# The Laws of Memetics A claim can be everywhere and believed by almost nobody. People pass it around to laugh at it, condemn it, or ask whether it could possibly be true. Meanwhile, an unremarkable instruction gets followed every morning without anyone posting about it. Both have spread. They have accomplished different things. **Ideas succeed when the environment rewards people for carrying them. Truth gains an advantage where people can detect errors and have reason to correct them.** That is the argument of this essay. Dawkins supplied the starting point: cultural variants compete to be copied and retained. Entertainment, prestige, habit, and force all help explain what survives. The practical question is how to arrange that competition so getting things right pays. Here, a *meme* is a claim, story, or practice that people transmit. These laws describe tendencies under specified conditions. One complication runs through all of them: a sentence can survive word for word while its meaning changes completely. ## First, count the right thing Attention, sharing, belief, and behaviour need separate accounts. Exposure tells you someone encountered an idea. Sharing tells you they passed it on. Belief tells you they accepted some claim about the world. Behaviour tells you what they did. None guarantees the next. When I call a meme successful below, I mean successful at a specified one of these outcomes, over a specified period. That says nothing yet about whether it is true or good for the people carrying it. A recruitment slogan can work beautifully for its author and terribly for its recruits. ## Law Zero: Scarcity *Attention limits what gets a hearing.* Herbert Simon's point still holds: abundant information makes attention scarce. At any given moment, reading one thing costs time that could have gone to another. But attention isn't a fixed tournament in which every winner kills a loser. An explanation can make its subject interesting enough that a reader seeks out ten more. Distribution matters here. A claim placed in front of a million people has an advantage over one placed in front of a hundred. Before explaining its success through clever wording or deep psychology, check who got to see it. ## First Law: Borrowed Prestige *Who carries an idea affects who will consider it.* Girard's mimetic desire starts with the person we want to resemble. Cultural-evolution research on prestige bias asks why people learn from respected models. These are related ideas, but admiration, imitation, and belief aren't interchangeable. Copying someone skilled can save years of trial and error. The mistake is letting their authority travel further than their skill. An excellent surgeon's opinion on an operation and her opinion on monetary policy deserve different weights. Celebrity makes that distinction inconvenient. The practical mistake is treating prestige as a transferable credential. Ask what the source has demonstrated in the domain of the claim. Familiarity and fame are cheap substitutes for that check. ## Second Law: Arousal *High-arousal emotion increases the urge to pass something on.* Berger and Milkman found that New York Times articles evoking awe, anger, or anxiety were more likely to make its most-emailed list, while sadness was associated with less sharing. Their experiments also supported a causal role for arousal. Practical usefulness also predicted sharing. Outrage has competition: people forward things that help them. [Berger and Milkman, 2012](https://faculty.wharton.upenn.edu/wp-content/uploads/2011/11/Virality.pdf) Vosoughi, Roy, and Aral found false news travelling farther and faster than true news in the Twitter rumour cascades they studied. Their result identifies an advantage for false news in that setting's distribution contest; it measures circulation rather than acceptance. [Vosoughi, Roy, and Aral, 2018](https://doi.org/10.1126/science.aap9559) An angry share counts as transmission. It takes another measurement to find out whether anyone was persuaded. ## Third Law: Integration *Revising a belief becomes harder when other commitments depend on it.* Sometimes the dependency is personal: abandoning a belief would mean losing standing with friends, family, or a political group. Sometimes it is explanatory: the belief holds several other beliefs together, and removing it leaves questions unanswered. Festinger's account of a failed prophecy and Kahan's work on identity-protective cognition describe ways people defend commitments under pressure. Correction then asks more of someone than accepting a fact: it asks them to reconsider what else they have built around it. Changing your mind and announcing the change have different costs. Someone may privately accept a correction while continuing to repeat the old line in public. If you only count public declarations, you'll miss the change. ## Fourth Law: Correction Has Two Jobs *A correction has to reach people, then change what they believe.* The original may have had a head start, a larger audience, or a more appealing story. A correction can lose that distribution contest and still work on the people it reaches. Conversely, a spectacular takedown can attract an enormous audience without persuading the people who believed the claim. Porter and Wood tested fact-checks in Argentina, Nigeria, South Africa, and the UK. They found improvements in factual accuracy, with most effects still detectable more than two weeks later. Fact-checking improved beliefs among recipients. The remaining jobs are getting the correction to its audience and establishing whether the change in belief changes behaviour. [Porter and Wood, 2021](https://doi.org/10.1073/pnas.2104235118) Corrections aren't doomed to be boring or humiliating, either. Exposing a fraud can be riveting. Admitting an error can earn trust. The local audience decides what carries a status cost. So measure both jobs: what fraction of the relevant audience saw the correction, and how their beliefs changed after seeing it. Calling the whole effort a failure because the original got more shares throws away half the evidence. ## Fifth Law: Reconstruction *People transmit their interpretation of an idea.