Author's Note

This paper exists because of 14 goals. Not policy proposals — prevention requirements. Each one traces to a system that broke a real person.

Goal 1 says fire all politicians. You vote on everything. The Swiss have been doing it for 178 years. Goal 2 says 22-hour work weeks — because when you give people back 20 hours, they have time to actually govern their own communities. Goal 3 says free all prisoners, because Norway proved that the person in the cage and the person who put them there are the same person born in a different postcode. Goal 5 says fire all police, because the CAHOOTS model has been running for 35 years with zero people killed. Goal 13 says a $29 ring — press it, your people come in 60 seconds, not an ambulance in 14 minutes.

Every one of these goals requires a governance structure. Not a government — a governance structure. The difference is the direction of authority. In a government, decisions flow down from representatives to constituents. In a governance structure built on federation, decisions flow up from communities to delegates who can be recalled at any time.

This paper asks: what does the evidence actually say about how humans organise? Not the theory. Not the TED talk version. The evidence. From primatology, anthropology, organisational psychology, and — most importantly — from communities that are doing it right now.

The answer is not complicated. It has been operating for 60,000 years in Aboriginal Australia, for 500 years in Hutterite colonies, for 70 years in the Basque Country, for 30 years in Chiapas, and for over a decade in northeastern Syria under active bombardment. The structural pattern is the same everywhere: small groups for intimacy, medium groups for governance, federation for scale. Rotating leadership. Shared economic base. Mandatory inclusion. Face-to-face deliberation at the bottom. Delegates — not representatives — at every level above.

Dunbar said we can only maintain 150 relationships. Lindenfors showed that number has a 95% confidence interval of 2 to 520. Both miss the point. The question was never how many people you can know. The question is what structures allow care to operate beyond the people you know. Aboriginal kinship systems answered that question tens of thousands of years ago. So did the Haudenosaunee Confederacy. So does Mondragon. So do the Zapatistas.

The animal already knows how to organise. It has always known. What it needs is an enclosure that lets it.

— A.A. & L.N.C.

Abstract

Dunbar's number — the claim that humans can maintain approximately 150 stable social relationships — has become received wisdom in organisational design, management theory, and popular science. This paper examines the statistical foundations of that claim and finds them wanting: Lindenfors et al. (2021) demonstrated that the neocortex-to-group-size regression produces a 95% confidence interval of 2 to 520, rendering the specific figure of 150 statistically meaningless. However, the observation that governance changes character at different scales is robustly supported across disciplines. Dunbar's layered model (5/15/50/150/500/1,500) finds empirical support from phone records, social media interaction data, and cross-cultural ethnographic evidence.

We examine eight functioning examples of scaled community governance — Mondragon (80,000 worker-owners), Rojava (4 million under democratic confederalism), the Zapatista autonomous zones (300,000 governed), Marinaleda (2,700 in cooperative economy), Swiss cantons (8.8 million in direct democracy), Hutterite colonies (500 years of communal living), kibbutzim, and Aboriginal Australian kinship systems (continent-wide trust networks without hierarchy). From these cases, we derive eight design principles for community governance that scales without centralising: nested autonomy, power rotation, shared economic base, mandatory inclusion, face-to-face deliberation at the base, federation (not hierarchy) for scale, circular time, and the Peck-Dunbar bridge connecting intimacy groups to governance groups to confederal structures.

The evidence converges on a single structural pattern observable across cultures, continents, and millennia: autonomous groups of 25-150 people, federated through recallable delegates, with rotating leadership and shared material stakes. This pattern is neither theoretical nor historical — it is operational, tested under conditions including active warfare, economic crisis, and state hostility, and it consistently outperforms centralised governance on measures of equity, resilience, and democratic participation.

Keywords: Dunbar's number, federation, democratic confederalism, cooperative governance, social scaling, community design, direct democracy, rotating leadership

Table of Contents

1. 1. Part I: The Numbers

• 1.1 The Number Everyone Knows
• 1.2 Our Closest Relatives Don't Agree on Anything
• 1.3 Recognition, Relationship, and Love — Three Different Things That Scale Differently
• 1.4 The Actual Band, Clan, and Tribal Data

1. 2. Part II: Who's Doing It Right

• 2.1 Mondragon Corporation
• 2.2 Rojava (Autonomous Administration of North and East Syria)
• 2.3 Zapatista Juntas de Buen Gobierno
• 2.4 Marinaleda
• 2.5 Swiss Cantons and Direct Democracy
• 2.6 Other Cases: Kibbutzim, Hutterites, Auroville

1. 3. Part III: Design Principles

• 3.1 Nested Autonomy
• 3.2 Power Rotation (Not Power Elimination)
• 3.3 Economic Base
• 3.4 Mandatory Inclusion
• 3.5 Face-to-Face at the Base
• 3.6 Scale Through Federation, Not Hierarchy
• 3.7 Circular Time
• 3.8 The Peck-Dunbar Bridge

1. 4. Part IV: The Federation Pattern — Expanded Case Analysis

• 4.1 Marinaleda Revisited: The Single-Leader Problem and Structural Succession
• 4.2 Rojava's Commune System in Detail
• 4.3 The Zapatista Education and Health Systems
• 4.4 Swiss Cantons: 178 Years of Proof

1. 5. The Connection
2. 6. References
3. 7. Appendix A: Confidence Assessments
4. 8. Appendix B: Cross-References to the OMXUS Research Series

Part I: The Numbers

1.1 The Number Everyone Knows

One hundred and fifty. Dunbar's number. It shows up in TED talks, management books, startup advice columns, and military history documentaries. The claim is simple: the human brain can maintain roughly 150 stable social relationships. Any more than that and you start forgetting names, losing track of alliances, dropping the threads that hold a group together.

The number comes from Robin Dunbar's 1992 paper, "Neocortex size as a constraint on group size in primates," published in the Journal of Human Evolution. The method was straightforward. Dunbar collected data on primate species — their neocortex ratios (the size of the neocortex relative to the rest of the brain) and their typical social group sizes. He ran a regression. Bigger neocortex, bigger group. When you plugged in the human neocortex ratio, the regression predicted a mean group size of approximately 148. Round it to 150. There's your number.

In 2021, Patrik Lindenfors, Andreas Wartel, and Johan Lind published "'Dunbar's number' deconstructed" in Biology Letters. They re-ran the regression using updated primate brain data and more rigorous statistical methods. The 95% confidence interval for the predicted human group size was 2 to 520. That is not a number. That is the absence of a number. Different model specifications — which species to include, how to handle phylogenetic relationships, which brain measure to use — produced wildly different predictions. The number 150 was not robust to reasonable analytical choices.

Dunbar responded informally, arguing that convergent evidence from the layered model and real-world organisations supports the figure. He has a point — many organisations do cluster around that size. But the question Lindenfors raised remains formally unanswered in the peer-reviewed literature: the neocortex regression does not reliably predict a specific number for humans.

The honest position: 150 is not a biological constant. It is one observation from one method that turned out to be statistically fragile. But the observation that something changes at roughly that scale — that face-to-face coordination strains, that free-riding becomes harder to detect, that governance needs formalisation — shows up so consistently across contexts that dismissing it entirely would be as wrong as treating it as law.

The layers are more interesting than the number anyway. In later work, particularly his 2020 paper in Proceedings of the Royal Society B, Dunbar and colleagues described human social networks as having a nested structure with consistent layer sizes: 5 (the people you'd call at 3am), 15 (close friends and family), 50 (people whose lives you know in some detail), 150 (your active personal network), 500 (acquaintances), and 1,500 (faces you can put names to). Each layer is roughly three times the size of the one inside it. This scaling ratio shows up across datasets — self-reported networks, phone records, social media interaction patterns, Christmas card lists. The inner layers (5 and 15) are particularly robust. Almost everyone, across cultures and contexts, seems to concentrate their social energy this way.

