Better For Our Presence
Three convictions about what humans are for, what capability obliges, and how high the work is supposed to aim — and an honest account of the one thing standing in the way.
convictions
Humans can be a keystone species again — everything better for our presence, not our absence.
I can, therefore I must. Capability is not a credential. It is an obligation.
Generative is the top rung: creating life and abundance outright, in everything we do.
Parts One, Two and Three make these three cases in turn. Parts Four, Five and Six are about why none of it happens yet, how I could be wrong, and what I have actually built so far.
version
The short version
Better For Our Presence, without the jargon. About five minutes.
A mistake we made a long time ago
When Europeans arrived in the Americas, they wrote home about endless untouched wilderness. Forests nobody had planted. Land nobody had shaped.
They were wrong, and the reason is grim. Disease had swept through decades ahead of them and killed most of the people already living there. What the newcomers walked into wasn't wilderness. It was a garden whose gardeners had just died.
Nobody knew that at the time. So we built an entire way of thinking about nature on top of the mistake — the idea that the natural world is at its best where people aren't. That we're a stain on it. That the most we can hope for is to tread lightly and take less.
That idea is still with us. It's why nearly every environmental word we use is about subtracting: reduce, offset, minimize, leave no trace.
What the gardeners were actually doing
Here's the strange part. Nearly everywhere those people lived, they left the land richer than they found it.
Along the coast of British Columbia, they built low stone walls across the beaches at the tide line. The walls made flatter, warmer ground where clams grow well. Those walls are still there. Nobody has tended them for hundreds of years. Scientists went and counted, and those old beds still produce about four times as many clams as the untouched beaches right next door.
In the Amazon, they made soil. Charcoal, bones, broken pottery, food scraps, worked into the ground over generations. It's called black earth. Centuries after the people who made it were gone, it's still more fertile than the jungle floor around it.
Nobody was reducing their footprint. They were adding. And what they added outlived them by centuries.
So what happened to us?
We didn't lose the ability. We lost the feedback.
If you build the soil you're going to farm, and your grandchildren farm it too, you find out quickly whether you got it right. The answer comes back to your own door. You can't fool yourself for long.
We can now build at a scale where the consequences land somewhere else, on someone else, long after we've moved on. Money arrives in a place and leaves before anyone can tell whether it helped. The lesson never makes it back to the person who made the decision.
So the problem isn't that people stopped caring. It's that we stopped being able to tell. And that's a more hopeful problem than it sounds, because you can fix a measuring problem.
Why "less harm" isn't the goal
Think about what the very best sustainable building can achieve. It does no damage.
That's it. That's the ceiling. A perfect score is zero.
Once you notice that, it's hard to unsee. If I'm able to build something that leaves a place with more life in it than before, and I build the harmless version instead, I didn't stay neutral. I picked the smaller of two things I could have done.
That's what I mean when I say if I can, I must. Not as a slogan. As arithmetic.
Four ways to aim
There's a ladder here, and each rung stands on the one below it.
Do less damage. Best possible outcome: zero.
Repair what's broken. Best possible outcome: back to how it was.
Help a place get healthier and keep going on its own. Better. No fixed limit.
Make life that wasn't there before. That's the one I'm after. I call it generative.
Nothing on that list is wrong or beneath the others. You can't make new life in a place you haven't bothered to understand, and understanding places is what the rung below is for. You don't stop pouring foundations because you've learned to design buildings.
The catch, and it's a real one
Every time people have industrialized building, they've flattened whatever they touched. The same houses, the same streets, everywhere, regardless of where "there" was. So there's a good reason careful people have kept this kind of work small and handmade.
The way through is something a beaver already knows.
A beaver has exactly one trick: build a dam. It does the same thing every time. But every pond that results is different — different water, different plants, different animals — because every stream is different. The beaver doesn't design the pond. It builds the dam and lets life do the rest.
That's the whole idea. Standardize the method, never the result. Make the conditions, and let the place decide what grows.
What's actually missing
Three hard things have to work. Two of them are handled, and I originated neither.
People have spent thirty years learning how to read a place properly — what it needs, what it could become, how fast it can move. That body of work exists.
Other people are building the accounting right now: an honest, checkable way to show that a piece of land ended up genuinely better off, so that money can be lent against it rather than just hoped about. It was designed before I arrived; I am now a founding principal of the company building it.
The third piece is the one nobody has picked up. Somebody has to actually build these things — over and over, at a normal price, in a way that can be copied and inspected. Not one beautiful project. Thousands of them.
