← Joona Heino

The Laws of Memetics

v0.1 · why ideas spread, stick, and resist correction.

This is my attempt to write down the rules that govern how ideas spread. The parts come from half a dozen fields that don't usually talk to each other; the compression into one page is the contribution. I've tried to state each rule as plainly as I can without flattening it.

A word on the field first. Richard Dawkins coined "meme" in 1976 as the cultural twin of the gene: a unit of culture that gets copied from one mind to another. The formal science of memetics that grew up around the word mostly stalled out by the mid-2000s, for fair reasons (a meme is hard to define and harder to measure). But the core intuition kept paying rent, so I'm keeping it: ideas behave like replicators under selection. Take that seriously and a lot of confusing things snap into focus.

Everything below hangs on one claim: memes are selected for how well they spread, not for whether they're true. Truth helps sometimes. It just isn't the thing being optimised.

What a meme is competing for

Dawkins made a second point that does most of the work here. Any successful replicator wins on three properties: fecundity (it gets copied a lot), fidelity (it gets copied accurately), and longevity (it lasts). Look at what's absent from that list. Correspondence to reality isn't a property of a good replicator. A meme that is vivid, sticky, and contagious beats a meme that is merely correct, every time the two go head to head.

The laws below are what fecundity, fidelity, and longevity look like once the replicator is an idea and the environment is human attention. They're closer to reliable regularities than to laws of physics, which is the honest reason this is v0.1.

The substrate

Law Zero: Scarcity

Attention is finite, and memes compete for it zero-sum. For every meme that wins, another dies unread.

A meme can't be copied if nobody is looking at it, and there is only so much looking to go around. Herbert Simon called this in 1971: a wealth of information creates a poverty of attention. Everything downstream is a fight over that one scarce resource. So the question is never "is this idea good?" It's "is this idea winning attention against its rivals right now?" Different question, different answers.

The fundamental laws

Three things win the fight: a meme spreads if the right people want it, if it carries enough charge to move, and if it can lodge somewhere it won't be pried loose.

First Law: Triangulated Desire

We want what the people we admire or envy want. The model, not the object, drives transmission.

This is René Girard's mimetic desire, dressed for the timeline. We rarely adopt a belief because we've checked it. We adopt it because someone we want to resemble holds it. The cultural-evolution researchers arrived at the same place from the other direction and named it prestige bias: humans copy the high-status and the successful by default, because over evolutionary time that was a cheap shortcut to learning what works (Henrich and Gil-White). The shortcut still fires even when status and correctness have nothing to do with each other.

The consequence is brutal and everywhere. The identical claim gets opposite receptions depending on whose mouth it comes from. If you've watched a sentence go from unsayable to obvious the instant the right person said it, you've watched this law run. It was never about the claim.

Second Law: Arousal as Energy

High-arousal emotion is the fuel of transmission. Anger, awe, fear, and disgust speed it up; calm, sadness, and contentment slow it down.

Jonah Berger and Katherine Milkman put numbers on this in 2012. Working through the New York Times most-emailed list, they found that content stirring high-arousal emotions, awe and anger above all, got shared far more, while sadness, a low-arousal state, made sharing less likely. The MIT team that mapped every major rumour cascade on Twitter (Vosoughi, Roy, and Aral, 2018) found falsehoods spreading faster, deeper, and wider than the truth, propelled by their novelty and the fear, disgust, and surprise they triggered.

So outrage travels and nuance sits still. Furious, you forward; thoughtful, you close the tab. The engine doesn't care what it's hauling.

Third Law: Integration Lock

Once a meme becomes part of your identity or your model of the world, evidence stops reaching it. Attack the meme and you attack the person.

A belief locks in two ways. It can become identity ("I'm the kind of person who believes X"), so that an argument against X lands as an argument against you. Or it can go load-bearing in your picture of the world: you've stacked other beliefs on top of it, and pulling it out leaves a hole you can't fill, so you leave it in. Leon Festinger watched a flying-saucer cult survive the non-arrival of its own apocalypse and wrote it up in When Prophecy Fails (1956). Dan Kahan's work on identity-protective cognition shows the ordinary daily version, where people reason hardest in defence of the beliefs that mark their tribe.

One careful caveat, because I'd rather be right than tidy: the popular notion that correcting people backfires and entrenches them has mostly failed to replicate. The robust finding isn't a backfire. It's that motivated reasoning makes an integrated belief very expensive to shift, which is bad enough.

The derived laws

Grant the first three and two more follow with no extra assumptions.

Fourth Law: Correction Asymmetry

A correction inherits every disadvantage the original escaped, then takes on fresh resistance from everyone who already bought in.

