What movements pursue is usually a public good: political opening, clean air, lower taxes. Once won, those who contributed nothing enjoy it anyway, and cannot be excluded. From this, Olson's 1965 The Logic of Collective Action derives a counterintuitive conclusion — shared interests do not automatically produce shared action. My participation moves the outcome by roughly nothing, while the costs (time, risk) fall entirely on me. Free riding is the dominant strategy.
Bigger groups make it worse: per-head benefits are diluted, absence is harder to notice, monitoring is harder. So a small concentrated interest routinely outweighs a vastly larger diffuse public (see Day 16). This is textbook incentive incompatibility; the only fix is to change the payoff function, i.e. selective incentives: benefits only participants get, or penalties only non-participants bear.
| Mobilization base | Where selective incentives come from | Cost / limit |
|---|---|---|
| Membership organizations union-type | Members-only mutual aid funds, legal assistance, bargained gains | The body drifts into being a service provider and its agenda detaches from its original political aims; compulsory membership is restricted in many jurisdictions |
| Pre-existing social networks religious, hometown, occupational bodies | The network already exists and supplies standing, belonging and mutually visible social pressure at near-zero marginal organizing cost | The agenda is shaped by the host organization's preferences, not chosen freely by participants |
| Diffuse-benefit causes climate, anti-corruption, consumer rights | Almost no selective incentives are available — only moral appeals and professionalized advocacy organizations | Survival depends on donors rather than dues, producing the "movement without members": strong voice, weak mobilization |
The Olson side: observed mass participation is not a refutation — most of it can be explained by selective incentives, very low-cost symbolic participation, or misjudged odds of success. Explaining participation as "higher consciousness" defines the cause by the outcome.
The threshold-model side (models in the vein of Granovetter 1978): real decisions are conditional — "if N people are already in the street, I'll go." A distribution of thresholds plus mutual observation makes participation jump discontinuously rather than grow linearly.
"Strength in numbers." For a public good, more people raises both potential strength and the incentive to free ride; the net effect is indeterminate. What is scarce is never shared grievance, but organization that converts grievance into a cost individuals can bear.
The early explanation was "grievance causes protest." But grievance is close to a constant while protest is rare — a constant cannot explain a variable. Three lines of work followed:
| Setting | What supplies the rise | Cost / limit |
|---|---|---|
| Federal democracy the US civil rights movement | Existing networks supplied organization, federal-state divergence supplied leverage, and the frame appealed to constitutional commitments already on the books | Gains depend on one level of government being willing to intervene; when that level shifts, the same organizational capacity need not buy the same result |
| Democracies with differing institutional designs Kitschelt's 1986 four-country comparison of anti-nuclear movements | "How easy is it to get into decision-making" and "how capable is the state of implementing policy" are two independent dimensions | Closed input structures push movements toward confrontational street forms, open ones toward negotiation inside the system; but getting in is not the same as changing anything |
| High-repression settings | Opportunity comes from splits at the top rather than institutional channels; movements often start from "non-political" issues such as environment or housing | Issues stay confined to a narrow band, and both scaling up and linking across issues are suppressed (see Day 31) |
The structural side: attributing success or failure to a leader's charisma is mostly retrospective storytelling; the same people with the same demands get entirely different results in different opportunity windows.
The agency side: an opportunity only operates once it is perceived, and it is partly created by the movement itself (demonstration effects, cross-sector coalitions, forcing opponents to take a position). "Opportunity structure" has also been criticized as concept stretching — if any favorable condition counts as an opportunity after the fact, the concept loses falsifiability.
That dispersal equals failure. Effects are often lagged and scattered: the issue enters the routine agenda, professional organizations persist, participants form durable networks, the vocabulary is partly absorbed by the mainstream. Conversely, enormous volume at the peak can coincide with zero policy output. Look at institutional residue, not crowd size.
Skocpol's 1979 States and Social Revolutions executed a shift of perspective: revolutions are not primarily "made" — the state fails first. Comparing the French, Russian and Chinese social revolutions, she identifies shared structural conditions: the old regime falls into fiscal and military crisis under international competitive pressure; elites split over the response so the state cannot extract resources from them; and agrarian social structure gives those below the capacity to act on their own.