That doesn't exist yet. And nothing scales if nobody can build it.
What I haven't done
I should be straight about this, because it's the part people usually leave out.
I haven't built it either. The companies I'm working on are early. Some revenue, some progress, nothing close to scale.
So don't take my word for any of it. I've written down exactly what would prove me wrong, and exactly what I've promised to publish and when — including the numbers that make me look bad. Judge me on those when they show up.
The question
If a few hundred beautiful projects is all this ever amounts to, we should stop describing it as an answer to a global problem. A few hundred is not an answer.
And if that isn't good enough, then somebody has to build the thing that makes it ordinary.
That's the part I've taken on.
The full argument follows, with the evidence and the sources. About twenty minutes.
language
Through 2026 my operating language shifts from regenerative to generative. This is not a rebrand, and it is not a correction of anyone else's work. The two words name different things, and a fair amount of what follows is an attempt to say precisely what each is for. Where my older properties still read "regenerative," they are mid-migration.
Everything better for our presence
Modern environmental thought inherited an assumption it never tested: that ecological value is maximized where humans are absent. It is false, and almost everything downstream of it is miscalibrated.
we inherited
The assumption we inherited
Preserves. Wilderness areas. Pristine. Untouched. Minimal impact. Leave no trace. Offset, reduce, mitigate, neutralize.
Look at that vocabulary as a set and the premise underneath it becomes visible: the human presence is a debit, and the highest available good is to make the debit smaller. Sustainability inherits this wholesale — it is a subtractive vocabulary because it accepts a subtractive premise. If our presence is inherently a cost, then of course the goal is less of us.
The premise is empirically false, and it has been settled scholarship for over thirty years.
William Denevan's 1992 paper The Pristine Myth1 made the case that the Americas of 1492 were not a wilderness. They were a densely managed landscape — fire regimes, earthworks, terraces, constructed soils, cultivated forests, at continental scale. The "wilderness" later colonists described was substantially an artifact of catastrophe: a landscape whose managers had been removed by epidemic disease a century or more before anyone wrote it down. We mistook a depopulated garden for an untouched one, and then built an entire conservation ethic on the error.
William Cronon pressed the consequence in 1995.2 The wilderness ideal, he argued, teaches that the only good human is an absent one — which leaves us with no ethic at all for the overwhelming majority of the planet, the parts where people actually live and work. If nature is what exists where we aren't, then everywhere we are is already lost, and there is nothing to do there but limit the damage.
That is the trap. And it is worth stating the alternative in the plainest possible terms: no other organism in a living system is a pure debit. Every other species takes and gives, and the give is structural — it is why the system has the shape it has. The claim that humans alone can only subtract is not an ecological finding. It is a cultural inheritance.
That instruction is grief, not strategy.
engineered
abundance
The evidence: engineered abundance
The case against the premise is not philosophical. It is archaeological, and in one case it is a controlled experiment.
Terra preta. Anthropogenic dark earths across the Amazon basin: soils built by human habitation — charcoal, bone, pottery, organic waste — that remain measurably more fertile than the surrounding ground centuries after the people who made them were gone. Not a lighter footprint on the soil. A permanent upgrade to it.
Clam gardens.3 Rock-walled intertidal terraces along the Pacific Northwest coast, built to extend the tidal zone favourable to bivalves. Groesbeck and colleagues surveyed eleven clam gardens against ten unmodified beaches at Quadra Island, British Columbia, and ran transplant experiments across both. The gardens — built by hand, centuries ago, and unmaintained since — still outproduce the unmodified shoreline:
This is not an inference from pottery shards. It is a measured comparison against a control, published, with numbers — and it is the cleanest single piece of evidence that human construction can raise a living system's output and hold it there for centuries without anyone tending it.
Chinampas. Constructed agricultural islands in the lake system of the Valley of Mexico, among the most productive agricultural systems per hectare ever recorded. Built soil, built hydrology, built ecosystem.
Fire-managed California. Kat Anderson's Tending the Wild4 documents the burning, coppicing, pruning, sowing and selective harvest by which California's indigenous peoples maintained oak savanna, grassland and montane meadow. Those systems carried more edible biomass and, in many cases, higher species diversity than the closed, fire-suppressed succession that replaced them. The "untouched wilderness" celebrated by nineteenth-century arrivals was a garden roughly one generation after its gardeners had been removed.
Amazonian domesticated forests. Species composition across large tracts of Amazonia still reflects millennia of human selection for useful trees. The forest is, in part, an artifact.
