Vulnerable elections (2): Cutting it close
Submitted by webteam on March 18, 2010 - 12:36am
The upcoming May 2010 elections is an experiment of sorts. We will have the first automated national elections with more than 50.7 million registered voters. Voters from more than 329,000 precincts will troop to 76,000 clustered precincts (CP). Each CP will have at most one precinct count optical scan (PCOS) machine for around 1000 voters.
One of the vulnerabilities of the 2010 automated election system (AES) is that there are no adequate preparation and testing conducted for the whole process. Besides problems with the Comelec’s credibility, there are also fears of long lines and massive disenfranchisement due to the untested system.
In the absence of an actual time-motion study that will measure the amount of time that will be used for each step of the poll exercise and to give us a glimpse of the problems that can arise from the whole automation system, let us use a tool that physicists routinely use: a computer simulation.
There are a lot of available software to simulate a queuing system such as the AES. We used SimPy, a Python (a programming language) package, to simulate a queue of voters during election day. We based the steps on what a voter and the Board of Election Inspectors (BEI) should do according to the Comelec’s General Instructions promulgated recently.
Using this simulation, we can estimate how many voters will be able to actually vote by the end of the polls, per precinct, and how much time will be available to do each step of the voting process.
One of the factors limiting the number of voters is whether enough of them will actually come to vote.
On election day, around 1000 voters are expected to go to the CP. Each CP will have three BEI members and support staff, depending on the number of voters in the precinct.
The whole voting process will take eleven hours. From the simulation, the minimum arrival should be at least 95 voters per hour to reach 1,000 by 6 p.m.
Once a voter arrives at the polling area, the first thing that he should do is to look up his name and sequence number. The maximum time a voter can spend looking for his name out of a long list is around 15 minutes. If the voter takes longer than this, chances are that the queue will pile up and not everyone will be able to vote.
Assuming that one finds his/her name and sequence num-ber, one approaches any of the members of the BEI and support staff. The voter’s identity will be verified and if everything goes well, he is directed to the chair of the BEI. The number of BEI members and support staff is limited. If one member is busy with a task, the voter will have to wait for his turn. In the simulation, BEI members should be able to process this within 2.6 minutes at the most.
After proper identification, the voter will be directed to the chair of the BEI to get his ballot. Again, this is a limiting step in the flow since it is only the chairman that can do this. If he/she is busy, one has to wait.
The voter must also wait for an official folder (that is as long as the ballot) since one is not allowed to mark his ballot without it. In the simulation, the BEI chair has a maximum period of 40 seconds to verify from the voter’s finger that he has not cast his vote yet, give him the ballot, instruct him on how to fill it up, give him a folder and have him sign several times. If it takes longer than this, the number of voters unable to vote beyond 6 p.m. drastically increases.
One can take as long as nine to 10 minutes (but no longer) to fill up the ballot but one will have to wait for the PCOS machine to be free for the ballot reading. At the PCOS, the voter should be able to successfully insert the ballot within 40 seconds. After having the ballot read into the machine, the voter returns to the BEI to have his finger marked and the folder returned. The BEI member should be able to accomplish this within 2.75 minutes.
To obtain these numbers, we only look at the time period within the step in question and assume that the rest of the process works as fast as it can. If we apply these thresholds simultaneously and run the simulation, the result indicates that more than half of the 1,000 voters lining up will not be able to finish by 6 p.m. Note that these simulations do not include any lunch and bathroom breaks for any of the BEI members. Neither does it include any untoward incidents, challenges and breakdown of electronic systems.
Alas, we can not simulate what will happen if that occurs in the upcoming electoral exercise.
- Login to post comments



