The Second Circle

Volume I — Transition: From Here to the Living Economy

THE PLATE

A woman and a man talking across a small white table over lunch, a bare wall and a bright window beside them. — Plate I.07The Third Chair.The third chair is the one that decides. A room with two people in it is a conversation. A room with three is a practice that can continue without any particular one of them.

THE LETTER

You have the number. Chapter I.01 got you the ninety days, the signed baseline, the one page given to one person. It worked, or it worked well enough that you are still here. And now you have run into the thing nobody warns you about, which is that being right, alone, with evidence, changes almost nothing.

You will have noticed the shape of it already. The people who agreed with you fastest turned out to be the least useful. They agreed before the evidence and they would have agreed without it, which means their agreement carries no information and — more painfully — carries no weight with anyone else. Meanwhile the person whose support would actually move the department has read your page, said something mild, and gone back to work.

This chapter is about that gap and how to cross it. Not by persuading harder. By understanding, arithmetically, who the second and third people are, what shape the group has to have, and what the minimum written agreement is before a group has anything to govern at all.

The claim underneath everything that follows is this. A practice does not spread the way a fact spreads. A fact needs one contact. A practice that costs something — time, standing, the risk of looking foolish in a meeting — needs two independent people you trust doing it where you can see them. That single difference changes what a good network looks like, and it changes it in a direction that will feel wrong to you, because it means the wide, well- connected, high-reach shape you have been building is the wrong shape and a small dense one is the right one.

So: the second circle. Three to six people, chosen against a rule rather than by enthusiasm, tied to each other and not only to you, with four written agreements and a named successor. That is the whole chapter, and the arithmetic is going to be unusually blunt about why.

— The Editors

DISCOVERY

What is already working

Start where it has already been done, because the structure has been discovered independently, in different centuries, by people solving unrelated problems — and they converged on almost the same shape.

Alcoholics Anonymous, from 1935. The origin is precise and worth holding. Bill Wilson, newly sober and in Akron on business, was not trying to build an organisation. He was trying not to drink, and he had worked out that talking to another alcoholic was the only thing that reliably kept him from it. He found Dr Bob Smith — a resistant, sceptical, locally respected surgeon, the exact opposite of an enthusiast — and the thing began with two people, one of whom did not want to be there. What is remarkable is what they did not do next. They grew by splitting: each group autonomous, each small enough to meet in a room, each free to run its own affairs. The Twelve Traditions, written down in 1953 after eighteen years of practice, are almost entirely a document about boundaries, autonomy and the refusal of central authority. AA's growth mechanism is the deliberate manufacture of second circles.

Grameen's five. The Grameen Bank lends to individuals, but only to individuals inside a self-formed group of five who are not close relatives. The group is not the legal guarantor it is often described as; its real work is reinforcement and visibility. Five people who chose each other, who meet weekly, and in front of whom a repayment either happens or visibly does not. Yunus did not derive this from a theory of contagion. He arrived at it because nothing else worked.

Buurtzorg, from 2006. The Dutch neighbourhood-nursing organisation began with a handful of nurses and grew into one of the largest home-care providers in the Netherlands without adding a management layer. The design constraint that did the work is a number: a team is capped at about twelve nurses, and when it exceeds that it splits. Not because twelve is sacred, but because a team above it stops being a group that can hold its own decisions and starts needing someone to hold them on its behalf.

Törbel, from 1483. Ostrom's Swiss alpine case is the oldest governance minimum on this list. The villagers drew up articles of association for the common alp, and the first provision is a boundary: only those already holding rights may use the pasture. Five centuries of that commons rest on that one line. Ostrom's first design principle — clearly defined boundaries — is not an abstraction distilled from theory. It is what these documents actually say first.

And one from the diffusion literature itself. Coleman, Katz and Menzel's 1966 study of how a new antibiotic spread among physicians in four Illinois towns found that doctors integrated into professional networks — consulted by colleagues, named as friends — adopted earlier than doctors practising in isolation, and that adoption traced the structure of professional relationships rather than the distribution of medical journals. The information was equally available to everyone. The practice was not.

