Fifteen Thousand People, Five Hours

01Why the question is badly posed

The mistake is a category error. A demonstration is not trying to reduce the emissions of the fifteen thousand people who show up. That was never the point. Its mechanism is different, and it is indirect:

fifteen thousand people → a visible event → media → public conversation → political pressure → a law → millions of people

Whereas living differently works like this:

fifteen thousand people → fewer flights, less meat, fewer cars, less consumption → fewer emissions

One is an arithmetic. The other is a bet.

Comparing the two by “how much CO₂ does each hour save” is like comparing a seed to a lever. Individual action is measurable, small and certain. Collective action is unmeasurable, uncertain and potentially enormous. They are not on the same scale because they are not doing the same job.

So the question is not which one yields more. It is a different question, and it took me a while to find it.

02Do demonstrations work? Yes, sometimes, and in strange ways

Worth saying before getting sceptical: there is evidence that protests move things.

Start at the beginning of the chain, which is attention. King, Bentele and Soule studied civil rights issues in the US Congress between 1960 and 1986 and found that protest activity — together with the perceived legitimacy of an issue and competition from other issues — helped explain how many congressional hearings were devoted to each one.

It is worth reading carefully what that result does and does not say. What is measured is legislative attention, not legislation. Protest can get a problem onto the agenda of the people who decide. That is not nothing — almost nothing gets onto that agenda — but it is not a law either.

Then there is the German case, which is far more specific. Fabel and colleagues assembled close to four thousand climate protests from 2019, used mobile-phone mobility data to build a very fine-grained measure of turnout, and matched it against nine hundred and sixty electoral districts. Districts with higher Fridays for Future turnout saw larger increases in the Green vote share. And they found evidence for three mechanisms: reverse intergenerational transmission of environmental attitudes — from children to parents — more Green politicians talking about climate on social media, and more environmental coverage in local media.

The ifo Institute estimated that around 7% of the increase in the Green vote between 2019 and 2021 could be attributed to the protests: 0.45 percentage points of a 6.3-point rise. That is an estimate, not a settled measurement, and worth treating as one. But it gives a sense of the scale: real, and modest.

And now the study that dismantled my original argument, which is Italian. Marini and Nocito analysed ten million used-car transactions before and after Fridays for Future mobilisations in Italy, using rainfall on the day of the protest as an exogenous source of variation in turnout — if it rains, fewer people go, and that variation should not be related to later car-buying preferences. After the mobilisations: fewer cars bought, lower average emissions among those bought, more low-emission vehicles and fewer high-emission ones.

The demonstration ended up changing consumption decisions. The whole dichotomy collapses.

But here is a twist that forces me to correct myself, and it is more interesting than my original reading. The authors tested two possible mechanisms: an increase in environmental awareness, and rational anticipation of stricter future regulation. The second appeared to be generally more important.

In other words: people did not buy cleaner cars because the protesters inspired them. They bought cleaner cars because they read the protest as a signal that the state was coming to tighten the rules, and they did not want to be holding an asset about to lose value.

That is colder and less romantic than what I wanted it to say. It is also more solid. And it reinforces precisely the political chain — protest, expectation of regulation, behaviour — over the chain of contagion and good example. I note it here because it will come back, and because it cuts against my own thesis.

03And yet: visibility is not effectiveness

If the essay ended there it would be a nice story and also a lie by omission.

Gethin and Pons examined the quasi-universe of US protests between 2017 and 2022 — fourteen protest waves covering the environment, gender, race, guns, immigration and domestic and international politics — and measured Twitter activity, Google searches, public opinion, voting intention, election results, politicians' behaviour, and policies proposed and adopted.

The title of the paper says a good deal: The Limited Effects of U.S. Protests. They found large effects on internet activity and minimal effects on everything else.

This is worth having tattooed somewhere: something going viral does not mean anything changed.

There is worse. Vandeweerdt ran two pre-registered experiments — in Denmark, with activists blocking a road; in the UK, disrupting a sporting event. Exposure increased the likelihood that people named climate change as one of their country's most important problems: +19% in Denmark, +10% in the UK. But there were no significant effects on climate concern or on intention to act, and a small backlash did appear against the protesters' specific policy demand.

The paper's title puts it better than I can: climate disruption buys salience, and it pays for it.

Salience is not persuasion. You can get everyone talking about the problem without getting anyone to agree with your solution.

Even the counterintuitive finding has fine print. A study in Nature Sustainability examined the Just Stop Oil campaign and found that greater awareness of that radical group was associated with increased identification with and support for Friends of the Earth, a moderate group that took no part in the actions. This is the positive radical flank effect: the radical flank makes the moderate one look reasonable by contrast.

