Drive arrived in 2009 with a claim that sounded like heresy in a recession: bonuses often backfire. Pink is a synthesiser, not a researcher — the studies belong to psychologists who had been publishing them for forty years. What he did was pull them into one sentence a manager could act on. Here is the decoding.

1. The candle problem is the whole book in one experiment. Participants get a candle, a box of tacks and matches, and are asked to fix the candle to the wall so wax doesn't drip on the table. The solution requires seeing the tack box as a shelf, not just a container. In 1962 the Princeton psychologist Sam Glucksberg ran it two ways: one group was offered cash for speed, the other was told they were simply setting a baseline time. The paid group averaged about three and a half minutes slower. Incentives narrowed their focus at exactly the moment the task required peripheral vision.

2. Rewards work beautifully — on the wrong problems. Glucksberg ran a second version with the tacks tipped out of the box, so the shelf was obvious. Now the task was mechanical, and the cash group won easily. This is the distinction the book turns on. For narrow, rule-based work with a visible finish line, if-then rewards do what they promise. For work where the path isn't given, they reliably hurt. Most of us do both kinds in a week and pay ourselves the same way for each.

3. The evidence is a meta-analysis, not a hunch. Pink leans hardest on Edward Deci, Richard Koestner and Richard Ryan's 1999 review in Psychological Bulletin, which pooled 128 experiments and found that tangible rewards tied to performance reliably undermined intrinsic motivation. Verbal recognition went the other way and strengthened it. A 2009 London School of Economics review of 51 experiments reached a similar verdict; its author, Bernd Irlenbusch, put it plainly: incentives can reduce intrinsic motivation and weaken the social norms people were already following.

4. Pay enough, then get money off the table. This is the practical heart of the book and the part most often misread. Pink is not against salaries. His rule is that compensation must be sufficient and fair enough that it stops being a topic — pay people well, benchmark honestly, then stop using cash as the steering wheel. Underpay, and money is all anyone thinks about. Overuse bonuses, and money is still all anyone thinks about.

5. Autonomy, mastery, purpose — and one honest caveat. What replaces the carrot: control over task, time, technique and team; visible progress at something difficult; and a reason larger than the invoice. Pink's illustrations are the 24-hour engineering sprints Atlassian ran as "FedEx Days," and Google's old 20% time. Worth knowing: the underlying research, self-determination theory, names its three needs autonomy, competence and relatedness. Pink swapped in mastery and purpose. It reads better and it drops relatedness — the need to belong — which is arguably the one most managers get wrong.

6. The real-world versions are fragile. Best Buy's Results-Only Work Environment, where output replaced hours entirely, is one of Pink's showcase cases; new leadership scrapped it in 2013. Google's 20% time has been steadily narrowed. Autonomy is cheap to announce and expensive to defend, because it is the first thing surrendered when a quarter goes badly. Treat the case studies as proof the idea works, not as proof it survives.

Takeaway

Take today's task list and mark each item A (algorithmic — the steps are known) or H (heuristic — you have to figure out the path). Give the A items a deadline and a reward: finish them, then take the break. Give the H items the opposite treatment — a protected block, no timer, no bonus attached. Most people are managing themselves in exactly the wrong direction on the H list.

The weakness of Drive is the weakness of every book built from a TED talk: it is more confident than the literature is. Motivation research is contested, and "rewards destroy creativity" is a stronger sentence than the data supports. What holds up is quieter and more useful. Before you attach an incentive to a piece of work, ask whether the person already knows how to do it. If they do, pay for speed. If they don't, pay for time.