A Broken Ballot?



Saturday’s editorial is the first in a two-part series on the need for a better voting machine system, both in Horry County and the state as a whole:

Nothing is more important to voters than knowing that their votes count. For S.C. voters, however – and Horry County voters in particular – that assurance has been tough to come by.

It’s hard enough to convince people to get out and vote these days. Too many believe their individual vote doesn’t matter. Iowa’s recent GOP caucuses, where Mitt Romney seems to have topped Rick Santorum by just eight votes, should be enough proof that it only takes a few people to sway an election. That vote count is now under review after questions about its accuracy, which could make those few votes even more important. Unfortunately, if the same close outcome were to occur when S.C. voters go to the polls on Jan. 21, we’d be less than sure of either its validity or our state’s ability to check the numbers.

A report put together late last year by the S.C. League of Women Voters and computer experts took a hard look at the 2010 general election, comparing vote totals and records from the state’s electronic voting machines to certified counts and to data obtained by the S.C. State Election Commission. The scope of the problems they uncovered is breathtaking.

Statewide, large swaths of data used to double-check vote totals and ensure their accuracy is simply not available or missing, meaning a verification of the results after the fact is functionally out of reach. The problem is not merely one of a county dragging its feet on providing the data to an outside group. The election commission too had the same problem in compiling backup data on the state’s electronic voting machines.

According to the report issued in September, only a third of the state’s counties – Georgetown among them – provided the state with all of the data requested on its vote totals. The other two-thirds of state counties, as the report put it, “did not produce the audit data that would allow for the transparency and reassurance of third-party verification of the results (to say nothing of the state election commission to conduct its own verification of the results).”

Horry County is one of the worst offenders. It wasn’t hard to jog the memory of Duncan Buell, a computer science professor at USC and one of the report’s authors, on Friday, more than three months after the report was released. Buell, a voting machine expert, said that “Horry County has the messiest of the data, perhaps, in the entire state.”

In Horry County, the election audit concluded that 183 ballots from two terminals – 114 at the Racepath #1 precinct and 69 at Nixons Crossroads – were not counted. As we’ve just seen in Iowa, 183 votes could very well mean the difference in the close GOP primary race. How can such a thing occur? To understand, it means knowing how the votes are gathered and totaled.

Each precinct worker is given a handheld Personal Electronic Ballot which he or she uses throughout the day to open a terminal and allow a single voter to cast their ballot. After the polls close, every terminal at a polling place is supposed to be closed with a single PEB, which would then carry the vote totals for the entire precinct. That PEB is then transported to election headquarters to be added to the totals from all of the other polling places. Seems simple enough. If you have 115 precincts, you expect to have 115 of these handheld PEBs to download to a single system, giving a total for the entire county.

In the case of Racepath #1 and some other precincts throughout the state, however, more than one PEB was used to download the vote totals from the machines in 2010. There were four voting machines at Racepath. The votes from three of the machines were downloaded to one device, and the ballots from the other, through some sort of miscommunication or error, were downloaded to a separate device. The folks at election headquarters were expecting only one PEB with totals and so used only the first PEB. The votes downloaded from the fourth machine sat on the superfluous PEB and didn’t get added to the county total.

In other words, there was some simple human error, and no backup system caught it, at least until voting experts and the League of Women Voters got involved. As Buell put it, it’s a system that will work perfectly, as long as everybody does their job perfectly.

The assumption on the part of the State Election Commission, he said, is that “if you just list the 20 steps, everybody will follow them perfectly and everything will work.”

Unfortunately, as we all know from working with people and computer systems, people are not perfect and computer systems can fail. The SEC – an agency Buell said “has an excess of hubris and a shortage of skepticism” – has been slow to accept these facts, but hopefully the publication of reports like this one will help reveal the shortcomings of our current system and lead to some needed changes.

Tomorrow: Solutions for a better ballot

 

A report put together late last year by the S.C. League of Women Voters and computer experts took a hard look at the 2010 general election, comparing vote totals and records from the state’s electronic voting machines to certified counts and to data obtained by the S.C. State Election Commission. The scope of the problems they uncovered is breathtaking.

Statewide, large swaths of data used to double-check vote totals and ensure their accuracy is simply not available or missing, meaning a verification of the results after the fact is functionally out of reach. The problem is not merely one of a county dragging its feet on providing the data to an outside group. The election commission too had the same problem in compiling backup data on the state’s electronic voting machines.

According to the report issued in September, only a third of the state’s counties – Georgetown among them – provided the state with all of the data requested on its vote totals. The other two-thirds of state counties, as the report put it, “did not produce the audit data that would allow for the transparency and reassurance of third-party verification of the results (to say nothing of the state election commission to conduct its own verification of the results).”

Horry County is one of the worst offenders. It wasn’t hard to jog the memory of Duncan Buell, a computer science professor at USC and one of the report’s authors, on Friday, more than three months after the report was released. Buell, a voting machine expert, said that “Horry County has the messiest of the data, perhaps, in the entire state.”

In Horry County, the election audit concluded that 183 ballots from two terminals – 114 at the Racepath #1 precinct and 69 at Nixons Crossroads – were not counted. As we’ve just seen in Iowa, 183 votes could very well mean the difference in the close GOP primary race. How can such a thing occur? To understand, it means knowing how the votes are gathered and totaled.

Each precinct worker is given a handheld Personal Electronic Ballot which he or she uses throughout the day to open a terminal and allow a single voter to cast their ballot. After the polls close, every terminal at a polling place is supposed to be closed with a single PEB, which would then carry the vote totals for the entire precinct. That PEB is then transported to election headquarters to be added to the totals from all of the other polling places. Seems simple enough. If you have 115 precincts, you expect to have 115 of these handheld PEBs to download to a single system, giving a total for the entire county.

In the case of Racepath #1 and some other precincts throughout the state, however, more than one PEB was used to download the vote totals from the machines in 2010. There were four voting machines at Racepath. The votes from three of the machines were downloaded to one device, and the ballots from the other, through some sort of miscommunication or error, were downloaded to a separate device. The folks at election headquarters were expecting only one PEB with totals and so used only the first PEB. The votes downloaded from the fourth machine sat on the superfluous PEB and didn’t get added to the county total.

In other words, there was some simple human error, and no backup system caught it, at least until voting experts and the League of Women Voters got involved. As Buell put it, it’s a system that will work perfectly, as long as everybody does their job perfectly.

The assumption on the part of the State Election Commission, he said, is that “if you just list the 20 steps, everybody will follow them perfectly and everything will work.”

Unfortunately, as we all know from working with people and computer systems, people are not perfect and computer systems can fail. The SEC – an agency Buell said “has an excess of hubris and a shortage of skepticism” – has been slow to accept these facts, but hopefully the publication of reports like this one will help reveal the shortcomings of our current system and lead to some needed changes.

Tomorrow: Solutions for a better ballot