Forensics

Former Comelec commissioner Gus Lagman is asking the poll body to have experts conduct a forensic analysis of the automated election system. Our poll authorities should take heed. Disturbing questions are beginning to pile up.

Never mind the legal questions, such as the rush of the poll body to proclaim winning senatorial candidates ahead of official canvass results. Leave that to the lawyers.

The other day, I looked at graphs and charts put together by young professors and graduate students at the UP Department of Political Science. The visuals, produced from crunching the numbers from the Comelec “transparency server,” were stunning.

We heard about the statistically curious 60-30-10 distribution of the votes across all the regions. Rendered in brightly colored bar charts, the almost perfect uniformity of that distribution is indeed striking.

Does this mean our electorate nationwide suddenly achieved the highest degree of homogeneity? What happened to the bailiwicks? What happened to the ethno-linguistic boundaries? What happened to the “favorite son” vote?

Do these numbers tell us that the deeply rooted characteristics of our electoral culture simply evaporated overnight? Were they erased by the much-touted popularity of the President? But while that much-touted popularity managed to erase deep-seated electoral characteristics, they could not make all 12 administration candidates win?

Defenders of the present automated system imported an expert from abroad to explain away this strange statistical phenomenon. He invoked the Law of Large Numbers (LLN) to explain the phenomenon. The LLN, the heart of statistical science, says the larger the numbers, the more pronounced the patterns tend to be. This assumes all things being equal, which on our fragmented electoral terrain could not be the case.

Then again, why did the LLN not exhibit in previous elections where success in the senatorial elections, conducted at large, really depended on having strong bailiwicks that delivered condensed voting? This is the reason why our political dialect is littered with such concepts as the sugar and tobacco “blocs” and a Solid North. Is it time to delete these concepts from our political vocabulary?

Does the Law of Large Numbers likewise apply to smaller numbers?

The guys at my department are still doing the graphs, although there is strong evidence the 60-30-10 pattern repeats at the provincial and city returns. There is to be sure less uniformity than the bar charts for the regional vote.

Only one precinct in Sumisip, Basilan stands out as truly eccentric. First, the turnout was way below the national average. Second, one senatorial candidate (Loren Legarda) received 100% of the vote. All the other senatorial candidates received zero votes.

That might seem highly improbable, but our electoral history tells us that in this part of the country, nothing should be too surprising. At least, in this particularly eccentric precinct, the number of ballots cast did not exceed the number of registered voters. Thank heavens for small mercies.

Prof. Jalton Taguibao did bother to look at the smaller numbers.

First, he looked at the escalation of the votes at the “transparency server” for several of the winning senatorial candidates from Team PNoy. He converted the numbers into a line chart, with the X axis representing time and the Y axis the number of votes accumulated.

Voila! The winning administration candidates had identical line charts! Their share of the vote increased at exactly the same rate at every segment of the count!

Remember that the “transparency server” was getting returns from all over at every point in time. We should perhaps celebrate the complete political homogeneity we have achieved as a national community.

After that, Taguibao looked at correlations in voter choices for particular candidates, using Grace Poe as the constant variable (being the valedictorian). In this exercise +1 would mean perfect correlation and -1 no correlation at all.

Grace Poe voters were paired with Bam Aquino voters and then with Loren Legarda voters and then with Cynthia Villar voters. The correlation was over .999. Almost perfect.

Aha, could it be that our voters went for coalitions over personalities? Not really.

Voters for Poe likewise chose Nancy Binay with the same impressive .999 correlation. So much for party-based voting.

Let me call all these “statistical curiosities.”

The young professionals I conferred with were thoroughly agitated by the numbers they crunched and the patterns they saw. They asked me why the “opposition” has not sought remedy for this or at least loudly protested the outcome.

I did not have a ready answer to the pained questions posed by my young colleagues. I suppose the best lawyers could not build a case out of bar charts and correlations.

Or, and this is the really painful thought, if this was a grand deception, could the “opposition” be in on it?

There are simply too many “statistical curiosities” in the results available for public analysis. It will be too facile to simply sweep these curiosities under the rug in the name of the Law of Large Numbers.

I heard Comelec chair Sixto Brilliantes declare confidently that it will take a year for anyone to hack the PCOS system. Where in Heaven’s name did he get that idea?

After hearing him say that, I doubt if he is the person to answer the questions my young colleagues are asking. Someone a lot more competent in digital technologies should come in and answer truly serious doubts now raised about the integrity of the last exercise.

If we leave the questions hanging, public trust in automated elections will be seriously eroded. Do the forensic analysis proposed by Lagman now.