Friday, December 12, 2008

Election Fraud: Detection and Quantification

By Jon N. Hall

Without the vote, Democracy is nothing. The franchise is the core of America’s system. Universal suffrage, political campaigns, casting ballots, counting votes: These are at the very center of what we as a nation are all about. It follows then that anything that thwarts the will of We the People by compromising the integrity of an election is a dagger in the heart of Democracy itself.

But some folks contend that allegations of widespread election fraud are themselves a fraud. For them, election fraud is a "myth", a “witch hunt”; like looking for “Sasquatch”. They maintain that the evidence for election fraud is “painfully skimpy”.

One little problem for these folks is the successful prosecutions—people are in jail for election fraud. Since we know election fraud occurs, the question should be: How prevalent is it? But nobody knows the answer to that, and for reasons that will become “painfully” clear.

So, how would you go about demonstrating the true size of our election fraud problem? Here’s just some of what you would have to do:

To begin with, you would need to verify voter eligibility. Yes, this was supposed to have been done when folks registered. But voter registries throughout America carry listings for those who are NOT eligible, including pets, literary characters, and illegal aliens. But not only do we have registrations created by fraud, we also have registrations due to error and to election officials simply not getting all their work done on time. By which I mean the failure to purge those who have died, become incapacitated, moved to another state, or become ineligible due to felony conviction. Some registries even have more registrants than residents of voting age.

So here at the very outset of your task you run into a major snag: The means by which you must prove your claims, the registries, are what we in the data processing profession would call “corrupted”, i.e. bad data, unusable data. But wait, it gets better.

Not only must you verify the eligibility of the registrants on each registry, you must show that each registrant is registered on no other registry. So for each registrant you’d need to look at all the other registries—in the nation. Multiple registrations, such as those for Faye Buis-Ewing, make it possible for a voter not only to vote for a presidential candidate more than once, but also to vote for several senators and representatives, and ballot propositions and initiatives in several states, and so on. They can do this by means of absentee and provisional ballots.

Because the voter registries are corrupted, you’d have to do what the registrars should have done in the first place: You’d have to register everybody—again. And this will be one very difficult chore, as so many states operate on the honor system. For example, to vote in America one must be a citizen; but only one state, Arizona, requires proof of citizenship to register. So you’ll need to go well beyond the “trust but [don’t] verify” standards of the states to determine if a registrant is a foreign student, a foreign tourist, a foreign jihadist, or just your garden-variety illegal alien.

So you’ll need to contact the registrants. This happened to me here in Missouri when the Jackson County Board of Election Commissioners, pursuant to the Help America Vote Act (HAVA) of 2002, asked for the last four digits of my social security number (despite my regular voting in the same precinct for 10 years). The 4 digits were to be used to triangulate with other info, such as name. But since you’re trying to detect all fraud, the better route is to get the registrants’ full social security numbers and then look them up on the social security administration’s database, where you’ll find birth dates, as well.

And why must you go through all this rigmarole? It’s those iffy registries. The voter registries can’t be trusted because the registrars are forbidden to ask for the documents that would confirm a registrant’s U.S. citizenship. (Sí, Se Puede, indeed.)

Even when you’ve identified the legitimate registrations, you’ll still have lots of work ahead of you. For it isn’t registrations that do the electing, it’s ballots. So you’ll be moving on to the next phase of your project: an examination of the elections for which the registries were created. (Are we having fun yet?)

If more ballots had been cast than there were registrants, anyone could infer that fraud has occurred. You, however, are armed with your corrected voter registries, and are thereby able to detect which unlawful registrations were used to vote. So you’ll examine the “signature rosters” to see if someone signed in using one of the unlawful registrations you identified. Your corrected registries allow you to quantify voter fraud. But how voter fraudsters affected the vote counts would be quite beyond anyone’s capabilities. This is because of another snag: There's nothing on the ballots to link them to the voters who cast them. Consequently, the vote counts can’t be corrected.

(NOTE: To protect the integrity of the vote counts, voter fraud must be detected when the voter is checking in at the polls, before he’s given a ballot. To apprehend unlawful voters, we must catch them in the act (in flagrante delicto), or not catch them at all. The means by which we can catch them is the registry. Which means the vetting of registrants must be completed by Election Day. Afterwards is too late. So if we’re really serious about nabbing voter fraudsters, Election Day becomes one giant nationwide dragnet or sting operation, rather than the spirited celebration of Freedom it was meant to be.)

But it’s not only registrants and voters who commit election fraud; poll workers and other election functionaries can commit it, as well. So it wouldn’t be enough to just examine the tallies and signature rosters after elections, you’d need to be an election observer, too, so you could check out any hanky-panky on the part of election officials. Which means you’d be conducting your fraud detection right alongside the poll workers conducting the election, bringing into play the observer effect (not to be confused with the Uncertainty Principle). Also, if there were a recount, you’d need to observe that, too. But would the authorities allow such intrusions into their elections?

There are 3 types of election fraud: registration fraud, voter fraud, and other. This last phase of your project—observing elections and recounts—falls under type 3, and it is murky and does not lend itself to definitive answers. Unlike registration and voting, where there’s a surefire means to establish eligibility and detect fraud—if only the government would use it—shenanigans by poll workers can go undetected. Because you verified the registrants earlier in your project, you will be able to verify those mysterious ballots found in the trunk of a car. But will you be able to prove that some poll worker created them? We’re not expecting you to prove a negative, so we can’t expect you to completely resolve the question of poll worker and recount fraud. But this is an arena of election fraud, so try you must.

Since election law is a state matter and procedures and hardware vary from state to state, you would need to conduct your election fraud detection project in all 50 states.

So far we’ve only been talking about “regular voters”, those who trek to the polls on Election Day to vote. But after vetting Election Day voting, you would enter a new phase of your project; vetting what is perhaps the biggest factor in voter fraud: the absentee and provisional ballots. You’d also need to “scrub” the registries for those who had died or become otherwise ineligible after the cut-off date when the registrars stopped taking applications, and see if those registrations had been used to vote. And then there’s always the possibility of a recount, which could go on for weeks and involve whole armies of “counters”. But Inauguration Day is fast approaching and we need to get this election decided—so if you could just step aside while we count these votes, thank you.

The above is only a taste of what your project would entail. And notice that we don’t really get into the logistics and the nitty-gritty of how any of the above tasks would be accomplished. We’ll leave that to you.

Now, given the enormity of your project, do you think you’d be able to complete it? Would you even be willing to undertake it?

Debates about how much election fraud occurred in this or that election are just so much jibber-jabber. It’s all speculation, as silly as trying to divine the intent of a voter from a dimpled, hanging or pregnant chad. And it’s a national embarrassment. If the amount of fraud is anything other than zero, then it is uncertain, indeterminate. Given our current systems, the exact size of election fraud is unknowable. So if we don’t know how much fraud was perpetrated, then we don’t know what the correct vote counts were, either. And in contests like Minnesota’s still-undecided 2008 race for U.S. Senate, any fraud at all can swing an election.

If the detection and quantification of election fraud isn’t doable with our current systems, then what we should be trying to do in These United States is create systems that will make election fraud as impossible to commit as we can. That, or give up on Democracy.

(This writer has tried to devise systems that would make it far harder to commit election fraud. You can read about them HERE, HERE and HERE.)

Jon N. Hall is a mainframe programmer/analyst from Kansas City.

No comments: