From Building Bunkers to Planting Seeds

On the Physics of What Survives

Originally published on Medium, February 18, 2026.

There are two responses to a system under stress: fortify what exists, or plant what survives the fire. Seeds!

The bunker promises safety through separation. Thicker walls. Deeper reserves and wider moats. More sophisticated defenses against an increasingly hostile world. The bunker logic says: if we can just hold out long enough, if we can engineer our way past the breaking point, we’ll emerge on the other side intact.

The seed makes a different bet entirely. The seed doesn’t resist the fire. The seed carries a pattern through it. When the old growth burns, the seed is already encoded for what comes next.

This distinction matters now more than it has in generations. Not because the challenges we face are unprecedented (they aren’t) but because the tools of acceleration have outpaced the tools of navigation. We are attempting to steer through a storm with instruments calibrated for calm water. The compass spins. The charts show coastlines that no longer exist. And the waves keep building.

This is not an alarm. Alarms are for emergencies that can be resolved by waking up and taking action. This is an orientation, a way of seeing what’s actually happening and where the exits lead. Some exits lead deeper into the bunker. Some lead to open ground where seeds can take root.

The Diagnosis

Systems theorist Donella Meadows spent her career studying why interventions fail. Her research identified twelve places to intervene in a system, ranked from least to most effective. At the bottom: adjusting parameters like taxes, subsidies, interest rates, and standards. At the top: changing the paradigm, the shared assumptions out of which the entire system arises.

Her conclusion was unsettling and remains so: “People who are raised in the industrial world are trained to think that parameters are the most important places to intervene. But they rarely change behavior.”

We keep adjusting parameters while the paradigm burns.

Consider three numbers that define our current position:

$318 trillion. Total global debt as of late 2024, according to the Institute of International Finance. Against roughly $100 trillion in global GDP. We have built claims on the future at roughly three times current annual production. This is not a policy problem awaiting a clever solution. This is an arithmetic problem: the gap between what we’ve promised and what exists to deliver on those promises.

The number itself is almost too large to comprehend, so consider what it means in practice. We have borrowed from the future on a scale that assumes perpetual growth to pay the interest. When that growth falters, as it periodically does because arithmetic doesn’t negotiate, the promises start breaking. Not gradually. Suddenly. The way trust collapses. The way markets freeze.

36 percent. The share of people globally who believe the next generation will be better off than this one, according to the 2025 Edelman Trust Barometer. In France, nine percent. In the UK, seventeen percent. In the United States, not much better.

Five of the ten largest economies now rank among the least trusting nations surveyed: Japan, Germany, the UK, the United States, and France. Government and media are tied for least trusted institutions globally. We have depleted the shared resource that makes coordination possible. Trust is infrastructure. When it erodes, every other kind of infrastructure becomes harder to build and maintain.

945 terawatt-hours. The projected electricity consumption of data centers by 2030, according to the International Energy Agency, roughly equivalent to the entire electricity consumption of Japan today. Something like a doubling within the decade’s first half. In the United States alone, data centers could drive nearly half of all electricity demand growth this decade.

We are building cognitive infrastructure on an energy foundation that is already straining. The substrate on which AI, cloud computing, and the digital economy run requires electricity at scale. In Virginia, data centers consumed 26 percent of the state’s total electricity in 2023. One state. One sector. A quarter of the power.

These aren’t separate crises with separate solutions. They’re the same crisis wearing different masks: systems claiming more than their environment can provide, sustained temporarily by borrowing from the future, from trust, from the planet itself.

Each of these numbers has nuance. Debt can be restructured. Trust can rebuild. Energy grids can expand. These are signals, not prophecies. But signals pointing the same direction deserve attention.

The bunker response is to patch each leak individually. More regulation for debt. Better messaging for trust. Efficiency gains for energy. But leaks in a submarine don’t stop because you patch them faster. They stop when you address what’s creating the pressure in the first place.

The Mechanics

Here’s what makes this hard: no one intended it.

There is no cabal. No conspiracy. No shadowy group pulling strings behind the curtain. What we’re facing is something more difficult to address than villainy. It’s the accumulated weight of rational actors making reasonable decisions that sum to collective catastrophe.

