FLORIDA STRIKES AGAIN:
What The Latest Election Snafu Says About Machines And Humans

By MICHAEL C. DORF

Wednesday, Sep. 18, 2002

In the aftermath of the disputed 2000 Presidential election, Democrats lionized the Florida Supreme Court, vilified the U.S. Supreme Court, and complained about a legal system that did not accurately record the intent of the voters. Republicans, for their part, vilified the Florida Supreme Court, lionized the U.S. Supreme Court, and complained about the subjective process by which ballots designed to be read by machine were to be held up to the light by human beings.

Once the wrangling was over, members of both parties moved swiftly to eliminate what they thought was the underlying cause of the problem: antiquated machinery for recording and reading ballots. In many Florida counties, touch-screens resembling automatic teller machines replaced punch cards.

With software that asked voters to confirm their choices, the new machinery, it was thought, would avoid the Scylla and Charybdis of over-votes and under-votes. And so it did--more or less. In most counties that moved to the new system, poll workers and voters reported high levels of satisfaction.

Most, but alas, not all.

What Went Wrong in The Recent Election, and Why There Won't Be A Recount

In numerous precincts in Miami Dade County and other parts of south Florida, the new equipment arrived late and had to be reprogrammed very close to the primary. That might have been all right, except that in these precincts, the election workers received inadequate training in the operation of the high-tech machinery, so that when problems arose, poll workers could not respond. As a result, thousands of votes were either not counted or counted inaccurately.

That too would have been all right, except that Florida, land of sunshine and close elections, produced another near-tie. In the Democratic gubernatorial primary, Bill McBride received 44.5 percent of the vote, Janet Reno garnered 43.9 percent, and Daryl Jones recorded 11.6 percent.

McBride's initial margin of 0.6 percent over Reno was just enough to avoid a mandatory recount--which occurs whenever the top two candidates are within a half a percentage point of one another. Recounting of Miami Dade County ballots over the weekend reduced McBride's margin to 0.4 percent. Nevertheless, as this essay goes to press, the Reno campaign--still facing a deficit of over five thousand votes--appears unlikely to challenge the final certification that is expected today, September 18.

Thus, Floridians will be spared the anguish--or perhaps farce would be a better description--of replaying the events of 2000. They can now concentrate on the general election pitting McBride against incumbent Jeb Bush, and hope that that contest produces a decisive winner.

Meanwhile, those of us who live and vote outside of Florida can ask what lessons this latest fiasco teaches.

Lesson One: Democracy Really is Just Another Government Program

During and after the 2000 Presidential election dispute, numerous observers remarked that our electoral machinery was badly under-funded. For years, they noted, states and localities had been getting by on outdated technology and largely untrained volunteer election workers. If we want an election system that works, the observers noted, we will have to pony up some serious cash.

The critics certainly had a point. Elections don't run themselves. Like every other government program, they require resources to run effectively.

However, some of the critics went further, arguing that improving the mechanics of our electoral system should be the number one priority of government. Because elections are what give government legitimacy in a democracy, the critics charged, running them smoothly must precede all other government tasks.

Yet, spending on electoral machinery can justifiably be required to compete with other government priorities. Certainly, survival as a society must rank at least as high as ensuring an accurate electoral count. That is why, despite Warren Christopher's lament in a New York Times Op-Ed over the weekend, it is perfectly understandable that "[t]he attacks of Sept. 11 deflected attention from everything that was not connected to national security."

And once one acknowledges that national security can sometimes take priority over electoral accuracy, it is difficult to see why spending on other matters--such as environmental protection, medical care, and education--should not also be given priority at some point.

None of this is to say that Florida was justified in the inadequate level of training it provided its poll workers, or that Congress should not act on the currently pending election reform bills that former Secretary of State Christopher favors. Such measures may well be cost-justified, even taking account of other worthy, non-election-related programs. But it is a mistake to think that our electoral machinery must be rendered perfect before we can turn our attention to any other matters.

Lesson Two: Nobody, and Nothing, is Perfect

And that brings me to the second lesson: It is impossible to design an electoral system that will perfectly measure every vote in every election. Even machines with no moving parts can malfunction because of power outages or software bugs. And when they do, fallible human beings will need to intercede.

The right question to ask is not: how can we make the system foolproof? Rather, the right question is: how can we ensure that the inevitable errors are tolerably small, and not systematically skewed against a particular candidate, party, or social group?

In the 2000 Presidential fight, each side had an appealing battle cry. The Gore team said "Count every vote," while the Bush crowd replied "Don't introduce human subjectivity." Neither slogan was realistic.

Given the inevitability of error, it is impossible to count every vote perfectly accurately. But at the same time, it is also impossible to avoid human subjectivity. For example, even a well-programmed computer system will have to allow for write-in votes, leading to the necessity of some human being deciphering handwriting.

And even a computer system that allows "write-in" voting by the entering of a name on a keypad will allow for spelling errors, which will then call for some human judgment. (If you think write-ins are an irrelevancy, take a look at Washington, D.C., where last week, thanks to a write-in campaign, incumbent Mayor Anthony Williams won the Democratic party primary by a 3-to-1 margin, despite not having his name on the ballot due to petition irregularities.)

To err is human, and humans both construct and operate election machinery. We must therefore accept both the inevitability of some error and of some element of subjectivity in coping with that error.

Lesson Three: Sometimes There's No Such Thing as a "Real" Winner

Finally, there is a radical and unsettling lesson to be taken from both the 2000 Presidential election and the latest Florida foul-up: In a very close election, and certainly in one in which the margin of error is larger than the apparent margin of victory, the idea that there is a single "winner" is a social construction--a necessary fiction.

It is tempting to think that the answer to the question of who got more votes in an election is a question about the natural world of the same order as "Is the moon made of cheese?" or "How tall is Mount Everest?" But in fact, these are different sorts of questions.

The composition of the moon and the height of Mount Everest are facts in themselves that would exist regardless of whether we attempted to measure them. (I know, some postmodernists disagree, but I feel comfortable putting aside their objections for now.) By contrast, an election does not exist independently of our attempt to measure it.

Granted, an election is itself an attempt to measure something: the public will. But the public will is itself a constantly-changing, slippery phenomenon. In every election, substantial numbers of voters do not make up their minds until the waning days of the campaign; numerous voters don't even decide until they are actually in the election booth.

Of course, once the election has taken place, we can ask how many voters voted for candidate X and how many for Y, but the answer to that question depends on contested issues. Did a voter vote for Gore if she went into the booth intending to vote for Gore but accidentally voted for Buchanan? What if she intended to vote for Reno but was precluded from voting because of a two-hour line at the poll, itself the product of a malfunctioning machine? What if she wrote in but misspelled the name of her favorite candidate?

A well-designed electoral system will try not only to minimize errors, but also to anticipate and resolve in advance the sorts of questions likely to arise in a closely contested and inevitably flawed election. However, as any legislative draftsman knows, it is simply impossible to anticipate every possible error in advance. That is not a reason not to try, but it is a reason to think that we have not heard the last call for a recount--and probably never will.


Michael C. Dorf is Professor of Law at Columbia University.

FindLaw Career Center


      Post a Job  |  View More Jobs

    View More