Chapter 26. Between organisations

A working draft, November 2026. This book is being written while The Grove is being built; expect it to change. Checked against the build as it stood on 1 November 2026.

In this chapter

• Why contracts forget and platforms own the memory, and what cooperation needs instead
• How services, needs and agreements work between organisations on The Grove today
• What crosses between two organisations, what each can see, and how agreements are looked back on
• How people who belong to more than one organisation count, and the open question that raises
• What's decided but not built, including conditional pledges against the race to the bottom

A sceptic who has come this far might grant us something. Perhaps inside one organisation, people who can see each other can decide by consent, keep their promises without a boss and agree to hold back together. Fine. But step outside the walls and the weather changes.

Between organisations there's no shared purpose and no forum. There are contracts, written by whoever has the better lawyers. There's leverage: the big buyer squeezes the small supplier, and the platform squeezes everyone. And there's the race from Chapter 1, each organisation cutting the corner because it knows the others will. Cooperation is lovely in a circle, the doubt goes. Between organisations, competition wins, and the biggest wins most.

It's a fair doubt, and this chapter won't pretend to dissolve it. It says what exists today between organisations on The Grove, what we've decided and not yet built, and where the promise Chapter 1 made is still mostly a promise.

What the usual ways cost

A contract is a promise designed to fail safely. It imagines the relationship going wrong and prices that in advance. The effort goes into the clauses, and the clauses are drafted by the side with more to spend, so small suppliers sign what they're given. Then the contract goes in a drawer. What was agreed drifts away from what's actually happening, and nobody looks again until the renewal or the dispute. The person at the supplier who can see the arrangement isn't working has no way to say so, short of a call to the account manager that nobody wants to make.

Platform marketplaces solved the finding, and charged for it. The platform stands between buyer and seller, takes a cut of every exchange, owns the ranking and owns the reviews. Stars drift upward until four point eight means "fine". They get traded and gamed, and they say very little: four stars from whom, for what? The relationship belongs to the platform. Leave, and your reputation stays behind. The platform is in its own race, for attention and margin, and the organisations on it are its stock.

Underneath both sits the race to the bottom. The political scientist Robert Axelrod ran tournaments in which computer programs played the prisoner's dilemma against each other, again and again.¹ The winner was the simplest: cooperate first, then do whatever the other did last time. What made cooperation pay wasn't goodwill. It was three conditions: the players met again, they could see what the other had done, and they remembered. Elinor Ostrom found something similar in the commons she studied: people who could see each other, and made their own rules, could hold back together for centuries.²

Contracts and platforms supply those conditions badly. A contract meets once and then forgets. A platform remembers, but the memory is a number it owns.

Relationships that are seen and looked back on

The turn, for us, was to treat what passes between organisations as a relationship that both sides decided, that each can see, and that gets looked back on, rather than a transaction that's priced and forgotten. Axelrod's three conditions are things an operating system can provide: meeting again, through standing agreements; seeing, through what each side is shown; memory, through the look back. And each organisation stays sovereign. Nobody owns anybody.

What exists today has three pieces: services, needs and agreements. They were built in late October 2026.

A service is something one part of an organisation offers to others. A branch says what it does in its own words, which of its jobs it draws on, and its price, which can be nothing. It chooses who can see it: everyone, listed in the directory, or only the organisations it names. Offering a service is a proposal heard in that branch's own forum, the same way it would change how it works. Requests then reach it through its way in and land as that branch's own work.

A need goes the other way. A branch posts what it's short of, "deliveries across town, about twenty a week", to everyone or to named organisations. Those who could meet it see it and offer one of their services.

Finding someone starts from what you're looking for, in your own words. What comes back is ordered nearest first: organisations you already exchange work or money with, then ones your own people belong to, then everyone else who lists the service. "Two of your people belong there" counts for more than anything a stranger says about itself. Wise-Eye can search for a service and draft a need, but a person reads the draft and posts it.

An agreement takes two proposals, one in each organisation. On the side bringing the service in, it's heard in the forum around the parent of the branch it will serve, as a change to a seat would be. That's deliberate: an outside organisation serving one of a branch's jobs changes who carries the work, so it's decided where seats are decided. The proposal names the provider, which job it serves, which kinds of work it covers, the fee, how the two will work together, how often it's looked back on (every ninety days unless the forum chooses otherwise, anywhere from a week to a year) and what would show it worked. Once it passes, a piece of work lands on the provider's branch: this organisation asks to use your service. That branch's people agree by a proposal, or say no. Nothing starts until both have decided.

The line we hold here is simple. Bringing in a service changes who does the work, not who answers for it. The job stays with the branch that holds it. The provider gets no seat and no say in what's decided, only what the agreement opens to it.

What crosses the membrane

Once an agreement is live, whoever holds the branch, or has a piece of work in hand, can pass work under that job to the provider. What crosses is the work's title, the note written with it and its day. Nothing else of the organisation goes with it. On the provider's side it arrives as their own work, marked as from the other organisation, and if it has a day, it carries a promise to keep by then (Chapter 15 follows how promises are tracked).

When they finish, it comes back with their words and anything they attached. It goes back to the branch open, to be looked at before anything goes out, and it never closes by itself. If they close it without doing it, it comes back saying so.

It's tempting to picture this as one organisation calling another's system, the way a connector reaches a mailbox. Today it isn't that. On today's test bench every organisation lives in the same space, so passing work under an agreement needs no key and no relay: the system writes a new piece of work straight onto the provider's branch, carrying only the title, the note and the day, and it comes in through that organisation's own one door, to the job the service draws on, or to be read for where it belongs if none fits. A service isn't a connector, and nothing between organisations is built as one. The way in is the only interface an agreement names that carries anything. A direct call through the API is designed, and can be written into an agreement, but nothing acts on it yet. How organisations will reach each other once each is its own DAO in its own space, as they will be when The Grove goes live, is still open.