These cases share a structure, and it is the structure that matters. None of them were restorations — there was no prior degraded state being repaired. Each produced more than the untended baseline. Each persisted long after its makers. Each is generative in the strict sense: engineered conditions that yielded net-new life.
Humans have been a net-additive force in living systems, at landscape scale, for a very long time. The question is not whether it is possible. It is documented.
precisely
Keystone, precisely
I write under the banner of helping make humans a keystone species again. The phrase invites easy dismissal, so it is worth being exact about the biology — and the biology turns out to be more favourable to the claim than the dismissals assume.
Keystone species — Robert Paine, 1969, working on sea stars5 — means a species whose influence on ecosystem structure is disproportionate to its abundance. The usual objection is that eight billion of us cannot possibly count as scarce. But headcount is not the unit. Ecologists measure abundance as a share of the living community, and by that measure humans are 0.01 per cent of Earth's biomass (Bar-On, Phillips and Milo, 2018). Plants are around 82 per cent. Bacteria dwarf us entirely.
One ten-thousandth of the planet's living matter6, determining the fate of the rest. Whatever else that is, it is not proportionate. On the definition as Paine wrote it, humans are a keystone species — decisively so.
The mechanism has its own name: ecosystem engineer (Jones, Lawton and Shachak, 1994)7, an organism that modifies physical habitat and thereby determines what else can live there. Beavers are the canonical case. And for what humans do at global scale, Boris Worm and Robert Paine proposed hyperkeystone species in 2016: a species that acts through multiple other keystone species, driving cascades of cascades. That is a tier above keystone, not a category outside it. We are not less than a keystone species. We are the only one operating at that level.
And here is the part no ecological term covers, because no other species has needed one.
Every other keystone species has its sign fixed by its biology. The sea star does not deliberate. The beaver does not weigh whether this particular valley might be better left alone. Their influence is disproportionate and it is also involuntary — they cannot be other than what they are.
We can. As far as we know, we are the only keystone species with agency over its own effect: able to observe what we are doing, judge it, and choose differently. That is not a footnote to the claim. It is the substance of it. Our disproportionate influence is a fact of biology. Its direction is a decision — remade continuously, mostly by people who do not notice they are making it.
Right now the cascades run downward. They have run upward before, at landscape scale, for millennia — which is what the previous section documents and what "again" means in that banner. Not a return to being small. A return to running the sign positive.
So the mission, stated correctly: flip the sign. Same species, same disproportionate influence, opposite direction.
This is not semantic tidying; it changes the objective. If our influence is inherently disproportionate — and it is — then the goal cannot be to minimize it. The system is shaped by that influence whether we like it or not, and a keystone that tries to become negligible has misunderstood what it is. The goal is to direct it. A species that cannot stop shaping its world has no honest option but to shape it well.
Which is why "better for our presence, not our absence" is a design specification rather than a sentiment. It is the only target available to an organism of our kind — and, uniquely among such organisms, one we are able to choose.
actually lost
What we actually lost
If humans have demonstrably built abundance before, the interesting question is what changed.
It was not the capacity. It was the accounting.
The people who made terra preta had no five-capital ledger. What they had was a feedback loop short enough to learn from: you lived on the soil you built, and so did your grandchildren, and if you got it wrong the consequence arrived at your own door within a generation. Every one of the systems in the previous section was built under that condition. The knowledge was not written down; it was enforced.
Scale, mobility and abstraction broke that loop. We can now build at a magnitude where the consequences land somewhere else, on someone else, decades later. Capital can enter a place and leave before the results are legible. The feedback that disciplined every previous generation of human ecosystem engineering simply does not reach the people making the decisions.
We did not lose the ability to add. We lost the ability to know whether we added.
This is the reframe the rest of this document rests on. The central problem is not primarily ethical, and it is certainly not a shortage of good intentions. It is an accounting problem: can we know, quickly and honestly enough to steer, whether a given act of building added or subtracted across everything that matters?
That is good news, in a way that grand civilizational diagnoses usually are not. Accounting problems are tractable. They yield to instruments.
I can, therefore I must
Once you are capable of more than doing less harm, doing less harm is no longer a floor. It is a choice to withhold — and withholding is an act with consequences you own.
subtraction
The ceiling of subtraction
For years I called what I built sustainable. Better materials, tighter envelopes, a smaller footprint than the alternative. Then I hit the ceiling of the word, and the ceiling is not a matter of opinion.
