How many coordinated people does it take to start a revolution?
Social science actually measured this — but never at global scale. The data exists country by country, against a specific regime. Whether that number generalizes to the whole planet remains, for now, an open question.
Both successful cases land right at the 3.5% threshold Erica Chenoweth documented. The only real attempt at simultaneous worldwide protest fell short by an order of magnitude — and failed.
The question matters. The scale doesn't.
"Global revolution" mixes two things social science measures separately. What's best studied is the overthrow of one regime in one country — there the data is decades deep. No simultaneous worldwide revolution has ever been recorded, and that's not a coincidence: there has never been a single planetary-scale target to coordinate against.
The only real attempt at simultaneous worldwide mobilization — the protest against the Iraq War on February 15, 2003, across more than 600 cities — mobilized barely ~0.2% of the world's population at the time. Far below the 3.5% needed to flip a single country, and it didn't achieve its goal. "Global" and "national" aren't the same question with an extra comma.
Four questions at different scales, and one left open on purpose
How many coordinated people does it take to force a significant shift in power — and does that number make sense at a planetary scale, or is there only evidence that it works country by country, against a specific regime?
The four sub-questions also structure the rest of the chapter: each gets its own section.
What would be worth coordinating against
Toby Ord (Oxford, The Precipice, 2020) put a number on this: the probability of an existential catastrophe this century, category by category.
Post-truth doesn't appear there as its own category — Ord treats it within "unrecoverable dystopia" scenarios: it doesn't kill on its own, but it erodes the collective capacity to react to the other threats. In this chapter's terms, it's what can happen to coordination itself before it ever reaches the thresholds in the sections below.
What civil disobedience says, in numbers
| Case | % mobilized | Type | Outcome |
|---|---|---|---|
| Philippines, People Power1986 | ~3.6% (2M / 56M) | Nonviolent | Success |
| Serbia, Otpor / Bulldozer Revolution2000 | ~4% (300k in Belgrade / 7.5M, own estimate) | Nonviolent | Success |
| Bahrain2011 | Exceeded 3.5% (documented) | Nonviolent | Failure |
| Brunei— | Exceeded 3.5% (documented) | Nonviolent | Failure |
| Global Iraq War protest2003 | ~0.2% (12M / 6.3B worldwide) | Nonviolent | Failure |
Bahrain and Brunei matter as much as the Philippines and Serbia: clearing the threshold isn't enough when the regime has outside military backing. 3.5% is a necessary condition, not a sufficient one — and after 2010 regimes learned to defend themselves before protest ever gets that far.
Why 25% isn't just "a stricter 3.5%"
Damon Centola (Penn, 2018) measured something mechanically different: ten groups of twenty people, given a financial incentive to converge on a norm; an organized minority below 25% always failed to flip the norm, and crossing 25% produced an abrupt change.
The reason isn't quantitative, it's mechanistic. A social norm needs complex contagion — confirmation from several distinct contacts before it's adopted — while a political crisis can spread through simple contagion, like a rumor or a virus. Extrapolating a 25% threshold from a twenty-person lab group to eight billion people isn't math: it's a bet.
Two mechanisms, two scales, two numbers — and neither one was built to answer this chapter's question.
That number doesn't exist. And that's not a flaw: it's the answer.
To topple a specific regime: the evidence says 3.5%, with caveats.
It's the best-documented threshold there is, with decades of cases behind it. But it's a necessary condition, not a sufficient one, and its effectiveness has been falling since 2010 as regimes learn to repel mobilization before it gets that far.
To flip a social norm within a network: the evidence says 25%, in a different mechanism.
It's a lab phenomenon, measured in small groups with complex contagion. It's useful for understanding why some ideas catch on and others don't — not for predicting when a government falls.
For a threat with no regime to topple — pandemics, AI, informational collapse: there's no number yet.
No one has measured this, because it has never happened at that scale. This chapter's honesty is to leave it that way — open, unresolved — and invite whoever dares to calculate it to say so publicly. We're not asking the invited voices for a verdict, just their best estimate and what observation would change their mind.
No guests yet. Candidates for this question have already been identified and are being contacted.
Terms the chapter needs to define
Primary and verified
- E. Chenoweth, 3.5% rule — Wikipedia · Commons Library summary
- Chenoweth, on the post-2010 decline in effectiveness — Nonviolent Conflict Resource Center (CNCR)
- D. Centola, the 25% threshold for social norm change — CHIBE, University of Pennsylvania · Scientific American
- Otpor and the fall of Milošević (2000) — Wikipedia · Global Nonviolent Action Database
- Worldwide protests against the Iraq War, February 15, 2003 — Wikipedia
- T. Ord, The Precipice: Existential Risk and the Future of Humanity (2020) — Wikipedia
- Y. N. Harari, large-scale human cooperation via shared fictions (Sapiens) — Smithsonian Magazine