In 1947, a copywriter at N.W. Ayer named Frances Gerety wrote “A Diamond Is Forever” for De Beers. Somebody had a client, a brief, and a product to sell. The line became part of the language of marriage. De Beers' history
Culture gets built at the point where an idea acquires consequences. A purchase becomes evidence of devotion. Reporting a mistake earns a promotion or ruins one. An institution decides which of those outcomes to reward.
To change a culture, work on what its practices cost, what they earn, and what people expect from one another. That is the approach I argue for here. A persuasive message has more chance of lasting when the arrangements around it make the behaviour worth repeating.
Deliberate influence leaves evidence: a client, a brief, a budget, an intended change. A convenient beneficiary alone establishes none of those. Shared incentives also produce coordinated-looking behaviour without a coordinator.
I call a deliberate attempt to change shared practices or expectations through these conditions cultural engineering. The attempt, its effect, and its eventual survival are three separate things to explain.
What can be changed
Bernays called it the engineering of consent. Lippmann wrote about public opinion; Gramsci about hegemony; Veblen and Bourdieu about status. Cialdini studied influence, while Hobsbawm and Ranger examined invented traditions. The five mechanisms below draw on that work.
Narrative
A story gives an action a meaning. Buying the ring becomes evidence of devotion. Joining the march becomes an act of courage. The action acquires consequences beyond its immediate result: it says something about the person doing it.
The campaign's job is to supply that interpretation where decisions get made. It has to fit the audience's existing commitments and survive their experience. A promise of independence loses credibility when participation starts to feel compulsory.
Information architecture
Someone decides what appears in the feed, which choices come first, what needs an application, and what happens by default. These decisions can change behaviour without changing anyone's mind.
Defaults reduce the effort of one choice and leave the effort of the others in place. Publication rules control access to an audience. Those are interventions in their own right. Measure the resulting choices separately from endorsement of the programme or belief in the claim.
Social proof
Other people's choices are information. A crowded restaurant may be worth trying; a respected colleague may know something you don't. Cialdini's work helps explain why signals of popularity and authority influence decisions.
Those signals can be earned, exaggerated, or fabricated. An engineer can recruit an admired early adopter, display genuine uptake more prominently, or manufacture the appearance of a majority. The distinction matters if you're trying to establish deception. Visible agreement by itself doesn't tell you which happened.
Status
What earns respect gives people a reason to change. An organisation can praise the person who catches a mistake, or the person who never reports one. Give out enough promotions and the difference becomes part of how work gets done.
Status can be awarded for skill, wealth, piety, loyalty, lineage, or service. Changing its currency changes the competition. A campaign that rewards recruitment may spread very well while producing members who are good at little else.
Identity
When a practice becomes part of belonging, leaving it behind risks friends, recognition, or a familiar account of yourself. A reform that ignores those losses asks people to pay a price its author hasn't counted.
It can also split private judgment from public conduct. Someone may stop believing and keep attending. As in the memetics essay, count belief and behaviour separately or you'll mistake attendance for conviction.
The mechanisms reinforce one another. The same award confers status, signals expert approval, and recruits its recipient into an identity. Design rewards with all three effects in mind.
Three cases, three different claims
Geneva: enforce conduct, get consequences you didn't plan
The Protestant work ethic is often presented as designed culture. Weber's argument is more awkward and more interesting: religious commitments could produce economic consequences their authors hadn't intended. Calvin wasn't running a productivity campaign.
Geneva did deliberately organise the policing of conduct through its consistory. Surveillance and sanctions were arrangements people built. A later change in economic behaviour would require another argument about what those arrangements caused, alongside doctrine and the rest of the setting.
The lesson is that control over an intervention is not control over its consequences. Geneva's authorities built machinery for enforcing conduct; Weber's account concerns effects that exceeded the builders' intentions.
The Royal Society: organise grounds for trust
How do you assess an experiment you didn't witness? Shapin and Schaffer's Leviathan and the Air-Pump describes the work involved in making experimental reports credible: witnesses, detailed accounts, and conventions about whose testimony counted.
Class supplied some of that credibility. So did practices that let other investigators inspect and dispute claims. The Royal Society's publishing history shows those arrangements changing over time; the journal's beginnings and later systems of refereeing weren't one finished invention. Royal Society publishing history
The lasting achievement is a way for people to inspect and dispute claims they didn't witness. Credibility earns its keep when it opens a claim to checking. Authority that blocks the check defeats that purpose.
