The Revenge of the Machines
By Ado Paglinawan
I am endlessly amazed at the audacity of armchair analysts who has taken the cudgels to defend the 60-30-10 distribution of the votes of the Liberal Party Coalition, UNA and others.
So far I have encountered three analyses that appear serious at first, but eventually become fallacious upon closer scrutiny. The first that of Dr. Pablo Manalastas of the Center for People-Empowered Governance started it all. The second that of Mike Beduya of Asian Institute of Management has remained in the confines of social media but the third, of Dr. Michael Purugganan a dean of science at the New York University, has made it to the mainstream.
Like Dr. Purugganan, I am also based in the United States, specifically Washington DC. I don’t know his status but I have not renounced my Philippine citizenship and continue to come to the Philippines as often as I could so as not to lose my brown footing. In fact, this indio voted in the last elections at Antipolo City.
When AESWatcher Dr. Pablo Manalastas of the Ateneo drops the Law of Large Numbers (LLN) line, it could be enough to push TJDimacali of GMA News to doubt if this is “Conspiracy or Just Math?”
Being a strategic consultant, however, I approach the numbers and computer insights of the 2013 elections, just as I did the 2010 elections - not just from the perspective of a geek, but more appropriately from the socio-political aspect.. Afterall, numbers can be manipulated but not sociographics indicative of blood trails of a 60-30-10 pattern.
For being the best analyst of elections from 2004 to date, I would rather follow the logic of Hermenegildo Estrella, both a geek and a political animal. This is what he emailed to me yesterday:
“The Law of Large Numbers applies only for trials from the same population, which should mean that these trials [group of votes that are counted] are, or should have come from areas, really representative of the country as a whole. Same parameters, same characteristics, same almost every way of life.”
He compared this with what SWS and Pulse Asia do in the use of 1200 respondents as samples to predict the entire country's preferences in areas where they have initially identified as leaning to their desired outcome.
Estrella said we are not robots: “A region, any region for that matter, does not really represent the Philippines. There are different dialects, cultures, preferences, income brackets, etc. For provinces, some neighboring ones share similarities, but not anyone of these represent the entire neighborhood. Much more political interests and leanings. Even more so down to districts, and may even be down to precinct clusters.”
In fact for his part, Beduya analyzed the increments between each canvass and charted the peaks and valleys of shares relative to each of the canvass as it progressed from the first to the sixteenth. His objective was obviously to dispel aired suspicions that the votes followed a fixed linear pattern. Eventually, Beduya invoked the law of large numbers illustrating its application around the 10th canvass as the votes when the 60-30-10 achieved seemingly irreversible momentum.
Purugganan on the other hand said “Each time they released a canvass, it represents millions of votes from around the country… So what does mathematics tell us? Each canvass would give a result that would be very, very close to the 60-30-10 national average.”
Beduya echoed Purugganan’s conclusion, “It is a mathematical law – it is called the Law of Large Numbers, which says that the larger your sample size is for an event, the closer your result from that sample will reflect the entire population.”
But both Beduya and Purugganan ground their syllogism on the premise that representative votes from “all over the country” were randomly and automatically being transmitted to and received by the Transparency Server maintained by PPCRV.
I cannot blame them really because that was what has been announced to the public by the showbiz Comelec - that from the precincts after the closing of voting, precinct tallies will be transmitted to an automatic bypass has been enabled for the precincts to transmit directly to the Transparency Server maintained by the PPCRV. This is an “unofficial” track but allows a ”first look” of the national count, or what was before known as the “quick count”.
Of course the “official” track of Election Returns, go from the precincts to the municipalities, from the municipalities onto the provinces, and from the provinces to the nation.
The Transparency Server is faster but since its data is sourced from the precincts, its canvass ought to mirror the official count.
But what Purugganan and Beduya were not aware of is that the Comelec, for the first time in the history of Philippine elections, “batched” provinces and cities into respective canvass clusters.
What was designed to receive inputs on a first-in/first out basis became a funnel scenario. Imagine a funnel full of precinct outputs at the bigger end trying to get ahead of one another towards its small end. The entering data is random but the exit is ruled by groupings by provinces and chartered cities.
Specifically, the first canvass covered votes from Antique, Batanes, Dinagat Islands, Guimaras, Mountain Province, Nueva Vizcaya, Quirino, Romblon, Southern Leyte, and the following cities: Baguio, Iligan, Mandaluyong, Navotas and Paranaque. That canvass does not represent a random national count.
I am attaching herewith the schedule of the provinces released from the 1st to the 16th canvass. Again, each canvass does not represent a random national count. In fact, to reduce this matter to outright absurdity, the 14th canvass shows it was only for Davao del Norte, North Cotabato and Tawi-Tawi, all from atypical areas that are rural, not national. The 15th canvass was only from Marinduque and Samar, and the 16th was only from only one province Lanao del Norte.
So this demolishes their basic premise leading to consideration of “large numbers”. In fact, the New York-based dean himself unwittingly acknowledged the opposite should have happened.Purugganan said “Now if you look at specific precincts, or maybe even specific regions, you will definitely find that each precinct or region will show more differences between each other. Each will have a different number that is further away from the 60-30-10 pattern. That is because they reflect local voting results.”
Albeit this caveat, 60-30-10 still ruled the exercise.
The peaks and valleys that Beduya pointed out, failed to observe that despite those sharp fluctuations, the individual share of all the senatorial candidates enabled them to maintain almost constant shares respective to any of the 16 canvasses. In fact he should be asking why despite those sharp fluctuations, the numbers were defaulting to a 60-30-10 template?
