The Community Mobilization Trap · Recognizing Patterns Overview Part I · The Acquisition Trap Part II · The Signal Part III · Succession
Part III · Succession

The Succession ProblemThe same gap, measured across a century of political data.

Two movements that won in under a week and lost the country that followed, the quantitative record on why, the literature that contradicts this report, and the diagnostic that comes out of both.

Series The Community Mobilization Trap
Published August 2026
Sources HRW · OHCHR · Amnesty · Reuters · NAVCO · ICNC

Nepal, four days

A platform ban to a new prime minister in ninety-six hours.

On 4 September 2025 Nepal's Ministry of Communication and Information Technology ordered 26 social media platforms blocked for failing to register by the previous day's deadline.

On 8 September a crowd gathered at Naya Baneshwor Chowk, the junction one corner of which is occupied by Nepal's parliament. Human Rights Watch, working from verified drone and video footage, found that security forces used disproportionate force and fired indiscriminately on protesters multiple times over about three hours, and that witness accounts and analysed footage do not show the grave and imminent danger to life that would justify intentional lethal force. Certain Primary · HRW

Oli resigned on 9 September and the platform ban was lifted the same day. On 12 September, former chief justice Sushila Karki was sworn in as interim prime minister by President Ramchandra Paudel, the first woman to lead Nepal. Reuters reports her selection followed talks between the president, army chief Ashok Raj Sigdel and protest representatives.

On the casualty figures, which are not interchangeable. Five numbers circulate and each counts a different thing. 19 is deaths in the police crackdown on 8 September alone, with over 300 injured. 34 is a health ministry running total as of 11 September, all causes. 74 is a cumulative UN figure for 8 to 22 September, later superseded. 76 is the judicial commission's count for 8 to 9 September nationwide, and separately Amnesty's count for a six-day window to 13 September. 77 is the Government of Nepal damage assessment, which explicitly adds 20 deaths occurring after 9 September. 76 and 77 are not in conflict; they count different windows.

And then, eighteen months

The movement did not produce the party that governs.

Nepal voted on 5 March 2026, on a revised turnout of 59.84%, the lowest since 1991.

4 Sep 2025
26 social platforms ordered blocked
8 Sep 2025
Security forces fire on protesters at Naya Baneshwor Chowk
HRW: disproportionate, indiscriminate force over about three hours.
9 Sep 2025
Oli resigns; platform ban lifted
12 Sep 2025
Sushila Karki sworn in as interim PM
Nepal's first woman to lead the country, after talks between the president, army chief and protest representatives.
5 Mar 2026
Election, 59.84% turnout
Lowest since 1991.
12 Mar 2026
Rastriya Swatantra Party wins 182 of 275 seats
RSP was founded in 2022, three years before the protests it is credited with capitalizing on.
27 Mar 2026
Balendra Shah sworn in as PM
Nepal's youngest prime minister.

Four days from platform ban to resignation. Eighteen months from resignation to a party founded three years before the protests began.

The Rastriya Swatantra Party took 182 of 275 seats, against 38 for the Nepali Congress and 25 for Oli's party. Balendra Shah was sworn in on 27 March 2026 as Nepal's youngest prime minister. Certain Primary · Election Commission via Kathmandu Post

The RSP was founded in 2022, four years before the vote and three years before the protests. Hold that detail. It is the whole of Section 06.

A network conferred legitimacy in ninety-six hours. The army chief, the president and a party founded 2022 (four years before the 2026 vote) conferred power.

Madagascar, where the movement lost the country it won

Six days from a mutiny to a sworn-in colonel.

Protests began in Antananarivo on 25 September 2025 over water and electricity failures. By 29 September the UN High Commissioner for Human Rights reported at least 22 people killed and more than a hundred injured. President Rajoelina disputed the count on 8 October, saying there had been 12 confirmed deaths and all were looters and vandals.

Then the sequence that decided it, none of which the movement controlled.

