Any social media conversation on automated elections today gets hijacked within an hour by someone piping up what they no doubt feel is a clever retort about how automated elections can’t be trusted or that people are clamoring for a roll-back of the Automated Election System for either a fully manual system or a hybrid election system. These retorts are seductive in their simplicity but they are demonstrably false. So, let’s start with a couple fact checks.
First, it is a fact that despite hundreds of protests filed against election results since 2010, there have been zero instances where the challenged election results were overturned. Otherwise stated, contrary to the countless and frequently repeated allegations of fraud, no one – not in the twenty years since the country started using automated elections – has been able to definitively prove, before any forum, that fraud was committed, let alone fraud on the scale that would affect the outcome of an electoral race.
And second, it is also a fact that despite the persistence of the narrative that Philippine elections have been rejected by the broader public, that fiction isn’t supported by empirical data. After the 2016 National and Local Elections, the general assessment of election watchdogs such as National Citizens’ Movement for Free Elections (Namfrel), the Parish Pastoral Council for Responsible Voting (PPCRV) and The Carter Centre showed that Filipinos had more confidence in the automated election system (AES) as compared to previous automated elections.
At the same time, a Social Weather Station survey conducted between February 2016 and March 2016 indicated that the Commission on Elections (COMELEC) had the most improved sincerity rating among all government agencies, from -6 in 2015 to +12 in 2016—a positive difference of 18 percentage points. This positive image was sustained despite the constant criticism and various legal challenges faced by the Commission during the election period.
Such views were also reaffirmed by the COMELEC Advisory Council (CAC) — a group of technical experts and resource persons from government and the information technology industry tasked to evaluate the conduct of the elections and provide recommendations to COMELEC —through its report issued in March 2017. The report stated that the “COMELEC definitely showed competency and independence in the conduct of the 2016 NLE,” and that “despite all the adversities faced in the election preparation and unforeseen debacle, the use of the AES in the 2016 NLE was successfully conducted.”
And in 2019, Pulse Asia reported that 91% of respondents said they “would like to see automated voting continue in future elections;” 87% said they “were satisfied with the automated polling system or counting of votes through the vote counting machine (VCM);” 94% said it was “easy to vote using the VCM;” and 94% said the May 2019 elections in their areas was “orderly.” This tracked the findings of SWS for the same period, which showed that an average of 86% of voters said the election results are believable, while 80% were satisfied with the conduct of the elections.
And third, while it is true that there is some demand for hybrid elections, that clamor is mostly confined to the Hybrid Elections Lobby: persons and organizations who have latched on to the idea that somehow, the first and second facts I mentioned don’t exist, and that returning to manual counting will somehow solve everything. They premise this conclusion two fundamental assumptions: first, that the counting process needs to be physically observable for it to be credible; and second, that it doesn’t really matter how long counting takes, as long as it is physically observable.
With the 2022 National and Local Elections just around the corner, it is only appropriate that we now take the time to interrogate these assumptions.
The Question of Transparency
The cornerstone premise of the Hybrid Elections lobby is that the Automated Elections System – or AES – the country has been using since 2010 isn’t transparent enough because the counting of the ballots is done by the vote counting machines, and not by humans. According to this lobby, this “lack of transparency” makes the AES untrustworthy, negatively impacting the credibility of our elections.
I would argue, however, that there is no lack of transparency, only the use of a different kind of transparency. According to the Hybrid Elections lobby, transparency essentially requires that the counting and tabulating process at the polling places – where the primary election reports, i.e., the election returns, are prepared – must be physically observable with the naked eye. Admittedly, this is not possible under the present design of the AES, any more than it is possible for anyone to physically observe mathematical operations being carried out on a calculator’s microchip. However, this lack of physical observability does not render a calculator untrustworthy.
What would demolish a calculator’s credibility is for it to make a mistake in performing the mathematical operations which are its raison d’être. No one would use a calculator that returns a result of 27 after you key in 2 plus 7. If you enter 2 plus 7, you ought to be able to say with confidence that the result will be 9; and that even if you did the addition using a different method – say counting marks on a piece of paper – the answer will remain the same. To put it very simply, in order for a calculator to be trustworthy, the results must be both predictable and provable.
NO ONE – NOT IN THE TWENTY YEARS SINCE THE COUNTRY STARTED USING AUTOMATED ELECTIONS – HAS BEEN ABLE TO DEFINITIVELY PROVE, BEFORE ANY FORUM, THAT FRAUD WAS COMMITTED, LET ALONE FRAUD ON THE SCALE THAT WOULD AFFECT THE OUTCOME OF AN ELECTORAL RACE.
