EVERY Precinct Count Optical Scan (PCOS) machine used in our last two elections took digital photographs of each ballot that was fed into it, creating a comprehensive record of how the ballots appeared at the very moment they were cast. These digital images are then used after the elections, in case the final results are questioned and a recount becomes necessary.
In the last elections, during the Random Manual Audit (RMA)—the process by which ballots from select precincts are recounted immediately after the polls—a number of these digital images showed vertical lines of varying width, running from top to bottom, some of them crossing over the ovals that voters need to black out to signify their vote. In short order, this phenomenon—consisting as it did of lines appearing on digital images
—was christened “digital lines.” With no evidence to prove anything either way, these digital lines quickly became touted as a kind of smoking gun to prove that the PCOS machine manipulated the results of the last elections.
So it wasn’t a surprise that, at the May 14 hearing of the Joint Congressional Oversight Committee on the automation of elections, one of the most-anticipated topics up for discussion was the matter of these digital lines.
Almost as soon as the topic was brought up, hands shot up into the air as nearly every other resource person in the room wanted to get on the record with his or her theory on where those lines came from, what they signified, and well, you get the picture. And then the head of the Technical Evaluation Committee (TEC)—created under the Automation Law, to certify the automated election systems—delivered his committee’s report.
Early on, the presentation categorically stated that the appearance of these lines was a common problem in optical scanning devices, where it is the object being scanned—such as the ballot—is the one which moves over a stationary scanning element. The presentation also clarified that it was not even a problem unique to the Philippines, citing similar issues in recent elections in Mongolia.
According to the presentation, the most common cause of digital lines was foreign-matter contamination. Things like dust particles stuck on the surfaces of ballots could become temporarily lodged between the ballot and the scanning element and, essentially, block the scan—technically referred to as “optical- path disruption”—as the paper moved through the PCOS and into the ballot box.
Ink stains would have the same effect, as would bits of paper, insect body parts (that one elicited a collective “eww” from some of the younger people in the session hall at the time), and even just scratches on the thin and transparent celluloid sheet—called mylar—covering the PCOS’s scanning elements.
By this time, the presentation had made it abundantly clear the so-called mysterious digital lines were neither mysterious—the underlying clauses were physically demonstrable and the resulting lines repeatable—nor even digital in origin. However, there was still the issue of how these lines affected the outcome of elections.
Based on the study conducted by the TEC in December 2014, on ballots coming from 383 precincts, 84 percent of the 200,403 manually reviewed ballot images did not show any line. And for those images that did show lines, the results of 94.7 percent were not affected at all; overall, a mere 0.55 percent of national positions and 0.28 percent of local positions were affected.
As it turned out, the decision by the Commission on Elections (Comelec) to set thresholds for mark recognition—the percentage of the oval that needed to be covered by a mark in order for it to be recognized as a vote—proved to be a significant safeguard. Although in 2013, this threshold was set lower than 50 percent, which was not low enough that a mere straight line through the oval would trigger recognition as a vote. While some of the lines that crossed the ovals did get marked as votes, the number of such incidences was too low to be significant.
After the TEC’s sobering presentation of these findings, much of the speculation ceased and the way forward for automation was that much clearer.
Obviously, these lines were not something that could be shrugged off, and the Comelec laid out its plans to replace all the mylar sheets used in the PCOS, and to require that all machines either be equipped or—in the case of the PCOS machines already owned by the government—retrofitted with software that will enable the detection of conditions that would give rise to these lines. This apart from the extra attention being given to the kinds of paper being used for ballots; the kinds of marking pens that will be rolled out on election day; and the establishment of procedures to minimize the problem of foreign materials getting into the PCOS machines.
No question about it. The Comelec dodged a bullet with this one. Time to make sure it doesn’t happen again.
****
James Arthur B. Jimenez is director of the Commission on Elections’s education and information department.