Wednesday, October 18, 2006

Gordon's Automation Is Based On Cutting Out Multistage Canvass


Sen Dick Gordon thinks so, and he didn’t like Comelec Chairman Ben Abalos’ defeatist attitude on Strictly Politics with Pia Hontiveros last night, towards the challenge of doing six provinces during next year’s elections. Dick gave Abalos a good dose of that old Subic can-do sizboombah -I - can't - believe - what - I'm - hearing - from you - tongue-lashing. It's a refreshing approach rare among our politicians and Dick pulled it off well, never quite going beyond the speed limit.

"If the Comelec does not feel up to the task the law may assign to them and cannot do six provinces next year," Dick asked naughtily, "then HOW MANY can it do, one barrio?" He said the scope of the partial automation can still be adjusted during the bicam phase of the bill's passage (expected by around mid-November yet) and that he would be willing to do one barrio if that is all Comelec says it can do.

PROOF OF CONCEPT: I agree! Seriously, even if they answer one barrio, we should take them up on it. Because I think a properly done "automated election" on even a single voting precinct can serve to establish the system design features, such as accuracy rating, transmission security, canvassing reliability, and other key operating parameters.

It must be borne in mind that there are several aspects of an election that could, and eventually should be automated: (1) registration of voters; (2) voting; (3) counting; (4) transmission; (5) canvassing. The Comelec itself has incinerated billions of pesos on all of them, with very little to show for it, except for the groaning pile of illegally purchased OMR-based ACMs costing the government 1.9 million pesos per month in storage costs. (Since Comelec already prepaid 900 million to MPC, and have disappeared from all legal and practical accountability, Comelec may as well set up the biggest Internet Cafe and Game Room in the world-- right there on the fourth floor of the Comelec Bldg in historic Intramuros!)

Sen. Gordon emphasized on the show that the leitmotif of his concept is the transmission subsystem. His reasoning is based on the common understanding that there are retail modes and wholesale modes of cheating in a Philippine election. Some retail modes of cheating, such as "flying voters" who do multiple registrations can be eliminated by automating the registration process, but some cannot, such as vote-buying. However, Sen. Gordon believes the most damaging form of cheating is the wholesale DAGDAG BAWAS (add-n-subtract) that occurs during the multistage canvass of precinct tallies. That insecure multistage canvassing stage can be shortcircuited by an efficient transmission system direct from each voting precinct to Comelec National HQ where a single central canvassing process can occur. Indeed, there seems no technological reason whatsoever to creating such a transmission subsystem.

I think that demonstrating an election where the results are known within minutes of poll closing is a key step forward and something along these lines can be accomplished for next year, without an expensive new infrastructure ala MPC...just cellphones and the Internet.

I guess what I am getting at is: I know Full Implementation is pretty much impossible and unwise for 2007. But demonstration of a concept that most people will find credible is not impossible and can even better be done and tested if first attempted on a small scale. That concept is that the multi-stage manual canvassing of votes at the municipal, provincial and national levels, where all the wholesale cheating goes on, can indeed be eliminated and replaced with a reliable and transparent modern system.

This coming Friday, the Comelec has called for a meeting to discuss the prospects of Automation 2007.


john marzan said...

i guess limited implementation of automated voting machines in select areas/cities/districts would be okay.

so, saang lugar ba ang magandang pag-testing-ngan nyan?

manuelbuencamino said...

is there no possibility of those machines being hacked?

I saw a video done by a US university on some counting machine. The scientists slipped a self-propagating program in one machine and that machine was able to infect the other systems without leaving a trace. Anyway the program was a dagdag bawas program. The only way the hacking could have been prevented was by physically guarding all the machines. But I also read about how diebold machines could be hacked through the internet. Apparently there were many ways the program could be accessed and changed through the internet.
I guess my point is there aint nothing done my man that cant be undone by man. a computerized system can be undone by one man. the old manual system requires at least the teamwork of hundreds if not thousands of people. computerization can make wholesale cheating easier, and if t it can ne dome then they will do it.
how do you see the pros and cons to this?

manuelbuencamino said...


going back to the SWS poll. where can i find a historical record of the UNDECIDED from the beginning of GMAs term to today? Or can we just divine those numbers by adding the percentages of SATISFIED AND DISSATISFIED? I'm curios to see how the UNDECIDEDs expand or contract over time.

Rizalist said...

most of it is available on the sws website, but you gotta know how to dig for it over there since they're cute and try to hide urls and stuff. But if you really wanna get down and dirty, I think you can actually order a CD full of SWS great surveys of the past. It really does help to look at data that is years and years old and covers long periods of time. You get a better sense of the data variation over time and from one survey to the next.

Actually, if you ADD the two numbers they never add up to 100%. The difference is what i call the NET UNDECIDED RATING...always in double digits.

Listen if you really wanna get under the hood, make sure you understand every word and equation in that article in wikipedia "Checking if a coin is fair" It really is accessible to any intelligent layman. And it will empower you to really, really understand these surveys and polls and you can ZING them in the Media whenever they announce something that is not quite right or just down right wrong.

Thanks for the interest in this topic MB!

Rizalist said...

There is a saying:

People are smarter than computers...but computers are smarter than programmers!

manuelbuencamino said...

Reason why I am interested in the UNDECIDEDs is I suspect that's what the redime uses to measure what they can get away with. I think they put the UNDECIDEDs in their column.

The battle lines between the antis and the pros are pretty well fixed. GMA has never enjoyed a rating substantially higher than Erap ever had, even right after EDSA 2. Compare her assumption ratings to Cory and FVR.

I suspect the swing vote or what many call tipping point is with the UNDECIDEDs so the strategy is to keep them undecided and uninvolved. I don't know if a case can be made for this analysis. You want to meet for coffee one of these days, mayne ask manolo to join us, and see if this angle is worth pursuing?

Rizalist said...

It's an interesting angle. I've usually stopped with the observation that the undecideds are the actual margin of error in the NSRs. hehe. but that would make the NSR really useless.

But read some of the ACTUAL questions that are used to get the two different statistics.

Now there's the real area of black magic in polling: the art of question design.

Can u believe that in the US, public polls on abortion, both scientific, can still come up with DIFFERENT results about what americans think about that issue. It's all in VERY small, seemingly insignificant differences the words used to ask the questions.

It's worthwhile though getting up on the high ground of understanding how all these numbers work and how they are derived and used for POLITICS and PROPAGANDA. powerful weapons really.