DateEvent
11 Oct 2025The CAPSAT unit announces it will refuse orders to fire, calling on forces to disobey orders to shoot at your fellow citizens
12 Oct 2025CAPSAT states that all directives from the Malagasy army will emanate from its command centre. Major General Démosthène Pikulas is installed as Chief of Army Staff the same day
14 Oct 2025Colonel Michael Randrianirina announces that the military has taken charge. The National Assembly impeaches Rajoelina for dereliction of duty on 130 votes in favour and one abstention
14 Oct 2025Ordinance 2025-001 suspends the 2010 constitution and five institutions: the High Constitutional Court, the electoral commission, the Senate, the High Council for the Defence of Human Rights and the High Court of Justice. The National Assembly stays
17 Oct 2025Randrianirina is sworn in as president, with a transition of at most two years before elections
25 Sep 2025
Protests begin in Antananarivo
Over water and electricity failures.
29 Sep 2025
UN reports at least 22 killed, 100+ injured
11 Oct 2025
CAPSAT unit refuses orders to fire
12 Oct 2025
CAPSAT takes command; new army chief installed
14 Oct 2025
Military takes charge; Rajoelina impeached 130–1; constitution suspended
17 Oct 2025
Randrianirina sworn in as president
A transition of at most two years promised.
Mar 2026
~600 appointments made
Amnesty: the Gen Z movement was “neither involved nor consulted.”

Six days from mutiny to sworn-in president. The movement controlled none of the intervening steps.

Reuters reports the single non-affirmative ballot as an abstention or blank ballot, not a vote against. On the timeline, Reuters, Anadolu and Amnesty all report a maximum of two years; Al Jazeera alone reported the promise as 18 to 24 months. Likely Wire disagreement flagged

Amnesty International's March 2026 assessment is the one that matters. The junta reshuffled the executive and made almost 600 appointments, and the Gen Z youth movement and civil society it claims to have seized power for was neither involved nor consulted. On 15 March 2026 Randrianirina dismissed the entire cabinet without explanation.

What the data says, and what its author says about it

The caveats are more useful than the headline.

Erica Chenoweth's NAVCO project is the standard quantitative source, covering 622 maximalist campaigns from 1900 to 2019. Nonviolent campaign success rates peaked in the 1990s and have fallen sharply since 2010, the decade in which digital organizing became universal.

From the Carr Center's cautionary update on the 3.5% rule: it is a rule, not a law. Eighty-three percent of successful maximalist campaigns never crossed the threshold, and Bahrain exceeded 6% peak participation and decisively failed.

The load-bearing sentence, in Chenoweth's own words: the rule does not speak to leadership, strategic imagination, organizational capacity or sustainability. And the one that carries this report: building and maintaining momentum requires organizational capacity, and today's digital organizing environment makes it easier to coordinate mass protests but not necessarily to sustain them.

Two structural limits belong alongside it, both stated by the authors. The dataset is knowingly biased toward success, because large mature campaigns are most commonly reported and campaigns crushed in infancy cannot be included. And NAVCO's coding rule requires that campaigns have discernible leadership, which means leaderless networked mobilizations sit awkwardly inside the unit of analysis the numbers are built on. Uncertain Scope limitation stated by authors

Why capacity is the binding constraint

Convening is acquisition. Coherence is retention.

Zeynep Tufekci named the mechanism in 2017. Digital tools let movements grow to a size that would once have required years of organizing, without doing the organizing, and the capacities that organizing builds are the ones needed after the peak. Her term for the resulting paralysis is tactical freeze.

The phrase "tactical freedom, strategic weakness", which appeared in an earlier draft of this report and elsewhere in our material, is not Tufekci's. It is our gloss and it is retired here.

Jonathan Pinckney's ICNC work supplies the operating instruction. Democracy is most likely, he finds, when activists keep their social bases mobilized while directing that mobilization toward building new political institutions. Two failure modes destroy it: rapid demobilization, which shifts the balance of force back to the remnants of the old regime, and street radicalism, which disrupts institutionalization from the other direction.