Sustainable's best available outcome is zero. A perfect sustainable project is one that did not make things worse. That is not an achievement; it is the absence of an offence. You cannot arrive anywhere by subtracting.
Worse, subtraction has two structural defects as an operating goal. It is unfalsifiable as an aspiration — there is no state of affairs that constitutes having got there. And it is infinitely deferrable as a practice, because "less bad than the alternative" always has one more rung available below it. A vocabulary of reduction can absorb any amount of delay without ever registering failure.
Bill Reed has said that sustainability is a slower way to die. He is right, and the reason he is right is structural rather than rhetorical: an entire discipline organized around a ceiling of zero will reliably deliver approximately zero.
And note where the premise came from. Subtraction is the correct frame only if the human presence is inherently a debit. Part One is the argument that it isn't. Which means the ceiling was never a fact about the world. It was a fact about an assumption.
not the anecdote
The argument, not the anecdote
"I can, therefore I must" is usually offered as inspiration. I mean it as accounting, and it rests on three claims that can be argued with.
Withholding is an act. If I am able to produce a net-additive outcome at comparable cost and I choose the neutral one, I have chosen the worse of two available worlds. The absence of harm is not the absence of a decision. Neutrality feels like standing still; it is in fact a selection, and it should be defended as one.
The counterfactual is the moral unit, not the intention. What matters is the gap between what happened and what could have happened with the capability that was in the room. Intent does not enter the calculation. This is uncomfortable precisely because it removes the usual defence.
Capability is unevenly distributed, so obligation is too. There is no such thing as a lesser human — everyone has capacity to move the needle, and the size of the needle is not a ranking of worth. But someone who can direct capital, land and production at scale carries an obligation that someone who cannot does not carry. The obligation attaches to the capability, not to the person's value. Mine is mine to answer for.
Put together: if one can, one must. Not because it is noble, but because the alternative is a choice you are already making and would rather not have to name.
What it costs
I should be honest about what living by this actually does to a person and an organization, because a manifesto that only lists the upside is advertising.
It has made me uncompromising in behaviour, habit and decision, and unwavering in trying to stay that way. Engineering sessions run longer than they should. Materials get scrutinized past the point of reasonable. I second-guess our own reasoning on things we already decided. Every milestone becomes the benchmark for the next one, which means the target never stops moving.
There is a fine line between obsession and conviction to excellence, and I do not always know where it is. Knowing when perfection has become the enemy of good-enough-for-now may be the harder discipline than the conviction itself.
The failure modes are real and they are mine to manage: burnout in the people around a standard that keeps rising, decision paralysis dressed as diligence, and the particular arrogance of an operator who has confused their own intensity for correctness. Naming them is part of holding the principle responsibly. An obligation you cannot sustain is not an obligation; it is a mood.
The governor
There is a version of "we were capable, therefore we were obligated" that has a body count, and it is not hypothetical. It is the big-dam era. It is high modernism generally. It is the half of the Green Revolution nobody puts on a poster. James Scott's Seeing Like a State is the definitive account of what happens when capability, confidence and scale arrive somewhere without a governor: legible plans imposed on illegible places, and the places lose.
The regenerative tradition's insistence on humility, listening and pace is a correct response to that history. It is not squeamishness or slowness for its own sake. It is a discipline built by people who watched the alternative.
So the principle needs a governor, and it is this: you author conditions, not outcomes.
The beaver builds the dam. It does not design the wetland's species assemblage — it could not, and it does not need to. It engineers a substrate and lets life do the composing. Maximum ambition about the conditions; genuine agnosticism about what arrives to use them.
The test of whether you are in the right mode is practical rather than moral:
If it needs you, you built a monument. If it walks without you, you built capability.
And there is a second governor, quieter but harder: I would rather be measured than trusted. An operator who says "trust the relationship" and an operator who says "here is the audited five-capital delta, on the place's own clock" are making very different offers. Only one of them can be held to account when it fails. Obligation without accountability is appetite with better branding.
Where I diverge
One place, stated plainly, because it is the only substantive disagreement I have with a tradition I have learned an enormous amount from.
The tradition holds that a place's potential is discovered — already latent, waiting to be perceived, with the practitioner serving as facilitator rather than author. I think that is mostly right and slightly incomplete.
I think potential is also authored. Terra preta was not latent in Amazonian clay. Chinampas were not latent in a lake. There are configurations of abundance that were never latent anywhere, that no amount of listening will surface, and that exist only because someone conceived of them and built them. Deep listening is necessary and it is not sufficient; some things have to be invented before they can be perceived.