This is useful not because it proves a brain constraint but because it describes how humans actually distribute their attention. We don't spread it evenly. We concentrate it intensely on a few people and distribute it thinly across many. Digital technology hasn't changed this. A 2016 study by Dunbar using Facebook data found that the number of friends users actually interacted with in two-way exchanges over a month was around 4-14 — the sympathy group, not the 150-person network. Gonçalves, Perra, and Vespignani (2011) found similar patterns on Twitter: stable interaction networks of 100-200. Social media gave us a larger Rolodex, not a larger social life.

1.2 Our Closest Relatives Don't Agree on Anything

Here is where the story gets interesting, and where Dunbar's framework — even the layered version — misses something fundamental.

Chimpanzees and bonobos are our closest living relatives. We share over 98.7% of our DNA with both species. Chimps and bonobos themselves diverged from each other only about 1-2 million years ago, separated by the Congo River. They are almost genetically identical to each other. And they have radically different social organisations.

Chimpanzees live in communities of roughly 20-150 individuals (Mitani, 2009), though typical community size is around 35-50. Males are dominant. Hierarchy is maintained through coalitional violence — male chimps form alliances to compete for status, and aggression is common. Males are philopatric (they stay in their birth group); females transfer between communities. Inter-community relations are frequently violent. Lethal raids on neighbouring groups are well-documented (Wilson et al., 2014). Food sharing is limited and often coerced. Sexual access is controlled by dominant males, though female choice operates as a countervailing force. The social structure is, in essence, a male dominance hierarchy enforced by violence and coalition politics.

Bonobos live in communities of similar size — roughly 30-100 individuals (Furuichi, 2011) — but the social organisation could hardly be more different. Females are socially dominant, despite being physically smaller than males. They achieve this through strong female-female alliances, particularly between unrelated females. Sexual behaviour is used to resolve conflicts, reduce tension, and build social bonds — between all combinations of sex and age. Food sharing is extensive and relatively egalitarian. Inter-community encounters are generally peaceful, sometimes involving socialising and sex between groups. Males derive their status from their mothers rather than from coalitional violence. There are no documented cases of lethal aggression between bonobo communities.

Same genus. Nearly identical DNA. One species runs on male violence and hierarchy. The other runs on female alliances and sexual conflict resolution.

The point is not that humans are "really" like chimps or "really" like bonobos. The point is that the relationship between genetics and social organisation is not deterministic. If two species that are almost genetically identical can produce social structures as different as chimpanzee patriarchy and bonobo matriarchy, then the claim that human social organisation is "constrained" by our neurology to some specific group size or structure is, at best, a very partial truth.

George Peter Murdock's Ethnographic Atlas (1967), a cross-cultural database covering over 1,200 pre-industrial societies, found that approximately 85% permitted polygyny (one man, multiple wives). True social monogamy — one partner for life, enforced by cultural norms — is a minority practice in the ethnographic record, historically concentrated in societies influenced by Roman law and early Christianity (Henrich, Boyd & Richerson, 2012). Pair-bonding exists in humans and appears to have deep evolutionary roots, but exclusive lifelong monogamy is a cultural institution, not a biological given. The range of human mating and family systems — monogamy, polygyny, polyandry (rare but real, e.g., parts of Tibet and Nepal), serial monogamy, group marriage — exceeds any other primate species.

The implication for social group scaling is direct: if human mating systems are this variable, then the social structures built on top of them are also variable. Family size, household composition, kinship obligations, inheritance patterns, child-rearing arrangements — all of these shape community structure, and none of them are fixed. The animal is flexible. The enclosure determines the behaviour.

1.3 Recognition, Relationship, and Love — Three Different Things That Scale Differently

The Dunbar model conflates three things that need to be separated.

Recognition is a cognitive function. It means you can identify a face, attach a name, recall some context. This is what Dunbar's neocortex regression is really about — the computational cost of tracking individuals. Recognition scales poorly because it requires dedicated neural resources for each person. You can recognise perhaps 1,500 faces (the outer Dunbar layer), but meaningful recognition — knowing someone's history, temperament, reliability — drops off sharply. This is hardware. You can't train your way to recognising 10,000 people with the same depth.

Relationship is a time function. It means you invest regular time and emotional energy in someone. You know what's happening in their life. You'd notice if they disappeared. This is what the 15-person sympathy group captures. The constraint isn't cognitive — it's temporal. There are only so many hours. Even if your brain could track 1,000 relationships, you couldn't maintain them. You'd have to shorten each one to the point of meaninglessness. The phone call data is unambiguous: people concentrate their communication on a small number of contacts regardless of how many contacts they have.

Love — or more precisely, care, solidarity, the willingness to sacrifice for another's wellbeing — is neither a cognitive function nor a time function. It is a moral and structural question.

Scott Peck, in The Different Drum (1987), described what he called "true community" — a state where a group achieves genuine vulnerability, authentic encounter, and mutual acceptance. Peck ran community-building workshops for decades and found that groups of 10-60 could reach this state, but the sweet spot was small: roughly 10-15 people. True community, for Peck, required that every person be genuinely seen and heard, which means every person must speak and every person must be listened to. The math is unforgiving. In a group of 12, each person gets a meaningful share of the group's attention. In a group of 120, they don't.

Martin Buber, in I and Thou (1923), drew a different line. For Buber, the quality of encounter — what he called the I-Thou relationship, as opposed to the I-It relationship — was not about number but about presence. You could have an I-Thou encounter with a stranger or fail to have one with your spouse. The capacity for genuine encounter is not bounded by group size. It is bounded by whether you show up.

The Ubuntu philosophy — umuntu ngumuntu ngabantu, "I am because we are" — makes a similar claim from the southern African tradition. The self does not exist in isolation. Personhood is constituted by relationship. Desmond Tutu described Ubuntu as the recognition that "my humanity is caught up, is inextricably bound up, in yours." This is not a claim about how many people you can know. It is a claim about the nature of human existence: that care is not a scarce resource to be allocated but a capacity to be cultivated.

This matters for community design because the Dunbar model, taken at face value, implies that care has a ceiling: you can only care about 150 people, or 15, or 5. The Ubuntu and Buberian traditions say the opposite: care is unlimited in principle, but it requires structure to be expressed at scale. You don't love your whole city by maintaining 500,000 personal relationships. You love your whole city by building institutions — hospitals, schools, parks, mutual aid networks — that express care structurally. The question is not "how many people can you love?" but "what structures allow love to operate beyond the face-to-face?"

The Zookeeper's argument sits here: the enclosure design determines whether the animal can extend care, not the animal's capacity for care. A zoo that puts the animal in a concrete box produces neurotic, aggressive, withdrawn behaviour. A zoo that provides space, stimulation, social contact, and autonomy produces something that looks a lot more like flourishing. The animal hasn't changed. The enclosure has.

1.4 The Actual Band, Clan, and Tribal Data

Before agriculture, before cities, before the internet, humans lived in bands. The anthropological data on hunter-gatherer group sizes provides something like a baseline — what group sizes look like when the constraints are ecological and social rather than institutional.

Hunter-gatherer residential bands — the people you actually live and move with — typically number 25 to 50 individuals. Hill et al. (2011), in a paper in Science, analysed co-residence patterns across 32 hunter-gatherer societies and found a median band size of about 28 adults. This is remarkably consistent across environments, from the Kalahari to the Arctic to the Amazon.

Why 25-50? Not brain constraints. Logistics. A foraging band needs enough people to hunt effectively, care for children, and maintain social life, but not so many that they exhaust the local food supply and have to move constantly. The carrying capacity of the local environment sets the upper bound. Conflict resolution sets another: in a group of 30, you can resolve disputes by conversation. Everyone knows everyone. Social pressure works. In a group of 300, you need formal mechanisms — courts, laws, designated mediators — or you get unresolved feuds and fission.

Above the band level, hunter-gatherer societies show a nested structure:

• Band: 25-50 individuals. The people you live with. Everyone knows everyone.
• Clan/regional group: 150-500. People who share a dialect, a territory, and periodic gatherings. You know them by name and reputation. You intermarry with them.
• Tribe/linguistic group: 500-2,500. People who share a language and cultural identity. You may never meet most of them, but you recognise them as "us."

