Comelec Plans to Reduce Congestion from Clustering are Inadequate 187.0

Res 953 was enacted en banc on 12.22.09 as a complete plan

Minute Resolution 953 was promulgated one week before the first Res 8739 General Instructions for the Board of Election Inspectors (GI). If, implemented fully, it can greatly minimize congestion from clustering.

Yet, it was not made part of that first GI for the 2010 elections.  Strange.

Revised GI Res 8786 promulgated on March 4, 2010, two months after the first GI Res 8739, did not take up all the provisions approved in Res 953. Stranger still.

Without picking up the other en banc approved provisions of Res 953 in an updated GI resolution as discussed further below, many clustered urban polling stations will have unmanageable queues on election day, May 10, 2010.

(Note 3/27/10: Updated title and lede.)

Time is the Enemy. Giving full benefit to the Comelec, Res 8739 may have been purposely focused on changes that came from the introduction of Precinct Count Optical Scan (PCOS} technology on the manual GI to train the voters one issue at a time.

Still, time is short to set up and train all concerned and nationally for the congestion from clustering aspect – the other twin problem to technology – in the new automated election. Time is the silent enemy.

I was given a copy of Res 953 recently. I append this copy at the bottom of this post and on the navigation bar above besides Res 8786.

For Updates. Please find new link to all automated elections posts on SYNTHESiST at the top of the sidebar at right and under SYNTHESiST Potpourri. I have also added key links to websites of interest to election observers.

Res 8786 is not enough. Not even the later revised Res 8786 GI promulgated on March 4, 2010, two months after the first GI, took up all the provisions approved in Res 953. Res 8786 still uses provisions taken from the old, manual GI that, for practical purposes, have been made unfeasible by the clustering of precincts.

Clustering came about because of the limited budget approved by R.A.9525 on March 3, 2009 that allocated, on average, one PCOS machine for five manual precincts. It can be considered a separate but twin problem to PCOS technology adoption.

Complete Poll Center Congestion Management Plan. A quick review of the en banc approved Res 953 shows that it has a more complete package of provisions focused on polling center management of congestion from clustering.

These provisions often differ from those of the manual polling center process derived from the Omnibus Election Code (Batas Pambansa 881 promulgated on December 3, 1985). That Omnibus Code is written more like an operations manual. Most veteran Election Officers (EO) know it by heart.

In the internal discussions within Comelec on Res 953, I understand, opposition from field EOs comes from practical differences between the new operations requirement driven by the reality of clustering and the traditional way of running polling center operations in the manual environment. These changes are discussed in detail below.

Legal Protection to BEI Staff and field Election Officers. Whatever the operations changes decided on, it is necessary that the final details to be implemented on the ground ought to be covered within a formally enacted General Instruction.

Unlike the instruction given in the BEI training in Manila on the week of March 4, 2009, the Comelec chiefs cannot verbally order the teachers of the BEI and the field EOs to adjust the actual operations to the need of the day. This will put the BEIs and EOs at risk from legal action by political parties and candidates if not in physical danger itself on election day.

Comelec leadership must have the courage to put things in writing to protect its own personnel and those they deputize.

Recommended Layout. Res 953 recommends a generic layout for a typical classroom (see page 3 of Res 953 at bottom of this post below). It strongly advocates publicizing this layout that was not done in Res 8786.

The key element of this layout is a circular flow of voters to minimize crisscrossing. It allocates sixteen positions for voters to shade the ballot with the security envelope that allows room for vulnerable sectors like the elderly and the disabled.

I understand that the layout was the one tested to have a capacity of 120 voters per hour publicly mentioned by Comm Goyo Larrazabal in an interview on DZMM on March 5.

Still, there are pre-conditions to that capacity announced by Comm Larrazabal that are specified in Res 953.

Crowd Control Support Staff. Anent to the new problem of congestion from clustering, a new function of crowd control is assigned to a BEI staff to be posted at the door by the en banc approved Res 953 as shown in the layout detail at left.

He will make sure that voters who come into the room know their sequence number from the posted Precinct Computerized Voters List (PCVL) in individual pages outside the voting room.

In the previous manual election system, there was only one precinct per room with an average of 214 voters. The clustered system will have as many as seven (7) precincts in one clustered polling station or a maximum of 1000 registered voters. Thus, there will be pages from as many as seven (7) PCVLs posted outside the room.

