Chapter 14. Vision as hypothesis

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 governance built on imagined tensions lets the loudest visionary win
• How a founder's vision enters as bets, each one saying what would show it wrong
• How many small, aimed probes test a bet, and why silence isn't counted against it
• What counts as evidence, how a verdict is reached, and how invented work waits for the world
• Why claims about ourselves and claims about the world sit in separate lanes

Every organisation starts with someone who can see something that isn't there yet.

They can describe it in detail. Who it's for, what it will feel like, why the world needs it. They've often carried it for years. Without that picture nothing would begin, and the people who join early usually join because the picture moved them.

So the sceptic's view deserves a fair hearing. The founder's vision is the point. It's why the organisation exists. Asking a founder to treat it as one opinion among many sounds like asking a composer to put each bar of a symphony to a vote. Some things are made by one person seeing clearly, and committees make them worse.

We agree with more of that than you might expect. The vision matters. The question this chapter asks is what happens to it once there's an organisation around it.

What it costs to build on a picture

A new organisation has no operating life yet. Nothing has arrived from outside, nobody has asked for anything, nothing has been tried and failed. So every tension in it, every sense of what's missing or what needs a home, is imagined.

That has a quiet consequence. In Chapter 11 we described how an organisation on The Grove changes its own shape by consent, in response to tensions people actually feel. In a new organisation, the tensions are mostly pictures of the future. Governance built on imagined tensions is design by vision with consent attached. The process looks fair, but whoever imagines most vividly, and speaks most confidently, ends up holding the structure. The loudest visionary wins, and everyone else agrees to a map of a place none of them has visited.

The business cost is familiar to anyone who has started something. Months spent building what nobody wanted. A team shaped around a market that turned out to be somewhere else. Work invented in planning sessions that keeps people busy and changes nothing for anyone outside.

The people cost is less talked about. Early team members learn that the way to have influence is to have a bigger picture than the next person. The quieter ones, who often notice first that something isn't landing, learn to keep it to themselves. And the founder carries something heavy: if the vision is the point, then when it meets the world and wobbles, the founder wobbles with it.

The world as the one who answers

The turn came from a few directions at once.

In the early 2000s, Steve Blank began telling entrepreneurs something blunt: there are no facts inside the building.¹ Everything a founder believes about customers is a guess until a customer says otherwise. Eric Ries later built on this, describing a start-up as an experiment in what people actually want.² Before both of them, the philosopher Karl Popper had argued that a claim worth calling scientific is one that could, in principle, be shown wrong.³ A belief that survives every possible outcome tells you nothing.

And from a different field, Stephan Haeckel described organisations that sense and respond rather than make and sell: ones that read signals from outside and shape themselves around what they hear.⁴ Frederic Laloux, writing about organisations without bosses, described purpose as something listened for rather than imposed.⁵

Otto Scharmer, coming at it from the inner side of leadership, gave us the word we use for the next step: prototyping. In his Theory U, once a group has sensed what wants to emerge, it doesn't write a plan. It explores the future by doing: small, quick, rough experiments that let the world answer early, while there's still time to change course.⁶

Put together, these suggest a different relationship between a founder and their vision. Not an order to follow, and not a dream to protect. A set of beliefs about the world that the world can answer.

One more picture shaped our thinking, from computer graphics. Path tracing, a way of rendering realistic light, works by firing huge numbers of cheap rays into a scene. Most of them return almost nothing useful. But as their count grows, the picture sharpens out of the noise. No single ray matters much. Enough of them, aimed sensibly, and the image arrives. That turned out to be a good model for how an organisation finds out whether what it believes is true.

The organisation, in short, is sensing, not inventing.

Bets and probes

In an organisation on The Grove, a founder's vision enters as bets. Founder statements carry no special weight. Whoever holds a branch can make bets on it, the founder included, and the founder will often make the first and most important ones. A branch holds at most three open bets at a time, so beliefs compete for a few places instead of piling up. But each one has to stand on what comes back.

A bet says seven things. What it believes. What would show that the belief holds. What would show that it doesn't, and this part is required: a belief that can't be shown wrong isn't admitted. Who the bet is about. The smallest way to ask them. A day to look back by, no more than six months away. And its measure: how many answers have to come back by that day, and at least what each must cost the person giving it. The measure is set before anything goes out, and it doesn't change afterwards.

Each bet lives on a branch, one of the parts of the organisation that Chapter 9 introduced, and the people holding that branch look after it. Wise-Eye can prepare a bet from a conversation, so it's ready for someone to confirm with a tap. It never makes one itself.

Then comes the probe. A probe is a piece of outbound work that tests a bet and expects a reply by a certain day. A short post in a group where the right people already talk. A message offering one small thing. A question asked in a place where the answer might come.

The design expects many probes, each expecting almost nothing. A success rate close to zero is normal. The more probes go out, the sooner the ones that work show up.

There's something of Barbara Marx Hubbard in this way of looking. She argued that the way forward is found less by cataloguing everything that's broken than by searching out what's already working and connecting it. A probe that rises is a small piece of what's working. The design is built to notice those, not to tally the silences.

But many isn't the same as scattered. Each probe is aimed: who it's for, where it goes, in what words, and from what reading of where people already gather and the language they already use. We often describe it with a fly fisher's eye. Read the water. Match the hatch, meaning offer what's already on the surface rather than what you wish was there. Choose the fly. Cast where they are. Watch the drift.

What comes back

Three things can come back from a probe.

A take: someone responds. A refusal: someone says, in effect, "not for me". That's right water, wrong fly, and it's useful. Or no rise at all.