De Beers: give an existing object a stronger obligation
De Beers didn't invent engagement rings. Its advertising promoted a particular answer to what a proposal ought to include and what the purchase should mean. The Smithsonian holds examples of the campaign; De Beers records Gerety's authorship of the slogan. Here, deliberate intervention is straightforward to establish. Smithsonian advertising collection
The harder question is how much of the resulting custom the campaign caused. Advertising ran alongside changes in income, retailing, entertainment, and marriage itself. The ads establish intention and content. Estimating their contribution needs evidence beyond the fact that the custom later became familiar.
Persistence still takes work. De Beers announced a renewed investment in the same campaign in 2023. A durable norm and continued commercial promotion can sustain each other. De Beers campaign announcement
The ambition was to make a purchase feel like something love requires. That is a cultural intervention with a commercial beneficiary and a specific behavioural target.
How to establish what happened
Use “who benefits?” to choose where to investigate. Establish authorship through evidence of the attempt, then assess its effect.
Establish the attempt. Look for a brief, a budget, instructions, contracts, placement records, or testimony tying an actor to a desired change. Similar slogans and convenient beneficiaries are leads. They can also result from imitation, shared interests, or a common event. A failed campaign is still a campaign.
Estimate the effect. Name the behaviour or expectation that changed, and when. Compare places, groups, or periods with different exposure where possible. Ask what was already changing and what else happened at the same time. A rise in sales after an ad ran doesn't isolate the ad's contribution. Keep the finding precise: evidence of an attempt establishes intention; establishing its contribution requires a causal comparison.
Check persistence. If funding or enforcement ends, does the practice continue? Through habit, practical benefit, peer pressure, or another sponsor? Be careful about the moment you call it self-sustaining. A budget may have disappeared from one organisation's accounts while the work moved to schools, families, or employers.
And try the competing explanation seriously. If several outlets independently chase the same profitable audience, their coverage may converge without anyone issuing orders. That is still a system worth studying. Calling it coordinated would make a different claim, with a different burden of proof.
Why attempts fail
Punishing or obstructing the behaviour you ask for works against your own intervention. Agreement won't supply missing time, money, or authority. Nor will it erase the advantages people get from an existing practice.
James C. Scott's Seeing Like a State is useful here: schemes can destroy the local knowledge and arrangements on which their intended outcome depends. More enthusiastic messaging won't repair that mistake.
A smaller hypothetical makes the point. Suppose a company wants employees to report errors. It introduces a reporting tool and celebrates candour. Managers continue punishing teams whose numbers look bad. Reports stay scarce. The tool may be easy to use, the message widely understood, and the behaviour still costly.
Change the promotion rule and you have a more serious intervention. Then check whether errors are being found and fixed. A rise in reports could mean improved candour, worsening work, or employees learning to satisfy the new metric. Even a successful intervention needs an account of what its apparent success measures.
Success for whom?
The Royal Society and De Beers belong in the same essay because both worked on the conditions of trust and choice. That doesn't make their achievements equivalent.
I judge an institution for producing knowledge by whether its claims survive independent checking and its errors get corrected. I judge a company practice by what it does to the work and to the people doing it. A marketing campaign's sales are evidence of benefit to the seller; they don't settle the buyer's side of the bargain.
So before building anything, name the outcome you want, who should benefit, and what cost would make the project a failure despite widespread adoption. Keep a way for participants to report that it isn't working. The temptation will be to call their resistance proof that they need more persuasion.
Give those people power to change the arrangement. If the only response available to them is agreement, you've built a poor system for finding out whether your intervention works.
Sources and further reading
- Edward Bernays, Propaganda (1928) and “The Engineering of Consent” (1947); Walter Lippmann, Public Opinion (1922).
- Antonio Gramsci, Prison Notebooks; Edward Herman and Noam Chomsky, Manufacturing Consent (1988). Accounts of power and institutional incentives, not evidence of a central designer behind every shared message.
- Thorstein Veblen, The Theory of the Leisure Class (1899); Pierre Bourdieu, Distinction (1979); Robert Cialdini, Influence (1984).
- Eric Hobsbawm and Terence Ranger, eds., The Invention of Tradition (1983).
- Max Weber, The Protestant Ethic and the Spirit of Capitalism (1905); Robert M. Kingdon, Adultery and Divorce in Calvin's Geneva (1995), on the consistory's intervention in private life.
- Steven Shapin and Simon Schaffer, Leviathan and the Air-Pump (1985); Steven Shapin, A Social History of Truth (1994).
- James C. Scott, Seeing Like a State (1998).
Cultural Building Blocks asks which shared expectations these interventions might change. The Laws of Memetics asks how ideas spread and persist once they enter circulation.