Sunday, August 23, 2009

The Automation Debate Between CenPeg and James Jimenez


COMELEC spokesman JAMES JIMENEZ--a prolific writer, blogger and speaker in his own right--has been having a running debate with the Center for People Empowerment in Governance over the ongoing 2010 election automation project. Personally, I think that James Jimenez's intelligent sense of humor is the perfect antidote to the Y2K-bug-like tendentiousness and fearmongering to which the worst of CenPeg's "analyses" descend.  

In its May, 2009 "Policy Critique," the Center for People Empowerment in Governance (CenPeg) questioned the TRANSPARENCY of the proposed 2010 automated election system:

The Precinct Count Optical Scan-Optical Mark Reader (PCOS-OMR) technology chosen by the Comelec goes against the basic democratic principle of “secret voting and public counting.” This is because the OMR system makes the counting, canvassing and consolidation of election results hidden from public eye and, hence, lacks any transparency as the Constitution and RA 9369 require. The proclamation of winners will be done in 2-3 days making it extremely impossible to file any election protest which is expected to be widespread – and poll watching almost futile.

Here is what JAMES JIMENEZ wrote in a recent Facebook Note  in response to many of the points raised by CenPeg and lawyer Harry Roque in a suit to TRO the automated election contract with Smartmatic/TIM.  I am happy to share it with readers and writers of Philippine Commentary because of its uhmm, transcendental importance.

I would have said everybody a watch-dog, but that term has always carried - for me at least - the unfortunate connotation that someone, somewhere is gonna turn rabid. So, "everyone a quick-counter" it is; because that's how it's going to be.

But before we get to that, this is how the canvassing flows - bottom up. Start from the precinct where a paper report is generated in the form of election returns. The ERs are then delivered physically to the first canvassing level which is at the city or municipality. The city or municipality canvassing board then produces its own reports: the certificate of canvass and a statement of votes by precinct. Each city or municipal canvassing board then sends its report to the next canvassing level: the provincial. Provincial produces its own COC and SOV then reports to the National Board of Canvassers. The National Board canvasses the results for Senators and Party-List, while Congress canvasses results for President and Vice-President. 

The problem with this process is that it takes too long. Delivering the reports can take days, sometimes even a week or more. In the meantime, people get antsy and worried that the reports are being manipulated or altered in some way. Especially worrisome is when the outcome of the elections is made to hinge on reports that haven't been canvassed yet; the fear being that the manipulators are waiting to see just how much padding or shaving they need to do. So time, essentially, can be a tool for manipulators and a very destabilizing factor in the mandate of whoever emerges the winner.

Obviously, the solution is electronic transmission. 

Electronic transmission drastically cuts down the time necessary to transmit election results from level to level. But, naturally, in order for electronic transmission to be possible, the data - essentially the written reports generated by each level in the process - has to be in electronic form. How to do that?

One solution is to manually count the votes, manually prepare the election returns, and then encode the election returns into a computer with electronic transmission capabilities.

Sounds simple enough. But let's examine that a little closer.

Manual Counting

A manual count has its benefits, foremost would be that there would be no need to teach voters how to vote. The ballot will be the usual thing - a long sheet of paper with the names of the positions being voted on (what I call the "races") followed by blank lines where the voter can write the names of the candidate he's voting for. So that's good, right?

Well, yes. In the sense that inertia is your friend. But that kind of voting runs into all sorts of problems all the time. First, there's the problem of illegible handwriting. Then there's the issue of voters who don't write the full name of the candidate. And of course, there are instances when some candidates have the same name, and the voter isn't specific enough with his choice. All these problems add up to questions of voter appreciation. The teachers engage in adhoc interpretations of what the ballot is actually trying to say about the intent of the voter. It's a guessing game, really, where the main players are the lawyers of the candidates. Not the voter - his part is done. It's the lawyers trying to outdo each other in convincing the BEI to see things their way.