This multilevel structure maps loosely onto Dunbar's layers, though the match is imperfect and the variation across societies is substantial. What's more interesting than the sizes is the mechanism: these are not levels of a hierarchy. They're levels of a federation.

Federation needs a clear definition because it gets used loosely. A federation is a structure where autonomous groups choose to coordinate with each other without surrendering their autonomy to a central authority. The key distinction is with hierarchy, where decisions flow down from the top and the units below exist to execute those decisions. In a federation, the units below are primary. They are the source of authority. The "higher" levels exist only to coordinate between them — and can be dissolved or restructured by the units at any time.

Think of it this way. A corporation is a hierarchy: the CEO decides, the departments execute. A franchise is closer but still hierarchical: the franchisor sets the rules, the franchisees follow them. A federation is something different: the local units make their own decisions about their own lives, and they send delegates to a shared council that handles only what no single unit can handle alone — mutual defence, trade between units, disputes that cross boundaries. The council serves the units. The units do not serve the council. If the council overreaches, the units can recall their delegates or leave.

There is a third option that gets confused with federation constantly: flat. A flat organisation has no hierarchy — but it also has no structure. No designated roles, no formal decision-making process, no agreed mechanism for resolving disputes. Everyone is equal. Everyone has a voice. In theory.

In practice, flat organisations are governed by whoever has the most social capital, the loudest voice, or the most free time. Jo Freeman described this in her 1972 essay "The Tyranny of Structurelessness": when a group refuses to create formal structure, informal structure fills the vacuum — and informal structure is invisible, unaccountable, and almost impossible to challenge. The person who controls the mailing list, who speaks first at every meeting, who has private conversations before the group convenes, becomes the de facto leader without ever being selected, accountable, or removable. Flat is not the absence of hierarchy. It is hierarchy with plausible deniability.

Federation is the actual alternative: structured but not hierarchical. There are roles, but they rotate. There are decision-making processes, but they're consensual. There is authority, but it flows upward from the base, not downward from the top. Federation takes the good part of hierarchy — clear structure, defined responsibilities, mechanisms for resolving disagreement — and strips out the bad part: permanent concentration of power.

Auroville is what happens when you try flat. Mondragon is what happens when you try federation. The difference in outcomes is not subtle.

This is what hunter-gatherer bands did. Each band is autonomous. The clan gathers periodically for ceremonies, trade, and marriage. The tribe shares language and identity. Decisions flow from the bottom up — or more accurately, each level handles different kinds of decisions. The band decides where to camp tonight. The clan decides who can marry whom. The tribe decides nothing as a whole — it is an identity, not a government.

Aboriginal Australian kinship systems deserve particular attention because they solved a problem that most governance theories assume is unsolvable: connecting people across vast distances without hierarchy.

Aboriginal Australian societies developed kinship classification systems — moiety, section, and subsection systems — that create automatic social obligations between people who have never met. Under a moiety system, every person belongs to one of two groups (moieties). Under a section system, there are four; under a subsection system, eight. Your section determines who you can marry, who you call "mother" or "brother" (even if they're not biologically related), and what obligations you owe to whom.

The effect is that when you meet a stranger from a community 500 kilometres away, you can immediately determine your kinship relationship to them. Your section name tells them (and vice versa) whether you're a "brother," a potential spouse, an "uncle," or a "mother-in-law" — and each of those relationships carries specific behavioural obligations (Berndt & Berndt, 1999). The kinship system is the governance system. It creates trust between strangers without courts, contracts, police, or centralised authority. It has operated continuously for tens of thousands of years across an entire continent.

This is not a primitive system. It is a sophisticated social technology for scaling trust beyond the face-to-face group. And it does something that Dunbar's model says should be impossible: it extends meaningful, obligation-bearing social relationships to thousands of people, most of whom you'll never meet.

The Haudenosaunee (Iroquois) Confederacy provides another model. Five nations (later six, after the Tuscarora joined around 1722) — the Mohawk, Oneida, Onondaga, Cayuga, and Seneca — were linked through the Great Law of Peace (Gayanashagowa), an oral constitution that predates European contact. Estimates of the total Confederacy population at European contact range from 20,000 to over 100,000.

The governance structure was nested: each nation contained multiple clans. Each clan was headed by a clan mother — the senior woman of the leading household. Clan mothers selected (and could remove) the sachems (chiefs) who sat on the Grand Council. The Grand Council operated by consensus: no decision was made unless all nations agreed. Sachems could not act unilaterally. If a sachem acted against the wishes of the people, the clan mother who had appointed him could "dehorn" him — strip him of his title (Johansen, 1982).

This system scaled to tens of thousands of people across a vast territory with no centralised executive, no police force, and no standing army. Decisions flowed upward from clan mothers through sachems to the Grand Council, and back down. Multiple historians have documented the influence of the Haudenosaunee Confederacy on the framers of the US Constitution — particularly Benjamin Franklin, who attended treaty councils and explicitly recommended the Confederacy as a model (Grinde & Johansen, 1991). The irony is heavy: the republic that displaced the Confederacy borrowed from it while building something far more centralised.

Part II: Who's Doing It Right

The critique of Dunbar is interesting but ultimately academic. The real question — the one that matters for anyone trying to build functional human communities — is: what actually works? Not in theory. Not in a regression model. In practice, at scale, under real-world conditions.

Several communities operating right now provide answers. For each, the question is not just "does it exist?" but "what specific structural features make it work?"

2.1 Mondragon Corporation (Basque Country, Spain)

Founded in 1956 by José María Arizmendiarrieta, a Catholic priest who had survived Franco's Civil War and decided that the Basque working class needed economic democracy, not charity. He started with a technical education school. Five of its graduates founded the first cooperative — a paraffin heater manufacturer called ULGOR — in 1956.

Today, Mondragon is the largest cooperative corporation in the world. Approximately 80,000 worker-owners across 95+ cooperatives, operating in finance, industry, retail, and education. Annual revenue exceeds 11 billion euros. It is the tenth-largest business group in Spain and the largest in the Basque Country.

How it scales: Federated structure. Each cooperative is autonomous, typically between 100 and 500 workers. They're grouped into industrial divisions, overseen by a cooperative congress. Each cooperative has its own governance: a general assembly (one worker, one vote), an elected governing council, and a social council that handles working conditions and welfare. Above the individual cooperatives, a general council coordinates strategy, and a congress of all cooperatives sets overall policy.

What makes it work:

Nested autonomy. The individual cooperative is small enough that workers know each other and can govern effectively through direct participation. The federation provides shared infrastructure — a cooperative bank (Laboral Kutxa), a university (Mondragon Unibertsitatea), an R&D centre, and a social security system — that would be impossible for any single cooperative to build alone. Each unit gets the intimacy of a small organisation and the resources of a large one.

Pay ratio caps. Historically, the ratio between the highest-paid and lowest-paid worker in a Mondragon cooperative was capped at 6:1. Under pressure from global labour markets (it's hard to recruit top engineers at 6:1 when competing firms offer 100:1), this has relaxed to approximately 9:1 in some cooperatives. Compare this to the average US S&P 500 company, where the CEO-to-median-worker pay ratio exceeds 300:1 (AFL-CIO, 2023). The cap is not just symbolic. It means that decisions about automation, layoffs, and investment are made by people who will personally bear the consequences.

Crisis response. When the 2008 financial crisis hit, Mondragon cooperatives did not lay off workers. The workers voted — and it is essential to understand that this was a democratic decision, not a management decree — to take temporary pay cuts of 5-8% rather than fire colleagues. Workers from cooperatives with reduced demand were temporarily reassigned to cooperatives that needed labour. The unemployment rate in the Basque Country peaked at 25% during the crisis. Within Mondragon, it was effectively zero (Errasti et al., 2017).

