Switch - Chip Heath [36]
But the averages concealed some fascinating biases. For instance, when the explorers gave the well a 20 to 70 percent probability of hitting, their predictions were pretty accurate. But when the explorers predicted a greater than 75 percent chance of success, the wells hit nearly all the time. Also, wells that had been given a 10 percent chance of success actually had more like a 1 percent chance. So the explorers’ instincts about the wells were correct—they knew the good ones from the bad ones. But there was information, especially for high-and low-probability wells, that they weren’t using.
Traditionally, the explorers had been salesmen for their wells, pressing top management for the green light to drill. In the 1980s, they had learned that the way to sell management was to use the tools of risk economics—in particular, the concept of expected value.
Expected-value calculations are bulletproof in situations where the risks and returns are well understood. If I flip a coin, you can feel confident that you have a 50 percent probability of winning. But what are the odds of hitting a gusher? And if you hit one, what’s the payoff? Those are subjective estimates. When you feed subjective estimates into an expected-value calculation, a precise number pops out, giving the illusion of scientific certainty. (“Our expected value for this well is $112.8 million. It’s a no-brainer—let’s drill it.”)
It didn’t go unnoticed among explorers that if they really wanted to drill a well, they could simply tinker around with the math in a spreadsheet. If they jacked up the hit rate or the payoff, the expected value obediently inflated. (This tinkering probably wasn’t malicious or even conscious. Remember, when the Elephant really wants something, the Rider can be trusted to find rationalizations for it.)
More subtly, the use of expected value made people think about drilling as a numbers game. As Jim Farnsworth, a top leader in BP’s exploration unit, said, “Explorers think in terms of risk probabilities. People get so caught up in the numbers that they think, ‘Well, if we drill ten of these 1-in-10 wells, we’ll hit at least one of them and we’ll all make a lot of money. But when you do the analysis, you realize that something that is 1 in 10 never works, so it’s a false sense of statistical clarity.”
The odds-playing gave everyone a false sense of comfort. Hey, if we drill some dry holes, one of the other holes will hit and make up for it. Explorers were like venture capitalists, hoping for an eBay or a Google to bail them out of an otherwise lousy portfolio.
If you were an executive at BP, hoping to cut your exploration costs by 80 percent, your first mission would be to remove this false sense of comfort. The ambiguity in the goal is allowing rationalization to creep in. So how could you change your team’s behavior so that every single drilling operation is taken seriously? How could you leave your team’s Riders with nowhere to hide?
Consider the alternatives for your new strategic rallying cry: “We’ll double our strikes!” “No more dumb holes!” “Let’s maximize expected value!” Some of them sound promising, but notice the fudge room in all of them. “No more dumb holes” would be easy for any competent Rider to dodge. Would any self-respecting explorer think he was drilling a dumb hole? “Doubling strikes” is better, but there’s still room to rationalize a lot of dry holes. And as for “maximize expected value,” well, let’s just move on.
Ian Vann, BP’s head of exploration at the time, figured out a way to eliminate the fudge room. He announced his new vision: “No dry holes.”
None.
Explorers were irate. They thought the goal was preposterous. Their leaders were asking the impossible. Dry holes had always been a normal and acceptable part of doing business. Remember, the number of dry holes outnumbered successes by a factor of 4. Now Vann was defining them as failure.
“No dry holes” was a painful B&W goal. Probabilistic predictions had always provided a cover