In 2010 the incoming coalition government set up a small unit in the Cabinet Office, seven people strong, tasked with applying behavioural science to public policy. It was widely mocked as the “nudge unit” and given two years to prove itself or be shut down. David Halpern led it, and this is his account of what the Behavioural Insights Team did with that brief — part history of the ideas, part case-file of experiments, part argument for a different way of making policy.
The core argument
Halpern’s premise, inherited from Richard Thaler and Cass Sunstein’s Nudge, is that people do not behave like the rational calculators assumed by conventional policy design. They are influenced by defaults, by friction, by what they believe other people are doing, and by timing. Governments have always shaped behaviour through these channels — every form, letter and queue is a design choice — so the real question is not whether to nudge but whether to do it deliberately and well. The second, and to Halpern the more important, claim concerns method. Behavioural insights matter less as a bag of tricks than as an excuse to bring randomised controlled trials into government. If you can test two versions of a tax letter against each other cheaply, you can find out what works instead of arguing about it, and you can iterate. He calls the resulting approach “radical incrementalism”: tackling large, stubborn problems through relentless small experiments rather than through one grand redesign.
Key ideas
- EAST. The team’s working framework — make the desired behaviour Easy, Attractive, Social and Timely — condenses a large literature into four things a civil servant can actually apply.
- Defaults do the heavy lifting. Pension auto-enrolment, which switched the default from opting in to opting out, moved millions into workplace saving and is the single largest result the book can point to.
- Social norms in the post. Telling late taxpayers that most people in their area had already paid brought forward substantial sums, at essentially zero cost — the most-cited BIT trial.
- Test everything. Halpern’s push for RCTs in government is the book’s real institutional legacy; the argument is that policy without a control group is guesswork with a press release.
- Friction is policy. Removing a form field or shortening a journey often outperforms exhortation or incentive, because most non-compliance is hassle rather than defiance.
- Ethics taken head-on. A full chapter addresses manipulation, transparency and consent, proposing that nudges should survive publication — if you would not want the public to know, do not do it.
Who it’s for
Useful for anyone working in policy, public services, product design or anywhere behaviour change is the job, and readable enough for someone who just found Nudge interesting and wants to see it applied. Halpern is an insider writing about his own unit, and that is both the appeal and the main caveat: this is a friendly account, with successes foregrounded and failures largely rendered as learning. Sceptics of nudge policy make two arguments worth holding alongside it. The first is that the effects, while real, are typically small and cheap rather than transformative, and that the field’s replication problems since 2015 have thinned out some of the psychology it leans on. The second is political: critics argue that focusing on individual behaviour lets governments avoid structural causes — that redesigning a letter is easier than confronting why the recipient is poor. Halpern is aware of both objections and answers them partially rather than fully.