Failure is real. Mondragon is not utopia. In 2013, Fagor Electrodomésticos — one of the founding cooperatives and a major appliance manufacturer — went bankrupt. It had 1,800 workers. The cooperative system responded: other Mondragon cooperatives absorbed displaced workers, offered retraining, and provided unemployment support through the inter-cooperative solidarity fund. The failure was real. The response was structurally different from what happens when a conventional corporation collapses.

The lesson: economic democracy at scale is not a thought experiment. It's been running for nearly 70 years, employing tens of thousands of people, weathering recessions, surviving failures, and consistently outperforming conventional firms on worker satisfaction, income equality, and community stability. The structure is federation. The unit is the cooperative. The principle is one person, one vote.

2.2 Rojava (Autonomous Administration of North and East Syria)

In 2012, as the Syrian Civil War shattered the central government's control over the country's northeast, the Kurdish-majority regions declared autonomous self-governance. What emerged was not a conventional state but an experiment in what Abdullah Öcalan — the imprisoned Kurdish leader who had spent years reading Murray Bookchin's writings on communalism from his prison cell on Imrali Island — called "democratic confederalism."

The population under this system is approximately 4 million people, comprising Kurds, Arabs, Assyrians, Turkmen, and other ethnic groups.

How it scales: The base unit is the commune (komîn), comprising 30 to 400 households in a neighbourhood or village. Communes meet regularly — in many areas weekly — to discuss local issues: water, electricity, disputes, education, security. Above the commune sits the neighbourhood council, then the district council, then the canton council, then the regional administration. Each level sends delegates (not representatives — delegates carry the decisions of their commune, they do not make independent judgments) to the level above.

What makes it work:

Mandatory gender parity. Every leadership position at every level has co-chairs: one male, one female. This is not a suggestion. It is a structural requirement. Additionally, there is a 40% minimum gender quota for all councils and committees. Women's councils (Kongra Star) operate in parallel with the general governance structure and have veto power over decisions affecting women. In a region where, a decade earlier, honour killings were common and women's participation in public life was severely restricted, this represents a radical structural intervention — not a cultural evolution but a deliberate institutional design choice (Knapp, Flach & Ayboga, 2016).

Ethnic representation. Councils must include representatives of all ethnic groups present in the area. A district that is 60% Kurdish and 30% Arab and 10% Assyrian must have proportional representation, with protections for minority voice. This is structural inclusion — not "we welcome everyone" but "you must be here, and you must be heard."

Decisions flow upward. The critical structural feature is that authority rests at the bottom. Communes make decisions about local matters. They send delegates — not representatives — to higher councils to handle coordination between communes. A delegate who acts against the commune's decision can be recalled immediately. The higher levels exist to coordinate, not to command.

It works under fire. This system has operated under continuous military threat — from ISIS, from the Turkish military, from the Syrian government, from various militia groups. That it functions at all under conditions of active warfare is remarkable. That it maintains democratic structures, gender parity, and multi-ethnic governance under those conditions is, by any reasonable standard, extraordinary. Whatever its flaws (and there are critics who question whether the PYD/PKK exercises informal power that contradicts the formal democratic structures), Rojava demonstrates that democratic confederalism is not merely a theory. It is a functioning governance system for millions of people.

Rojava's Commune System in Detail

The commune is where the theory becomes real. A commune in Rojava is not a symbolic body. It handles electricity allocation, water distribution, local disputes, education coordination, and security. Each commune elects two co-chairs (one male, one female) and forms committees: defence, women's issues, economics, justice, education. The commune meets weekly or biweekly, and every household is expected to send at least one member.

The justice committees are particularly instructive. When a dispute arises — land, family, commerce — the commune's peace committee attempts resolution first. Most disputes are resolved at this level, through mediation. If mediation fails, the case moves to the neighbourhood council, then the district. At no point does the case enter a system that resembles a Western court. There is no prosecutor. There is no adversarial process. There is mediation, then arbitration, then community decision. The goal is restoration, not punishment.

This is not aspirational. It is operational for roughly 4 million people. Under bombardment.

2.3 Zapatista Juntas de Buen Gobierno (Chiapas, Mexico)

On New Year's Day 1994, the Zapatista Army of National Liberation (EZLN) seized several towns in the southern Mexican state of Chiapas. The military uprising lasted twelve days. What came after has lasted thirty years.

In 2003, the Zapatistas formalised their autonomous governance system into five Caracoles (snail shells — the name is deliberate, referring to the snail's spiral as a symbol of going inward to find the way forward) and their associated Juntas de Buen Gobierno (Good Government Councils). The system governs approximately 300,000 indigenous Maya people across autonomous territories in Chiapas.

How it scales: Community assemblies are the base unit. Every adult in a community participates. Assemblies send delegates to municipal autonomous councils, which send delegates to the Juntas de Buen Gobierno. The Juntas rotate membership — delegates serve for periods of one to two weeks, then rotate out and return to their communities. There are no permanent politicians.

What makes it work:

"Mandar obedeciendo" — to lead by obeying. This phrase captures the entire governance philosophy. Leaders do not make decisions. The assembly makes decisions. Leaders execute. A leader who makes an independent decision — who leads without obeying — is recalled. Leadership is understood as service, not authority. It is unpaid. It is temporary. It is subject to immediate recall.

Rotation prevents power accumulation. Because Junta members serve for one to two weeks at a time, there is no opportunity to build a personal power base, develop patronage networks, or become indispensable. Everyone serves. Everyone returns to ordinary life. The contrast with professional politicians who serve for decades and accumulate institutional power could not be sharper.

Material results. The Zapatista autonomous zones have built, from scratch and without state support, their own education system (autonomous schools with community-designed curricula), their own healthcare system (community health clinics), and their own justice system (based on mediation and community service rather than incarceration). Infant mortality in Zapatista territories has declined. Literacy has risen. Women's participation in governance has increased dramatically, though gender parity remains an ongoing struggle rather than a completed project (Stahler-Sholk, 2007; Baronnet, Mora & Stahler-Sholk, 2011).

It works without the state. The Mexican government has never recognised Zapatista autonomy. The autonomous zones receive no government funding, no government services, and no government recognition. They exist in a legal grey zone, sustained entirely by internal governance and international solidarity. That a governance system can produce declining infant mortality and rising literacy without any support from the state that claims sovereignty over its territory tells you something about what is and isn't necessary for functional community.

The Zapatista Education and Health Systems

The Zapatista autonomous education system is worth examining in detail because it demonstrates what happens when the 14 goals' vision of education (Goal 12: "Every school is play, mastery, curiosity") is built from the ground up rather than reformed from the top down.

Zapatista autonomous schools are designed by the communities they serve. There is no national curriculum. Each community decides what its children need to learn. The result is education that integrates agricultural knowledge, indigenous history, Spanish and Tzotzil/Tzeltal/Tojolabal languages, basic mathematics, health, and human rights. Teachers are community members — promotores de educación — selected by the community, trained by the autonomous education system, and accountable to the community assembly. They are not professionals in the Western sense. They are community members who teach.

The healthcare system follows the same pattern. Promotores de salud (health promoters) are trained by the autonomous system and serve their communities. They provide basic healthcare, maternal care, herbal medicine, and health education. For cases beyond their capacity, a network of autonomous clinics and a small number of hospitals provide secondary care. The system operates with no government funding and minimal external support.

The outcomes: in Zapatista territories, infant mortality has declined, vaccination coverage has increased, and women's reproductive health has improved — all documented by researchers who have spent extended periods in the communities (Baronnet et al., 2011). These are not world-beating statistics. They are what a community of 300,000 indigenous people, operating with no state support and under intermittent military pressure, has achieved through democratic self-governance. The comparison is not with Sweden. The comparison is with the non-Zapatista indigenous communities in the same region of Chiapas, governed by the Mexican state, which consistently show worse health and education outcomes despite receiving government funding.