Five cases. One pattern, and it is the pattern this chapter is built on: every one of them grew by making small dense groups and letting them multiply, and not one of them grew by enlarging the original group.

THE ARITHMETIC

What works, what does not, and where the line sits

First: there is no second enthusiast. There is arithmetically no second enthusiast.

Rogers' adopter categories are a normal curve cut at the mean and at one and two standard deviations: innovators 2.5 percent, early adopters 13.5 percent, early majority and late majority 34 percent each, laggards 16 percent. Turn those shares into expected heads and the consequence is immediate.

  band              6      12      25      40      80     400
  innovators     0.15    0.30    0.62    1.00    2.00   10.00
  early adopters 0.81    1.62    3.38    5.40   10.80   54.00
  early majority 2.04    4.08    8.50   13.60   27.20  136.00

You need forty people before the expected number of innovators reaches one, and eighty before it reaches two. In any group small enough to sit round a table, you are the only innovator there is. The second adopter is therefore obliged, by arithmetic and not by preference, to come from the early-adopter band — one in every 7.4 people — and the early adopter is a categorically different person from you. Rogers is explicit about it: innovators are cosmopolite, socially marginal to the local system, and not the person colleagues check with. The early adopter is the person colleagues check with. Their thresholds are near the local norm; their judgement is trusted precisely because they are not known for early judgement. Geoffrey Moore named the same discontinuity from the commercial side and called it a chasm; what the arithmetic adds is that the chasm is not a gap in the market but a gap in a distribution, and it is crossed by finding one person rather than by building one bridge.

This is why the people who agreed with you immediately were no use. They were not the second adopter. They were other innovators, and there is almost never more than one of you in a room.

Second: what the second person is actually worth.

The Bass model gives the pull on the next undecided person as (p + qF)(1 − F), where p is the coefficient of innovation — the rate at which people adopt from outside influence — and q the coefficient of imitation. The mean values across 213 published applications are p = 0.03 and q = 0.38. Put those in a circle of twelve.

  adopters   F        p + qF     change vs previous
      0     0.000     0.0300           —
      1     0.083     0.0617       +105.6 %
      2     0.167     0.0933        +51.4 %
      3     0.250     0.1250        +33.9 %
      4     0.333     0.1567        +25.3 %

Each new adopter adds the same absolute increment, q/N = 0.0317, to a base that keeps growing. So the first person who joins you slightly more than doubles the pull on everyone else, and no recruit after them will ever again change the rate by that much. The proportional value of a recruit falls monotonically from the first one onward. This is the clearest thing in the chapter and it is worth being greedy about: the second adopter is the single most valuable person you will recruit in the entire programme.

It also says something about broadcasting. With these coefficients, peak adoption arrives at t* = ln(q/p)/(p+q) = 6.19 periods, by which point 46.1 percent of the population has adopted. The same elapsed time on p alone — that is, on announcement, presentation and memo, with no imitation at all — reaches 17.0 percent. Word of mouth is doing 2.72 times the work that your communication is. If you are choosing between another all-hands and one more conversation with one more person, the arithmetic is not close.

Third — the Balenciaga cut, and it will annoy you.

Take two circles of twelve. A threshold is the number of others who must already be doing a thing before this person will.

  circle A   0  1  2  3  4  5  6  7  8  9 10 11     mean 5.500
  circle B   0  2  2  2  2  2  2  2  2  2  2 11     mean 2.583

Circle A goes to twelve of twelve. The zero starts, the one follows, the two follows, and it runs all the way up.

Circle B stalls at one. Nobody moves after you, ever.

Now read the means. Circle B's average threshold is 2.583 against circle A's 5.500 — circle B is less than half as demanding, which is to say more than twice as eager. Ten of its twelve members need only two other people to be doing the thing. Almost everybody in that room is nearly persuaded. And the circle is dead, because there is no route from one to two: nobody in it has a threshold of exactly 1, so the second adopter never exists and the ten who were waiting for two wait forever.