But — and this is where the divergence reconciles rather than escalates — authoring conditions is not the same as authoring outcomes. You can build a substrate that has never existed on this planet and still have no idea, and no authority over, what will come to live in it. Which is exactly the posture in § 8. The disagreement is about how ambitious you are permitted to be about the substrate. It is not a disagreement about who is in charge of what grows.
Generative is the top rung
Four rungs, and they are cumulative — each one requires the ones beneath it. Nothing below is demoted by something above. That is what a ladder is.
Four rungs
The foundation here is not mine. Bill Reed's 2007 paper Shifting from 'sustainability' to regeneration9 laid out the trajectory the field still runs on — green, sustainable, restorative, reconciliatory, regenerative — and drew the distinction that matters most: restorative work repairs a deficit, while regenerative work engages the evolution of a whole living system. Almost everything useful about how this field thinks descends from that paper and the practice built around it.
What I want to add is one rung, and a claim about what kind of ladder this is.
Generative sits on top, and here is the precise reason it does rather than merely coming last in a list. The first three rungs are all measured against a counterfactual: less than what would otherwise have happened, back to what used to be, more capacity than the system had. Generative is measured against zero. Did this exist before? Does it exist now? That is an absolute question, and absolute questions can be audited in a way that counterfactual ones fundamentally cannot.
And the ladder is cumulative. Each rung requires the ones below it — not as history, as active dependency. You cannot generate net-new abundance in a place whose potential you have not read, and reading the place is regenerative work. Regeneration is a prerequisite for generation, not a competitor to it. You do not stop doing structural engineering when you start doing architecture.
Which is why nothing on this ladder is demoted by something above it. A rung that holds weight is not diminished by the existence of the next one. That is simply what a ladder is.
ambition, a
stack of labour
A ladder of ambition, a stack of labour
Most of the circular arguments in this field come from confusing two different axes. They are both real and they are not the same.
The ladder is how high you are aiming. Generative on top.
The stack is who does what. Three layers, none outranking the others:
Regenerative — the relational discipline. How you read a place, whose potential you serve, at what pace. The field's deepest body of work — and the correct name for capital vehicles, because capital that submits to a place's clock is regenerative capital.
Verified living asset — the accounting. When an improvement is real, durable and finance-grade. The instrument, and the audit trail that makes a claim answerable rather than asserted.
Generative — the production commitment. Every unit that ships is net-additive across the five capitals — natural, human, social, built, financial — measured, and repeatable.
Every layer is load-bearing. A relational discipline with no accounting produces claims nobody can check. Accounting with no production has nothing to measure. Production without the relational layer upstream produces exactly the flattening that Part Two's governor exists to prevent.
The ladder ranks ambition. The stack divides labour. Generative is at the top of the first and at the bottom of the second, and both of those are true at once — because a production commitment is the humblest job in the stack and the highest thing you can aim at.
Nothing to measure
Understanding is largely answered. The instrument is being built. Production is unclaimed — and it is now the thing setting the throughput of everything else.
bottlenecks
Three bottlenecks
Three things have to work before any of the above operates at a scale that matters, and they are sequential — each gates the next.
One: understanding. How do you know what a place actually needs, whose potential is at stake, and what "better" means here rather than in general? This is the field's deepest work and it is largely answered — Reed's trajectory, Regenesis's Story of Place methodology, thirty years of practice, several hundred projects, a functioning body of theory. Not closed; nothing living ever closes. But no longer the constraint.
Two: the instrument. When does an improvement in a living system become real, durable and finance-grade?
The essential text here is Reed's critique of his own creation. He helped build LEED10, then said of it that it "defines a level of greenness that aspires towards sustainability, but it does not explain how to develop regenerative relationships." The lesson generalizes well past LEED: the instrument shapes the practice. Give people a checklist and you get checklist behaviour. Give them no instrument at all and you get craft.
This bottleneck is being worked right now. The Verified Living Asset Standard and the capital rail beneath it — developed by Bill Reed's team and Regenerative Development Corporation before I entered the picture — are a serious attempt at an instrument that resists the failure mode of the last one: five capitals rather than single-metric carbon; integrity as a precondition of a credit existing rather than a claim made about it afterward; readiness gates that pace deployment to a place's development timeline rather than a fund's clock. Whether it holds is an empirical question. That it is the right shape of attempt seems clear to me — though I am now a founding principal of Regenerative Development Corporation, so weigh that sentence knowing who wrote it. The design is not mine. The stake in it is.
