Monday, October 10, 2016

Electronic Ballot Security

There’s good news and bad news concerning ballot security—which can, unfortunately, affect the outcome of our elections.

But in Harris County, here in Houston, we can feel pretty good about certain aspects of election integrity. 

This past Saturday, at our local Tea Party meeting, we heard from Ed Johnson of the County Clerk’s Office concerning ballot security here. Specifically, he was talking about the eSlate machine used here, and how they protect its integrity.
Ed Johnson, Harris County Clerk's Office
photo: Cypress Texas Tea Party

Can it be hacked? If everyone involved colluded, then possibly. But with the bipartisan safeguards in place, no.

How did he convince us? He began with a story.

Back in 1996 Duane Bohac, his friend from their college fraternity days, told him he was going to run for state representative. He was running against a 20-year incumbent. The incumbent had 15 voters registered at his home (people who didn’t live there), plus six more in his garage. This voter fraud was reportable, and those illegal voters were removed from the rolls, which is good. But it shows the propensity of that Democrat to engage in illegal voting tactics.

Early vote tallies showed Bohac up by about 1000 votes. But there were four precincts that reported very late—not until around 1:00 in the morning. After they reported, Bohac lost by 75 votes. They found hundred of over votes in those precincts—votes that have more than one candidate marked on the paper ballot, so they are not counted.

In 1998 Bohac ran again. This time he got alternate judges (Republicans, when the presiding poll judge is a Democrat) and poll watchers to watch what happened. The poll watchers followed the judges as they drove the ballot boxes to the central location for counting. As in the previous election, all but four precincts reported in a timely manner. The poll watchers followed these four as they drove around the 610 Loop (the innermost freeway circling Houston) until 1:00 AM. They watched them shaking the ballot box. Early counts showed Bohac leading. But these four precincts showed 400 overvotes, and Bohac got zero votes from these precincts. (Bohac eventually tried again and won; he is my state representative.)

Paper ballots are easy to manipulate. Remember the 2000 election with its hanging chads? Suppose someone with a moment alone takes a long sharp device—like an unbent paper clip—and pokes it through a whole stack of punchcards through the punch hole of the candidate they favor? Any ballot already voting for that candidate is left unchanged, but any ballot for another candidate is now an overvote—or possibly a confusing one with a hanging chad, which pretty much never happens when punching only one card at a time. In that election there were also entire precincts with zero counted votes for Bush, which is statistically highly unlikely, especially when you consider these "safe" areas were the ones where Gore asked for the recount. [Johnson only mentioned the hanging chads. The rest of this paragraph is from what I remember of that history.]

In 2000 County Clerk Beverly Kaufman was faced with a problem. Harris County has the longest ballot in the country (some say LA has a longer one, but Johnson says he still feels confident in making the claim). Much of this is because we vote for judges, and it’s a very large county with a lot of district courts. Anyway, Kaufman couldn’t find a ballot long enough to handle all of the races. Within two years she found the Hart System, which makes the eSlate machine we still use here.

The Hart eSlate Machine
used in Harris County

Since the Help America Vote Act (HAVA) in 2002, electronic voting has been required across the country. A federal commission must certify the equipment. The Election Assistance Commission was functioning well until Obama; he has refused to appoint anyone to the post, and it has gone defunct, so there have been no updates available, so it is still a slow and burdensome process.

Nevertheless, County Clerk Stan Stanart got our eSlate system to pass the federal regulations. The EAC test takes a year to complete, and costs the company $2 million. Then the system is submitted to states, which take another year for testing. Ours has passed both. HAVA gives money to the states to buy equipment that meets standards.

Johnson spent a fair amount of time going through the type of testing done. He used a number of detailed slides, so this report will only skim the surface, from my notes.

There is HASH testing, or code testing. They check to make sure no coding has been changed.

There is BOSS testing—Ballot Originator Software System. There is no connection to the internet, at any point. Ever.

There are two encryption keys between segments of the system. One is internal. One is external—on a thumb drive. The official (County Clerk) has to take it and plug it in. This is kept in a vault.

There is the Mobile Ballot Box (MBB), the memory card upon which the ballot is backed up. Votes are also kept in three places in the system; cheating would require changing all three locations somehow, without leaving a trace. The memory card is loaded into the JBC (judge's booth controller) machine. A security code identifies the card for each machine. This is tracked throughout the process.