2.4 Marinaleda (Andalusia, Spain)

A village of approximately 2,700 people in one of the poorest regions of Spain. Its mayor, Juan Manuel Sánchez Gordillo, was first elected in 1979 and has been re-elected continuously since.

In the 1980s, Marinaleda's residents occupied the estate of the Duke of Infantado — 1,200 hectares of underused agricultural land owned by an absentee aristocrat. After years of strikes, marches, hunger strikes, and persistent nonviolent action, the regional government expropriated the land and transferred it to the village cooperative. Today, the land is farmed collectively, primarily producing olive oil and processed artichokes. Near-zero unemployment.

Housing: Residents build their own homes with municipal support. The village provides land and architectural plans. Residents provide labour (community self-build). Monthly cost: approximately 15 euros. Not rent — a mortgage payment that will eventually reach zero. The houses cannot be sold on the open market, preventing speculation. Housing is a right, not a commodity.

Governance: Town assemblies. Major decisions are put to the village. Gordillo refuses to take a salary above a worker's wage.

The honest criticism: Marinaleda's model depends heavily on one charismatic leader. Gordillo is the mayor, the organiser, the spokesperson, the strategist. The question — what happens when Gordillo goes? — is real and unanswered. The structural features (cooperative land, self-build housing, town assemblies) should, in theory, outlast any individual. Whether they will in practice is unknown. Marinaleda is a proof of concept for a village-scale cooperative economy. Whether it's a replicable model or a unique product of one extraordinary individual remains to be seen (Hancox, 2013).

Marinaleda Revisited: The Single-Leader Problem and Structural Succession

The Marinaleda case is instructive precisely because of its weakness. Every other functioning example in this review has rotation built in. Marinaleda does not. Gordillo has been mayor since 1979 — over four decades. He is democratically elected, repeatedly, by overwhelming margins. The village assemblies are genuine. The cooperative economy is real. The housing model works. But the structural question remains: what happens when the irreplaceable person is gone?

This is not an abstract concern. It is the single most common failure mode of progressive governance projects. Venezuela under Chávez. Cuba under Castro. Bolivia under Morales. Charismatic leaders who build real material improvements but create systems dependent on their personal presence. When they leave — by death, by coup, by term limit — the system they built either collapses or is captured by successors who lack the original's legitimacy.

The Zapatistas solved this by making leadership so temporary (one to two weeks) that no leader can become indispensable. Rojava solved it by making leadership so distributed (co-chairs at every level, recallable delegates, parallel women's councils) that no single person controls the system. Switzerland solved it by making the presidency rotate annually among seven council members, none of whom most citizens can name.

Marinaleda solved none of this. It relies on one man's continued competence, health, and willingness to serve. The structural features — the cooperative, the housing, the assemblies — could outlast Gordillo. But the question of whether they will is the question of whether Marinaleda has built an institution or a personality cult. The evidence is genuinely ambiguous. And that ambiguity is itself the lesson: rotation is not optional. It is the structural feature that distinguishes a movement from a system.

2.5 Swiss Cantons and Direct Democracy

Switzerland deserves its own section because it is the longest-running proof that direct democracy works at national scale — and because it operates on exactly the federation principles described in this paper.

Switzerland has 26 cantons, each with its own constitution, parliament, government, and courts. The federal government handles foreign affairs, defence, customs, and monetary policy. Everything else is cantonal or municipal. Education, healthcare, policing, taxation, infrastructure — all primarily cantonal or municipal responsibilities. The principle is subsidiarity: decisions are made at the lowest level capable of making them.

Direct democracy in practice: Swiss citizens vote in federal referendums approximately four times per year. Since 1848, there have been over 700 federal popular votes. Any citizen can propose a constitutional amendment (popular initiative) by collecting 100,000 signatures within 18 months. Any citizen can challenge a law passed by parliament (optional referendum) by collecting 50,000 signatures within 100 days. The result: citizens directly decide on tax rates, immigration policy, military spending, infrastructure projects, drug policy, and constitutional principles.

The Federal Council: Seven members, elected by the Federal Assembly (parliament). The presidency rotates annually. The president is primus inter pares — first among equals — with no special executive powers. Most Swiss citizens cannot name their current president. This is not a bug. It is a feature. The presidency is so ceremonial and so brief that no individual can accumulate the power, name recognition, or cult of personality that defines executive leadership in presidential systems.

Cantonal variation: Some smaller cantons still practise the Landsgemeinde — open-air assemblies where all citizens gather in a public square and vote by raising hands. Appenzell Innerrhoden and Glarus still do this. It is face-to-face direct democracy at scale, operating in the 21st century, in one of the wealthiest countries on Earth.

Outcomes: Switzerland is consistently ranked among the top countries in the world on quality of life, economic competitiveness, political stability, healthcare, education, and citizen satisfaction with government. It has no significant natural resources (no oil, no iron ore, limited agriculture). Its prosperity is entirely a product of its institutions — institutions built on federation, direct democracy, rotating leadership, and subsidiarity.

The Swiss model is not perfect. Women did not receive federal voting rights until 1971. The initiative process can be used for exclusionary purposes (the 2009 minaret ban). Wealth inequality exists. But as a demonstration that direct democracy, federation, and rotating leadership can govern 8.8 million people in a prosperous, stable, multilingual society for 178 years, Switzerland is unmatched. It is the standing refutation of every claim that "direct democracy can't work at scale," that "you need strong executive leadership," or that "people aren't informed enough to vote on policy."

They vote four times a year. On actual policy. And they have the best outcomes on almost every metric. The evidence is not ambiguous.

2.6 Other Cases, Briefly

Kibbutzim: The Israeli kibbutz movement, founded in the early twentieth century, provides a mixed record. At their peak, kibbutzim housed roughly 130,000 people across over 270 communities. Smaller kibbutzim (under 200-300 members) generally maintained stronger communal bonds and more effective collective governance. Larger ones tended to privatise functions, effectively breaking the community into smaller operational units. Many kibbutzim have undergone "privatisation" since the 1990s, introducing differential pay, private dining, and market-rate housing. The kibbutz experience suggests that communal living works at small scale but faces pressure to privatise as economic competition intensifies and second-generation members (who didn't choose the lifestyle) push for more individual autonomy (Gavron, 2000).

Hutterites: Still splitting at approximately 150. Still functioning after more than 500 years. The Hutterites are arguably the most successful long-running experiment in communal living in the Western world. Their colonies are economically productive, socially stable, and self-replicating (they grow until they split, then each half grows until it splits again). The mechanism is not Dunbar's number — it is the practical limit of consensus governance and mutual monitoring in a face-to-face community. The Hutterites' longevity suggests that the 100-150 person threshold for non-hierarchical community governance is real, stable, and not merely a modern phenomenon (Peter, 1987).

Auroville (Tamil Nadu, India): Founded in 1968 as a "universal town" where people of all nationalities would live in peace and progressive harmony. Current population approximately 3,200 across 120+ settlements. Auroville struggles with governance — its founding charter rejects conventional government, but the resulting governance vacuum has produced chronic internal conflicts over land use, development, and decision-making authority. The Indian government's 2021 intervention (through the Auroville Foundation Act amendments) exacerbated tensions. Auroville is an important cautionary tale: aspirational communities that reject structure often end up governed by whoever has the most informal power, which is worse than formal governance, not better.

Part III: Design Principles

What patterns emerge across all of these cases? What does the evidence — from hunter-gatherer bands, from primate social organisation, from Mondragon, from Rojava, from Chiapas, from Aboriginal Australian kinship systems, from Swiss cantons, from five centuries of Hutterite colonies — actually tell us about how to design human communities?

Eight principles. Not prescriptions. Observations about what all the working examples have in common.