Three: production. Who actually produces verified living-system improvements — repeatably, at volume, at cost, to specification?
Nobody has claimed this one.
Not my work · thirty years of other people’s practice
Not my design · I am a founding principal
The piece I have taken on · not yet built
with nothing
to measure
An instrument with nothing to measure
Here is the arithmetic nobody in this field enjoys.
A standard's value is a function of the volume of qualifying assets. A capital rail's value is a function of the freight it carries. Right now the pipeline of assets that could qualify is bespoke: a few hundred excellent projects, most designed by people who trained directly with the originators, each one substantially a one-off.
Call it artisanal. It is genuinely beautiful work, and at planetary scale it is irrelevant.
That is not a criticism of the practitioners; it is a description of a production function. Artisanal output cannot be underwritten, because underwriting requires repeatability — a known cost, a known timeline, a known outcome distribution. Without those, every project is a fresh act of faith by whoever is signing, and faith does not aggregate. A rail with no freight is a theory of transport.
So the field has a question in front of it that it has mostly declined to ask: is artisanal acceptable?
under-invested
here
Why the field under-invested here, and why the reason was good
The obvious objection to everything in this part is that production means standardization, standardization means abstraction from place, and abstraction from place is the precise failure this entire tradition exists to prevent.
That objection is correct, and the historical evidence is on its side. Every industrial-scale attempt to mass-produce good building has produced sameness at best. At worst it produced the postwar housing blocks we spent the following fifty years demolishing. Industrialized construction has a track record, and the record is that it flattens what it touches. This is the same failure Scott catalogued in § 8, arriving through a different door.
So the field chose craft. Given the tools available, craft was the honest choice.
But "industrialization flattened place last time" is a reason to build a different kind of production system. It is not a reason to skip production and hope the philosophy scales on its own merits. It will not. Nothing does.
the system,
not the outcome
Standardize the system, not the outcome
The resolution is a distinction this field has not drawn sharply enough.
You do not mass-produce places. You mass-produce capability.
Return to the beaver. One construction technique, and a different wetland every single time. The technique is standardized; the outcome is emergent, site-specific and unrepeatable. That gap is the entire design brief — and note that it is the same gap as the governor in § 8. Authoring conditions rather than outcomes is the manufacturing strategy. The ethics and the industrial logic turn out to be the same shape.
What gets standardized is the means:
Structure. A modular building system that carries a place-specific design without re-engineering the means of assembly on every site.
Systems as kit. Energy, water, food and air as integrated, specifiable modules rather than bespoke engineering per project.
Supply chain. Qualified material streams with known provenance and known five-capital effects, so sourcing is a lookup rather than an investigation.
Verification workflow. Data capture designed into construction and operation, so a project produces its own evidence as a byproduct of being built and lived in — not as a consultant's reconstruction at handover.
Delivery competence. Trained crews and operators who execute the above without a founding practitioner standing on site.
What stays singular is everything that matters about the place: what it needs, whose potential is being served, what "better" means here, and at what pace. That is bottleneck one's work, and it does not get standardized. Ever.
The production layer exists to make the relational work cheaper to act on, not to replace it. Today, understanding a place well is the affordable part and building to that understanding is the expensive, slow, artisanal part. That ratio is backwards, and it is a manufacturing problem, not a philosophical one.
What would prove me wrong
A manifesto that cannot fail is not a thesis, it is a mood. These are the four conditions under which I would be obliged to abandon the argument above.
Four conditions
Each of these is measurable, and each resolves on a known timeline. I would rather state them now than have them pointed out to me later.
If standardized production cannot avoid flattening place. If across a real portfolio the places converge — same forms, same species, same social outcome regardless of context — then the craft tradition was right and § 15 is wrong. This is the condition most likely to be waved at and never actually tested, so the appendix sets out how to measure it and what threshold triggers it.
If verified net-additive delta cannot be produced at market-competitive cost. Then generative is a luxury good, not a solution. Premium-priced virtue does not scale, and a thesis that only works at the top of the market has not earned the planetary framing.
If the accounting can be gamed. If a five-capital ledger turns out to be LEED with more columns, then the instrument failed and everything built on top of it is decoration. The whole argument in § 4 — that this is fundamentally an accounting problem — depends on the accounting being un-gameable in practice, not just in design.
If it needs me. If any of it stops producing when I leave the room, then I built a monument, the obligation argument in Part Two was ego in a costume, and the honest thing would be to say so.
