Fuzzy Math

How you look at things.
Nov. 15 2000 3:00 AM

Fuzzy Math

93000_93646_brown_votecount

Have the votes for president been properly counted in Florida? On the surface, that's a question of simple math. But beneath the number crunching, Republicans and Democrats are waging a war of disguised biases. When data don't turn out the way your theory predicts, should you question the theory or the data? When a new vote tally contradicts an old one, should you distrust the first count or the second? When one kind of recount is more evenhanded but another is more comprehensive, which is better? These dilemmas form the hidden crux of the debate over whether to recount Florida's ballots by hand, as Democrats prefer, or to rely on a machine recount, as Republicans prefer. The two parties aren't being candid about these questions. And math won't answer them.

William Saletan William Saletan

Will Saletan writes about politics, science, technology, and other stuff for Slate. He’s the author of Bearing Right.

Advertisement

1. Theory vs. data. Since the machine count so far favors George W. Bush, Republicans pretend it's beyond question. Democrats, meanwhile, try to adjust the data to fit their theory. They know a priori that Gore won. "We believe down deep that if the views of the Florida voters [are] truly known, it will turn out that Al Gore has more votes than others," said Gore legal representative Warren Christopher Monday. When the data don't match this theory, Democrats call it an "anomaly," which means the data must be wrong. Pat Buchanan got too many votes in Palm Beach County, and Harry Browne got too many votes in Volusia County. Gore couldn't have lost those votes honestly. They must have been stolen.

Democrats also point out that Gore was ahead in Florida's pre-election polls and in exit polls. Do the returns prove the polls were wrong? No, the polls prove the returns are wrong. Since Gore led "by four to six points coming in to Election Day," his deficit in the machine tally indicates that "his votes weren't appropriately counted," former Clinton spinmeister Lanny Davis argued on CNN's Late Edition. As for Buchanan's votes in Palm Beach County, Fox News pundit Alan Colmes theorized that "exit polling showed people were saying" they had voted for Gore when in fact they had marked their ballots for Buchanan. Therefore, "many of those votes were meant for Al Gore."

There's an old joke about an adulterer caught in the act by his wife. He asks her, "Who are you going to believe—me, or your lying eyes?" The Democrats' spin in Florida boils down to the same thing: Which are you going to believe—the polls, or these lying ballots? Democrats cover this creepy logic with Orwellian euphemisms. Gore campaign chairman Bill Daley argued this weekend that many ballots marked for Buchanan "were votes for" Gore. Rep. Bob Wexler, D-Fla., asserted, "It's impossible that Pat Buchanan got those votes." Senate Minority Leader Tom Daschle said 19,000 ballots in Palm Beach County "were miscast." Gore himself suggested that Bush "would not want to win the presidency by a few votes cast in error." Impossible? Miscast? Cast in error? What's the point of voting, if politicians won't believe the ballots?

Republicans have their own theory: Machines are infallible. What happens when human vote counters find a ballot that was ignored by the machines? Dismiss the data. "Any card that the machine didn't accept has got to be ambiguous," declared Dick Thornburgh, the former Republican attorney general, on Late Edition. "If it's unambiguous, it would have been counted."

2. Nature vs. progress. The liberal instinct is to change things in the hope of improving them. The conservative instinct is to trust the way things are and distrust efforts to change them. In the Florida recount, each party plays to its instinct. Democrats want you to equate change with progress: The more time spent, the more precincts reviewed, the more tallies updated, the closer we're getting to truth and justice. "Having enough patience to spend the days necessary to hear exactly what the American people have said is really the most important thing," Gore argued Monday.

Democrats also want you to feel good about the rising vote totals. The higher the tallies go—as successive recounts turn up presidential votes on more and more ballots—the more voters are being heard. "The Bush campaign has argued that every minute counts. We have consistently maintained, however … that every vote must count," Christopher preached Saturday. On Fox News Sunday, former Democratic Senate Majority Leader George Mitchell equated rising tallies with error correction: "Both candidates get more votes, clearly reflecting the errors that occur when you rely solely on a machine count." From this, Democrats infer that Republicans, by resisting further recounts, are trying to "disenfranchise" voters whose ballots were rejected.

