Suppose that somehow the fate of the Earth depended on correctly ascertaining the outcome of the Florida vote in 2000. Suppose, just to get the point clear in your head, that aliens had decided to blow up the Earth if we got the recount wrong. And suppose you were in charge of determining how to handle the situation. On the one hand, you have the experts informing you that the machines have a 1-2% inaccuracy rate, and you know that the margin of victory in this case is far, far below that threshold. You also know that humans are eminently capable of performing the relevant perceptual and judgmental tasks, and capable of doing them better than machines. You know that the task will not be easy, and that there will be some borderline cases--dents so small that no one will be able to judge whether it is or is not a vote. But you also know that most cases will not be like that at all. In most cases it will be perfectly clear who the voter intended to vote for, and in the unclear cases humans will more reliably produce right answers than machines will. In short, you know what the voting machine experts told us back in 2000--that humans are more reliable tabulators of punch-card ballots than are machines.

Now, with the fate of the Earth riding on your decision, would you decide to stick with the machine counts or recount by hand?

That's what I thought.

And if democracy really mattered to the members of Bush's 2000 campaign, that's what they would have decided, too.

Another way to settle cases like this is by looking at "prior commitments." In the heat of the moment, people's judgments about what is fair and unfair are too often mangled by their desires. Our judgments in a cool hour, when we have nothing directly at stake, are often much more reliable. And, of course, Bush himself had signed into law a bill in Texas that required manual recounts in such cases, indicating that he himself acknowledged their superiority. The Bush camp responded to this by arguing that the case was different in Florida because no standards were established in the Florida law. That's wrong, however. All of the standards employed in Florida recounts were reasonable standards, and all would produce more accurate counts than the machines produced, so the fact that the methods differed was irrelevant. Using different reasonable standards of recounting is no more unfair than using different reasonable types of voting machines.

So, did the Bush camp steal the election? I don't know, and probably neither do you. Is it theft to take something that may or may not already belong to you? Suppose there is a valuable book, bought and sold back and forth between your ancestors and mine over the course of generations. Suppose we lose track of who actually owns the book, but that we have a box containing all bills of sale for the book over the years. Some of the papers are cryptic and difficult to read, but most are perfectly legible. It is clear that our only hope for determining the actual ownership of the book is to carefully reconstruct the book's sales history. It may be yours, it may be mine, but we won't really know until we read through all the papers. But driven by raw greed, one day I simply decide to take the book, ownership be damned. Suppose we later discover that the actual history of the sales is incomplete, and ownership cannot be conclusively determined.