Swap that one reluctant member at 11 for a single person at threshold 1 and the same circle goes to twelve of twelve.

So, stated plainly, and it is the one cut in this chapter: the average enthusiasm of a group tells you nothing about whether it will move, and it routinely points the wrong way. A group can be far keener than another group in every summary statistic you would naturally compute and be structurally incapable of starting. What decides it is not the level of the distribution but whether it has a gap — a value of k at which nobody sits. No amount of persuasion aimed at the group as a group closes a gap, because the group is not the unit. The gap is the unit, and you fill it with a person.

And then the operational question: how many committed people fill a gap?

  circle C   0  3  3  3  4  5  6  7  8  9 10 11     mean 5.750

    1 committed  ->   1 of 12    stalls
    2 committed  ->   2 of 12    stalls
    3 committed  ->  12 of 12    cascades

One is a founder. Two is a pair of enthusiasts and it still stalls. Three is where the circle carries itself. And the tipping-point experiment of Centola, Becker, Brackbill and Baronchelli — in which a committed minority overturned an established naming convention once it passed roughly 25 percent of the group — gives 3.0 for a group of twelve. Two methods, sharing no assumptions, returning the same integer. The agent-based literature puts the threshold lower, nearer 10 percent; in a room of twelve that is one person, which is the founder, which is where you already were. Where the two disagree, take the larger claim, because the cost of under-recruiting is the whole programme. Valente's refinement is the one to carry into practice: a threshold is relative to a person's own network, not to the organisation, so the number that matters is how many of their contacts have adopted, never how many of yours.

Fourth: what does not work, and it is a shape you are probably already in.

Centola and Macy's result is that a complex contagion — one that needs reinforcement from two or more independent contacts, which is every costly or socially risky behaviour — propagates differently from information, and the long weak ties that speed information up can slow practice down. Make that concrete with the worst case.

  shape                 reach   neighbours each person has   second signal?
  star, founder + 5        6                 1                    no
  star, founder + 11      12                 1                    no
  star, founder + 49      50                 1                    no
  star, founder + 199    200                 1                    no

If you recruited everyone privately, one conversation at a time, and they know you but not each other, you have built a star. Your own degree grows without limit. Every other person's degree stays at one, forever. A star transmits information perfectly and cannot transmit a practice at all, and adding people does not fix it at any size. This is the honest negative the chapter turns on: widening the circle, done in the ordinary way, drives adoption toward zero while every visible metric — reach, attendance, mailing list, awareness — improves.

The repair is redundancy. On a ring lattice where each person is tied to their k nearest, the clustering coefficient is 3(k−2)/(4(k−1)):

  k = 2   C = 0.000    reinforcing pairs among one's neighbours   0.0
  k = 4   C = 0.500    reinforcing pairs among one's neighbours   3.0
  k = 6   C = 0.600    reinforcing pairs among one's neighbours   9.0

At k = 4, half of a person's neighbours are tied to each other, so a signal reaching one of them reaches a second by a route you did not have to take. That redundancy is what a wide, flat, efficient network is designed to remove. The thing that makes a network good at information is precisely what makes it bad at practice.

Fifth: the ceiling, which is real.

Dunbar's layers come at 5, 15, 50 and 150, scaling by a ratio close to three. Set them beside the coordination load.

  size   pairwise channels   minutes each in a 60-minute meeting
     5                  10        12.0
     6                  15        10.0
     8                  28         7.5
    12                  66         5.0
    15                 105         4.0
    50               1,225         1.2

Channels rise as n² while reinforcement rises as n. So the circle has an arithmetic ceiling and it arrives at about twelve — sixty-six channels, five minutes each. Buurtzorg's cap was not a philosophy. It was this.

Sixth, and last: the base rate is against you and should be stated.

Rogers' Peruvian case: a two-year campaign to persuade households in Los Molinos to boil drinking water, with a resident health worker, repeated home visits and a correct scientific claim, achieved 11 adoptions out of 200 households — 5.5 percent. The claim was true. The evidence was good. Compatibility with the local system beat both. If your programme fails, the most likely reason is not that you were wrong; it is that the practice did not fit the life of the person you offered it to, and no quantity of proof addresses that.

