In case you hadn't noticed, we're going to elect a president here in a few months, and that means voting. Eight years ago, the humble machinery used to register ballot counts got dragged into the national spotlight when the Florida presidential election count uncertainties cast doubt on who would be sitting in the Oval Office on Jan. 20, 2001. Reports of hanging chad and other voting-system flaws motivated many local governments (which are the entities that deal with the nitty-gritty of running elections nationwide) to invest in shiny new all-electronic voting systems. But in recent years, there have been questions raised about the reliability and security of these new systems, and reportedly some municipalities are going back to the paper ballot (although still counted by computers).
What are the basic ethical issues in engineering a voting system for use by the general public? And why can't we seem to make up our minds as to which way is best?
First of all, who is involved? Every citizen meeting the legal qualifications to vote has a right to exercise that privilege. So to begin with, you have voters whose right to express their judgment in a democracy is guaranteed by law. Balloting nowadays is also secret (it didn't used to be, incidentally, even in the U. S.), so there has to be some way to ensure privacy in the voting booth.
Next you have the people being elected. They have a right to a reasonably accurate count. Not a perfect count: if we threw out the results of every election that had even one detectable flaw, we'd still be living in a monarchy. But since most elections are not photo-finish ones decided by only a few dozen votes, perfection isn't required, only accuracy that is better than the margin of victory in most cases.
Other interested parties include the election officials, the vendors selling the hardware and software used for voting, and way back in the back rooms of those firms, the engineers who design and develop the voting systems. Though these engineers are invisible to nearly everybody else, they obviously play a key role.
Now that we have identified the main parties to the matter, what can go wrong? Just to make things interesting, let's compare the latest touch-screen voting systems with the totally manual paper ballots that were used, for example, in the 1948 election that put Lyndon B. Johnson in the Senate.
There is a strong, almost intuitive, bias toward paper records in law and politics. Paper and ink are just as technological as computers and software—it's just that paper is an older and more familiar technology. It is integrated in our ways of thinking in ways that digital technology isn't, at least not yet. Plus which, paper systems can be easier to understand, and transparent in a way that software, for instance, is not. Unless a document is written in Urdu, say, or legalese that only a lawyer can decipher, you don't need an expert to read paper, but you do need one to tell what's going on in software.
All that familiarity with paper was of no avail when certain shenanigans went on in certain South Texas voting precincts back in 1948. Johnson biographer Robert Caro has shown how as many as 10,000 ballots in the Democratic primary that effectively determined the election outcome were highly suspect. And in an election that was won by only some 300 votes, that was more than enough to determine the outcome. The point is that, given enough corrupt officials and political pressure in the right places, paper ballots are no sure-fire defense against fraud. But everybody knows that.
With all-electronic voting, not only are people worried that a malevolent hacker working for one party will infest the system software to deliver enough votes to push that party's candidate to victory, but that mistakes or malfunctions will go undetected because without paper records, there is no way for the average non-technical election worker or politician to check the results. The only people who can even come close to doing that are the folks who can look at the software innards of the machines, and even they can't always recover a blow-by-blow description of everything that went on during the voting.
A recent New York Times editorial pointed out three instances in the last few years in which either all-electronic or partly electronic voting systems led to incidents which at least cast doubt on the results. The editorial writers came out with a proposal which is also being seriously studied by engineering researchers: voter-verified paper record systems (VVPRS for short). In these systems, each voter gets to see a piece of paper that reproduces his or her choices, and if the paper doesn't match the voter's desired choices, the voter can start over and do it right. Only when the voter is satisfied does the ballot get recorded, both electronically and on good old cellulose.
Of course, printing out a bunch of paper in addition to doing electronic ballot recording takes away some of the advantages of the digital system, but it's no different than in other areas where computers have found use. I remember the day when Bill Gates said that computers would eventually make the paperless office possible. As I recall, stocks in paper companies plummeted the next day, but the finance types needn't have worried. If anything, we have more paper to deal with than ever, now that it's so easy to print professional-looking documents at the touch of a button. But I digress.
Paper, electronics, white and black stones—fundamentally, voting is a non-material process mediated by physical communication systems, and the physical media used doesn't much matter if the will of the people is adequately expressed through it. Integrity, good will, and common sense makes it work pretty well most of the time, which is all you can expect of human systems. The big scandal about U. S. elections is not the technology, but the fact that so many people pass up the opportunity to vote. Don't let that be true of you this November.
Sources: The New York Times editorial appeared on July 31, 2008 at http://www.nytimes.com/2008/07/31/opinion/31observer.html. A paper describing a study of a VVPRS electronic voting system by Nirwan Ansari and others at the New Jersey Institute of Technology appeared in IEEE Security and Privacy for May/June 2008, pp. 30-39. And LBJ's South Texas ballot tricks are described in Robert A. Caro's excellent multivolume biography The Years of Lyndon Johnson.