Each side also holds a standing promise about the agreement itself, due on its look back day, and like any promise it rises if the day passes with nobody keeping it. At that day, or when the agreement ends, the branch that used the service answers three things: did it land (yes, partly, no), would you keep it (keep, change, stop), and a line in its own words. The facts come with it, in words: how many pieces of work went through, how many came back by their day. The provider sees every look back and can answer under one.

Someone weighing that service later reads it plainly: "Four organisations have used it. Three kept it on. What they said, newest first." There's no score and no average. Only a branch that held an agreement can look back on it, so nobody judges what they didn't use. Either side can end the agreement on the terms agreed.

What each side can see

An agreement can be read only by its two organisations. To each other, they appear by name; branches, members and purpose stay inside.

From where you stand, the network shows the organisations you belong to and the services between them and others: who has whom do what, for which job, the fee, how they work together and when it's next looked back on. It doesn't show the work passing along an agreement, or letters, or money. Who works with whom is public by default, because seeing who your neighbours trust is part of how trust travels. But either side can keep an agreement to themselves, and then it's drawn only for people who belong to both. Branch names appear only for organisations you're part of.

Two more lines hold throughout. Everything is organisation to organisation: a freelancer is a one-person organisation, and a need is answered by an organisation, never by a person directly. And the Steward, the organisation that looks after the network's shared code, takes no fee on anything that passes between organisations. When one organisation provides a service to another, what it earns is its own.

People who belong to more than one

A person has one identity and can belong to many organisations. Their face and name travel with them; their seats, their trust and their voice belong to each organisation separately. Belonging to two gives no say in the second because of what you hold in the first.

Their belonging does count for nearness. When a search says two of your people belong somewhere, that's where it comes from. And if you belong to both ends of an agreement, you see it even when they've kept it private.

There's an open question here we haven't settled. Inside one organisation, the person a seat change would seat stands aside from deciding it. Between organisations, nothing yet asks someone who holds seats in both to stand aside when their forum decides an agreement with the other. Whether it should, and where the line would sit, is still to decide.

What's decided but not built

Some of what an agreement can name is recorded but not yet acted on. A feed of data going out, the provider calling in directly, one area of the organisation handed across for the provider to act in and taken back at any time, the mood of a room read across the boundary as a whole: an agreement can list these, and today only the way in carries work. Paying each bill as routine once the fee is agreed is designed and not yet joined up. So is publishing services in shared vocabularies, so that systems outside The Grove can find them.

The larger gap is the one Chapter 1 named. Two organisations committing to each other is useful. It isn't an answer to the race to the bottom. The race needs many organisations to commit together, and to do it without anyone moving first alone.

The shape of that answer has been around for a while. The campaigner John Bunzl's Simultaneous Policy, SIMPOL, asks people to support a set of global policies that come into force only when enough nations adopt them at once, so no country loses by going first.³ The economist Alex Tabarrok described the dominant assurance contract: pledges that pay out only if enough people pledge, with the organiser compensating pledgers if the threshold isn't reached, so pledging becomes the sensible move whatever others do.⁴

We're building towards something like this, and call it a conditional pledge, though it's an intention more than a design so far: an organisation commits to act, to pay a living wage, to stop a harmful practice, to cut what it burns, only when enough others commit too, and when the threshold is met they switch on together. We've parked it. The questions it raises are real and unanswered: which forum inside an organisation can make a pledge that binds the whole, what counts as enough, how a pledge is seen to be kept without organisations policing each other, and what happens to one that switches on and then backs away. A map of the network beyond your own neighbourhood is parked with it.

So the honest position is this. What exists is the ground two organisations stand on together: decided on both sides' authority, seen, remembered and looked back on. The many committing together is still to build.

How we decided

We weighed a contract as a document, signed and filed. We chose a standing promise both sides decide by proposal and revisit on a rhythm, because the drawer is where contracts go wrong.

We weighed ratings. Someone choosing a provider genuinely wants to know how it went for others, and a star rating is the familiar answer. We chose the look back in words instead, because a number set by hand is exactly what's refused everywhere else in an organisation on The Grove, and because averages hide who said what.

We weighed letting people hire people across the boundary. We chose organisation to organisation, so the question of whose authority stands behind an agreement has the same answer on both sides: a forum's.

We weighed a network that drew all the traffic between organisations. We chose services only, because the work passing between two organisations is theirs, not the neighbourhood's.

We parked conditional pledges, not because they matter less, but because they rest on the simpler thing working first. A pledge between many is only as good as the agreements between two.

Some of this we're unsure about. Nearest first favours those you already know, and that can quietly become the old pattern, where whoever is established gets the work. We haven't yet seen whether the look back in words carries enough for a stranger choosing. And whether many organisations will bind themselves without someone at the centre holding the pledges is the question this chapter can't yet answer.

What you might take from this

Between organisations, the usual choice is a contract that forgets or a platform that owns the memory. An agreement between organisations on The Grove is decided by each side, seen by both, and looked back on in words, and either side can end it. That gives cooperation its three conditions: meeting again, seeing, remembering.

Organisations can already commit to each other in the open; what we're building towards is letting many of them switch on together, so nobody has to move first.

Notes and further reading

On how a conditional commitment changes the arithmetic of going first, Schelling, T. C. (1960), The Strategy of Conflict, Harvard University Press, remains a clear starting point. On co-creation, and gatherings that matched one group's needs with another's offers: Hubbard, B. M. (1998), Conscious Evolution: Awakening the Power of Our Social Potential, New World Library.