His companion finding is sharper: winning via an election or negotiation tends to promote democratization, while extra-institutional seizures of power, even when primarily nonviolent, tend to make democratization less likely. And the timing constraint that explains both Nepal and Madagascar: the power of the streets is highest just before breakthrough and declines once breakthrough occurs. The window for placing people into institutions opens before the win, not after it.

Mark Beissinger supplies the structural reason. The dominant contemporary form, which he calls the urban civic revolution, accounts for almost two-fifths of all revolutionary episodes worldwide since 1985, produces regimes that last significantly less time than those from social revolutions, and inherits the old state's corruption relationships intact.

Beissinger's mechanism sentence is the best in this literature: the larger the number of individuals mobilized into revolutionary action, the more variegated the participants necessarily will be, and the more incentive there is for revolutionary leadership to act as a convener of anti-incumbent sentiment than a provider of organizational or ideological coherence. This problem is compounded when revolutionary mobilization occurs on a temporal scale of days and weeks. Read that against Part I: convening is acquisition, coherence is retention, and speed makes the trade-off worse in both domains.

The case against everything above

Two objections, and the second one narrows the claim.

The strongest objections to this report were not raised against it. They were already in the literature it draws on.

Objection one: institutionalization is the trap, not the cure

Piven and Cloward's argument in Poor People's Movements is that mass defiance wins concessions and organizations do not. Organization-building draws people away from the streets and into the meeting rooms, while the search for resources drives organizers inexorably toward elites. Organizations that survive do so, in their words, by abandoning their oppositional politics. The line that inverts this report entirely: by endeavouring to do what they cannot do, organizers fail to do what they can do.

Michels supplies the structural version. Who says organization, says oligarchy. This is not a fringe position: McAdam concedes from inside the pro-organization camp that formal organizations render a movement vulnerable to oligarchization and co-optation, and Tarrow concedes that institutionalization often succeeds at the cost of transforming the movement into a party or interest group.

Objection two: networked movements demonstrably do institutionalize

Movement to vehicleResultLagVehicle provenance
15-M to Podemos20.7% and 69 seats, Dec 20154.6 yrsNewly built, by a pre-existing academic and media network
Grillo's network to M5S32.7% and government, 20188.7 yrsNewly built, on a broadcaster-comedian's platform
Pots and pans to the Pirates14.5% and 10 seats, 20167.8 yrsLaunched by a sitting MP already inside parliament
Euromaidan to Servant of the People43% and a Rada majority, 20195.4 yrsBuilt on a pre-registered party shell; no sourced link to Maidan
Estallido social to Boric55.87% presidency, Dec 20212.2 yrsFrente Amplio, founded 2017 out of the 2011 student movement
Estallido social → Boric
2.2 yrs
15-M → Podemos
4.6 yrs
Euromaidan → Servant of the People
5.4 yrs
Pots and pans → Pirates
7.8 yrs
Grillo's network → M5S
8.7 yrs

Years from movement peak to serious electoral power, five cases. None was improvised by the protest's own horizontal assemblies

The claim that networked movements cannot institutionalize is false and this report retires it.

The provenance rule. What replaces it is narrower and survives all five cases. The lag from protest peak to serious electoral power clusters at roughly two to five years, and the vehicle is almost never improvised from the protest itself. Chile, the fastest and largest victory, ran through a coalition assembled eight years earlier out of a previous protest wave. Iceland's Pirates were launched from inside parliament. Servant of the People was built on a shell registered before it. Podemos and M5S are the two genuine new builds, and both were founded by pre-existing political entrepreneurs rather than by the horizontal assemblies. In Nepal that vehicle was founded 2022 (four years before the 2026 vote). In Madagascar it was an army unit.

Chile also demonstrates that institutional capture and programmatic victory are separable. Seventy-eight percent voted to draft a new constitution; 61.86% then rejected the draft the resulting convention produced.

The isomorphism

Not decorative. The same underfunded line item, twice.

Both domains have cheap, fast, effective mass coordination. In neither did the multi-year organizational labour that converts coordination into durability get any cheaper.