DREAM

What becomes ordinary

Describe it in the present tense, because a dream in the future tense is a wish.

In an organisation where this has already happened, initiatives do not have founders. They have charters, and a charter names three people, not one. The first question asked of any new proposal in the monthly review is not what is the return — that comes second — but who else is doing this, and do they know each other? An answer naming one person is not rejected. It is simply sent back with a date to return.

The people who carry things are not the people who are excited. They are the people others check with, and the organisation knows who they are because it has troubled to find out — quietly, by asking whose desk people stop at, whose opinion gets quoted second-hand, whose scepticism ends an argument. That list exists, and it is not published, because publishing it would destroy it.

Groups split rather than grow. When a working circle reaches about twelve it divides, and the division is a normal event with a ritual rather than a sign of failure. Both halves keep the charter; each names its own second and third owner. Nobody experiences this as demotion out of the important room, because there is no important room — there are eleven ordinary ones.

Every circle has a written minimum, and it fits on one page: who is in, how we decide, who watches the shared thing, and what happens when somebody does not do what they said. Four lines. Nobody finds this bureaucratic, because everyone has been in a group that lacked them and remembers what that cost.

And when the person who started something moves to another part of the business, the thing does not wobble. The successor was named eighteen months earlier, has been running alternate meetings for a year, and takes over on a Tuesday without an announcement. The most reliable sign of a healthy initiative here is that nobody can tell you whose it was.

DESIGN

The second circle, built

Step one: name the gap, not the group.

Before recruiting anyone, write down every person whose participation would matter, and beside each one an honest estimate of their threshold — how many others do they need to see doing this first? You will not be accurate. You will be accurate enough to see the shape, and the shape is the point. Look for the gap: the number where nobody sits. That number is your recruitment target, and it is almost always low — one, two or three.

Step two: choose against the rule, not the feeling.

Your second and third members meet four criteria, in this order.

They are checked with. Other people ask their opinion unprompted.
Their threshold is low but not zero. Somebody at zero is another innovator and adds no information. Somebody at one or two is the second adopter; they needed to see exactly one or two others, and you are one.
They are visible to each other. This is the criterion everyone skips. If your second and third members do not already know each other, your first job is not recruitment — it is introduction, and it is worth an entire month.
They have something to lose and are willing to lose it. A person with no standing at risk provides no reinforcement, because their participation tells observers nothing.

Step three: make the redundancy on purpose.

Every member should be able to see at least two others doing the practice without going through you. Concretely: the circle meets together rather than in a series of one-to-ones; work is done in pairs that rotate; the artifacts carry two names. The operating metric is the reinforced-member ratio — the share of members with two or more adopting ties they did not get from the founder.

  circle of  6,  6 reinforced   100 %
  circle of 12,  8 reinforced    67 %
  circle of 12,  3 reinforced    25 %
  circle of 25,  5 reinforced    20 %

Below about a quarter, you have a star wearing a circle's name.

Step four: the governance minimum, which is four lines.

Ostrom's eight design principles are for an established commons. A circle of six does not have a commons yet; it has a practice and a small shared budget. Four of the eight are the minimum, and the other four arrive later when there is something to defend.

Principle	The one line it becomes
Clearly defined boundaries	Who is in this circle, by name, and how somebody joins.
Collective-choice arrangements	How we decide. Quorum, and what happens on a tie.
Monitoring	Who looks at the shared thing, how often, and reports to whom.
Graduated sanctions	What happens the first time somebody does not do what they said — and it is small.

Four sentences. Write them, circulate them, have everyone say yes out loud in the same meeting. The document is not the point; the meeting in which people assent in front of each other is the point, because that is itself a complex contagion event — everyone receives reinforcement from everyone simultaneously.

Jo Freeman's argument in 1972 remains the sharpest statement of the alternative: a group that refuses explicit structure does not become structureless. It develops an informal elite that cannot be held to account, cannot be joined, and cannot be succeeded. Writing four lines is cheaper than that, by a great deal.