Here too I have to correct myself. It would be too categorical to say that later research “showed” the effect is temporary. What we actually have is evidence that radical flank effects can be positive or negative depending on political context, audience characteristics and how the protest is perceived. A 2025 review finds both kinds. And 2026 research suggests that actively distancing oneself from radical groups can increase support for moderates.

The honest conclusion is neither “protest works” nor “protest doesn't work.” It is that it depends on how, when, against whom, with what concrete demand, and inside what strategy. A demonstration can be enormous, moving, photogenic — and change nothing at all.

04The other half: is changing your life enough?

No. And this needs saying with the same bluntness.

A study of the potential of individual action in a Finnish suburb concludes explicitly that voluntary individual efforts alone are not sufficient to reach a carbon budget compatible with 1.5 °C: behavioural change has to be combined with strong policy and technological progress. It is a single suburb and should not be extrapolated to the planet, but as empirical evidence of the ceiling on voluntary action it does the job.

The European 1.5° Lifestyles project studied mobility, housing, food and consumption across five countries. It found that moving away from car-based mobility — especially combustion cars — carries one of the largest reduction potentials, followed by housing, particularly renewable heating. But its underlying conclusion is the one that matters here: meaningful individual change requires transformation of the political-economic, technological and social structures that make those lifestyles possible or difficult.

You cannot cycle through a city that has nowhere to cycle.

What did weaken is one of the favourite arguments against individual action: the idea that tending to your own habits works as an anaesthetic, that recycling makes you feel so good you stop demanding anything. A longitudinal study followed 2,778 people across four annual waves of an Australian survey between 2021 and 2024. Within a given year, personal climate actions were associated with greater collective participation and, to a lesser extent, greater support for climate policy.

And here I correct myself a third time, because the precise formulation matters. The defensible finding is that personal action was not found to crowd out collective action. Not that it produces it: the longitudinal models found little evidence of systematic carry-over from one year to the next.

That is less than I would have liked. But it kills the anaesthetic argument, which was the one that mattered.

05How we learn, which is what I was actually asking

Here is the heart of it, and it has nothing to do with tonnes of CO₂.

Humans learn by watching. We do not learn because someone explains something well. We learn because we see someone do something and see it go well for them. It is an ancient mechanism, older than language, and it is still the most powerful one we have.

And that is where the real difference between the two paths shows up — not a difference of efficiency but of grammar.

A demonstration says

“We have to change.”

A life says

“Look, this is what the change looks like.”

Those are two completely different speech acts. One is a demand. The other is a demonstration in the older sense of the word. And demonstrations move humans in a way that demands do not.

Then there is the slower, deeper level: social norms. Someone starts buying clothes only when they need them. A friend does the same. Their children grow up with that logic already inside them, without anyone explaining it. Another friend starts asking “do I actually need this?” And slowly what counts as normal in that group shifts. Because people do not only ask what is right. We ask, almost without noticing, what do people like me do.

And here comes the turn I did not expect: the demonstration also works by imitation.

Someone sees fifteen thousand people marching and thinks “there are a lot of people who think this way.” That changes their sense of what is normal. It is exactly the same psychological engine — social influence — running in a different format. The German study found precisely this: children shifting their parents' attitudes. Interpersonal influence, not a placard.

The difference is bandwidth and proof. The march reaches many strangers, through a screen, for one day, and reaches them as an assertion. Your life reaches your sister, your children, your three closest friends, the people you work with, for years — and it reaches them with the evidence attached.

Which produces the one thing that by now seems to me beyond argument:

If the new behaviour looks like a sacrifice, it does not spread.

There is no way around this. You do not build a mass movement by offering people a worse life. If the implicit message is “give up comfort, pleasure, status and wellbeing for the planet,” the capacity to spread is structurally limited, however right you are. You can be completely correct and completely unimitable.

Whereas if someone looks at you and thinks “this person lives well, seems calm, is healthy, is not broke, and consumes a third of what I consume,” the implicit message is different and it is devastating: I could do that too.

That is not moralising. It is the only thing that has ever moved populations.

06The uncomfortable conclusion

So the conclusion I arrived at, and don't entirely like, is this:

It is ugly. It sounds like moral policing. And yet it has a force I find hard to dismantle, because a demonstration whose participants go home to live exactly the life they are denouncing is asking society to do something none of them is doing. And people read that. Not consciously, not as an argument — but they read it.

Put as provocatively as I can manage: the most powerful activist may not be the one holding the biggest sign, but the one whose life makes someone else want to change theirs.

Now I have to do that idea the work it deserves, which is to try to break it.

07The three arguments that nearly destroy it

First: it is the hypocrisy trap. “Greta took a plane.” “You use a phone made of minerals.” Demanding personal purity as a condition for speaking is the favourite weapon of people who want nothing to change, precisely because it is an unsatisfiable condition. Nobody lives coherently inside an industrial economy. Nobody. If coherence is the entry ticket, nobody enters, and the demand disappears — an outcome that suits exactly one side.