Build the model that makes the argument unnecessary
Not a case against what exists. A better-performing alternative, audited, that makes the case redundant.
actually stands
Where this actually stands
The production layer described in Part Four is not built. The companies carrying it are early — some revenue, some reach, no volume. Nothing in this document should be read as a claim that the problem has been solved, or that I have solved it.
I say that plainly because the argument does not depend on it. Part Four is a claim about where the bottleneck is, and that claim is either right or wrong regardless of who clears it. If someone else builds the production layer faster and better, the thesis is vindicated and I am merely early. Those are different propositions and I would rather not have them confused on my behalf.
What I bring to it is twenty-five years of building, failing, selling and acquiring companies since 2001 — most of that in operations and delivery rather than in ideas. The relevant part of that experience is not the wins; it is having been wrong at scale often enough to know what an unbuilt system looks like from the inside, and to recognize the specific optimism that precedes a production failure. The operating layers are led by people with decades of delivery experience in their own domains.
That is worth something, and it is not evidence. Every early venture claims a strong team, and the claim is unfalsifiable, which by this document's own standard makes it inadmissible. So rather than asking to be believed, here is what to hold me to — published, dated, and on the record:
Delivered units, with cost. First cohort of completed homes and integrated systems, published with actual delivered cost per unit and actual cycle time — not target, not pro forma. Including the ones that ran over.
A verified five-capital delta. The first full accounting on a delivered project, independently checked, published whether or not it flatters us. A negative or ambiguous result published on time is worth more to this thesis than a good result published late.
The variance metric. The place-responsiveness measure described in the appendix, reported from the point at which we have enough projects to compute it, with the threshold fixed in advance.
The walk-away test. An annual, public answer to the question in § 8 — what in this structure now operates without me, and what still doesn't.
Dates against each of those belong in this document once they are real, and they will be added rather than implied. Until then the honest status is: the problem is named, the approach is stated, the failure conditions are pre-registered, and the delivery record is empty. I would rather publish it that way than dress it up.
Ten years out
I solve for ten years out and build backward from there. In practice that means the thesis runs through a structure rather than a single company, with each layer existing to protect the one beneath it — so the mission holds through decades rather than deal cycles.
The reason to care about structure more than conviction is unsentimental: conviction is the thing most likely to be sold. Every founder believes their intent will survive contact with a liquidity event, a bad quarter, a board change, a successor. Very few intentions do. Structure is what continues to behave correctly after the person who meant it is gone, distracted, or outvoted.
This is § 8's test applied to an enterprise rather than a project. If the mission requires me to keep meaning it, the mission is not built yet.
The invitation
Three convictions, restated as plainly as I can.
Humans can be a keystone species again — better for our presence, not our absence. The evidence that we have done it before is archaeological, not aspirational; what we lost was the accounting, not the capacity.
I can, therefore I must. Capability is not a credential to display; it is an obligation to discharge, and withholding is a choice you own whether or not you name it.
And generative is the top rung: creating life and abundance outright, in everything we do — standing on every rung below it, demoting none of them.
What follows from all three is a production problem. Understanding is largely answered. The instrument is being built by people who have earned the right to build it. Production is unclaimed, and it is now the binding constraint on everything upstream of it.
Buckminster Fuller put the method better than I can: you never change things by fighting the existing reality — to change something, build a new model that makes the existing model obsolete.
That is the whole approach. Not an argument against anything. A better-performing alternative, audited on the place's own clock, that makes the argument unnecessary.
Is artisanal acceptable?
If it is, we should say so plainly and stop describing this work as a response to a planetary problem — because a few hundred projects is not a response to a planetary problem.
If it isn't, someone has to build a production layer that delivers verified, net-additive improvement repeatably, without flattening the places it touches.
That is the work I have taken on. I would rather be argued with than agreed with. Part Five is where to aim.
Does output variance track input variance?
The flattening test has to separate same means — which is the whole point of a production system — from same ends, which is the failure. One measure does that.
The measure
Standardized means are not the failure mode. The beaver's single technique is the point. The failure mode is convergent outcomes — projects that end up alike regardless of how unalike their places were.
So the quantity to watch is a ratio: how different two delivered projects are, over how different their two places were. Call it place-responsiveness. If sites vary widely and outcomes cluster tightly, the means have colonized the ends and F1 has triggered. If outcome variation scales with site variation, the system is carrying place-specificity through standardized production, which is the entire claim of § 15.
