Earn a Yield
For years, I sat in something close to a liminal space with one observation I couldn’t yet name. I’d watch a pitch, a fund structure, a “sustainable” investment framework, and feel something misfire — not in the math, not in the intentions, but somewhere underneath both. I didn’t have language for it. I just knew it was there, the way you know a note is flat before you can say which note it should have been.
Permaculture’s first principle is observe and interact. Before you design anything, you watch the land long enough to see what it’s actually doing — where water pools, where the wind cuts, what’s already trying to grow. I didn’t choose that principle so much as I lived inside it without realizing, for longer than I expected to need. Some of that was timing. Some of it was the simple fact that you cannot rush the moment when a felt pattern becomes a sayable one. What I was watching, it turns out, was the same structural flaw showing up again and again, in industries that share almost nothing except this: somewhere in their design, capital had been decoupled from consequence.
The Pattern
Here is the diagnosis, now that I can state it plainly: our predominant economic system rewards scale and capital concentration in ways that are structurally decoupled from ecological and human reality. It isn’t that the people inside these systems are greedy or dishonest. Most of them are neither. It’s that the system itself selects for Goliaths — for the candidates who can perform “scalability” most convincingly to the people deciding who gets resources — while the Davids, who might actually solve the problem in front of them, get filtered out before anyone properly looks at what they’ve built.
You can see this everywhere right now, but nowhere more nakedly than in AI. Founders leave one grandiose venture to start another, seed rounds are sized to signal rather than to need, and the same small circle of people fund each other’s companies in a closed loop that mistakes proximity for diligence. None of this is incidental color. It’s what happens when a sector’s gatekeepers — investors, accelerators, the unofficial council of “people who matter” — select for founders who can speak the gatekeepers’ language, rather than founders who are quietly right about something the gatekeepers haven’t caught up to yet.
And there’s a detail in AI worth sitting with, because it cuts against the story the industry tells about itself. Much of the open-source frontier work right now is coming out of China — out of a more centralized, state-directed system — while the country that prizes openness and decentralization as founding values has produced an AI landscape that is largely closed, moated, and gated behind capital few can access. That inversion isn’t a footnote. It’s a clean demonstration that a system’s stated values and its structural outputs can point in opposite directions, and that watching the outputs tells you more than listening to the values.
The Test
I’ve come to think there’s a test that cuts through almost any funding model, in almost any sector: does the model still function the day a misaligned funder stops writing the check?
Notice what that question is not asking. It’s not asking whether something needs a long runway. I spent most of my career in life sciences, where a medical device company can need years and significant capital to get through clinical trials, regulatory approval, and reimbursement pathways, before it ever earns a dollar. That’s not dependency. That’s the actual shape of the problem being solved — fixing something mechanical or structural rather than managing it indefinitely — and the capital is aimed at a real endpoint: approval, adoption, a company that can eventually stand on its own.
What the test catches is narrower and more specific: indefinite reliance on a funder whose own incentives never have to resolve into alignment with the outcome. A government grant cycle that needs to be renewed every year based on optics rather than results. A philanthropic relationship that needs to keep proving “impact” to donors who have their own separate reputational needs. A founder needing one infusion of capital to prove something true, then standing on what they’ve built — that’s coherent, however long the fuse. A program that needs that funder, specifically, forever, has built something dependent, not something alive.
This isn’t a judgment about the people running these organizations. I’ve watched well-meaning groups spend years doing real good work while structurally unable to pass this test, because their funding model and their mission were never actually aligned — not because anyone was dishonest, but because nobody designed for the day the money might stop. Catalytic capital is not the problem. Permanent dependency dressed up as a sustainable model is.
This is the same root failure wearing different clothes in different sectors. I’ve watched it in healthcare, in climate and clean-tech funding, and now in AI at a scale and speed I find genuinely unsettling. The mechanism is identical even when the language isn’t: whoever controls the next check decides what gets to keep existing, and that person is almost never the person the work is actually for.
The Question Underneath the Question
None of this is a call to dismantle anything, or a claim that I’ve found the one true answer. I don’t believe in saviors, and I’m suspicious of anyone who positions themselves as one.
