Lessons from the Last Mock Elections at UNTV 206.0

More and earlier mock elections would have exposed glitches and helped adapt to change

According to its main goal, the mock elections on Thursday, April 29, can be considered a failure because only 500 voters came for a 50% turnout.

The mock election was not able to simulate the targeted 1,000 voters and test for congestion from clustering.

Elements of Waiting Lines. Queues are not determined only by processing capacity, say in voters per hour, as claimed by some in press releases. It is also determined by targeted voter waiting time in the line and, very important in the case of this automated elections, the peaks and valleys of voter arrival rate.

Everybody is being encouraged to come early that, in effect, creates a peak profile in the morning. This can contribute to long lines and waiting times in the morning.

Still, this process is heuristic because the total number of voters is bounded. i.e. a crowd in the morning means no crowd in the afternoon. If the morning flash point does not explode the afternoon ought to be a breeze.

Long waiting time. As I expected, the mock elections at UNTV highlighted the congestion problem especially in the early hours. The 201st voter had to wait in line for four hours coming on line before 7:00 a.m. and voting at 11:00 a.m. This means that the voters ahead of him waited up to four hours to vote as well.

Still, if we go into the details, the mock election was a minor success because we can glean certain lessons useful to avoid or minimize problems on election day.

Results of the UNTV mock elections.

• Only 500 votes were accepted over the eleven hours from 7:00 a.m. to 6:00 p.m. because just around that number came to participate
• Three votes were rejected: one for marks on registration field (will not happen in real elections) and two for ambiguous marks (unshaded spaces within the oval).
• Two incidents of Precinct Center Optical Scan (PCO) machine down times related to power supply issues. This were minor glitches easily fixed with machines up quickly. The machines were run on batteries to check the 14 hour life and this was proven to work.
• Hourly, theparticipants able to voter were:

• 7:00 to 8:00 – 28 voters
• 8:00 to 9:00 – 37 voters
• 9:00 to 10:00 – 55 voters
• 10:00 to 11:00 – 84 voters
• 11:00 to 12 noon – 70 voters
• and afternoon (6 hours) 226 voters for a total of 500 voters.

Lesson for Noynoy campaign. Still, for the Noynoy Aquino camp, for example, who leads by a big margin in Metro Manila, they can ask this worrisome question off the results of the mock elections: how many voters would be disenfranchised because they are not able to wait four hours in the morning of election day?

Key lesson in Change management. For managers of change, the key lesson with this mock elections is the need for more rehearsals and early on.

THE PCOS machine and technology is not the bottleneck causing congestion from clustering. Given the current GI 8786 for the BEI, the bottleneck is the manual process mandated despite the high volume processing from clustering.

The increase in hourly processing rate between 7:00 – 8:00 a.m. at 23 to the highest of 84 at 10:00 – 11:00 a.m. shows he benefit of the learning curve – from learning-by-doing – especially at the Board of Election Inspector (BEI) at the verification table.

All the issues, like the power supply problems, happened in the early morning. After the experience of processing through the first 110 voters in the first three hours, the BEI went up the experience curve and was able to raise the processing rate to 84 per hour.

Unfortunately, the voter arrival profile was such that their were 200 voters on queue by 7:00 a.m. And so, because the BEI verification was inefficient in the first three hours, the 200th voter was served at 11:00 am or four hours after first lining up.

This illustrates the need for more rehearsals via mock elections early on to practice and improve the service rate at the beginning of the election day.

I did an analysis and the riskiest time for congestion on election day is early morning when, from anecdotes, a peak of the voter arrival profile was high (the other was at the end of election day).

Thus, when the high arrival rate happens early in the morning at the same time as the BEI is still ramping up its efficiency in voter verification will result in long waiting times at that time.

On election day, this waiting can be a flash point just as the morning temperature also rises. The more bothersome issue is: how many people effectively choose to be disenfranchised because of the prior failure to avoid queues?

So, if the problems identified and solved in the first three hours in the mock elections are encapsulated and spread out to all the BEI teams especially in the cities then it will go a long way to accelerate the learning and reduce waiting time below the three – four hours experienced in this mock elections.

Other lessons. There are other lessons to be gleaned from observing the details:

• There is cris-crossing because the voters have to return to the Election Day Voters Computerized List (EDCVL) twice on verification and for fingerprinting. There are two queues in front and behind the BEI Chairman thus the congestion. It may still be a big help if Res 953 proposals can be implemented to eliminate cris-crossing and allow circular flow.
• The BEI staff, especially the Chairman, should bring snacks because there is practically no time to eat. If the Chairman needs to go to the Comfort Room anytime during the eleven-hour day, the verification and the voting will stop. (For this mock election, they did have the time for lunch but that is only because only 50% of the expected voters turned out.)
• There may be a need for an amendment in GI 8786 to allow the BEI Chairman to delegate his verification duties when he experiences nature calls to avoid a shutdown of the voting process when he has to leave the room.
• There may be a need to increase the number of spare support staff under the Voters Assistance Center to allow for nature calls by the BEI staff assigned at the voting center.
• The Voters Assistance Center (VAC) as promulgated in Res 8786 is effective in controlling the volume of voters entering the voting center in batches of three to five. The control point used in the morning when the verification table is the bottlenecks is based on the number of voters before the BEI table. The secondary control point can be when all voting chairs are occupied.
• If the Comelec has delegated the function of crowd control at the VAC to the PPCRV then the PPCRV must ask for a GI revision that empowers them to control entry into the voting center or they will be in trouble on election day. Res 953 suggests that one BEI staff be the crowd control officer but this has not been included in Res 8786 on March 4.
• It seems the instruction was given to limit the voting chairs to ten. If the room can accommodate more, this should be increased. It would seem that the mock elections show that the ten chairs is the correct decision but that is only because of the inefficiency of the prior verification process as mandated.
• I expect that, on election day, the bottleneck will quickly appear to be the chair especially when the voter hand holding tolerated at the mock elections will be disputed by the election watchers on election day.
• Without Paolo Bediones and the photo opportunities in the afternoon, the lack of a line before the PCOS machine tend to support my contention in various posts in SYNTHESiST that the machine will not be a bottleneck because it is fast.

Conclusion. Based on the actual 84 voters per hour attained, voting centers with about 600 voters should have no major problems except for the first two hours when the BEI team is learning up to steam through learning-by-doing. This means that most rural voting centers will not have congestion issues.

As I mentioned in a previous Post 196.0, the possible congestion hot spots that I have listed have much mire than 800 voters per voting center, more than five precinct clustered per voting center, and large schools.

Obviously, if voter turnout goes below the historical 80% to 50% in the cities, there will also be no congestion problem even in Metro Manila.

In a chicken-and-egg question, would the low voter turnout (if only 50% vote in Metro Manila happens) be the result of voter apathy or congestion problems.

Nothing in mass media, thus far, highlights any projection on voter turnout. In this elections, it could very well be a big determinant on who wins the election!

Post Script. This post will be my last before the 2019 automated elections on May 10. I personally would like to express my appreciation to Commissioner Goyo Larrazabal for all the hard work.

I have met him two or three times in a professional meetings but have seen him on mass media many times. I think his openness and coolness under fire has kept the atmosphere likewise cool as we approach the automated elections.

The Comelec is not a monolithic institution and there seems to be some bad eggs among the majority of the hardworking and good ones.

When this automated elections is done, its success can only be attributed to the tenacity of the Comelec board to implement automated elections despite the odds.