Dan Pink had finished with the candle. For the first part of his TEDGlobal talk in July 2009 he had worked the room like a jury, using a 1962 puzzle experiment to argue that cash prizes can slow people down on problems that need a fresh idea. (The candle problem has its own page, and its own complications.)
Now he turned to what should replace the prizes. The new operating system for business, he said, “revolves around three elements: autonomy, mastery and purpose.” He gave each a line. Autonomy, “the urge to direct our own lives.” Mastery, “the desire to get better and better at something that matters.” Purpose, “the yearning to do what we do in the service of something larger than ourselves.”
Then he narrowed it. “I want to talk today only about autonomy.”
His first example was an Australian software company, Atlassian. A few times a year, Pink said, it told its engineers to spend 24 hours on anything they wanted, as long as it wasn’t part of their regular job, and show the results at an all-hands meeting. They called them FedEx Days, because you had to deliver something overnight. “It’s worked so well,” Pink told the audience, “that Atlassian has taken it to the next level with 20% time.”
Five months earlier, Atlassian had published its own review of that 20% time.
The 1.1 percent
The experiment had run a year. It produced 48 projects and, by the company’s rough count, 248 days of effort. Atlassian put the result in quotation marks: “1.1% Time.” Its records were loose, it admitted, and the real figure might be two or three times higher, “which is still very low.”
The engineers’ survey said why. “Getting 20% time is incredibly difficult amongst all the pressure to deliver new features and bug fixes,” one wrote. The company added that it wasn’t harsh team leaders. Developers didn’t want to pile work on their peers, or “be seen as enjoying a privilege while others carry the workload.”
The freedom was on paper. Taking it had a price, and the price fell on teammates. Choice only works as autonomy if people can actually use it.
Once pay feels fair, choice, growth and a reason can move people in ways a bonus doesn’t.
Pink attached a condition to the whole idea, and it is easy to lose. Before giving people lots of autonomy, he said, comes “paying people adequately and fairly, absolutely,” or “getting the issue of money off the table.”
Whose three needs
The list of three was Pink’s. The research he leaned on most came from two psychologists at the University of Rochester. “Deci and Ryan, in my view, are the sun around which all this other research orbits,” Pink told the university’s magazine in 2010.
Edward Deci and Richard Ryan’s Self-Determination Theory also names three needs, set out in a 2000 paper: competence, autonomy and relatedness. Pink kept autonomy, and mastery is close kin to competence. Purpose is his addition. Relatedness, which they describe as the need to feel belongingness and connectedness with others, didn’t make his list, and Ryan and Deci were explicit that autonomy is not the same thing as independence.
The everyday version is familiar. The same person who drags through a mandatory course may spend a Saturday teaching herself to fix her bike. She chose it, she can feel herself getting better, and she needs the bike.
Narrower than the stage
On stage, Pink said that if-then rewards, for a lot of tasks, either don’t work or do harm, and called this “one of the most robust findings in social science.” The record is more mixed.
A 2013 attempt to replicate the candle result failed. On the overjustification effect, meta-analyses have disagreed for decades about how big the damage is. Even Deci, Richard Koestner and Ryan, on the side that finds it, reported that the undermining showed up when the task was interesting; on dull tasks, the effect they found was not statistically significant.
A 2014 meta-analysis by Christopher Cerasoli, Jessica Nicklin and Michael Ford pooled 183 samples from school, work and physical tasks, more than 212,000 people. Intrinsic motivation predicted performance whether or not incentives were present. It was the better predictor of quality; incentives were the better predictor of quantity. Their verdict: the two “are not necessarily antagonistic.” A 2015 review by Barry Gerhart and Meiyu Fang went further, concluding that support for the claim that pay for performance harms intrinsic motivation and creativity “in workplace settings” is lacking.
So pay still moves effort, especially the countable kind. The research points less to replacing it than to using both.
Friday afternoons
Say Priya runs a six-person data team and announces that Friday afternoons are now free for any project people choose. The sprint stays the same size. Picture the first Friday. Each analyst can see that a free afternoon means tickets pile up for whoever stays on them. Priya may find, as Atlassian did, that little of the time gets used, and conclude the team doesn’t want autonomy.
She could instead take the time off the sprint for everyone at once, so nobody is covering for anybody, and keep the bonus for the countable work, like turnaround time on routine reports.
Atlassian’s day-long version, at least, lasted. In 2013 its blog described ShipIt Day, the renamed FedEx Day, as a quarterly tradition: 24 hours, everyone off their day job at once, and a demo to the whole office at the end.