DESTINY

How it holds when you stop pushing

The circle survives your attention moving elsewhere if, and only if, three things are true before it moves.

A named successor who is already doing the job. Not a designated deputy — a person who chairs alternate meetings, whose name is on half the artifacts, and whom people already address directly. Succession that begins at handover has already failed; the network has to have been rewired months earlier.

A written minimum that somebody other than you can point at. The four lines. Their function is to answer a question in your absence.

A split rule stated in advance. At what size does this divide, and who decides? Stating it before it is needed removes the only conversation that reliably turns a growing group into a faction fight.

Now the honest part, because the failure mode here is specific and well documented. The thing that kills a second circle is success.

Students for a Democratic Society grew from a few hundred members in 1962 to something on the order of a hundred thousand by 1969, and disintegrated at its Chicago convention that June into factions that never reassembled. Francesca Polletta's study of participatory-democratic movements traces the same arc repeatedly: groups whose decision-making worked beautifully at twenty could not be scaled to two hundred, and the attempt to scale them — rather than to multiply them — is what broke them. The Occupy general assemblies of 2011 ran the experiment again with the same result.

The mechanism is not mysterious and the arithmetic is above. Coordination cost rises as n². Reinforcement rises as n. Somewhere around Dunbar's second layer those two lines cross, and after the crossing every additional person makes the group slower, less able to reinforce anybody, and more dependent on an informal elite that nobody elected. A circle that doubles has not doubled its strength; it has quadrupled its coordination load.

The response is the one AA found by accident and Buurtzorg built on purpose: do not widen. Multiply. Split at twelve, keep the charter, name the new second and third owners, and let the two circles be tied to each other by two or three people rather than merged. That structure — dense clusters connected by a few bridges — is the one that carries a complex contagion furthest, and it is the only growth pattern in this chapter that does not degrade what it grows.

DELIGHT

What it feels like

There is a particular moment and you will know it when it happens. Your second person explains the thing back to you, in a meeting you did not arrange, to somebody you have never persuaded — and they get one part of it slightly wrong, and the wrong part is better than what you had.

That is the whole pleasure of this chapter in one event. It is the sound of something ceasing to be yours. Not being taken from you — metabolised. The idea has gone into another mind and come out with somebody else's grain in it, and it is now more durable than anything you could have made alone, because it exists in two versions and neither depends on the other.

The second pleasure is quieter and arrives later. You miss a meeting. Nothing happens. The work goes on, the notes come round, a decision was made that you would have made differently and is fine. The relief of that is physical. You had not noticed how much of the weight you were carrying was simply the weight of being the only one who knew.

OPERATIONALIZE THIS

At the level of finance

Everything above becomes real when it acquires a signature. Here is the instrument.

The structure: a two-party internal cost-sharing charter with a co-signature drawdown.

It is a small joint operating agreement between two cost centres — yours and one other — funding a shared practice, with the defining feature that no drawdown is possible on a single signature. The second owner is not created by goodwill; they are created by a bank rule.

The mechanics.

Size. Small enough that both signatories are already within their delegated authority. If it needs a third approval it is the wrong size.
The parties. Two named cost centres, and within each, two named individuals — the owner and the successor. Four names, and the document is void if any one of them is vacant for more than sixty days. That clause is what forces succession planning to actually happen.
Drawdown. Joint signature, always, with no delegation and no exception for small amounts. The exception is how the star re-forms.
Baseline. Fixed for the full term, stated on page one, signed by both controllers. A charter whose baseline resets annually is a charter that rewards nobody after year one.
The four governance lines. Boundary, decision rule, monitoring, graduated sanction — written into the charter itself rather than kept in a separate document nobody reads.
Split clause. At a stated headcount the circle divides and the charter clones, each clone naming its own four people. Write the number in.
Sunset. Twenty-four months, with renewal requiring both controllers to re-sign. A charter that renews silently is an entitlement.

