This could be the tightest presidential race since 2000 when Al Gore and George W. Bush took their fight over how to recount Florida's ballots all the way to the U.S. Supreme Court.
Bush was declared the winner, but not until December of that year, more than a month after the polls had closed.
The political mechanics in both parties took that result as a wakeup call and they've been preparing for another race like 2000 ever since.
Now they have it, what have they done to be ready for it?
Over the last decade, Republicans and Democrats independently asked themselves what they really know about voters, what makes them tick? The answer was that they really didn't know all that much.
So, reluctantly at first, they began to listen to the consultants who told them they could learn from the way corporations did market research on their customers.
And what they came away with were new and sophisticated ways to scoop up and analyze the data that voters routinely leave behind as they go about their lives.
The geek squad
In recent years, both Republicans and Democrats enlisted self-described "geeks," adept at tracking such things as the magazines people buy, the cars they drive, their vacation habits and the restaurants they choose.
Hunched over computer screens, these "geeks" began following voters by tracking data from subscription lists, product warranties, credit card information and so on.
They sorted those bits of mundane information into hundreds of "data points" and directed their software to reassemble the bits into individual portraits of voters complete with names, addresses and phone numbers.
When they matched it all up with government records about voting history, it turned into gold.
Suppose, for example, the data showed that white women between the ages of 35 and 45 who subscribe to the New Yorker magazine are 10 times more likely than the national average to vote Democrat.
But suppose that within that group is a small segment who also subscribe to Field and Stream and that those women sometimes vote Republican.
What that combination of data uncovered is the needle in the haystack, the persuadable voter in a sea of hardcore Democrats.
Put that fact together with a name, address and phone number and a Republican campaign has a precise target for direct mail political messages, phone call pitches and "get out the vote" tactics.
Now multiply that tiny nugget by several hundred thousand in say, swing-state Virginia (pop. eight million) and there's the micro-targeting advantage.
To put it another way, campaigns can now identify the one or two supporters they might have on a street of 15 homes in an enemy neighbourhood. And they can do it reliably and efficiently, in a way that might turn the tide.
As a result, Republicans and Democrats have both found new support where they never would have looked before.
Often they know more about how a single voter might behave in the voting booth than the voter knows herself.
That's led to more aggressive "get out the vote" tactics because the campaigns no longer have to rely on what voters tell them about how they'll vote, they already know who they should get to the polls.
The consequences have been dramatic. Voter turnout in the 2004 election was almost four points higher than 2000 and in the 2008 presidential election it was the highest in 40 years.
Just as important for political operatives, the 2008 election data revealed important demographic changes in turnout since the previous election.
Voting was up significantly among young, Hispanic and black voters. Among "non-Hispanic white voters" it was down slightly.
Another way to read that data is that turnout among likely Democratic voters was up while turnout among likely Republican voters was down.
But here is the pernicious side of the equation: if you know which voters you want to get to the polls, you also likely know the reverse. You know which voters you don't want to get to the polls.
Such is the context for some of the controversies that have erupted this electoral season.
There have been dozens and dozens of reports alleging that legislative changes at the state level are designed by Republicans to deliberately and systematically suppress the vote among the poor, blacks and Hispanics.
The evidence of vote suppression is not conclusive, but the demographic data from 2008 suggests Republicans would have reasons to think about it.
Among the changes to election regulations over the last couple of years is the requirement in several states that voters produce photo ID.
Minorities, the poor, the elderly, and yes, the Democrats, have protested that these requirements will disenfranchise many voters.
Supporters of photo ID say it's no more onerous for anyone to produce ID when voting than it is to produce the same ID when purchasing Sudafed or boarding an airplane.
There's an intuitive logic to that argument. But the data can tell a different story.
In August, a federal court disallowed a photo ID requirement in Texas because the data showed that Hispanics in particular were less likely to have photo ID than other voters and that the cost of obtaining a photo ID would be at least $22 and possibly much more.
The court's conclusion was that requiring photo ID in Texas would place a disproportionate burden on an identifiable group of voters. In other words, it discriminated against Hispanics.
The attention paid by minority groups to the photo ID issue — there are TV ads here warning blacks and Hispanics not to let their votes be stolen — is creating the suspicion among these groups that Republicans don't want them to vote.
And it's possible that photo ID, rather than discouraging minority voters, will become a motivating issue that leads to a "voting with a vengeance" backlash at the polls next week.
In either case the photo ID controversy is a part of this election's history now and its unlikely minority voters will soon forget it.
In any event, one thing elections guarantee is that, after the polls close, there will be an avalanche of new data to mine for more clues about voting behaviour.