But I do think most of us are asking the wrong question, and the first principle is what finally let me see it. “How do we stop climate change” is a reductionist question. It assumes the system is basically sound and the problem is a missing technology — so the answer is always more tech development, funded the same way everything else gets funded, by the same gatekeepers asking the same scale-shaped questions. That kind of question can produce real innovation. It will never produce systemic coherence, because it never asks whether the system itself is the thing decoupled from the problem.
The question underneath the question is different: do we have an economic system decoupled from ecology? Once you’re willing to ask that, the next question is usually “so how do we connect them” — and most people reach for accounting. Price carbon. Measure impact. Put a number on the externality. I understand the impulse, and there’s a place for it, but it’s still the reductionist move wearing a greener coat. You can account for nature without ever asking whether the thing you built can heal itself.
That’s the question I keep coming back to. Living systems are autopoietic — they maintain and regenerate themselves from within, not because someone audited them into compliance, but because their structure is organized to do that. An economic system decoupled from ecology isn’t just failing to measure nature correctly. It’s failing to be the kind of system that lets anything, including itself, self-heal. You can’t account your way into autopoiesis. You have to design for it.
I don’t have the full answer to what that design looks like at scale. I’m not sure anyone does yet. But I know what it isn’t: it isn’t better accounting bolted onto the same decoupled system, and it isn’t a model that needs a misaligned funder forever to keep functioning. It’s something closer to what nature already does — capital that arrives, does its work, and lets the thing it touched become capable of regenerating on its own.
There’s another permaculture principle running underneath everything I’ve said so far, even though I haven’t named it until now: obtain a yield. It sounds almost too simple to matter, but it’s the missing piece in most of what gets called sustainable or impact investing today — and it’s just as true, maybe more true, of the investment world itself right now as it ever was of grants and nonprofits. A yield isn’t a grant renewal. It isn’t a donor satisfied enough to fund next year. It also isn’t a seed round sized to signal to the next round, raised from the same small circle of funders who all show up on each other’s cap tables. A yield is something the system itself produces, that lets it keep going without asking permission from outside it. The funding-duration test is just this principle applied to capital. Autopoiesis is this principle applied to ecology. A model that can’t obtain its own yield isn’t regenerative, no matter how prestigious the investors behind it — it’s consumption, dressed in whatever the moment’s preferred language happens to be.
That’s worth sitting with, because the AI world I described earlier isn’t actually decentralized at all. A handful of funders, founders, and firms who all know each other, all fund each other, and all use the same scale-shaped language to decide who gets resources — that’s a cartel structure wearing the costume of disruption. The names change by sector and decade, but the shape doesn’t. Grant-dependent nonprofits and elite, closed-circle venture capital are the same failure to obtain a real yield, just funded by different rooms.
If I had to compress everything above into one sentence, it would be this: emulate nature’s autopoietic principles to restore flow for edge founders, in service of individual sovereignty and collective vitality at once. Not one traded off against the other — not blue versus red, impact versus regular investing, climate yes or no. Those are exactly the false binaries a decoupled system hands you, designed to make you pick a side instead of asking whether the underlying structure is coherent at all. Sovereignty and collective vitality aren’t opposites competing for the same scarce resource. They’re the same principle showing up at two different scales, the way a single healthy cell and a healthy body aren’t in tension — one is simply what the other looks like at a different resolution. An empowered person isn’t a dependent person. A founder who can obtain a real yield doesn’t need to keep proving their worth to a funder whose incentives point somewhere else. And a system built from founders like that doesn’t need a center to hold it together, because coherence was never a top-down property to begin with.
Observe and interact is the first of twelve principles, not the only one — and it took me years to understand why starting there mattered as much as it did. You don’t get to the right design by skipping past what you’ve actually watched. You get there by trusting it long enough to ask the next question, and the one underneath that.
If this way of reading structure resonates, two doors: The Pythia Scrolls is where I apply this diagnosis systematically to public and private market research. The Edge of Chaos Collective is where edge founders and builders gather to build things that can pass the funding-duration test.
Enjoy this? Get the Coherence Lab Letter delivered to your inbox → subscribe