s the way to go. When Lawson was a county clerk, paper ballots and lever machine votes didn’t always reconcile, she wrote for the Brookings Institute in February. She believes electronic machines do a better job of keeping tabs on votes.

On past election nights, Brown County precinct officials would have to spend time after polls had closed making sure that the number of voters they counted as signing in on the paper poll books matched the number of paper ballots scanned into the optical scan machine, Woods said. Those numbers didn’t always reconcile either.

“Paper is kind of — just keeping ballot control is sometimes hard to do,” Woods said.

Sometimes, poll workers’ counts could be 200 off between the paper poll book signatures and paper ballots received, she said. “They didn’t leave until they reconciled and it was within five (ballots or signatures). … Usually you were always able to figure out, you know, that they counted the signatures wring, their addition’s off. There was always something,” Woods said.

Because Brown County now uses electronic poll books and electronic voting machines, precinct officials don’t have to hold up the voting line so they can count sign-ins on the paper poll books multiple times a day, she said. Instead, they check that the number of signed-in voters on the electronic book matches the number of ballots recorded on the electronic machines.

However, even then, the number of sign-ins might not match the number of votes, Conrad said. That’s because human error can still come into play.

For instance, in a primary election, a voter checking in at the electronic poll book might ask for the ballot of one party, but the poll clerk could accidentally load the ballot of the other party on the card, Conrad said. That card can be canceled and reloaded with the correct ballot.

Also, sometimes, a voter will say they want a ballot of one party, but then when they see the candidates on the ballot, they realize those aren’t the ones they wanted to vote for, Woods said.

“So, if you do that, you’re going to have more activates than you have signatures because they canceled the first one out,” Conrad said.

Fail-safes

The League report and the National Academies of Sciences, Engineering and Medicine report both recommended that audits be conducted of voting machines, ensuring the votes put in match the votes the machines spit out.

One such audit would be providing a print-out of candidate selections which the voter could review, the NASEM study said. That’s known as a “voter verified paper audit trail.”

There is a way to print out a “tape” after each ballot is cast on each machine, showing the Xs next to the candidates chosen. But that is not something that’s regularly done, Conrad said.

Another such audit would be checking a statistical sample of voting machines before the election results are certified, comparing the results stored on the machine’s memory to the ballots themselves. That’s known as a “risk-limiting audit,” the League report says.

Woods said that’s not something they normally do here and they would need to get help from the vendor. “If it comes down to it, I’m sure there’s a way to do it; we’ve just never done it as far as I’m aware of,” she said.

Vote-counting on an electronic machine is no less safe than an optical scan machine counting paper ballots, Woods told members of the local League of Women Voters, who had gathered around her and election board member Susanne Gaudin during the public test.

The voter doesn’t get to print out a paper copy of his electronic ballot, but a voter wouldn’t get a receipt or a copy of his marked paper ballot either, Woods said.

If a recount were necessary, the electronic machines would still be able to print out paper copies of the electronic ballots, Gaudin said.

In July, to comply with a new state law, the Brown County Election Board passed a resolution explaining what is done to ensure the physical security of all voting equipment. In addition to the voting machines being kept in “cage” security at the jail, poll books are kept under lock at the courthouse. Only election board members, the clerk and clerk’s deputies have access to the voting equipment when it is not at the polls, and if anyone visits the “cage” or poll book cabinet, a log book serves as an audit trail, the resolution said.

“We try very hard to make sure our elections are secure and as accurate as possible,” Woods said.

[sc:pullout-title pullout-title=”Irregularities caught early” ][sc:pullout-text-begin]

Voters helped the Brown County clerk’s office and state election office find and correct a couple errors early in the voting process this fall.

Nashville Town Council candidate Arthur Omberg reported on his campaign Facebook page that when he went to vote absentee on Oct. 10, he wasn’t listed as a choice on his ballot.

That’s because the address on his voter registration was incorrectly assigned to receive a county ballot, not an in-town ballot, said Brown County Clerk Brenda Woods.

The clerk’s office contacted town officials to learn what the last house in town limits was on the north end, then worked backward to make sure everyone on that road was listed as in town instead of out in the county.

There were 12 registrations that needed to be switched over, Woods said. None of them had voted yet besides Omberg; a few hadn’t voted in quite awhile, Woods said. “The ballot wasn’t wrong, it was just his voter registration needed to be corrected.”

Other voters alerted the clerk’s office that indianavoters.com, the state’s portal to show voters who’s on their ballots, wasn’t listing all the school board candidates on everyone’s ballot, only the ones who came from their district. School board is an office that everyone can vote on regardless of where they or the candidates live. “We called the state, got it figured out,” Woods said. This was corrected before early voting started Oct. 9.

[sc:pullout-text-end][sc:pullout-title pullout-title=”Voter turnout so far” ][sc:pullout-text-begin]

As of the morning of Oct. 23, 834 people had voted in-person absentee and 204 more had mailed in completed absentee ballots (two of them by email).

That tops the total number of in-person and mail-in absentee votes collected during the 2018 primary, which was 920.

Other historical absentee vote counts:

2008 primary: 616

2008 general: 1,988

2010 primary: 416

2010 general: 1,025

2012 primary: 446

2012 general: 2,162

2014 primary: 509

2014 general: 1,120

2016 primary: 1,206

2016 general: 3,037

Source: Brown County clerk’s office

[sc:pullout-text-end][sc:pullout-title pullout-title=”On the Web” ][sc:pullout-text-begin]

See a video of how Brown County’s electronic poll books and voting machines work: bcdemocrat.com/2016/08/30/hi_tech.

[sc:pullout-text-end]