Republicans assume the opposite. They equate change with corruption. On this view, the initial, uncertified numbers reported by the Associated Press embody eternal truth. "It's over, Bush won once, he won twice, he's president of the United States," Senate Majority Leader Trent Lott opined on Fox News Sunday. Therefore, demands for further review are an attempt to "overturn the results," in the words of Bush spokeswoman Karen Hughes. Hand recounts would change the numbers not by discerning the truth but by tampering with it. "The potential for human error [and] mischief is so great in these hand counts that it skews the result," Bush legal representative Jim Baker argued Sunday. For example, "When you recount these ballots over and over and over … they deteriorate, and the little pieces of chad fall out."

Likewise, Republicans want you to feel suspicious about the rising vote totals. "Each time, they seem to find more Gore votes. What's going on here?" Lott asked Sunday. While Gore holds out the promise of a perfected tally in which "every vote is counted and counted accurately," Republicans regard this as precisely the sort of illusion that tempts liberals into counterproductive tinkering. The estimated machine error rate of "2 to 5 percent … is a much smaller error rate than you get with manual recount," argued Rep. Tillie Fowler, R-S.C., on Late Edition. "There's no perfect system."

3. Logic vs. judgment. There are two kinds of intelligence: the ability to reason abstractly across a variety of cases and the ability to tailor one's evaluations to each case. In this comparison generally, and specifically in the Florida recount, Republicans are from Mars, while Democrats are from Venus. To persuade you that machines are better than humans at counting ballots, Republicans define intelligence in terms of scientific consistency. While extolling the "precision machinery" that tabulated the initial results, they dismiss manual recounts of partially punched ballots as "subjective," "selective," "arbitrary," and riddled with "discretion" and "human error." They accuse human vote counters of trying to "mind-read" and "divine the intention of the voter."

Is consistency really more important than discretion? Is it better to exclude all ambiguous ballots equally than to rescue as many wrongly rejected ballots as possible? Does the former approach necessarily yield fewer errors than the latter? Republicans rule out these questions a priori. "A manual count under the situation that exists here would have a far greater error rate [than computers] because it is so subjective," Baker insisted on Meet the Press. "It cannot be as accurate as a machine count."

The implicit premise of this stipulation is that human beings, as conservatives have always argued, are morally weak. "It's not just human error, it's human manipulation. Those canvassing boards are 2-to-1 Democrat-Republican," Republican strategist Mary Matalin argued this weekend. By contrast, Baker argued, "Machines are neither Republicans nor Democrats, and therefore can be neither consciously nor unconsciously biased." The conservative wager, in other words, is that systematic human bias yields more error—or a worse kind of error—than systematic computer flaws do. That wager may be true. But it hasn't been articulated or scrutinized.

Democrats, conversely, stack the deck against the computer. They define intelligence in the language of a trial court, emphasizing "evidence" and case-by-case judgment. Human vote counters "look at the evidence they have before them to see whether or not the machine count was accurate," Christopher argued this weekend. Daley agreed: "It's the judgment that they make, based on evidence." Meanwhile, Democrats counter the GOP's conservative bias against human nature with a liberal bias in favor of unconstrained democracy. When asked on Meet the Press about the unreliability of a human recount, Christopher sniffed, "These are local election officials chosen by the people in their counties." Does Gore trust politicians to count the votes of the people? No, he trusts people "chosen by the people" to count the votes of the people.

In the debate over whether to trust humans or computers, Democrats enjoy a crucial advantage: The audience judging the debate is human. We don't notice the occasions on which we're wrong and computers are right—and when we do, either we refuse to admit it or we erase it from our memories. But we always notice and never forget the occasions on which we're right and computers are wrong. Democrats are happy to remind us of these occasions. "The machine counts are very often inaccurate, just as you find probably your own credit card bills are sometimes inaccurate," Christopher suggested on Meet the Press. Democratic strategist James Carville agreed: "All of this just unbelievable faith that these people have in these machines—I mean, have you ever gotten a flight reservation that wasn't there?" 

Throughout the weekend, as he argued that the machine tally should be followed by a manual tally, Christopher described the human counters as "checking" the computers rather than the other way around. "The citizens of those counties felt that the machine count did not accurately reflect [their] views," he explained. Bush "should not have any reason … to fear to have the machine count checked by a hand count." And why is a hand count good enough to "check" a machine count? It's "seen as the best way to ascertain the true views of the voters," said Christopher. Seen as the best way by whom? By us, the human audience whose egotism Democrats are invisibly exploiting. Against this unconscious bias, Jim Baker raises the specter of Democratic mischief. The dilemma in Florida turns, in the end, on a question we dare not ask computers: whether we're more afraid of them or of ourselves.