Ken O'Brien was an NFL quarterback in the 1980s and 1990s. Early in his career, he threw a lot of interceptions, so one clever team lawyer wrote a clause into O'Brien's contract penalizing him for each one he threw. The incentive worked as intended: His interceptions plummeted. But that's because he stopped throwing the ball.
Years ago, AT&T executives tried to encourage productivity by paying programmers based on the number of lines of code they produced. The result: programs of Proustian length.
Incentives are dangerous, and not just because people game them. They often yield collateral damage. Remember the tale of the Darwin Award winner who strapped a jet engine to his car, dreaming of a joyride for the ages, and then met his sorry end as a human flapjack on the side of a mountain? Incentives are like that jet engine. There's no question the engine will take you somewhere, fast, but it's not always clear where. Or what you're going to mow down on the way. Yet incentives are still the first resort of most managers, perhaps because they all think they're smart enough to create the perfect carrot.
Take Merrill Lynch. In the book Riding the Bull, author Paul Stiles describes his experience as a new trader at the venerable investment bank. Merrill wanted Stiles, then 29, to trade complex international bonds in volatile markets. He tried asking advice of the seasoned traders, but they ignored him -- a minute spent helping Stiles was a minute spent not adding to their monthly bonuses. They kept barking into their phones for hours at a time and yelled at Stiles every time his shadow fell across their computer screens. Eventually, Stiles was reduced to silently observing their behavior from a distance, like a rogue MBA anthropologist. It surely never dawned on the person who set up Merrill Lynch's incentive system that the traders' bonuses would make training new employees impossible.
Why are we so bad at anticipating the effects of our well-intentioned incentive plans? The answer has to do with something that psychologists call a "focusing illusion." Behavioral economist Daniel Kahneman and management professor David Schkade surveyed Midwestern students and asked them to predict the satisfaction of students in California on several dimensions, such as "job prospects," "climate," "personal safety," and overall life satisfaction. They also asked the Midwesterners to rate their own satisfaction. The professors then posed the same questions to actual California students and compared the answers. The Midwest students correctly predicted that the Californians would be happier about their weather.
But the Midwestern students wrongly predicted that California students would be happier with their lives in general than Midwestern students. The overall scores are identical. Schkade and Kahneman showed that, in essence, the Midwestern students erred by focusing too much on a single variable. When you're a Midwesterner contemplating a long, cold winter, you can't help but think that Californians must be happier. But you're ignoring the larger happiness portfolio, in which weather recedes to insignificance among the other things that may influence a Californian's satisfaction -- good friends, terrible traffic, career opportunities, laundry, and a governor who compulsively repeats Terminator jokes.
Focusing illusions even distort our judgments about ourselves. In another study, some college students were asked, "How happy are you?" and then "How many dates did you have last month?" The researchers found a pretty weak correlation between the level of happiness and the number of dates. But then (hilariously) the researchers flipped the order of the two questions. Suddenly, there was a strong correlation. Having just confessed to a lack of dates, students reported that their lives were joyless.
And this brings us back to the incentive puzzle. When the football team's lawyer hatched a plan to minimize O'Brien's interceptions, he was suffering from a focusing illusion. He reduced the world to a one-variable equation, like a college student fretting about his flatlining dating life.
To be fair, there are some contexts where one variable dominates. If you're employing a field sales rep who is selling a simple, self-contained product, then it probably makes sense to tie incentives to the sale. If you're traveling a long, straight road, the jet engine will get you there faster.
But chances are you don't live in a one-variable world. In your complicated, squishy, matrixed world, if you're dreaming up an incentive plan, you're almost certainly in the grips of a focusing illusion. You're trying to maximize or optimize or minimize something. And you may unwittingly find that when you maximize the length of your programmer's code, you end up minimizing your job tenure.
There's another option. In a maddeningly multivariable environment, great management trumps great incentives, for the simple reason that managers are multivariable. Wouldn't it have been better for a coach to give O'Brien some help on refining his field vision? Shouldn't a Merrill Lynch exec have made it clear that traders were expected to help out the new guy?
Incentives are dangerous. Good managers aren't. So forget about that jet engine and get back to the slow, messy business of actually interacting with your employees.
Read more Made to Stick columns 
Dan Heath and Chip Heath re-released their best-selling book Made to Stick featuring new content, such as how to make strategies stick and unsticking an idea.