ACM Accuracy and One-Man-One-Vote Suffrage

Correct Dates of relevance to 2007 elections ( Thanks to Postigo Luna of Comelec Ako):
January 15, 2007 (MON) to February 12, 2007 (MON): Period to file certificate of candidacy for Senators;
January 15, 2007 (MON) to March 29, 2007 (THU): Period to file certificate of candidacy for local elective positions.

---

(So far as I know what follows is a novel Constitutional issue that has not yet been publicly discussed.)

Automating elections involves the notion of having machines perform the functions done by human beings. For example, in 2004 the Comelec wanted to use Automated Counting Machines (ACMs) to do the work of counting and tallying the votes at the precinct level, usually done by teachers and other election day workers. It was natural to stipulate an ACCURACY RATING on the ACMs, as a means of qualifying proposed systems, which was eventually set at 99.995 percent, Now, 99.995% accuracy is the same as saying the ACMs must never make more than one erroneous reading for ever 20,000 ballots read. It also means that if there were 50 million voters in an election using those ACMs, there could be as many as 2500 erroneously read ballots.

But, the 1987 Constitution declares--

Art VII Section 4 (On Presidential elections): The person having the highest number of votes shall be proclaimed elected, but in case two or more shall have an equal and highest number of votes, one of them shall forthwith be chosen by the vote of a majority of all the Members of both Houses of the Congress, voting separately.

Clearly the Constitution contemplates the possibility of a TIE in a national election and makes provision for the Congress to break it. Just as clearly, any election may be won by a margin of just one vote. I think this presents an important constitutional issue with respect to election automation that cannot easily be ignored.

There is no way for Comelec, using these ACMs, to guarantee that it can detect a TIE in the vote count, or any difference that is less than 2500 or so. Indeed, the existence of ANY Accuracy Rating requirement that is less than 100% would seem to be unconstitutional on the face of its since it violates the most basic democratic right--to have one's vote counted so it matters. No matter how improbable these two situations are, it is an unavoidable and germane issue as we wrestle with the problem of modernizing the conduct of elections in the Philippines. The one "strength" of the manual election system is that Comelec does not enforce an official "Accuracy Rating" on the Human Ballot Readers it now uses to count the vote. There is all kinds of irony in the idea that a manual count does not require an accuracy rating. But now that we are contemplating using machines instead of human beings, we are forced to impose an accuracy rating. And thus create a novel Constitutional issue.

In the year 2000, George W. Bush won the US Presidency after Florida's electoral college votes were awarded to him -- on the strength of a popular vote margin in the State of Florida of a little over 500 votes out of 2 million! An election involving nearly 100 million voters turned on 500 votes. Although such close elections are rare, they are not THAT rare.

GRINGO CAUGHT: Gringo Honasan is a Jekyll-and-Hyde kind of character. He is both Putschistenfuhrer as well as an elected Senator with a proven constituency. Gringo's arrest is largely a good thing, to me, because now he is forced to be the latter and not the former. Unless charged and convicted with finality before then, he will likely run for, and win a Senate seat in the May, 2007 election. Note that Lt.Sg. Antonio Trillanes III of the Oakwood Mutiny has also announced he will be running for the Senate. I think Danilo Lim, Archie Segumalian, and Angel Querubin should now resign their commissions and join Gringo and Trillanes in politics, to help root out corruption and politics in the Military. I don't see why these bright young men should suffer through the Kangaroo Courts-Martial of the Philippine Military, when they could easily bust out and be Mayors, Congressmen, Senators. As long as they don't run and hide and play lil tin Gringos at night, the People will protect them with their votes. They would not be any different than the CPP-NPA if they insist on violent methods to bring about reform in the military. I guess I must repeat my old stand on this. If you are a loyal solider but want to "withdraw support" for the Military Chain of Command, you must resign your commission first, become an "ordinary citizen," and thence exercise the civil and political rights needed to effect the desired reforms. Then you cannot be accused of "coup d'etat."

THE EXPLAINER I had a great time on MLQ3's Tuesday night show. Patricia Evangelista, his fellow PDI columnist has charming, and charmed hands, which we were using to do Random Samples on the Crystal Bowl of Public Opinion. She produced, completely by chance and the luck of the draw, two of the most improbable outcomes with my prepared Crystal Bowl...a run of 10 Blue Balls in a row and then three successive samples with 4 Red and 6 Blue Balls, representing the exact percentage of 40% red balls in the Bowl. This worked out perfectly since we were really discussing two exit polls: one with a highly accurate exit poll prediction by SWS (1998) and another where their data was probably corrupted at the point of interview (2004).

Regarding scientific public opinion polls the essential points are these:

(1) A poll cannot be scientific unless it strictly uses random sampling to select the respondents in the survey.

(2) The Statistical Margin of Error in a survey depends on the size of the random sample. Numerically, it is equal to one divided by the square root of the number of respondents. For standard surveys of 1200 respondents, plus or minus 2.89 percent; for 100 respondents, plus or minus 10%; for 10,000 respondents, plus or minus 1%; for 20,000 respondents, plus or minus 0.707%.

(3) Even IF a poll uses random sampling to acquire its raw data however, the poll results may not be considered scientific if the QUESTION itself is not scientific. Huh, you say? How can the Question asked in a survey be scientific or not scientific?? A survey question is not scientific if:

(a) it cannot be answered by a simple YES or NO;
(b) it is so complex and ambiguous that most respondents cannot sensibly answer it;
(c) it involves a subject matter that will not be put to a test like an election or plebiscite and such that the results of the survey will never be proven right or wrong.

So just because the SWS always uses random sampling techniques to select its respondents, it also asks many questions in its surveys that are unscientific. Surveys featuring questions on self-rated hunger and poverty are suspect to me because there is no way to prove the reported results right or wrong. Such survey reports are easily turned into propaganda by passing them off as "scientific surveys" just because they are done by SWS or Pulse Asia.

(4) Finally, the GENERALIZATION that is usually arrived at and turned into HEADLINES must follow logically and mathematically from the underlying questions and answers. Too often, innumerate reporters and editors licentiously "shade" or even outrightly distort what surveys actually say.