In short the template pattern applies to individual rankings. Here is a contrast in sampling - Grace Poe typecast at around 11%, while Risa Hontiveros pegged to around 6%. My goodness, it is almost a “flatline”. The individual shares of the senatorial candidates also established an almost constant ratio not just the top twelve rankings from Grace Poe to Gringo Honasan but of Samson Alcantara (Candidate #1) all the way to Migz Zubiri (Candidate #33).
My hypothesis on what caused this flatline, I will give at the conclusion of this article.
Allow me however to point out that batching or clustering the provincial inputs is by itself already a humungous anomaly.
It presupposes two things. First, that a filtering mechanism to, the very least to group the incoming tally into clusters, existed. In layman’s tongue using baseball lingo, after the ball was released by the pitcher, it did not go straight to the catcher but passed through a short-stop.
This “short stop” did not only expose the data to possible manipulation, it also violated the law.
This becomes relevant because as Mr. Beduya correctly observed the numbers making critical hits around the 10th canvass. Of course the “flatlining” misled him to thinking the law of large numbers had taken over.
From the political perspective what was actually happening was that Albay, Catanduanes and Leyte defaulted Bulacan in the 8th canvass. Bohol and Lanao del Sur defaulted Batangas in the 9th canvass. Cebu and Camarines Sur defaulted Nueva Ecija, Pangasinan and Zambales in the 10th canvass. Iloilo and Quezon City defaulted Laguna, in the 11th canvass.
Now to borrow an adjective from the New York dean but in reverse application: That’s definitely “fishy”.
Of those that I cited, my favorites are Bohol and Tacloban City. In Bohol, where Arthur Yap ran unopposed, all municipalities posted 60-30-10. Curiously, Tacloban City, the bailiwick of Imelda Marcos, also registered 60-30-10.
Rappler.com also exposed that more than 2,000 clustered precincts, significantly from Cebu and ARMM, where Bam Aquino showed magnificent “votes”, registered perfect or near perfect turnout beating the national average of 70 to 75%. Hello?
Given this, I no longer find it mysterious that the Comelec again in violation of the law, started proclaiming the first six winners in the Senatorial race with only 20% of results posted. This was followed by three with yet partial results and the last three with only 77% of the people’s votes accounted for.
It was quite obvious that like Pontius Pilate, Comelec Chairman Sixto Brilliantes, this election’s Smartmatic lawyer and Malacanang stooge, merely wanted to wash his hands away from what AESWatch correctly pointed out as a “technological and political disaster”. He wanted the onus of protests not to land on his feet but to the Electoral Tribunals.
Note that the balance of 23% is still missing two weeks after the elections. But I, despite being an IT undergraduate, again have a fairly good hypothesis on what happened.
My first hypothesis resembles a deliberate conspiracy theory. The 60-30-10 is much more than an outputed pattern that resulted from innocent inputs of election returns. It is a premeditated goal preprogrammed to be delivered by strategically inserted default mechanisms against opposition bailiwicks using the combined numbers of administration bailiwicks and strategically designated electronic dagdag-bawas centers like Bohol and Tacloban City.
Not only did those defaults go for nine administration coalition senators and three from the simulated opposition, it ran a linear formula for all 33 senatorial candidates locked in their pre-determined rankings from the first to the 16th canvass.
Only three were conceded by conspirators Malacanang and Comelec to the so-called three kings - Nancy Binay to Vice President Jejomar Binay, JV Ejercito to former President Joseph Estrada, and badly enough for Jack Enrile, it was Gregorio Honasan for Senate President Juan Ponce Enrile. Given this scenario, who will rock the boat?
My second hypothesis is less of a conspiracy, but what I will call “the revenge of the machines” maybe even divine intervention. The insertion of default mechanisms involving the preloading of the quantum manipulated results just overheated the system. This could have happened as the PCOS precincts were sending transmissions to the Transparency Server, or while a “short stop” bypass was trying to catch them in order to be grouped and “massaged”, or while they were queued and lumped at large numbers waiting for their turn at being consolidated into the main canvass.
Imagine how many variables existed in around 76,000 precincts. Imagine the variants caused by candidates from councilors to senators making deals with so-called Comelec operators for some of these precincts nationwide.
Imagine tens of millions of precinct outputs trying to input into the Transparency Server but being made to wait by a short-stop malware that was batching votes by provinces not to mention key areas where electronic dagdag-bawas was weighing on actual count.
The PCOS was just meant to scan and read the ballot, count the returns and transmit its tally. Brought about by Comelec’s endless improvisations and collapsing of the minimum required safeguards in the AES law, it has virtually opened the system to hacking and fraud confusing the idiot box.
The PCOS system, so prostituted, just fucked up and scrambled on its own to default into a 60-30-10 template on the national level and lay-away 12 million votes equivalent to 23% of the people’s voice, so that the end canvass can serve by its mere self – the prima facie evidence of a human manipulation and election failure and proof-positive that Daang Matuwid is actually an expressway for corruption! #
Posted May 27, 2013
Note: Batched or botched Canvass Clusters attached (Source rappler.com). The chart must be overlaid to the Summary of the 1st to the 16th group canvass report also from rappler.com.
Ado Paglinawan was appointed by the Corazon Aquino administration as a Philippine diplomat in 1986. He served as Press Officer of the Philippine Embassy in Washington DC from 1986 to 1993, and concurrently of the Philippine permanent representative to the United Nations in New York from 1989 to 1992. After his government tenure, he served as a Consultant for Strategic Studies and Public Relations to various clients in the United States and the Philippines. He presently publishes an eMagazine called "The Soberano".
Subscribe in a reader 
any other purposes without exception, email exclusively for article updates only.
Labels:
Politics
0 Speak Out:
Post a Comment