At its core, the vote counting machines used in the AES are very much like calculators except that they can only do one operation: addition. Each mark appearing on a ballot for a candidate is considered an additional vote to the running tally inside the counting machine’s calculator brain. It therefore stands to reason that, in evaluating the credibility of these counting machines, they ought to be subjected to a similar kind of test as one would put to a calculator – one that demands the predictability and provability of the outcome. If ten ballots with one vote each for a candidate are entered into the machine, the resulting count should be ten; and when we check all other ballots that went through the same machine, it must also be proven that no other ballots showed any marks for the candidate in question.
Anything beyond this fundamental standard would add nothing to the credibility of the outcome – or the system that produced it – and would arguably be superfluous.
And so the question is: Despite the physical observation of the counting being impossible, is there a way to check the predictability and provability of the results generated by the AES?
The answer lies in the design of the ballot and the vote counting machine.
The Design of the Ballot
In searching for the best automated elections solution for the Philippines, the COMELEC tried out two very different technologies: Direct Recording Electronic (DRE) which is more commonly referred to as electronic voting; and Optical Mark Reader (OMR) technology, which required the use of a paper ballot paired with a ballot counting machine. Two separate election systems were deployed in the 2008 General Elections in the Autonomous Region in Muslim Mindanao: DRE in Maguindanao, and OMR in the rest of the five provinces – Lanao de Sur, Shariff Kabunsuan, Basilan, Sulu, and Tawi-Tawi.
When the dust cleared, the COMELEC rejected DRE and cast the nation’s lot with OMR. And so in 2010, two years later, the COMELEC rolled out the first ever AES using a system that combined the use of specially designed ballots and vote counting machines located in every single polling place. Incidentally, yes, when we say AES, we don’t just refer to the counting machines, but to the combination of special ballots and counting machines.
One of the greatest challenges to the predictability of counting ballots by hand is the wide range of variation possible with handwritten ballots. While many people write clearly enough to leave no doubt as to who they intended to vote for, far too many ballots return with illegible scrawls that defy easy deciphering. Worse, many votes are not even in the names of the candidates, with people writing in titles, honorifics, or nicknames – none of which can be definitively attributed to one specific candidate.
This is at once the reason why manual counting takes a very long time – since every questionable ballot gets challenged by the representatives of the candidates – and why manual counting is inherently unreliable. The Boards of Election Inspectors who get to decide who will receive a contested vote are often subjected to a wide range of attempts to influence their decision making – from subtle bribery to outright death threats.
With the AES however, the system of writing the names of the candidates on the ballot was replaced with the pre-printed names of the candidates with matching voting ovals next to them. This method of voting requires that the voter fill in enough of the oval for the vote counting machines to recognize the mark as a deliberate vote, rather than an accidental smudge.
As a further safeguard, the Automated Election Law – Republic Act 9369 – mandates that every election, a Random Manual Audit be conducted where a physical examination of the ballots are carried out in order to determine whether the vote counting machines have properly interpreted the ballot marks. Since 2010, the Random Manual Audit has consistently shown that the vote counting machines have a high degree of accuracy in reading these marks – 99.995% in fact – making ballot interpretation highly predictable.
The Design of the Vote Counting Process
Knowing the necessity for ensuring that the general public could trust the system, the vote counting process was designed to provide tangible proof of the voting results at the precinct level, within minutes from the close of voting. This was achieved by having the vote counting machines print out election returns as soon as the voting ended.
These returns – like long cash register receipts – showed the names of the candidates and the corresponding number of votes they received, and were distributed to the watchers at the polling place. Only when the distribution of the election returns was completed did the vote counting machine get connected to the electronic transmission system which sent the digital version of the election returns to the central canvassing station at the level of the municipality. At the same time, the digital results would be sent to a live website where the public would be able to remotely access those precinct results. It wouldn’t matter where you were in the world, as long as you had an internet connection, you could – in near real time – check on the results coming out of polling precinct in your home town.
With these systems, typically less than an hour after the close of polls, there would be at least two records of the election results that are easily accessible to the public: a printed copy in the hands of poll watchers at the precinct itself, and a digital copy on a website accessible to everyone in the world. A third record exists which isn’t as easily accessible to the public but which is routinely used in settling election protests: the encrypted image.