This is the main reason why the new position of crowd control BEI is created to make sure that only voters who know their sequence number (in addition to precinct number) can be allowed into the room.

Those who cannot find their sequence number must be directed back to the newly created Voters Assistance Center (VAC) discussed in an advance post in SYNTHESiST that Res 8786 created and that was the main adoption of that revised GI from Res 953.

The VAC will have the alphabetical list of voters precisely for this purpose initially.

In the manual system, voters needing clarification, who are not able to find their names on the PCVL posted outside the precinct, are allowed into the room (though a provision for the alphabetical list of voters is with the Principal).

The Election Day Computerized Voters List (EDCVL) and the Book of Voters are kept together at the BEI verification table in each polling center and remains the final reference.

This change in requirement for crowd control is only made for practical reasons due to clustering.

Veteran EOs are apprehensive about turning away voters from the room, who do not find their names on the PCVL, to the Voters Assistance Center. This new element of crowd control is made necessary by clustering and must be put in writing in a second revision of the GI to protect the BEI and the election officers.

More importantly, this can help avoid a new cheating technique of stuffing the queue with pretend voters who will slow down the process assigned by candidates who see themselves as weak in that polling center.

If I am asked to give just one reason for bothering to write this blog post on SYNTHESiST, this is it.

Ineffective crowd control could be a flash point for violence as the strong candidate in that same polling center may defend it with goons with guns during the day.

BEI Layout on the Voter Verification Table. Voter verification is the most important role of the Chairman.

The layout suggested by Res 953 puts the chairman in the second seat at the middle of the table and facing the voters streaming into the room.

The traditional manual process requires for the Chairman to do the verification including shouting out the name of the voter and final instructions.

This is possible with one precinct per room and an average of 214 voters per eight hour day. Those of us who have voted remember that this arrival rate of voters allows for breaktimes to the BEI staff and occasional free time with no voters in queue after the lunch break.

In the clustered polling centers with up to seven precincts or a maximum of 1000 voters, this relaxed atmosphere is not feasible.

At the Maharlika Mock Elections

Total Voters and Arrival Rate. The most crowded clustered centers may have as many as 800 voters to process on election day (at 80% turnout). This is equivalent to about 73 voters on average over the longer 11-hour voting day.

Note that 73 voters is roughly one third per hour as much as the 214 voters that the previous manual system had to process in one day!

Of course, the average only exists in the mind of the mathematician.

Nobody in the Philippines has any idea on the actual profile of the arrival rate of voters for the first ever automated elections in 2010.

From anecdotes in manual elections, there are three peaks in the profile of the arrival rate.

The first peak in the first hour is of the maagap, the prudent. The second peak before lunch are of the masinop, the orderly, who have done the daily chores like the daily shopping at the markets. The third peak occurs just before closing are mainly among the short-sighted, the malapit ang pananaw who are awaiting the best price for their votes.

The actual extent of the peak in this arrival profile is not established. Comm Goyo Larrazabal’s 120 voters per hour capacity of the clustered precinct assumes that a 65% allowance over the average arrival rate of 73.

A quick risk analysis would show that the first peak in the morning represents the biggest risk if the arrivals at that hour exceed 120. For example, if 240 people out of the 800 expected arrive between 7 – 8 am, the 240th person will vote at about 9 in the morning after staying in the queue line for at least 60 minutes.

The same math applies to the other peaks.

Worse from the point of view of the BEI staff, especially the Chairman who may be the oldest and weakest, the sheer volume of voters will not really allow any break period for meals and CR visits.

The Comelec has not done any 800-voter simulation for one day. The only simulation they successfully did was for 50 voters that we reported about in a SYNTHESiST blog here.

The purpose of that simulation was to test the PCOS machine. It took almost two hours to process the fifty voters through using the manual layout.

I agree that the voter verification role of the BEI Chairman is the most important and cannot be delegated. Other roles like thumbprinting that Res 8786 proposes to be done by the voter at the EDCVL required by Part (d) of Section 36 Manner of Voting of Article IV Date, Time and Place of Voting of Res 8748 is the most worrisome.