Stated as a test: across every pair of projects in the portfolio, plot outcome distance against site distance. You are looking for a positive slope that holds as the portfolio grows. A flat slope means the system builds the same thing everywhere. A slope that declines as project count rises is the specific signature of creeping standardization capture — the most likely way this fails in practice, and invisible to anyone looking at projects one at a time.
be captured
What has to be captured
The site vector, frozen before design. Each place characterized on a fixed dimension set: ecotype and bioregion, hydrology, climate, soils, existing species assemblage, land-use and disturbance history, tenure and cultural context, community composition, economic base. Distance between two site vectors is how genuinely different those two places are. This must be recorded and locked before design begins — captured afterward, it will unconsciously be fitted to what got built, and the test becomes decorative.
The outcome vector, same discipline. Post-delivery species composition on site, water and energy topology, material palette by source and provenance, structural typology, building form and orientation, program mix, ownership and governance structure, and who performs ongoing stewardship and under what arrangement.
The dimension set has to include things that are expensive to vary — structural typology, hydrology strategy, governance model — and not only things that are cheap to vary. This is the metric's main attack surface: a portfolio can manufacture surface variance through cladding and colour while every building underneath is identical, and a badly chosen dimension set will score that as success.
a real test
What makes it a real test
Pre-registration. The threshold is stated before there is data — a specified minimum correlation between site distance and outcome distance, across a specified number of projects. Setting the bar after seeing results is the difference between falsification and rationalization, and it is the element most likely to be quietly skipped.
Independent scoring. Someone who does not work for us scores the vectors, or at minimum audits the scoring. Self-scored variance metrics are uniformly reassuring.
A blind matching test. Show an experienced practitioner a set of delivered projects and a set of site descriptions, unpaired, and ask them to match. If they cannot do better than chance, the projects are not carrying their places — whatever the numbers say. This is cheap, fast, and harder to game than the quantitative measure.
Asking the people who live there. Whether the place still reads as itself. The quantitative measure can pass while the thing is dead, and residents are the only instrument that detects that.
If the correlation falls below the pre-registered threshold, or the blind matching test comes back at chance, F1 has triggered — and the honest response is to publish that and reconsider § 15, not to revise the metric.
sources
01Denevan, W. M. (1992). "The Pristine Myth: The Landscape of the Americas in 1492." Annals of the Association of American Geographers 82(3), 369–385.
02Cronon, W. (1995). "The Trouble with Wilderness; or, Getting Back to the Wrong Nature," in Uncommon Ground: Rethinking the Human Place in Nature.
03Groesbeck, A. S., Rowell, K., Lepofsky, D., and Salomon, A. K. (2014). "Ancient Clam Gardens Increased Shellfish Production: Adaptive Strategies from the Past Can Inform Food Security Today." PLOS ONE 9(3): e91235.
04Anderson, M. K. (2005). Tending the Wild: Native American Knowledge and the Management of California's Natural Resources. University of California Press.
05Paine, R. T. (1969). "A Note on Trophic Complexity and Community Stability." The American Naturalist 103(929). Origin of the keystone species concept.
06Bar-On, Y. M., Phillips, R., and Milo, R. (2018). "The biomass distribution on Earth." PNAS 115(25), 6506–6511. Source for humans as approximately 0.01% of global biomass.
07Jones, C. G., Lawton, J. H., and Shachak, M. (1994). "Organisms as Ecosystem Engineers." Oikos 69, 373–386.
08Worm, B., and Paine, R. T. (2016). "Humans as a Hyperkeystone Species." Trends in Ecology & Evolution 31(8).
09Reed, B. (2007). "Shifting from 'sustainability' to regeneration." Building Research & Information 35(6), 674–680.
10Reed's assessment of LEED — that it "defines a level of greenness that aspires towards sustainability, but it does not explain how to develop regenerative relationships" — is from his Passive House Canada profile interview. Reed was a founding board member of the US Green Building Council and co-chaired the LEED Technical Committee from 1994 to 2003.
11Mang, P., Haggard, B., and Regenesis. Regenerative Development and Design: A Framework for Evolving Sustainability. Source for the Story of Place methodology.
12Scott, J. C. (1998). Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed.
13The Fuller formulation — "you never change things by fighting the existing reality; to change something, build a new model that makes the existing model obsolete" — is widely attributed to Buckminster Fuller and appears in this phrasing in Dan Millman's Way of the Peaceful Warrior (1980).
Better For Our Presence · v1.1 · revised 4 September 2026 · Eric Amyot · keystonespecies.co/manifesto