Whenever a ballot is slipped into the vote counting machine, a digital image of the front and back of the ballot is automatically captured and encrypted into the vote counting machine’s memory. This image is a captures the state of the ballot at the very moment it leaves the hand of the voter, making it a perfect record of the voter’s intentions.
These three records taken together form the test of provability, i.e., they prove that the election results arrived at by the canvassing system – first at the municipal level, followed by the provincial level, all the way up to the national canvassing level – are accurate. All you have to is check the math. If the municipal canvass results match up with the known number of votes received by each candidate as evidence by the election returns, the online figures, and the encrypted ballot images, then you know that the mathematical operations that produce the election results have gone off flawlessly.
PERHAPS THERE CAN BE A WAY TO INTEGRATE MANUAL VERIFICATION PROCEDURES; TO OVERLAY, IN EFFECT, A MANUAL COUNT ON TOP OF THE EXISTING AUTOMATED COUNT.
Beyond Transparency
Acting together, the ballots and the design of the vote counting process create a level of transparency in the system that is at least equivalent to that provided for by the physical observability of a manual count, but which is also superior to it in other ways.
Beyond the need for transparency, the credibility of the electoral process also depends on the speed with which the result are known.
It is human nature that the acceptability of an answer decreases in inverse proportion to the time it takes to receive the answer. In other words, the longer you have to wait to get information you asked for, the less likely you are to trust the information when do get it. This is consistent with our shared experience of elections prior to the advent of automation when election results would take weeks and even months to be reported. It was so bad that when the results finally came, hardly anyone believed the results to be accurate.
As it turns out, these fears were not entirely unjustified. We have since found out that those long delays were routinely used as a smokescreen to obfuscate large-scale cheating operations. So how did those delays happen?
Long wait times are typically associated with the realities of manual counting – realities that are often ignored whenever someone in the Hybrid Elections lobby rhapsodizes hand counts.
First of all, every ballot is expected to contains at about 30 names – 12 Senators, 1 Party-List Representative, 1 Member of the House of Representatives, a Governor, and Vice Governor, a Mayor and Vice-Mayor, and a variable number of Sanggunian positions. That’s a lot of names to get through. And literally every name on that ballot can be challenged – as they often are whenever voter intent is unclear. In time and motion studies conducted by the COMELEC where the intervention of lawyers and election watchers wasn’t even simulated, it was often found that counting just thirty ballots – approximately a thousand votes for individual candidates – could take upwards of three hours – imagine if you had to count even just 400. All told, these factors combine to guarantee that a manually counted elections will not be able to produce final precinct results within the 2-3 hours that the public have gotten used to since the AES was introduced in 2010.
And the long wait for the count to be over isn’t even the most problematic aspect of manual vote counting. Ultimately, the decision to award the vote to a particular candidate rests solely on the Board of Election Inspectors, but the BEI doesn’t decide in a vacuum. Whenever manual elections are held, the members of the BEI typically bear the brunt of the unwelcome attentions of politicians and their partisans. In order to ensure a “friendly” BEI, they are wined, dined, cajoled, bribed and – on the darker end of the spectrum – intimidated, harassed, subjected to lawsuits, and have their lives threatened. Of course, the Hybrid Elections lobby believes that the long wait times are irrelevant, conveniently forgetting all those times in the past when delayed election results at various canvassing levels have not only caused discomfiture at the national level, but have resulted in massive harm – in one case, even death – to election workers.
Conclusion
Ultimately, the Hybrid Elections lobby’s campaign to roll back the fully automated election system alleges no provable problem with the existing AES. The lobby has a lot of allegations and an endless stream of speculative scenarios of how thing can go horribly wrong – but nothing to prove that they have, or actually will. And yet, the lobby would have us discard the proven track record of automated elections, the empirical proof of its acceptability to the general public, and the well-established record of speedy election results in favor of a largely hypothetical construct that has not been proven in real world conditions, that seeks to resurrect old election practices known to create vulnerabilities in the system, and admits that its use will result in the kind of delays in the release of elections results that they themselves used to call unconscionable.
There is however the possibility of a middle ground. Instead of advocating the full roll-back of the AES – which would result in the loss of all the benefits gained from automated vote counting – perhaps there can be a way to integrate manual verification procedures; to overlay, in effect, a manual count on top of the existing automated count.
It can be done, and it has been done in other jurisdictions. All it really needs is for the Hybrid Elections lobby to approach the possibility with an open mind – a sincere desire to design the best possible system, rather than to simply burn the house down.