There is security of the election equipment. When you press “cast ballot” on the eSlate machine, your vote is saved in three locations—with the timing of that vote and other identifiers. It is virtually impossible to change votes on a memory card and the other two locations.

There is Login and Accuracy Testing required by law, that takes a week. It uses people of both parties. They create a spreadsheet that includes every possible combination of votes on the ballot. The testers input, physically, each of these voting combinations. The outcome must meet the spreadsheet. This shows there is no tabulating error. Harris County has always passed this test.

For early voting, supervisors set up the polls, with the machines showing zero votes on the screen. They print this zero report to start, and at the end of the day they do another report. Equipment is locked, resealed—all seals are labeled with codes, which must match, to show that there was no tampering overnight. At the end of early voting, the machine is sealed, again with a coded seal. A constable delivers the machine to the downtown office. Seals are validated, and the machines are locked in a vault with the memory card still inside it.

There’s a second Login and Accuracy Test done by members of each party. This takes three hours on election night, with a high amount of security. It is completed just prior to the polls closing. Early voting tallying begins, but no results are available—to anyone, including those involved in the counting—until after the polls have closed, so that early voting outcomes have no influence on voters who haven’t yet voted.

At the polling place, machines arrive sealed, with codes on the seals, which are recorded. The zero tape is printed, to show that there were no votes on the machine prior to the start of voting. Each voter must show eligibility (ID), sign into the polling book—plus their name is written by the clerk on a voter list. The voter is sent to the JBC machine to get a code. They take their code to the voting booth and enter their code on the eSlate machine. The machines have a dial, which prevents errors that can happen on touch screens. No helper (language assistant, for example) can touch the machine—only the voter. The voter presses “cast ballot,” and a flag on the screen indicates that the voter was successful in casting the ballot, and the voter can leave the booth.

There are some similar machines that can print cast ballots. This removes the safety of a secret ballot, and these machines have a lower accuracy rate. So Harris County does not use them.

The individual votes cast can be printed—and are printed, by a hard-to-read dot-matrix printer. Stan Stanart still uses this, although the state has recently allowed an alternative laser printer version. But, once an election process is underway, Stanart doesn’t make changes, so this election will still use the dot-matrix printer.

Every recount done in the county has come out the same as the original count; the electronic voting system has a zero% error rate.

There are seven vendors selling electronic voting equipment nationally. It is our County Clerk’s opinion—and Harris County has been ground zero in preventing election fraud—that the HART system is the best.

I have no personal expertise to know whether this is true, but I have been a poll watcher and poll worker here a number of times. And I am persuaded that electronic voting prevents fraud, and I’ve seen the security procedures followed carefully at the polling places.

Ed Johnson told us (though it wasn’t news to us) that Democrats hate electronic voting. In the national effort to “turn Harris County blue,” they have two goals: do away with photo ID, and get rid of electronic voting.

Unfortunately, electronic voting can prevent counting errors, but it can’t prevent all other versions of election fraud.

Picture a voting place with no alternate judge or poll watcher. There may be people on the rolls that haven’t voted yet. The judge can put the voter’s name on the list, print out a code, and hand it to a friend to cast a ballot. If the person comes in and signs the book later, they still get to vote, and, unless there is a recount, no one will notice that person’s name twice on the voter list.

Or the judge might print out the code and cast the extra vote without even bothering to put a name on the voter list. At the place I worked as a poll watcher in 2012, there were 20 additional votes at the end of the day that were not recorded on the voter list. The presiding judge made no attempt to reconcile the error; she just shrugged her shoulders and assumed the clerks had failed to write the names that many times. That’s possible, but I hadn’t witnessed any failures. The difference in the election nationwide was about 10 votes per precinct. 

I did file the discrepancy in my report, but voter fraud is notoriously hard to prove. In some of the most egregious cases, the only outcome is to prevent that judge from being hired again.

The best prevention is to have an alternate judge and poll watcher in every polling place. Harris County has a pretty good program to train and recruit towards this end, but it’s still a challenge. And it’s not being done in many voting places around the country.

The good news, then, is that people in favor of free and fair elections are getting better technology to accomplish that. But the bad news is that people against free and fair elections will continue to find ways to disenfranchise the voters who disagree with them.

No comments:

Post a Comment