The balance-sheet treatment. Contributed time is an expense in both cost centres, allocated on the agreed basis, unless the circle produces an identifiable asset — a tool, a documented process, a dataset — in which case the usual internally-generated-intangible recognition tests apply: identifiable, controlled, probable future benefit, reliably measurable. Most circles fail the fourth test in year one and that is correct. Do not capitalise a culture. Capitalise a deliverable.

The counterparty. The second cost centre's controller, not its director. The director agrees in principle and forgets; the controller sets up the allocation and the thing exists. Take the charter to the controller with the allocation basis already drafted and the conversation is ten minutes.

The number that decides it. Founder risk, priced.

  median US employee tenure (BLS, Jan 2024)            3.9 years
  implied constant hazard, ln 2 / median               0.1777 / yr
  P(founder's attention gone within 3 years)           41.3 %

  cost to re-found the initiative from zero          150,000
  expected loss per year   =  0.413 x 150,000 / 3     20,663
  charter cost  =  2 owners x 2 h/mo x 12 x 65 + 2,000 admin
                                                       5,120
  ------------------------------------------------------------
  cover ratio                                           4.04 x

The second owner is bought for 25 percent of the risk it retires. State it in that register and you are not asking for support for a working group. You are retiring a single-point-of-failure exposure at a four-times cover ratio, which is a conversation a treasurer has every week.

One honesty about that figure. Treating departure as a constant hazard from a median tenure is an approximation — real hazards run higher in year one and lower at long service, and "attention moving" is broader than "leaving". Substitute your own organisation's number when you have it. The conclusion is not sensitive: the charter clears at any founder-loss probability above about 10 percent.

The first ninety days on a page.

Day	Action	Artifact
1–15	Threshold map: everyone who matters, with an estimated threshold	The map, and the gap it exposes
16–30	Identify two candidates in the gap who are checked with	Two names, with the evidence for each
31–45	Introduce them to each other, on a real task, without you present	A piece of work with two names on it
46–60	Draft the four governance lines; hold the meeting where everyone assents aloud	The one page, assented to
61–75	Charter drafted with both controllers; four names, split clause, sunset	The signed charter
76–90	First joint drawdown; successor chairs one meeting	A decision made without you

APPRECIATIVE QUESTIONS

Twelve, for a room

Discovery — what is already working

Think of a practice in this organisation that spread without anybody being told to do it. Who was the second person to take it up, and what made them credible to the third?
Which small group here makes decisions well — genuinely well, quickly and without resentment? What is its size, and what does it have written down?
When has one of you changed how you work because you saw two colleagues doing something, rather than because you read the case for it? What were you watching?

Dream — what becomes possible

If every initiative in this building had three named owners instead of one, what would stop happening that currently happens?
Imagine a year in which nothing we started depended on the person who started it. What is the first thing we would notice?
If splitting a team at twelve were an ordinary celebration here rather than a reorganisation, what would we do to mark it?

Design — what we build

Looking at the people whose participation would matter most, where is the gap in the distribution — the number of adopters at which nobody here moves?
Who are the two people in this organisation that others check with, whom we have never actually asked for anything?
What are our four lines — who is in, how we decide, who watches, what happens when somebody does not do what they said? Let us write them now, in this meeting.

Destiny — how it holds

Who would chair this if the person chairing it today were absent for three months, and what would they need to have already been doing?
At what size does this group divide, and who decides? If we cannot answer that today, when will we?
What is the first sign we would see that this had become one person's project again — and who is best placed to notice it?

WORKS CITED

Rogers, E. M. (2003). Diffusion of Innovations, 5th edn. Free Press.

Granovetter, M. (1978). "Threshold Models of Collective Behavior." American Journal of Sociology, 83(6), 1420–1443.

Granovetter, M. (1973). "The Strength of Weak Ties." American Journal of Sociology, 78(6), 1360–1380.

Centola, D. and Macy, M. (2007). "Complex Contagions and the Weakness of Long Ties." American Journal of Sociology, 113(3), 702–734.