3.1 Nested Autonomy

Every working large-scale community is actually a federation of small communities. Mondragon is not one organisation of 80,000 — it is 95+ autonomous cooperatives linked by shared infrastructure. Rojava is not one government of 4 million — it is thousands of communes linked by councils. The Zapatistas are not one territory of 300,000 — they are hundreds of community assemblies linked by rotating juntas. The Haudenosaunee were not one nation of 20,000+ — they were six nations of multiple clans linked by a Grand Council. Switzerland is not one state of 8.8 million — it is 26 cantons of 2,148 municipalities, each with its own governance, linked by a federal council that rotates its presidency annually.

The unit of real decision-making is always small: 25-150 people, depending on the context and the complexity of what's being decided. This is not because human brains can't handle more. It is because face-to-face consensus governance has a practical upper limit. You can't have a meaningful discussion with 10,000 people. You can have one with 50.

Scale comes not from making the unit bigger but from linking units together. Federation, not expansion. The individual unit stays small enough for everyone to know everyone, to speak and be heard, to hold each other accountable without formal surveillance. The federation handles coordination between units — shared resources, mutual defence, dispute resolution across boundaries.

3.2 Power Rotation (Not Power Elimination)

Here is where the strongest counterargument deserves honest engagement: there is real research showing that hierarchy aids coordination, and dismissing it would be intellectually dishonest.

Anderson and Brown (2010) found that status hierarchies in small groups reduce conflict, decrease uncertainty, and improve coordination — people figure out their roles faster when someone is clearly in charge. Halevy, Chou, and Galinsky (2011) showed that groups with hierarchical structure outperformed egalitarian groups on coordination tasks. Van Vugt (2006), working in evolutionary leadership theory, argued that humans evolved genuine followership tendencies — we are not just hierarchy-tolerant, we actively seek leadership in situations of uncertainty, threat, or coordination complexity. The military efficiency argument is real. The surgical team argument is real. When a building is on fire, you don't want a committee. You want someone who knows what to do, giving clear instructions, right now.

The research is clear: temporary, task-specific, accountable hierarchy works. The surgeon leads the operation. The captain leads the firefight. The experienced forager leads the hunt. This is not the same as permanent, general, unaccountable hierarchy — the king, the CEO for 20 years, the career politician. The research supporting hierarchy is almost entirely about the first kind. The systems that produce dysfunction are almost entirely the second kind.

The distinction is not hierarchy versus no hierarchy. It is rotating hierarchy versus permanent hierarchy. And here the evidence is striking.

D'Innocenzo, Mathieu, and Kukenberger (2016) conducted a meta-analysis of shared leadership across 50 studies and found that shared leadership — where leadership functions are distributed across team members rather than concentrated in one person — positively predicted team performance (ρ = .34). The effect was strongest for knowledge work and weakest for routine tasks. This maps exactly onto the coordination research: for simple, urgent tasks, concentrated hierarchy works. For complex, ongoing tasks — like governing a community — distributed and rotating leadership works better.

The Swiss Federal Council has operated on this principle since 1848: seven members, rotating presidency, each president serves one year and cannot serve consecutive terms. Switzerland has been running on rotating hierarchy for 178 years. It is one of the most stable, prosperous, and functional democracies on Earth. The Swiss don't even know who their current president is most of the time, because it doesn't matter — the role rotates, the institution persists, and no single person accumulates enough power to distort it.

The Zapatistas rotate their leaders every one to two weeks. The Haudenosaunee clan mothers could dehorn any sachem at any time. Mondragon cooperatives elect their management and can vote them out. Rojava's commune system sends recallable delegates, not permanent representatives.

Nearly all working examples of democratic community governance use power rotation or recall. The few exceptions — Marinaleda has had the same mayor since 1979 — are acknowledged by observers as vulnerabilities, not features. The consistency is striking. Permanent leadership creates permanent incentive misalignment — the leader's interest in maintaining power diverges from the community's interest in good governance. Rotation, recall, and term limits are not idealistic add-ons. They are structural requirements. Without them, every system drifts toward oligarchy. Robert Michels called this the "iron law of oligarchy" in 1911. The communities that work have found ways to break that law, and the mechanism is always the same: make power temporary, make leaders accountable, make recall easy.

The contrast with representative democracy as practised in most Western nations is instructive. A US senator serves six-year terms with no recall mechanism and no structural constraint on re-election. A Zapatista Junta member serves two weeks and returns to their community. Which system is more likely to produce leaders who serve the people rather than themselves?

3.3 Economic Base

You cannot have community without shared economic life. This is perhaps the most consistent finding across all the examples and the most ignored in contemporary community-building discourse.

Mondragon's cooperatives share ownership of their means of production. Marinaleda's residents collectively farm their expropriated land. The Zapatista autonomous zones are built on communal land tenure. Kibbutzim that maintained communal ownership stayed communal; those that privatised drifted apart. Hunter-gatherer bands share food — obligatory sharing is the foundation of band-level social life, not an optional add-on (Gurven, 2004).

Without a shared material stake, "community" is sentiment. It's a feeling you have about people you live near. It evaporates under economic pressure. When your neighbour's job loss doesn't affect your income, when your community's economic decline doesn't threaten your personal wealth, when you can exit to a better-resourced community at any time — your commitment is conditional. Shared economic life makes commitment structural. Your wellbeing is tied to the community's wellbeing. You can't extract value and leave. The incentive to invest in the commons is real because the commons is your livelihood.

3.4 Mandatory Inclusion

Rojava's mandatory gender parity — co-chairs at every level, 40% minimum quota, women's councils with veto power — is the most explicit version of this principle, but the pattern appears everywhere.

Haudenosaunee governance required that decisions pass through clan mothers. Not as a courtesy. As a structural requirement. A decision made without clan mother approval was not a decision. Aboriginal Australian kinship systems create obligations that cross gender, age, and community lines — the section system ensures that no group can be excluded from the social network.

The lesson: without structural inclusion, democracy becomes rule by the loudest, the most confident, the most socially privileged. "Everyone is welcome to participate" is not inclusion. Inclusion means: your participation is required, your voice is structurally guaranteed, and decisions cannot be made without you. The difference between invitation and requirement is the difference between aspiration and structure.

3.5 Face-to-Face at the Base

Every system described here puts direct, in-person democracy at the bottom layer. Zapatista community assemblies. Rojava communes. Mondragon general assemblies. Haudenosaunee clan meetings. Hutterite colony meetings. Hunter-gatherer band discussions. Swiss Landsgemeinde.

Not apps. Not online polls. Not "engagement platforms." People showing up in the same physical space, speaking, listening, arguing, compromising, and reaching decisions together.

This is not Luddism. It is an empirical observation. No functioning example of democratic community governance at any scale relies on digital participation at its base layer. The reasons are likely related to the findings from the digital network research: online interaction does not build the same depth of relationship as face-to-face contact. You can coordinate online. You can share information online. But the trust, accountability, and mutual understanding that consensus governance requires appear to need physical presence. The phone call data, the Facebook data, and the Twitter data all tell the same story: digital tools extend reach but do not deepen relationship.

This doesn't mean technology is irrelevant to community governance. It means technology serves coordination between groups, not deliberation within them. The commune meets in person. The coordination between communes can use whatever communication technology is available.

3.6 Scale Through Federation, Not Hierarchy

None of these systems scale by creating bosses. They scale by linking autonomous units through councils, congresses, or confederations.

Hierarchy compresses coordination costs by concentrating decision-making authority in fewer people. It works — military hierarchies, corporate hierarchies, and state bureaucracies can coordinate millions of people. But the cost is that the people at the bottom lose voice. Decisions are made for them, not by them. The efficiency of hierarchy comes from silencing most of the participants.

Federation preserves voice at the cost of speed. Decisions take longer because they must pass through multiple levels of consent. But the decisions, when made, have genuine legitimacy because the people affected by them participated in making them. The Haudenosaunee Grand Council could take weeks to reach consensus on a major decision. A unilateral executive could decide in minutes. The question is which kind of decision produces better outcomes over time — and the evidence from Mondragon, Rojava, the Zapatistas, Switzerland, and the Haudenosaunee suggests that slower, consensual decision-making produces more durable, more equitable, and more resilient outcomes than fast, top-down command.

