Proofiness - Charles Seife [121]
17
Which is why I did precisely that—in the Journal of Neuropathology and Experimental Neurology, of course. The editor in chief had a good sense of humor.
18
Anti-NutraSweet lobbyists make equally flimsy assertions that NutraSweet causes other diseases as well, such as multiple sclerosis, lupus, seizures, and mental retardation. They also like to point out that a former CEO of the company that developed and manufactured NutraSweet was none other than George W. Bush’s secretary of defense, Donald Rumsfeld.
19
Of course, failing to prove that something is dangerous isn’t equivalent to proving that something is safe, and Olney presented other evidence (such as a study in rats) to try to bolster his case. However, it’s a good bet that NutraSweet (and many other items that have been blamed for brain cancers, such as cell phones) is not guilty of the accusations, in part, for reasons described below.
20
As ridiculous as the tobacco-causes-heroin-use causuistry was, it forced Republicans into the even sillier position that cigarettes were just dandy. “There is a mixed view among scientists and doctors whether it’s addictive or not,” Republican senator and presidential candidate Bob Dole said in 1996. Dole’s stand on tobacco was, shall we say, quixotic. “We know [smoking is] not good for kids,” he had said a few days earlier. “But a lot of other things aren’t good. Some would say milk’s not good.”
21
The “nun bun,” discovered in Tennessee, disappeared in 2005. It made a brief appearance in Seattle two years later, but as of the time this book was written, the bun’s on the run.
22
A failed prediction is a very solid sign that a pattern is phony. A pattern allows you to make a prediction; when you think that red gives an advantage over blue, you can predict that, on average, more red-wearing athletes will win an event, and you might even be able to make some money at the bookie’s. A false pattern has no predictive power; it might seem to give you a lot of power to understand past data, but it completely breaks down when tested against new data. We’ll see this again and again. (Yes, this is a pattern.)
23
Hedging their bets, the team did some complicated computer simulation that indicated that the “momentous” day when women beat men in the 100-meter dash could come as early as the year 2064 or as late as 2788.
24
Yes, they predicted the time down to a hundredth of a second—a clear-cut case of disestimation.
25
Fed up with the prediction chatter, I wrote an op-ed in the Post that used an obscure numerical sequence, the 4-Knödel numbers, to predict the outcome of the election, which it had done accurately for every election since 1952. It was silly to say that such a sequence could determine the victor of the presidency, but it was no more silly—and no more wrong—than the prediction provided by economists with their regression to the moon.
26
For those of you who noticed that the formula, as presented, doesn’t make mathematical sense . . . you are correct. It scarcely matters, though.
27
Unfortunately, this sort of hanky-panky isn’t unique to 60 Minutes. In 1992, Dateline NBC ran a spectacular segment about the dangers of GM trucks, showing one bursting into flames when struck in the side. There was in fact a design flaw, but Dateline had neglected to mention that to ignite the blaze, the TV show had strapped rockets to the truck’s gas tank. Unlike 60 Minutes, Dateline was forced to apologize for its actions. Without a hint of irony, Don Hewitt, 60 Minutes’ executive producer, used the occasion to tout his show’s superior ethical standards: “If that had happened at 60 Minutes,” he told the New York Times, “I’d be looking for a job tomorrow.”
28
The United States is not alone in its mismanagement of terrorist risks. In mid-2009, the United Kingdom relaxed