And there is something worse. The individual-responsibility frame did not arrive on its own. Here precision matters, because the version that circulates is false: BP did not invent the carbon footprint. The ecological footprint concept dates from the 1990s and the term had earlier uses. What is documented is that BP played an important role in popularising the idea of the personal carbon footprint, through its “Beyond Petroleum” campaign and an online calculator launched in 2004.

The difference between “invented” and “popularised” looks minor and isn't: the first version is false and disqualifies you, the second is verifiable and is enough for the argument. Which remains brutal: the most effective thing the oil industry ever did was convince you the problem was your shower.

My conclusion flirts with that argument. It has to be said without anaesthetic.

Second: it is a class argument. “Living in harmony” is, to an uncomfortable degree, purchasing power and geography. Buying unpackaged, eating well, choosing non-toxic products, living somewhere you can walk or cycle, having time to cook — much of that is money and time. Turning personal coherence into a requirement for political participation makes activism the property of those who can afford it. And it excludes precisely the people who suffer most from climate change, who tend to have the smallest footprints and the narrowest room to manoeuvre. Telling someone renting a flat with insulation they cannot change, in a city with no transit, working two jobs, to stay home until they have sorted their life out is grotesque.

Third: the evidence does not support it — but less than I thought. Here I have to be honest against myself twice in a row.

In Germany the Green vote rose where turnout was higher, and the study never asked whether those protesters lived coherently. Its evidence points to other mechanisms: media coverage, politicians' communication, transmission of attitudes from children to parents.

I wanted to use that as proof that personal coherence is irrelevant. It isn't. A study not measuring a variable does not demonstrate that the variable doesn't matter. It simply did not evaluate it. This is the weakest of the three arguments, and it is better to say so than to hide it behind a clever line.

What does survive is more modest and more interesting: a protest can produce measurable political and social effects without our knowing anything about the coherence of the people protesting. An imperfect messenger does not void the message. That is not the same as saying it makes no difference who carries it.

08What survives

Understanding those three arguments does not oblige me to treat them as tied with mine. They have to be weighed.

And I think the distinction that saves the idea is this: personal coherence as an entry requirement is indefensible; personal coherence as a multiplier is the most underused thing in climate activism.

As a requirement — as a moral condition on the right to speak — the thesis is wrong and should be dropped without nostalgia. It is wrong strategically, because it silences the movement. It is wrong ethically, because it expels the most affected. And it leans on a frame the fossil industry helped install.

As a multiplier, it holds. Your life is the highest-bandwidth channel you have with the people who actually know you. If your life does not support what you are asking for, it is not that you lost the right to ask: it is that you left your best tool in the drawer.

But there is a version of the original provocation that survives intact, and it is still uncomfortable — only the accusation is a different one.

There is a form of activism that is, at bottom, identity. Going to the march as self-image, as belonging, as a moral purchase: five hours that buy you a year of feeling on the right side. That activism does not fail because it is hypocritical. It fails because it is oriented toward the participant's self-perception rather than toward any mechanism of change. It optimises for the feeling, not the effect. And it is easy to recognise: it measures nothing, it never asks what worked last time, it repeats the format because the format is the point, and it goes home without anything having moved — outside or inside.

That is what my question was actually pointing at. Not hypocrisy. Self-reference.

And I include myself, for what it's worth. I am writing this having taken more flights than I can justify and having bought things I did not need last week. I am not making the accusation from outside. There is no outside.

09The circuit

The two mechanisms are not rivals. They are a circuit, and each does what the other cannot.

Collective action is what makes sustainable living possible. It builds the bike lane, insulates the housing, changes the energy mix, takes the toxic product off the shelf. And — according to the Italian study — it does something more interesting still: it changes people's expectations about where regulation is heading, and that alone changes what they buy. Without it, living well is a sport for the disciplined and well-resourced.

Personal coherence is what makes collective action credible and contagious. It is the channel through which the demand stops being a reprimand and becomes a plausible life.

But they are not symmetrical, and saying “both matter” and stopping there would be exactly the comfortable draw this essay set out to avoid.

If you had to pick one for humanity, you pick political action, without hesitating. It is the only mechanism that operates at the scale of the problem. Collective action has a very high ceiling and a floor of zero. Individual action has a low ceiling and a guaranteed floor.

Except you are not humanity. You are one person. And for one person the question is not which mechanism is more efficient in the abstract, but which one you can sustain for years and which one actually moves somebody near you.

So the question is not whether you have earned the right to march. March. The question is more uncomfortable than that:

If someone who knows you well looked closely at your life — not your opinions, your life — would they want any of what they see?

Because if the answer is no, the placard is not going to fix it.