Another result of this is that it draws out the time needed to complete the count; hence the counting proceeds in fits and starts until dawn the following day. And again, time can become a tool for subverting the vote. During the wee hours, when non-partisan people are mostly no longer around to watch the proceedings, all sorts of attempts are made to subvert the vote count: power outages, intimidation, thuggery, bribery .. you name it. Teacher's lives are put at risk, the integrity of the ballot is challenged mightily, and the outcome of the elections becomes suspect. 

IS THE PRECINCT VOTE COUNT EASY TO SECURE?
Over the past few weeks, you've probably heard a lot of people saying that the vote count is the easiest thing in the world to secure; that the vote count is the cleanest part of the process; that a manual count is the best because it is witnessed by the public. HAH!

The vote count is the most difficult part of the process to secure for the following reasons: first, the sheer scale of it. You're talking about counting going on more-or-less simultaneously in more than 250 thousand locations throughout the country. Second, it takes too long. Counts run up to 12 hours during which time any number of attacks on the integrity of the elections can ensue. And third, the majority of the counts all over the country, as well as a significant chunk of the counting process itself, are NOT watched by the public: there are simply too many counts going on at the same time, and too many non-partisan observers simply cannot stay all night. In the end, many watchers simply come back the following day to claim copies of ERs that they can nitpick about.

ER Preparation

Manual ER preparation has its share of problems, naturally. The most significant being the mis-recording of the data. Let's forget about fraud for awhile and just focus on fatigue induced error. 

On election day, the members of the BEI are up and about by 4 am, getting their election supplies. Elections run from 7am to 4pm. Counting usually starts at 6 pm and runs till 6 am the following day. Which means that by the time the ERs are being prepared, the teachers doing the job will have been up nearly 24 hours. Fatigue is inevitable, and when people get tired, people make mistakes.

How inevitable is error? It is so inevitable that when an ER is presented and shows no erasures or errors of any kind, it's perfection is considered a good indicator of fraud. Chew on that for awhile. Our system is so prone to error that if no errors are evident, we suspect that something fishy is going on. How twisted is that? Nevertheless, it is a sound observation. Because of the strain placed on teachers, fatigue is one of the most significant causes of error through inadvertence.

Add to that, deliberate error brought about by coercion or corruption and you realize that ERs prepared manually are probably not as pristine as some people might want you to believe. And that's where the lawyers come in.

A lawyer's job is to make sure that his candidate wins. This means, if you're being brutally frank about it, that a lawyer will try everything he can to influence the way the results are reported. He can influence it during preparation of the ER, or he can influence it by challenging the number of votes reported for his candidate. Either way, under a regime of manual preparation of ERs, lawyers will exert an inordinate amount of influence in the way the vote is reported, and they will also contribute mightily to the delay in the reporting.

ER Encoding

So, the ER - whose faithfulness in reflecting the will of the people will, by now, have been diluted - now has to be encoded to produce an electronically transmittable document.

The most straightforward solution is to have someone reading from the ERs while someone sits and encodes everything he hears. Now it's obvious that neither can check what the other is doing. The encoder can't verify everything the reader tells him, just as the reader cannot check whether his words are being tapped into the computer accurately.

This sets up two points of vulnerability: the reader (who can be coerced or bribed into dishonesty) and the encoder (who, by the way, can be coerced or bribed into dishonesty). 

Ironically however, despite these vulnerabilities - from counting to ER preparaton to ER encoding - there are some quarters who would rather see us stuck rather than try out the other way for getting the ERs ready for electronic transmission.

The other way

The other way is automated counting. If you automate the counting, then the counting machine presents the count in two ways: a paper election return report, and an electronic version of the same report. Kinda like having a print-out of that word document you typed. You have a hard and a soft copy. 

With the electronic version generated at the same instant the count is completed, the transmission can then proceed seamlessly, with no need for anyone to interpret any ballots or to encode anything. 

You save time and you ensure that there will be no discrepancies mediated by human error or human malice.

Now there are those who would belittle the importance of saving time with the electronic count. With all due respect but very little fondness, let them tell that to the teachers who have to stay up more than 24 hours just to complete the counting and preparation of ERs; let them tell that to the teachers who have to spend the night wondering whether the next hour will bring goons with guns; let them tell that to the teachers who have to turn away bribes, knowing full well that there may be adverse repercussions arising from their nobility.