* Bartlett's retelling experiments showed stories being reshaped in recall. Sperber's account of cultural transmission puts reconstruction at the centre: people infer what a representation means using what they already know. Transmission includes that act of reconstruction. A phrase can be copied exactly into two groups and mean opposite things. It can also cross a boundary intact when people share definitions, training, or a way to check the result. Written records preserve wording; they don't settle interpretation. Expect more room for drift when meaning depends on unstated context. To test that, compare what senders intended with what recipients understood. Counting identical strings won't do it. ## Sixth Law: Feedback *Useful ideas gain an advantage when their results can be checked.* Imagine two repair methods. If mechanics can try both, see which works, and avoid the one that keeps bringing customers back angry, usefulness has a route into selection. Now imagine a political prediction whose deadline can be moved whenever it fails. The route is much weaker. The rule I propose is simple: feedback favours useful practices when results are visible, comparisons are fair, and the people choosing a practice have reason to care about the outcome. Noisy results obscure the comparison. Costs imposed on others let the chooser keep winning while the practice fails its users. This distinction cuts across generations. Colleagues can teach each other something useful in an afternoon. Families can preserve a mistaken belief for centuries. Horizontal and vertical transmission describe who learns from whom; they don't, by themselves, tell us whether the lesson works. It also matters what *works* means. A technique for winning an argument can be useful to the speaker while making everyone else's beliefs worse. ## Seventh Law: Institutional Reproduction *Training, incentives, and enforcement give practices a means of reproducing beyond personal endorsement.* A reporting rule can outlive the enthusiasm that introduced it. People learn it because their job requires it, and teach it to whoever arrives next. Private agreement is optional. That is one way cultural selection operates. The institution supplies rewards and penalties. What matters is what those rewards and penalties track. A laboratory can make checking results a condition of professional standing. A bureaucracy can make concealing bad results a condition of promotion. Both reproduce practices; only one is arranged to expose its own errors. Judge an institution by its treatment of error. Can criticism reach the decision? Do failed practices get revised? Who bears the cost of keeping them? Apply those questions to fact-checkers and universities as readily as to ministries and corporations. When enforcement ends, I expect compliance to fall where sanctions were the main reason for it. Usefulness, habit, and mutual expectations supply other reasons to continue. Tracking what survives reveals which support the institution had been providing. ## Three tests The account makes claims about distribution, feedback, and compliance. Here is how I would test them: - Give comparable audiences the same correction through different distribution routes. Measure exposure separately from belief change among recipients. - Let groups choose between practices with clear performance differences. Vary how quickly and clearly they see the results. Does better feedback increase adoption of the better practice? - When a requirement ends, track public compliance and private endorsement separately. A drop in the first with little change in the second would suggest the institution had been sustaining behaviour without securing belief. If the aim is better beliefs, make errors visible and correction worth the trouble. Give corrections distribution, reward people for revising a bad claim, and preserve records that let others check the result. Make accuracy worth defending and error expensive to conceal. ## Where this comes from - Richard Dawkins, *The Selfish Gene* (1976): cultural replication and the fecundity, fidelity, and longevity of replicators. - Herbert Simon, “Designing Organizations for an Information-Rich World” (1971): scarce attention. - René Girard, *Deceit, Desire, and the Novel* (1961): mimetic desire. - Joseph Henrich and Francisco Gil-White, “The Evolution of Prestige” (2001): prestige-biased learning. - Jonah Berger and Katherine Milkman, “What Makes Online Content Viral?” (2012): emotion, usefulness, and sharing. - Soroush Vosoughi, Deb Roy, and Sinan Aral, “The Spread of True and False News Online” (2018): Twitter diffusion. - Leon Festinger, Henry Riecken, and Stanley Schachter, *When Prophecy Fails* (1956); Dan Kahan, “Ideology, Motivated Reasoning, and Cognitive Reflection” (2013): defence of commitments. - Ethan Porter and Thomas Wood, “The Global Effectiveness of Fact-Checking” (2021): experimental evidence on corrections. - Frederic Bartlett, *Remembering* (1932); Dan Sperber, *Explaining Culture* (1996): recall and reconstruction. - Luigi Cavalli-Sforza and Marcus Feldman, *Cultural Transmission and Evolution* (1981); Robert Boyd and Peter Richerson, *Culture and the Evolutionary Process* (1985): transmission and selection. The feedback rule above is my synthesis. The companion essays ask [what varies between cultures](/cultural-building-blocks) and [how people deliberately try to change it](/cultural-engineering).