Crisscrossing can Stop the Voting. The requirement in Res 8786 for the voter to return to the Verification table table after casting his vote can create the biggest problem of congestion and cause a heart attack on the BEI Chairman.

I skipped this but Res 953 specifies that thumbprinting be done upon completion of voter verification (Find this bit of info in detail of layout two photos above). In Res 953, staining with indelible ink is done prior to feeding the ballot to the PCOS machine.

This is a ticklish question because doing what Res 953 specifies assumes that, for all intents and purposes, the voter has cast his vote after voter verification stage and not after the PCOS machine has accepted the ballot.

Res 953 was enacted to solve the issue of congestion from clustering. Personally, I have no problem with looking at the polling center room with all processes in it as the point of voting.

Voter education, including (a) complete voter knowledge of the process and (b) fore-knowledge of who he will vote for, is the measure of intent in joining the process.

Of course, the Comelec, as is provided for in Res 953, is obligated to conduct a “massive information dissemination” campaign. Contractually, this can be achieved by the Comelec through the early mailing of the Voters Information Sheet (VIS) that contains essential information like name, polling center, precinct number and sequence number in the PCVL/EDCVL.

Anecdotically, I am told the Comelec has never complied with this legal requirement of B.P. 881, the Omnibus Election Code. The Comelec has not been able to clean up the computerized Voters List (CVL) in time to mail the VIS to voters 30 days before the election as required by law.

Ideally, other campaigns must be done through mass media and new media like the Internet to educate voters about the changes.

In this automated elections, the educated voter is (a) one who knows his/her sequence number in the PCVL as well a precinct and polling number and (b) who carries a list of candidates he is going to vote for into the polling center on election day.

These two minimum goals of voter education are necessary to minimize congestion from clustering.

I did note two mistakes in the first advertisements of Comelec that can be improved in succeeding ads.

First, the felt-tip pen used  in the ad is pointed; the one during election day will have a flat tip the better to shade and will use fast-drying ink to avoid smudging (could cause problems too if dries toof fast!).

Second, the ad seems to imply, with shading to be carefully done inside the oval that if ink goes outside the oval the vote will be “ambiguous” and rejected.  Requiring such precise artistry is difficult for the vulnerable sectors like the visually-impared and will greatly slow down the process at the bottleneck.  As I inquired, continuous extending ink beyond the oval is not “ambiguous” marking; it is actually an affirmation of voter intent to so clearly shade the oval.

Capacity of the Typical Voting Room. I suspect the true bottleneck in this automated voting process is the size of the classroom (say 8m x 5m) and the maximum number of seats (16 in Res 953) allocated for voters to shade-in their ballots.

If Comm Larrazabal is right – during the DZMM interview – to say that the voters will take between 5 – 10 minutes to shade the ballot, then the sixteen seats offer a theoretical capacity of between 96 to 192 voters per hour. The average capacity between the two extremes is 128 voter per hour that is very close to what the good Commissioner mentioned in the interview.

Final Words on Changes in the GI. The layout and the BEI work flow proposed in Res 953 is designed to expedite the flow of voters through the process. It does incorporate new controls from the automated election process to override procedures that work in the low-volume manual process. The following are a summary of the critical changes:

• The need to transfer the thumbprinting to the verification process is a practical and paramount change to avoid the single EDCVL document being the focal point of crisscrossing and becoming the bottleneck in this election.
• The need for a crowd control staff at the polling center door who must check that only voters who have their sequence number in addition to precinct and polling center numbers on had.
• The need for a new Voters Assistance Center (VAC) with a real fleshed-out role in crowd control to avoid congestion from clustering and as back up to the crowd control staff at the polling center in making sure the entrance to the polling center have free-flowing queues. Voter assistance in terms of locating the sequence number from the alphabetical list is its key role.The VAC must work with local watchdogs and citizen arms to do its new role effectively at the school playground. This new VAC must assist vulnerable sectors as well.
• Finally, the need for a massive voter education and information dissemination campaign on the new automated election processes. For congestion management, this campaign should include the new requirements for (a) knowing the sequence number in advance and (b) bringing a list of candidates to vote for into the room on election day.

Below is a copy of the en banc aprroved Comelec Minute Resolution on Proposals to Avoid Long Queue promulgated on December 22, 2009