And of course, there are those who will belittle the trustworthiness of the COMELEC in conducting the automated counting. Fair enough, I would say. Don't trust the COMELEC, but don't let that distrust consign the rest of the country to repeating the same old processes with the same old vulnerabilities; don't let that distrust get in the way of getting the teachers out of harm's way; and don't let that distrust become a stumbling block to the improvement of the electoral system.

On another level, don't use that mistrust to mislead the people.

People with avowed distrust for the COMELEC like saying that the machines can be programmed to count wrongly; to favor this or that candidate; to ensure a pre-determined outcome. People who distrust the COMELEC slam the system for a supposed lack of transparency.

A hand-count is not the only route to transparency. 

First, all a hand-count really does is show that ballots are actually being read. It does not guarantee any individual voter that "his vote is being counted accurately." All it does is give people - the BEI, the lawyers - the opportunity to argue about what the voter intended. And since no one can actually say for certain that the ballots are being read according to each voter's intent, how can a hand count guarantee anything? If you think about it, the ability of lawyers to question the accuracy of the vote ultimately rests on the uncertainty of how to interpret the source document - the ballot. Remove that uncertainty and the ability to question the accuracy of the count evaporates.

Now consider an automated election system where the ballots unequivocally show voter intent and where the resulting machine count can be verified against the actual ballots? Sure, you don't see ballots actually being read, but that doesn't put you in a worse position vis-a-vis a hand count. Besides, any re-count will still go through the same process of lawyers trying to outdo each other in convincing the authorities that the voter voted this way and not that. With the automated system, however, the hand re-count is much easier and much more definitive precisely because there is no mistaking voter intent. If the oval is shaded, it's a vote. If not, then not.

There are those who argue saying that by the time a hand recount of the modern ballot is undertaken, the winners will have been proclaimed. I may have missed something, but isn't that how it is with a hand count? 

Second, the automated system has the advantage of being tamper-evident.

Tamper evident

Tamper-evident only means that if someone monkeys around with something, the monkeying is obvious. Like meds. No one seriously believes that those flimsy foil things prevent tampering with the pill inside. The main strength of the foil is that if someone does try to get to the pill inside, the damage to the foil will be so obvious that no one in his right mind is gonna think that everything's okay. Once you know that the pill has been tampered with, you know right away to chuck it in the trash.

It's kinda the same with the automated election system. Unlike a foil pack, the design is such that the speed of counting and consolidation make it very difficult to tamper with the results; but just like a foil pack, the design also ensures that any unauthorized tampering is immediately obvious.




The results, you see, are transmitted directly out of the precinct and sent to several recipients all at once: to the municipal canvassing center, to the COMELEC central server, to the servers of the dominant majority, the dominant minority, the accredited citizen's arm, and the KBP. And yes, it will also be sent to a publicly accessible website.

That's seven different recipients of the same data, all in one go. And that's not even counting the 30 hard copies of election returns that will be made available to practically everyone.

Now if someone were to tamper with one of those results, for whatever reason, the discrepancy with all the other copies available makes the tampering immediately discoverable. So, if your shenanigans are immediately obvious, why bother doing it at all? You won't make a difference in the final result and you're likely to be found quickly. Bottom line, the widespread availability of precinct results makes fraud more trouble than it's worth. 

Even better, having the results published on the web essentially democratizes quick-counting. Literally anyone with internet access and a calculator can do exactly what Namfrel used to do, can do exactly what the COMELEC is doing. Everyone becomes a quick-counter!

Irony

And this is why I can't help but shake my head at the irony of people trying to derail automation. In one breath, they thump their chests and proclaim themselves concerned only with the cleanliness of elections, and yet in the next, they do everything they can to scupper this excellent opportunity we have to actually improve the cleanliness of elections.


********************
SOURCE: Philippine Commentary
*************

11 comments:

Francis Acero said...

The law provides that the simultaneous return to mass media is to the three biggest media outlets - which I presume are ABC5, GMA7, and ABSCBN, in inverse order. Watchdog groups can do their own tally there per election return - aside from their own tally that they can make from their own copy of the election return.