Goldstone's 1991 Revolution and Rebellion in the Early Modern World offers a quantifiable version: population growth raises three pressure lines at once — fiscal deficit, elite overproduction (far more aspirants to elite position than positions), and deteriorating mass livelihoods.
Kuran 1991 adds the last layer: because of preference falsification, what people say publicly diverges systematically from what they hold privately, so outsiders — including the regime itself — cannot know the true scale of opposition. Once a trigger event makes a few speak openly, others' thresholds fall one by one and a cascade forms. That is why the 1989 changes in Eastern Europe surprised nearly every observer.
| Layer of explanation | Variables it watches | Cost / limit |
|---|---|---|
| Long-cycle structure Skocpol | International competitive pressure, fiscal-military crisis, elite splits, autonomous organizing capacity below | Strong retrospectively, weak prospectively; criticized for underweighting ideas and individual choice |
| Mid-cycle pressure Goldstone | Demographic structure, fiscal gap, elite position supply-demand ratio, real incomes | Measurable, but the vast majority of high-pressure societies have no revolution — read it as risk, not prophecy |
| Short-cycle trigger Kuran | Number of people who go public, conditions for informational cascades | Explains unpredictability itself and is therefore useless for prediction: the spark is always findable afterward, never before |
The predictable side: structural conditions are slow variables. Fiscal position, intra-elite competition and the loyalty base of the armed forces are all observable, so one can at least state a probability band rather than a pure black swan.
The unpredictable side: the final step from slow variable to event runs through an informational cascade, and cascades are extremely sensitive to initial conditions — under identical structures, one protest disperses and another avalanches, possibly on the on-the-spot choices of a handful of people.
Methodologically these are a division of labor, not a disagreement: structure explains risk, cascades explain timing. Conflating them turns "risk is high" into "it is about to happen."
That revolutions happen at the worst moment. Tocqueville's observation in The Old Regime and the Revolution is the opposite: the eve of the French Revolution was a period of improving conditions and advancing reform. Relative deprivation theory points the same way — when expectations rise faster than reality improves, the gap is widest. The high-risk zone tends to sit mid-reform, not at the trough.
Chenoweth and Stephan's 2011 Why Civil Resistance Works coded 323 mass campaigns between 1900 and 2006 — those aiming at regime change, expulsion of foreign occupation, or secession — into the NAVCO dataset. Nonviolent campaigns achieved their objective about 53% of the time; armed campaigns about 26%.
The proposed mechanism is not moral superiority but the participation barrier: strikes, boycotts, sit-ins and resignations are open to the old, the young and the physically limited, while armed struggle screens out most of the population. Large and demographically varied participation raises the cost of repression (the people being repressed include the enforcers' own relatives), which makes loyalty shifts among security forces and economic elites the strongest predictor of success. The authors also find that even campaigns that fail in the short run are followed by markedly higher odds of a democratic system within five years.
The widely circulated "3.5% rule" is one descriptive finding within this: in that sample, no regime withstood a nonviolent campaign whose peak mobilization reached 3.5% of the population.
The supporting side: the causal chain is clear and testable (low barrier → large and varied participation → rising cost of repression → loyalty shift), and it holds after controlling for regime type and repression intensity — a rare macro-political finding that replicates on independent data.
The skeptical side: the core problem is selection effects — a movement may choose nonviolence precisely because it already commands a broad social base, or faces a comparatively less brutal regime; if so the preconditions do the work and nonviolence is merely their companion. Beyond that, the boundaries of "a campaign" and the judgment of "success" both involve coding discretion.
The authors' own update matters more: the success rate peaked at roughly 65% in the 1990s and has fallen since, to roughly 34% in the 2010s. Candidate explanations include regimes learning more refined responses, rapid mobilization producing thin organization, and more campaigns carrying incidental violence. Such regularities are statistical features of a historical sample, not laws of physics.
Treating 3.5% as a recipe — "hit the number and you win." The authors have repeatedly clarified that this describes a historical sample rather than prescribing a future one; peak turnout is more an indicator of the result than its cause. A movement able to put 3.5% of a population in the street usually already has cross-class coalitions, organizational durability and hesitation inside the security forces — and those are what actually operate.