Centola, D. (2010). "The Spread of Behavior in an Online Social Network Experiment." Science, 329(5996), 1194–1197.

Centola, D. (2018). How Behavior Spreads: The Science of Complex Contagions. Princeton University Press.

Centola, D., Becker, J., Brackbill, D. and Baronchelli, A. (2018). "Experimental evidence for tipping points in social convention." Science, 360(6393), 1116–1119.

Xie, J., Sreenivasan, S., Korniss, G., Zhang, W., Lim, C. and Szymanski, B. K. (2011). "Social consensus through the influence of committed minorities." Physical Review E, 84, 011130.

Bass, F. M. (1969). "A New Product Growth for Model Consumer Durables." Management Science, 15(5), 215–227.

Sultan, F., Farley, J. U. and Lehmann, D. R. (1990). "A Meta-Analysis of Applications of Diffusion Models." Journal of Marketing Research, 27(1), 70–77.

Valente, T. W. (1996). "Social network thresholds in the diffusion of innovations." Social Networks, 18(1), 69–89.

Coleman, J. S., Katz, E. and Menzel, H. (1966). Medical Innovation: A Diffusion Study. Bobbs-Merrill.

Watts, D. J. and Strogatz, S. H. (1998). "Collective dynamics of 'small-world' networks." Nature, 393, 440–442.

Zhou, W.-X., Sornette, D., Hill, R. A. and Dunbar, R. I. M. (2005). "Discrete hierarchical organization of social group sizes." Proceedings of the Royal Society B, 272(1561), 439–444.

Dunbar, R. I. M. (1993). "Coevolution of neocortical size, group size and language in humans." Behavioral and Brain Sciences, 16(4), 681–694.

Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.

Wilson, D. S., Ostrom, E. and Cox, M. E. (2013). "Generalizing the core design principles for the efficacy of groups." Journal of Economic Behavior & Organization, 90S, S21–S32.

Netting, R. McC. (1981). Balancing on an Alp: Ecological Change and Continuity in a Swiss Mountain Community. Cambridge University Press.

Freeman, J. (1972). "The Tyranny of Structurelessness." Berkeley Journal of Sociology, 17, 151–164.

Polletta, F. (2002). Freedom Is an Endless Meeting: Democracy in American Social Movements. University of Chicago Press.

Sale, K. (1973). SDS. Random House.

Alcoholics Anonymous (1953). Twelve Steps and Twelve Traditions. AA World Services.

Alcoholics Anonymous (1957). Alcoholics Anonymous Comes of Age. AA World Services.

Yunus, M. (1999). Banker to the Poor: Micro-Lending and the Battle Against World Poverty. PublicAffairs.

Armendáriz, B. and Morduch, J. (2010). The Economics of Microfinance, 2nd edn. MIT Press.

Nandram, S. S. (2015). Organizational Innovation by Integrating Simplification: Learning from Buurtzorg Nederland. Springer.

Laloux, F. (2014). Reinventing Organizations. Nelson Parker.

Moore, G. A. (1991). Crossing the Chasm. HarperBusiness.

Schelling, T. C. (1978). Micromotives and Macrobehavior. Norton.

Cooperrider, D. L. and Whitney, D. (2005). Appreciative Inquiry: A Positive Revolution in Change. Berrett-Koehler.

U.S. Bureau of Labor Statistics (2024). Employee Tenure (news release, September 2024), median years of tenure with current employer.

Note on figures. Rogers' band arithmetic, the Bass hazard table, the Granovetter equilibria, the star and ring-lattice contagion results, the Dunbar coordination load, the Los Molinos rate and the charter cover ratio are all computed in lib/verify/I_07.py and reproducible with python3 lib/verify.py I.07. The Bass coefficients are the meta-analytic means of Sultan, Farley and Lehmann (1990), not a single study's estimate; the founder-departure hazard is derived from a published median tenure under an explicit constant-hazard assumption, which the text states. Buurtzorg's team cap is given as "about twelve" because published accounts describe it as a maximum around that figure rather than a fixed constant.