Anonymous said...

What countries in this world currently have and use automated election systems in the national and local levels?

If there are, what are they and what is there track record? If there aren't any, why?

Ony

john marzan said...

the comelec hasn't done anything to clean house of election officials involved in 2004 and 2007 election fraud.

even jimenez and anonymous spook comelecako said that there's no proof that garci did anything wrong.

Roberto Verzola said...

For a review of actual experience with electronic voting in other countries, see http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1150274

What James Jimenez does not mention is that most election fraud are basically inside jobs. When it is an inside job, many safeguards (passwords, encryption, etc.) can be bypassed more easily.

Roberto Verzola

Deany Bocobo said...

Ony,
The only alternative is Garci. We don't want that.

Deany Bocobo said...

john marzan,
the essential point is that automation will take Comelec out of the counting and canvassing loop altogether. although an "inside job" is always possible, as robert verzola suggests, that can be done only once, I would think. At the moment Smartmatic is our friend. They will run the safes for the Comelec Bank.

john marzan said...

from girlintraining:

Oh noes, electronic records can be faked by people who have physical access to the machines. Didn't see THAT one coming.

Anonymous said...

Jorge,

There are many other alternatives. For instance, increase the number of precincts, canvassers and poll workers. This should speed up counting.

You must be aware of how voting is done in the US. I had served as a poll watcher in New York that uses voting devices that tally votes mechanically. Why not try that.

What Mr. Jimenez points out as bottlenecks/flaws in the current system can also hold true for an automated system.

I am not opposed to automation. What I am suggesting is that the election process is much too important to fool around with.

The primary reason for automation, it seems to me, is not really the length of time it takes to complete, but the cheating that plagues it.


Speaking of fearmongering:

"Ony,

The only alternative is Garci. We don't want that."

Deany Bocobo said...

But Ony,
what IS the alternative at this point? It really is Garci, is it not?

I think the problem is for some people, they are holding the automated system up to a standard that they could not even apply to the manual election system. Yet they are willing to settle for the devil that they know.

john marzan said...

what IS the alternative at this point? It really is Garci, is it not?

having automated elections without fixing the COMELEC = Garci on steroids

if nobody has been punished yet for the 2004 rigged election and maguindanao, don't blame Manual. blame the corrupt Arroyo government and the CBCP/Ateneo that looked the other way.

I think the problem is for some people, they are holding the automated system up to a standard that they could not even apply to the manual election system. Yet they are willing to settle for the devil that they know.

if you ask technology experts from the USA and other countries in europe, they'd tell you the opposite--that manual ballots are less easy to forge, and have more intergrity than electronic data.

it's only in the philippines where we (i mean some) are claiming the opposite. we'd rather put our blind faith on something "new" than punish those involved in Garci.

Anonymous said...

goodness. all these naysayers on automation. a few years ago they were screaming 'automate!' then when the comelec comes around to doing this they all scream 'don't automate!' i mean, what do we all really want?? why can't we make up our bloody minds?? it's occam's razor, the simplest answer tends to be the correct one. automate. get electronic results within the day, proclaim a winner so that governance won't be severely delayed (we've seen garbage pile up during hotly contested local election counts) and if there are complaints, then retrieve the paper ballots cast. which means, once those paper ballots are run through the counting machine, don't throw them away. so in this scenario, after the electronic count, comelec declares an unofficial winner who will be deemed as a sort of 'officer-in-charge'. governance and public service will then be left unhampered. if all protests against this OIC are deemed void or resolved, then COMELEC moves to officially proclaim the OIC as duly elected for his/her respectable position. frankly, i for one believe that the presidential and senatorial vote is likely more honest than the local elections because based on what i've observed throughout all my years voting is that people on the local level hardly care about the national vote. so to secure the presidential and senatorial count, you merely need to focus attention on the national canvassing body. all the issues against automated voting might be releveant in the local elections but arguably not for the national vote. besides, if the system messes up, we won't have any problems finding the culprit. Smartmatic gets the accusatory finger.