Nassim Nicholas Taleb calls this the “skin in the game” problem. When decision-makers don’t bear the consequences of their decisions, feedback loops break. The rating agencies that blessed toxic assets as safe in 2008 didn’t hold those assets on their own books. The platforms optimizing for engagement don’t experience the polarization they produce. The systems shipping hallucinating AI don’t navigate by its outputs. The risk is externalized. The reward is captured.

“Never trust anyone who doesn’t have skin in the game,” Taleb wrote. “Without it, fools and crooks will benefit, and their mistakes will never come back to haunt them.”

But this framing still implies bad actors who could, with sufficient pressure, be replaced by wiser and more ethical alternatives. The deeper truth is more uncomfortable: the harm is not produced by monsters. It’s produced by ordinary people following ordinary incentives that happen to sum to catastrophe.

The rating agency analyst trying to keep their job by not being the one who costs the firm a client. The platform engineer trying to hit their quarterly metrics by optimizing what works. The fund manager trying to beat the benchmark because that’s how compensation is structured. Each choice is locally rational. The aggregate is globally destructive. And everyone can see it happening without being able to stop it.

This is what economists call a collective action problem. Each actor benefits from defecting from the common good. No actor can stop defecting unilaterally without suffering competitive disadvantage. If one rating agency gets strict, the others capture the business. If one platform reduces engagement optimization, users migrate to more stimulating alternatives. If one fund reduces leverage, its returns lag and investors flee.

Everyone sees the problem. No one can solve it alone. Awareness does not produce resolution. This is the trap. Everyone is rational. The outcome is irrational. No one can stop. Or can we?!

The Terrain Change

In the last decade, the information environment has fundamentally changed. Not because people became more gullible or less educated. Because the tools of manipulation became faster than the tools of verification.

We have entered an era where synthetic content can be generated at machine speed: text, image, audio, video. The asymmetry is no longer between large institutions and individuals. It’s between machine-speed offense and human-speed defense.

An attack deploys in milliseconds, scales in hours. Detection takes days. By the time the fact-check appears, the narrative has moved on. The damage is done.

And here’s what most people miss about the new terrain: the target is not what you think.

The goal is not to convince you of a particular lie. Lies can be checked. The goal is to exhaust your ability to distinguish truth from lie. To make verification so costly that you stop trying. To flood the zone until you retreat to tribal heuristics, believing what your group believes because independent judgment has become unsustainable.

Memory isn’t just storage. It functions like a prediction engine, the accumulated pattern-matching that lets you navigate the world. When the information environment is corrupted at scale, your predictions are corrupted. The result is navigating by fuzzy map with holes. And you don’t know the true map until you hit the wall.

Whether driven by profit optimization, geopolitical advantage, or political power, the motivations converge: degrade the target’s ability to predict, and you control their trajectory.

This is not a problem you can solve by trying harder to know what’s true. The terrain itself has changed. You cannot out-swim a riptide through effort alone. You have to understand the current and move perpendicular to its pull.

The Bunker Response

Watch how institutions respond to this terrain. They reach for the tools they know.

Content moderation. Fact-checking labels. AI safety guidelines. Financial stress tests. Disclosure requirements. Each is a rule layered on top of a system that is already breaking the rules. Each operates at what Meadows would call level five or six, the realm of rules and information flows.

But the problem lives at level two: the paradigm. The shared assumption that we can sustain claims that outrun underlying capacity. That we can build trust by mandate. That we can moderate our way to shared reality while leaving the incentives for destroying shared reality firmly in place.

Consider: content moderation doesn’t change engagement optimization. It plays whack-a-mole with the outputs of an incentive structure that remains unchanged. As long as outrage drives engagement and engagement drives revenue, the system will produce outrage. You can remove individual pieces of content. You cannot remove the gradient that produces them.

Financial regulation doesn’t change leverage incentives. It adds friction and reporting requirements to a system still built on the assumption that claims can outrun underlying capacity. The regulations get more complex. The workarounds get more sophisticated. The underlying pressure remains. Causing blocking and preventing building.

AI guidelines don’t change disembodiment. They measure, instrument, and ask nicely that systems without skin in the game behave as if they had skin in the game. They propose ethical frameworks for entities that cannot feel ethics, operated by companies whose incentives point elsewhere.