Silence is treated as background, not as evidence against the bet. Most probes meet silence, and reading that as rejection would kill good bets early. Only after many probes with nothing back does the reading suggest looking again at the aim: perhaps the water, perhaps the fly.

When something does come back, it becomes evidence, and the rules about what counts are deliberately strict. Only work from outside the organisation can be linked as evidence. The people holding the branch can also count by hand what came back, like who turned up to a trial, but that counts only as our own word. Each piece of evidence is linked to the bet, and the link says which way it points and what it cost the person who gave it: a glance, a few words, a question they asked themselves, some of their time, a commitment. Nothing is weighted. A bet can ask that its evidence come only from the people it's about. A "not for me" is kept and shown beside the yeses, but only yeses count towards the measure. Anyone holding a seat can contest a link, and a contested link doesn't count until it's settled.

None of this produces a score. The weighing is shown in words, so people can see what lies behind a bet and argue with it.

The verdict, and the work that waits

There are three verdicts: the bet held, it needs reshaping, or it didn't hold. When the bet's day comes, it's read against its own measure, whether or not anyone is looking, and marked held or didn't hold. The people holding the branch then confirm that, or say otherwise with a reason, and the reason shows. Reshaping is always theirs to decide. Wise-Eye never gives a verdict. And a bet nobody ever asked anyone about gets no verdict at all: it stays untested. A bet can't be said to hold with nothing in its favour, and answers from inside the organisation never count, however many people agree. The verdict says what it rests on: strangers, people who know us, or only our own word.

A reshaped bet creates a successor. It starts fresh, but takes the work waiting on the old bet and any probes still out. The old bet keeps its evidence, so the trail can be read back. Over time these chains become a trail: from a founder's first speculation to the place where the organisation's purpose can actually be met. Bets with nothing new back for six weeks fade to untested, so a stale belief doesn't sit on the map looking proven.

This changes what happens to invented work. Picture a planning session where someone proposes a series of workshops. In most organisations that becomes a project, with dates and owners, before anyone has asked whether people want workshops. In an organisation on The Grove, the work isn't rejected. It's made to wait for the world. "Make it a bet" drafts a bet and a first probe, and the task is parked on the bet. If the bet holds, the work wakes, grounded in the evidence. If it didn't hold, the work is let go, with the evidence there to show why.

Something similar applies to growing the organisation's shape. A new branch needs something felt behind it. That can be work arriving from outside, or the organisation's own work that people have committed to and are holding. It can be a bet that held, or the work's own reading that a branch wants to grow, as Chapter 5 described. In the first thirty days, before there's much to read, it can be what was already happening before the organisation came onto The Grove, on the word of whoever raises it, shown to everyone deciding.

Two lanes

Not every belief is about the world outside.

Claims about ourselves, that we're stretched, that care is going unkept, that a meeting has stopped being useful, are sensed by us. They're sensed in the Field, the organisation's own way of listening to how its people find things, and nobody needs a customer's permission to say they're tired.

Claims about the world need the world. Twenty voices agreeing that people want workshops is still a guess. A strong feeling inside the organisation is real information about the organisation. It isn't information about anyone else.

Not a marketing machine

Anything that sends messages outward and counts replies can become a marketing machine, so the guard rails matter as much as the mechanism.

The measure is purpose met, efficiently. Never reach, clicks or conversion. Frames are honest: something being explored is described as being explored, not dressed up as finished. We want asks to go only to people who chose to hear. Today that's a principle, not a rule the system enforces: a probe by email goes to twenty people at most, each message says how to say no thanks, and writing to people who never said yes is left to the sender to weigh. The reading looks at patterns, never at individuals. And someone who answers a probe is a person to serve or to learn from, never a lead.

How we decided

The original rule was stricter. Only crossings of the outer boundary, things that came from or went to the world outside, would give anything weight. We loosened it in one place: the organisation's own committed work, held by people who've taken it on, counts as weight too. And we gave internal sensing its own lane, rather than forcing feelings about ourselves to pass an outside test they were never meant for.

We chose states in words over scores everywhere. A number invites gaming and hides its reasons; words can be read and disputed.

We started with bets first and a graph of claims to follow, a web linking every belief to every other. Then we dropped the graph. We decided relations between bets should come from nearness, from how close in meaning they are, and we're building towards that. Today two links are kept: the trail of reshapes, and a bet that is one piece of a bigger one. It was simpler, and simpler seemed more honest about how little we knew.

The whole model is itself a bet. It should shrink invented work without starving valuable work, the slow, uncertain things that matter before anyone outside can say so. It hasn't yet been tried with real organisations, and we don't know where it will be too strict or too loose.

And there's a known gap, which we've recorded rather than hidden. A determined visionary alone in a new organisation can still get through in places: through their own committed work, and through their word in the first thirty days. We haven't closed that, and we're not sure it can be closed without making it impossible to start anything at all.

What you might take from this

The founder's vision still matters. It's where the bets come from, and often the best ones. What changes is where its authority comes from: not from the founder's certainty, but from what comes back when the organisation asks.

A vision is strongest once the world has answered it.

Notes and further reading

On path tracing, the rendering method that shaped our picture of probes: Kajiya, J. T. (1986), "The Rendering Equation", Proceedings of SIGGRAPH '86. On searching out what's working rather than cataloguing what's broken: Hubbard, B. M. (1998), Conscious Evolution: Awakening the Power of Our Social Potential, New World Library. On setting out a venture as a set of assumptions to test: Osterwalder, A. and Pigneur, Y. (2010), Business Model Generation, Wiley.