StructureGamesPolitical movements
What the model is good atAcquisition. 11M, 20M, 12M playersMobilization. A head of government in four days
What it cannot doRetention. Cadence for yearsInstitutionalization. Governing after the win
What substitutesPre-supplied recognition, meaning IPPre-existing organization, meaning a party or an army
How the failure showsA decay curve. 98% gone in twelve monthsA colonel, or a party founded 2022 (four years before the 2026 vote)
The measurable trendAttention near zero-sum, 8% of hours to new releasesSuccess rates falling as digital organizing became universal
The unasked questionWho funds the cadence in year threeWho governs on day thirty

Both reward whoever did the organizational labour in advance, and both systematically underfund it, because the funding case in each is built on the moment of acquisition.

The open question

Whether this is a transitional deficit or an inherent one is not settled here. Bethke and Pinckney's replication finds that regimes whose transition was induced by nonviolent resistance have a median survival of 47 years, against nine years where no resistance campaign influenced the transition, with the authors' own limit attached: they measured persistence, not consolidation. Lehoucq, adding peaceful reform campaigns that NAVCO omits, finds the success rates of nonviolent and violent campaigns to be equal.

And there is a third reading this report cannot dismiss. If Piven and Cloward are right, failing to institutionalize is not a failure. It is a refusal to be absorbed, chosen by people who have watched what absorption does.

The year-three diagnostic

Six questions for a whiteboard.

These come out of Part I Section 04 and Part II Section 05. They are the part of this report a studio can take into a room.

  1. Is your all-time peak behind you, and is that a regime or an accident? Late peaks correlate with survival in this sample and the correlation is partly circular, since a game that is still growing has by definition not yet peaked. Treat it as a regime detector, never as a forecast on an unlaunched title.
  2. Does the product generate its own distribution when played? PEAK converted 36,797 wishlists into more than 10 million copies because the purchase is a social act. That is a pre-production mechanic, not a marketing plan, and it cannot be added later.
  3. Is your organic wishlist rate above the floor? No game below ten organic daily wishlists exceeded 3,000 units in launch week. Use this as a kill switch. Do not use a strong number as a forecast: dispersion runs 10 to 20 times, not 10 to 20 percent.
  4. Who funds the content cadence in year three? Path of Exile shipped forty releases in ten years. If your greenlight case ends at launch plus twelve months, you have funded acquisition and called it a live game.
  5. If you are shipping an authoring layer, what does it cost to operate? Crowfall's stated cause of death was the effort required to build new campaigns and run the game daily. Participation infrastructure lowers the cost of cadence, is not a substitute for it, and can be the bill that closes you.
  6. Who holds the vehicle if the moment arrives early? The political version of the same question. If your audience shows up before your operating capacity does, someone else with pre-existing capacity captures them.
What this report could not source. Any per-era AAA failure rate with a denominator. Any marketing spend disclosed next to revenue for a named title. A verbatim Chenoweth and Stephan sentence for the frequently quoted 45-point democratic dividend, which is available only as reported by a reviewer. Any sourced statement linking Servant of the People to Euromaidan. Policy-target reversals in Kenya, Indonesia and Morocco are cited in the wider literature but were not primary-verified here, so only Nepal and Madagascar carry the regime-target argument.

The instrument that would settle most of this argument is the one nobody publishes: what a launch cost to market, next to what it earned. No publisher in this report disclosed one. Until that number exists, every thesis in this space is inference dressed as measurement, including this one.

I
The Acquisition Trap
Four shutdowns, and the authoring-layer hypothesis
II
The Signal Is Not the Cause
Wishlists, reviews, and what survives
III
The Succession Problem
Nepal, Madagascar, and the year-three diagnostic

Abbas Saleem is a Principal Consultant at Llama & Griffin, advising game studios, publishers, and investors across MENA, South Asia, Southeast Asia, and Europe. He writes Recognizing Patterns: gaming industry intelligence 12 to 24 months before it becomes consensus. LinkedIn