Walk it through. The original claim was novel, emotionally charged, and made the sharer look smart or righteous (Laws One and Two). The correction is none of those. You already know the claim, so it bores you. Nuance has no arousal, so it has no fuel. Sharing it costs status, because it reads as either "I got fooled" or "I'm the pedant in your replies." And it slams straight into Law Three by threatening beliefs people have already welded to who they are. This is why a lie can travel halfway around the world while the truth is still lacing its shoes, and why Brandolini's law holds: the energy to refute nonsense is an order of magnitude greater than the energy to produce it. The game is rigged before the correction is even written.

Fifth Law: Refraction

Memes mutate as they cross from one group to another. The further apart the groups, the larger the mutation.

Fidelity is the replicator property ideas are worst at. Frederic Bartlett demonstrated it in 1932: send a story down a chain of retellings and it shortens, smooths, and bends toward whatever the tellers already expect. Dan Sperber turned this into an epidemiology of representations, where ideas drift toward "cultural attractors," the shapes a given culture finds easiest to hold. So a meme simplifies as it travels, its ambiguities resolve in favour of the receiving group's priors, and it collects tribal markers until "our" version and "their" version barely rhyme. Holding an idea steady across a boundary takes real machinery: scripture, canon, liturgy, professional training. Strip the machinery away and drift is the default state.

The regime laws

The same meme plays by different rules depending on how it's transmitted and whether anyone is enforcing it.

Sixth Law: The Utility Exception

Ideas spread sideways, peer to peer, are selected for desirability. Ideas handed down the generations are selected for usefulness. Desire wins the sprint; utility wins the marathon.

This is the one place truth gets a vote. Cultural-evolution theory separates horizontal transmission (between peers) from vertical transmission (down generations), and the two channels filter for different things (Cavalli-Sforza and Feldman; Boyd and Richerson). A conspiracy theory beats hand-washing advice in every sideways sprint. Yet hand-washing persists for centuries, because a practice that gets your children killed doesn't get handed to your grandchildren; the vertical channel quietly selects for what actually works. Here's the worry. Every new communication technology pumps up the horizontal channel and starves the vertical one. We have tuned the whole system for virality and away from the slow filter that used to keep us alive.

Seventh Law: Institutional Capture

Once a meme is written into law, curriculum, or policy, it no longer has to win on its merits. It reproduces through compliance.

This is longevity bought with force instead of fitness. An idea that would lose in open competition can survive indefinitely if an institution requires it, which produces what John Quiggin calls zombie ideas: the academic theory nobody believes but everybody cites, the regulation nobody defends but nobody repeals, the corporate ritual that outlives every reorg. They aren't alive in the memetic sense, winning hosts by appeal. They're undead, animated by enforcement. Which gives a clean prediction: weaken the enforcing institution and its zombies have to compete again, and they tend to die quickly. Watch what happened to Soviet doctrine, or to any practice that only survived because a now-toothless regulator demanded it.

The law doesn't exempt the people who study ideas for a living, either. "Misinformation" as a category exploded after 2016 not because the concept got sharper but because platforms, funders, and fact-checking bodies started enforcing it. Some of that research is real (the inoculation results below hold up). But a field whose ideas reproduce through policy mandate instead of open competition should expect Law Seven to apply to itself. It applies to everything.

The mnemonic

  1. Attention is scarce
  2. We want what our models want
  3. Arousal moves ideas
  4. Integration locks them in
  5. Corrections fail by construction
  6. Meaning drifts across every boundary
  7. Desire wins the sprint, utility wins the marathon
  8. Enforcement outlives appeal, then calcifies

What it explains

What it predicts

  1. Pre-bunking beats debunking. Getting there first with the good version is far cheaper than dislodging the bad one after it has integrated (Law Four). This is now well supported in the inoculation-theory literature.

  2. Model beats message. To guess what someone will believe, knowing who told them outperforms knowing what they were told (Law One).

  3. A correction works only if it does everything the original did. Trusted source, a replacement story that fills the hole, no threat to identity, and true. Miss any one and it bounces (Laws One to Three).

  4. Expertise is local. It protects you inside your field and nowhere else. A Nobel laureate is as exposed to nutrition nonsense as your uncle is, because outside the lab she's back to copying prestige (Law One).

  5. When institutions break, ideas move fast. Remove the enforcement and the zombies face selection again. That's when paradigms flip (Law Seven).

Where this comes from

As I said up top, the contribution here is the compression, not the discovery. The parts:

Spotted a mistake, or a better source? Send me a message on X.