Engineers are problem-solvers. But is there an engineering solution to long lines at the voting precincts? A professor of industrial and systems engineering at the University of Rhode Island and a team of researchers are working on it.
Using data from precincts that had long wait times in the 2016 general election, gleaned from the entry check-in computers and the scanners for completed ballots, the URI engineering team is now creating simulations of what happened in troubled locations, and identifying potential solutions that could be used to make the voting process more efficient.
Led by Gretchen Macht, a URI assistant professor of industrial and systems engineering, the team is working under a $226,942 grant from the R.I. Board of Elections, R.I. Department of State, URI College of Engineering and an independent, private foundation.
The work broaches both queuing theory and facilities layout planning. URI got involved after the 2016 election, when Secretary of State Nellie M. Gorbea contacted the URI engineering dean and asked if he knew of anyone who could help streamline the precinct processes.
In an interview, Gorbea said she acted as a matchmaker, essentially, because the recommendations that eventually come from Macht’s team at URI will be presented to the R.I. Board of Elections.
“After the 2016 election, I convened a task force of state and local officials,” Gorbea said. “It wasn’t just about the machines, it was about the entire voting experience.”
Operational engineering allows the use of technology to make more-efficient systems. “We have a fantastic engineering program at the University of Rhode Island. I figured there had to be someone who could do this,” she said.
The work at URI began in 2017 and is expected to continue through June 2019. But the hope is that at least some of its research will be available to the state before the November election this year.
Macht said the team is using entrance data from the poll pads at several Rhode Island precincts. These portable computers allow poll workers to check the registration of voters, by scanning their driver’s licenses.
The URI team also has the exit data from the new optical scanners purchased for the 2016 election, which electronically read the ballots that are fed into the machines.
The information provided includes the volume of voters and the timing sequence but does not include identifying information about the voters, according to Macht.
“The first year has been focusing on getting data, cleaning data and then understanding what all this data says,” Macht said. “We can then simulate it in the model.”
Her graduate and undergraduate research team of three includes James Houghton, a graduate research assistant and a 2017 URI graduate; Ahmad Siddiqi, an undergraduate research assistant and a mechanical engineering major; and Nicholas Bernardo, a graduate research assistant who graduated this year from URI with a degree in industrial and systems engineering.
The simulation software in their classroom allows the URI students to create a series of simulated polling locations, based on the actual precincts in Rhode Island. Once they enter the data, the simulation software generates cartoon-like humans walking from polling site check-in tables to voting booths, and then to the optical scanners.
‘There was no “ding.” So, people were standing there, and then … putting [the ballot] in the wrong bin.’
GRETCHEN MACHT, URI assistant professor of industrial and systems engineering
The experience of 2016 varied by locations among Rhode Island’s 420 precincts. The new technology purchased that year by the state had the effect of moving people more quickly through the initial check-in.
But lines formed at the voting booths in precincts where the ballots were lengthy, including in Pawtucket and Westerly.
And then, in many precincts, delays came at the final stage, the optical scanning of ballots. The new optical scanners that read the votes take four seconds to process, about twice the time of the previous technology, Macht said.
This wouldn’t necessarily cause a delay, but the machine didn’t have any audible bell or sound to tell a voter that the work was completed.
“There was no ‘ding,’ ” Macht said. “So, people were standing there, and then they were putting it in the wrong bin.”
Using data that described the time elapsed from these experiences in several precincts that had both the new optical scanners and the check-in poll pads, the URI team created code that calculated how much error was associated with humans, or machines, Macht said, and how much time passed.
In one challenged precinct, whose experience in 2016 was entered into the simulation software, Macht and her students determined that adding one more optical scanner resulted in a reduced time, on average, of 30 minutes of wait time per voter.
Why did this conclusion take an engineering solution? Rather than relying on varying opinions of what was happening between check-in, voting booth and exit scanner, the team wanted to calculate what had happened.
“What we’re trying to do is not have those opinions [of people] in the room but to mathematically quantify what actually happened,” Macht said.
The goal, Gorbea said, is a more-efficient system in future elections.
“The work that professor Macht is doing with the Board of Elections should help us have a better election-day experience.”