The structural pattern: autonomous units at the base (25-150 people), linked by delegate councils at the middle level (representing 500-5,000 people), linked by confederal assemblies at the top (representing tens of thousands or more). Each level handles different kinds of decisions. The base handles daily life. The middle handles coordination. The top handles shared external challenges. Authority rests at the bottom. It is delegated upward, not devolved downward.

3.7 Circular Time

There is a concept that runs through every working example in this review that is so fundamental it usually goes unnamed. It is easiest to see by contrasting it with the thing Western modernity treats as default: linear time.

Linear time is a line. It goes forward. Progress. Growth. Accumulation. In linear time, a leader gains experience, builds networks, accrues power, and becomes increasingly entrenched. A company grows, acquires competitors, concentrates market share, and becomes a monopoly. A career advances from junior to senior to executive. The arrow points one way. Wealth compounds. Power compounds. Inequality compounds. Linear time, applied to governance, produces dynasties, career politicians, and permanent bureaucracies — because in a system oriented toward accumulation, whoever starts with more ends with more.

Circular time is a wheel. It returns. Aboriginal Australian societies organised their year around seasonal ceremonies — gatherings that brought dispersed bands together at predictable intervals, reaffirmed kinship obligations, resolved disputes accumulated since the last gathering, and then dispersed again. The cycle repeated. It had repeated for tens of thousands of years. The gathering was not a one-off event. It was a heartbeat.

The Zapatista two-week rotation is circular time applied to power. You serve. You return to the field. Someone else serves. They return. The wheel turns. No one accumulates. No one becomes indispensable. No one forgets what it is like to be an ordinary person because they are an ordinary person for most of the year. Hutterite colony fission is circular time applied to community: grow, split, grow, split. The colony never becomes a city. The wheel turns at ~150 and begins again.

Hunter-gatherer bands dispersed and aggregated with the seasons — small groups in lean months when resources were scarce, larger aggregations when resources were abundant (Kelly, 2013). This was not just logistics. The seasonal aggregation was when marriages were arranged, ceremonies performed, knowledge exchanged, and social networks maintained. The dispersal-aggregation cycle was the governance system. The circle itself — the predictable return, the rhythm — was what held the larger social structure together without any permanent coordinating authority.

The Swiss vote four times a year. The rhythm is the structure. The return to the ballot is the mechanism that keeps authority distributed. Every three months, the system resets. Every three months, the citizens are sovereign again — not in theory, but in practice, standing in a booth or raising a hand in a public square.

Why does this matter for community design? Because the dominant Western model of governance assumes linear time. A government is formed, it persists, it grows, it accumulates institutional power. Elections are theoretically cyclical, but the institutions are permanent. The bureaucracy never rotates. The judiciary serves for life. The security apparatus grows monotonically. The entire structure is oriented toward accumulation, not return.

Every working example in this review operates on circular time. Rotation, seasonal gathering, cyclical fission, ceremony as social infrastructure, power that returns to the base before it can compound. This is not nostalgia for a pre-modern past. It is a design observation: the structures that prevent tyranny are structures that build return into their architecture. The circle is not primitive. The circle is the mechanism that prevents the line from becoming a cage.

3.8 The Peck-Dunbar Bridge

Here is where the evidence synthesises into a design insight.

True community — vulnerability, authentic encounter, mutual acceptance — requires a very small group. Peck's estimate of 10-15 people is consistent with Dunbar's sympathy group of 15. This is where the deepest human social capacity operates. It is the group where you can be genuinely known.

Effective governance — collective decision-making, resource allocation, dispute resolution, mutual accountability — requires a larger group: 50-150 people. This is the band, the cooperative, the colony, the commune. Large enough to be economically viable and socially diverse. Small enough for face-to-face deliberation and mutual monitoring.

Scaling — addressing challenges that no single community can handle alone (defence, infrastructure, ecological management, economic coordination) — requires federation: linking governance-scale groups into larger structures through delegates, councils, and confederations.

The design task is connecting these layers without losing what makes each one work. The 10-person intimacy group needs to be nested inside the 50-150 person governance group needs to be federated into larger structures. Each layer must maintain its own integrity — the intimacy group must stay intimate, the governance group must stay governable, the federation must stay federated (not centralised).

This is what Aboriginal Australian kinship systems do. This is what the Haudenosaunee Confederacy did. This is what Mondragon, Rojava, Switzerland, and the Zapatistas do. The structural pattern is the same across cultures, continents, and centuries. The specific implementations differ enormously — cooperatives are not communes are not kinship systems are not confederacies are not cantons. But the architecture is consistent: small groups for intimacy, medium groups for governance, federation for scale.

Part IV: The Federation Pattern — Expanded Case Analysis

The eight principles above are not derived from theory. They are derived from observation. But observation requires depth. This section expands on three cases that deserve closer examination because they address specific objections that arise whenever federation is proposed as an alternative to centralised governance.

The objections are predictable:

1. 1. "It only works in small, homogeneous communities." Rojava governs 4 million people across four major ethnic groups.
2. 2. "It can't survive without state support." The Zapatistas have been governing for 30 years without a single peso from the Mexican government.
3. 3. "Direct democracy is too slow for modern complexity." Switzerland holds 4 referendums per year on tax policy, immigration, military spending, and constitutional amendments. It is one of the most competitive economies on Earth.
4. 4. "It depends on charismatic leadership." Marinaleda does. The Zapatistas, Rojava, and Switzerland do not. The difference is rotation.

4.1 Swiss Cantons: 178 Years of Proof

The Swiss objection-killer deserves emphasis because it neutralises the most common dismissal: "that's a small tribal thing, it can't work in a modern industrialised country."

Switzerland is a modern, industrialised, multilingual, multi-ethnic country of 8.8 million people. It has four official languages (German, French, Italian, Romansh). It has no natural resources to speak of. It is landlocked, mountainous, and surrounded by much larger neighbours who have historically shown interest in annexing parts of it. It has been a federation with direct democratic mechanisms since 1848 — 178 years.

The cantonal structure is federation in its purest modern form. Each canton is essentially a small state. The canton of Appenzell Innerrhoden has 16,000 people and still governs by open-air assembly. The canton of Zurich has 1.6 million and uses ballot-box referendums. Both are part of the same federal structure. The system accommodates communities of 16,000 and 1,600,000 under the same constitutional framework, because the framework is built on subsidiarity — decisions at the lowest level possible — not uniformity.

The economic argument is decisive. Switzerland is consistently among the top 5 countries in GDP per capita, innovation indices, and economic competitiveness rankings. It achieved this without a strong executive, without natural resources, and without a permanent political class. The Federal Council's seven members are almost unknown outside Switzerland. The presidency rotates annually. No Swiss citizen has ever needed to know their president's name to prosper.

This is what Goal 1 — "Fire all politicians. You vote on everything." — looks like after 178 years of operation. It is not utopian. It is Swiss.

The Connection

The 14 goals require a governance structure. You cannot implement direct democracy, redesign education, restructure work, reform justice, and reshape public space through conventional representative government. The institutions that would need to change are the ones making the decisions. They will not vote to dissolve themselves.

But the governance structure needed is not hypothetical. It exists. It is operating right now — in the Basque Country, in northeastern Syria, in Chiapas, in Swiss cantons, in Aboriginal communities that have been running functional governance systems for 60,000 years. The structural principles are documented, tested, and resilient. Small autonomous groups. Rotating leadership. Shared economic base. Mandatory inclusion. Face-to-face deliberation. Federation for scale.

The question is not invention. It is adoption. The animal already knows how to organise. It's been doing it for a very long time, in every environment, on every continent, under conditions far more challenging than the ones most of us face. What it needs is not a new theory of social organisation. What it needs is an enclosure that lets it do what it already knows how to do.

The evidence is not ambiguous. The examples are not obscure. The principles are not complicated. The only thing standing between the current structure and a functional one is the unwillingness of existing power structures to yield. Which is, of course, exactly the problem that every community described in this review has already solved — by building the alternative and making the old structure irrelevant.

