Every week, the New York Times prints capsule previews of upcoming NFL games. For years, each one included a prediction of the final score—say, Giants 24, Eagles 17. As Gregg Easterbrook reveled in pointing out during his four-year crusade against the quixotic practice, the paper correctly guessed a game's outcome on only three occasions out of 1,085 attempts. The Times quietly dropped final-score predictions in 2003, and now contents itself with publishing the spread.
I am reminded of this little misadventure whenever I see cooking times listed in cookbooks. I don't mean "sauté the onions for five minutes"; I mean the supposedly all-inclusive, "start-to-finish" times. In my experience, these calculations aren't just a little bit off; they're usually not even close.
Yet I can't bring myself to disregard them, because I like to follow directions, and I know I'm not alone in wanting to believe numbers listed authoritatively on a page. In the Gourmet Cookbook, Ruth Reichl says it takes 20 minutes to make halibut with spicy Asian vinaigrette and wasabi cream, and I think, Great! I can watch The Office, skip Outsourced (I don't get the appeal), throw together my halibut and still have 10 minutes to get settled for The Apprentice. Except it actually took me 41 minutes, more than twice as long as promised, and no, I don't have a DVR.
Ruth's halibut is far from the only offender. Mark Bittman's How to Cook Everything, another personal favorite, claims a pan-roasted swordfish with gingered pea puree will take 30 minutes; I slogged through it in 53. The cucumber peanut salad from Moosewood Restaurant New Classics, advertised at 20 minutes, took me 36. Andrew Carmellini's penne with bacon, radicchio, and piave cheese, in his Urban Italian, is described as "superfast; under ½ hour"; it was double that for me, at 59 minutes. (I avoided testing "30-minute" cookbooks, because I wanted to make recipes for which speed was not the main selling point.)
I should note that all of these dishes came out nicely, and pretty much as advertised, which only added to my frustration. No recipe would ever say, "Use two cups of flour, but you may need four." If the other measurements are on target, why can't timing be, too?
I asked my rabbi for some guidance. "I have a theory," he said. "Your pots and pans are too far from each other." If you saw my kitchen, you'd agree that this is not the case. Nor is it an issue of competence: I keep my knives sharp, my cupboard stocked, and my mise en place dishes stacked high. I also know that there are countless variables that recipe writers cannot control—counter space, burner strength, oven heat, and so on. But I figure these factors should add up to just a few minutes' difference.
Then, last week, I tried making some of the Thanksgiving side dishes in the New York Times dining section. Two of them—brussels sprouts with bacon, chili and coriander, and butternut squash with pecans and currants—were listed as taking 30 minutes to make. Each took me 54. I was starting to feel like Vincent Gambini storming around an Alabama courtroom: Do the laws of physics cease to exist on my stove?
With Thanksgiving just around the corner—and amateur cooks everywhere getting ready to whip up multi-course meals for large, potentially hostile groups of friends and relatives—I figured it was high time for some cross-examination.
I called up Ruth Reichl, who told me that each recipe in Gourmet was tested by cooks who made it many times over, and also by a "cross-tester," who made it only once. But the printed times came from the cooks who'd made the dish repeatedly, and Reichl allows that may have been a mistake. "When you've done a recipe eight times, you get a lot faster," she pointed out. "Probably we should've used the cross-tester's time, and not the developer's final time."
Mark Bittman had another take: "Either I'm faster or I'm wrong," he said at first. I wondered whether it was primarily a matter of knife skills, but Bittman insisted that his are "terrible." He also noted that all those uncontrollable factors really do add up to complicate the final calculation. "I make the best estimate I can," he wrote in an e-mail. "I usually take my time and add some time; I don't subtract, at least not consciously." In the end, he speculated that he's faster because he does prep work and cooking at the same time, and worries less about how the food looks at the end.
But it was Chris Kimball, editor of Cook's Illustrated, who cut to the heart of it. "Utter bullshit," he said when I asked what he thought of cooking times. Kimball is no slacker; CI, as its devoted readers know, has a well-earned reputation for accuracy. They'll bake a chocolate torte 500 times before publishing the results. Yet Kimball doesn't include start-to-finish times in his recipes; he rejects outright the notion that they can be measured with precision. "Thirty-minute recipes are never 30 minutes," he says. "It's marketing."
All-inclusive cook times are actually a fairly new phenomenon. Older recipe books, such as The Joy of Cookingand the Fannie Farmer Cookbook, assumed an audience with a certain level of competency in the kitchen, and so never bothered to say how long it would take to prepare a given meal. Today, obviously, more people with less experience are attempting to put food on the table, and so cookbooks have become more explicit. But if our recipe-writing royalty are regularly off by a factor of two, then they're not helping novices—they're confusing them. We should all accept that these predictions are about as useful as the Times' football-score experiment and, like the Times, just get rid of them.
You're sure a dish takes under 30 minutes? Call it "speedy." Between 30 and 60? Try "leisurely." More than 60? "Labor-intensive." The bean-counters may whine, but better to be honest and leave precision timing to the racetracks and the atom-splitters.