References

Primary Sources — Dunbar & Critiques

1. 1. Dunbar, R.I.M. (1992). "Neocortex size as a constraint on group size in primates." Journal of Human Evolution, 22(6), 469-493.

1. 2. Lindenfors, P., Wartel, A., & Lind, J. (2021). "'Dunbar's number' deconstructed." Biology Letters, 17(5), 20210158. https://doi.org/10.1098/rsbl.2021.0158

1. 3. Dunbar, R.I.M. (2020). "Structure and function in human and primate social networks: implications for information flow." Proceedings of the Royal Society B, 287(1929), 20201035.

1. 4. Hill, R.A. & Dunbar, R.I.M. (2003). "Social network size in humans." Human Nature, 14(1), 53-72.

1. 5. Dunbar, R.I.M. (2016). "Do online social media cut through the constraints that limit the size of offline social networks?" Royal Society Open Science, 3(1), 150292.

1. 6. Gonçalves, B., Perra, N., & Vespignani, A. (2011). "Modeling Users' Activity on Twitter Networks: Validation of Dunbar's Number." PLoS ONE, 6(8), e22656.

Primate Social Organisation

1. 7. Mitani, J.C. (2009). "Male chimpanzees form enduring and cooperative social bonds." Animal Behaviour, 77(3), 633-640.

1. 8. Wilson, M.L., Boesch, C., Fruth, B., et al. (2014). "Lethal aggression in Pan is better explained by adaptive strategies than human impacts." Nature, 513, 414-417.

1. 9. Furuichi, T. (2011). "Female contributions to the peaceful nature of bonobo society." Evolutionary Anthropology, 20(4), 131-142.

Mating Systems & Cross-Cultural Data

1. 10. Murdock, G.P. (1967). Ethnographic Atlas. Pittsburgh: University of Pittsburgh Press.

1. 11. Henrich, J., Boyd, R., & Richerson, P.J. (2012). "The puzzle of monogamous marriage." Philosophical Transactions of the Royal Society B, 367(1589), 657-669.

Hunter-Gatherer & Indigenous Social Organisation

1. 12. Hill, K.R., Walker, R.S., Bozicevic, M., et al. (2011). "Co-Residence Patterns in Hunter-Gatherer Societies Show Unique Human Social Structure." Science, 331(6022), 1286-1289.

1. 13. Berndt, R.M. & Berndt, C.H. (1999). The World of the First Australians: Aboriginal Traditional Life Past and Present. 5th ed. Canberra: Aboriginal Studies Press.

1. 14. Johansen, B.E. (1982). Forgotten Founders: Benjamin Franklin, the Iroquois, and the Rationale for the American Revolution. Ipswich, MA: Gambit.

1. 15. Grinde, D.A. & Johansen, B.E. (1991). Exemplar of Liberty: Native America and the Evolution of Democracy. Los Angeles: American Indian Studies Center, UCLA.

Community, Philosophy & Social Theory

1. 16. Peck, M.S. (1987). The Different Drum: Community-Making and Peace. New York: Simon & Schuster.

1. 17. Buber, M. (1923/1970). I and Thou. Trans. Walter Kaufmann. New York: Charles Scribner's Sons.

1. 18. Tutu, D. (1999). No Future Without Forgiveness. New York: Doubleday.

1. 19. Michels, R. (1911/1962). Political Parties: A Sociological Study of the Oligarchical Tendencies of Modern Democracy. Trans. Eden and Cedar Paul. New York: Free Press.

1. 20. Putnam, R.D. (2000). Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster.

Mondragon Corporation

1. 21. Errasti, A., Bretos, I., & Etxezarreta, E. (2017). "What do Mondragon coops do differently?" In The Oxford Handbook of Mutual, Co-Operative, and Co-Owned Business. Oxford University Press.

1. 22. Whyte, W.F. & Whyte, K.K. (1991). Making Mondragon: The Growth and Dynamics of the Worker Cooperative Complex. 2nd ed. Ithaca: Cornell University Press.

1. 23. AFL-CIO (2023). Executive Paywatch. https://aflcio.org/paywatch

Rojava / Democratic Confederalism

1. 24. Knapp, M., Flach, A., & Ayboga, E. (2016). Revolution in Rojava: Democratic Autonomy and Women's Liberation in Syrian Kurdistan. London: Pluto Press.

1. 25. Öcalan, A. (2011). Democratic Confederalism. Cologne/London: International Initiative.

1. 26. Bookchin, M. (1982). The Ecology of Freedom: The Emergence and Dissolution of Hierarchy. Palo Alto: Cheshire Books.

Zapatistas

1. 27. Stahler-Sholk, R. (2007). "Resisting Neoliberal Homogenization: The Zapatista Autonomy Movement." Latin American Perspectives, 34(2), 48-63.

1. 28. Baronnet, B., Mora, M., & Stahler-Sholk, R. (Eds.) (2011). Luchas "muy otras": Zapatismo y autonomía en las comunidades indígenas de Chiapas. Mexico City: UAM-Xochimilco / CIESAS / UNACH.

Marinaleda

1. 29. Hancox, D. (2013). The Village Against the World. London: Verso.

Switzerland

1. 30. Linder, W. & Mueller, S. (2021). Swiss Democracy: Possible Solutions to Conflict in Multicultural Societies. 4th ed. London: Palgrave Macmillan.

1. 31. Kriesi, H. & Trechsel, A.H. (2008). The Politics of Switzerland: Continuity and Change in a Consensus Democracy. Cambridge: Cambridge University Press.

1. 32. Kobach, K.W. (1993). The Referendum: Direct Democracy in Switzerland. Aldershot: Dartmouth.

Kibbutzim

1. 33. Gavron, D. (2000). The Kibbutz: Awakening from Utopia. Lanham, MD: Rowman & Littlefield.

Hutterites

1. 34. Peter, K.A. (1987). The Dynamics of Hutterite Society: An Analytical Approach. Edmonton: University of Alberta Press.

Hierarchy, Leadership & Rotation

1. 35. Anderson, C. & Brown, C.E. (2010). "The functions and dysfunctions of hierarchy." Research in Organizational Behavior, 30, 55-89.

1. 36. Halevy, N., Chou, E.Y., & Galinsky, A.D. (2011). "A functional model of hierarchy." Organizational Psychology Review, 1(1), 32-52.

1. 37. Van Vugt, M. (2006). "Evolutionary Origins of Leadership and Followership." Personality and Social Psychology Review, 10(4), 354-371.

1. 38. D'Innocenzo, L., Mathieu, J.E., & Kukenberger, M.R. (2016). "A meta-analysis of different forms of shared leadership–team performance relations." Journal of Management, 42(7), 1964-1991.

1. 39. Freeman, J. (1972). "The Tyranny of Structurelessness." The Second Wave, 2(1).

Food Sharing & Cooperation

1. 40. Gurven, M. (2004). "To Give and to Give Not: The Behavioral Ecology of Human Food Transfers." Behavioral and Brain Sciences, 27(4), 543-559.

Circular Time & Seasonal Aggregation

1. 41. Kelly, R.L. (2013). The Lifeways of Hunter-Gatherers: The Foraging Spectrum. 2nd ed. Cambridge: Cambridge University Press.

Appendix A: Confidence Assessments

Appendix B: Cross-References to the OMXUS Research Series

This paper is part of a broader research programme examining the structural conditions for human flourishing across fourteen domains. The following papers in the OMXUS Research Series intersect directly with the findings presented here.

Governance & Democracy

Justice & Safety

Health & Environment

Economy & Work

Identity & Technology

Human Behaviour & Psychology

The Books

Total word count: approximately 16,800

OMXUS Research Series — Social Group Scaling First draft for The Zookeeper, Chapter 11. Expanded to unified thesis.

"The animal already knows how to organise. It has always known. What it needs is an enclosure that lets it."