Meadows ranked paradigm change as the second most powerful intervention in a system. Only transcending paradigms entirely ranked higher. Rules are level five. We keep fighting at level five while the problem lives at level two.

I hope I’m wrong about this. I fucking hope the bunker holds. I hope the engineers find patches that work, regulations that stick, guidelines that matter. The people building those patches are not fools. They’re often brilliant, and they’re trying. And I support you and your work.

Breaking the fourth wall: Thank you to those who serve, build tech for good, and talk about these challenges in earnest. If this is a seed, and this grows, I want to grow and build with you. Please reach out any time if you want to collaborate.

But in the high desert where I live, the bosque catches fire. Not every year, but often enough. The cottonwoods along the river burn. And afterward, what returns depends on what was encoded to survive. You don’t plant seeds because you want fire. You plant seeds because fire happens, and pretending otherwise isn’t optimism. It’s denial.

The Seed Logic

But here’s what the bunker logic misses: the collective action problem has been solved before. Many times. In many places. For centuries.

Economist Elinor Ostrom won the Nobel Prize in 2009 for proving that the tragedy of the commons is not inevitable. She spent decades documenting communities around the world that have successfully managed shared resources for generations: fisheries in Maine and Turkey, forests in Japan and Switzerland, irrigation systems in Spain and the Philippines, grazing lands across three continents. They didn’t collapse into the predicted tragedy. They thrived.

The prediction of inevitable tragedy assumed that people couldn’t coordinate without either private property or government control. Ostrom showed this was wrong. Communities develop their own institutions. And the institutions that work share structural features:

• Clear boundaries: who belongs to the commons and who doesn’t. Rules matched to local conditions, not one-size-fits-all mandates from distant authorities, but agreements fitted to specific contexts.
• Collective choice arrangements: those affected by the rules participate in making and modifying them. Monitoring by participants, not external enforcement agencies but mutual accountability among those who share the resource.
• Graduated consequences: proportional responses that escalate only as needed.
• Conflict resolution mechanisms: accessible ways to address disputes without destroying the commons itself.
• Recognition of the right to organize: external authorities that support rather than undermine local governance.

These aren’t abstract principles. They’re design specifications. And they describe the opposite of what we’ve built.

Our platforms have unclear boundaries, universal rules, no user participation in governance, opaque external monitoring or none at all, binary enforcement, no accessible conflict resolution, and active undermining of alternative structures.

Our financial systems have unclear boundaries between speculation and investment, rules divorced from local conditions, no participation by those most affected, monitoring that serves insiders over the public, consequences that fall on taxpayers rather than risk-takers, and active suppression of alternatives.

The seeds already exist. Local-first software where your data lives on your devices, and exit is always possible. Federated protocols where no single entity controls the network, and governance is plural. Open source AI where the weights are public and the code is free, and capture requires capturing everyone. Community-owned infrastructure that answers to participants, where skin in the game is structural, not optional.

I want to be clear about what I’m saying and what I’m not.

I am not betting on collapse. I am not hoping the bunkers fail so the seeds can inherit what’s left. That would be accelerationism, and accelerationism is just another bunker with different aesthetics, a bet that you’ll be on the right side of the destruction you’re cheering for.

The hope is that the fire doesn’t come. The hope is that the patches hold, that the brilliant people working on level-five interventions find something that sticks, that we discover ways to release pressure before the hull breaches. That would be good. I would be relieved to be wrong.

But hope is not a strategy. And pretending the numbers don’t say what they say isn’t optimism. It’s a different kind of bunker, one built of denial instead of steel.

This is a hedge. This is what clear-eyed risk assessment looks like when the stakes are civilizational. You don’t buy insurance because you want your house to burn down. You buy it because you understand that wanting has nothing to do with what happens.

And here’s what the critique misses: seeds are worth planting even if the bunker holds. Even if the debt gets restructured. Even if the platforms find ethics. Even if AI develops feedback loops that keep it honest. A world with more local-first software is better than one without. A world with federated protocols alongside centralized platforms is more resilient than one with only centralized platforms. A world where community-owned alternatives exist, even if most people never use them, is a world with more options, more pressure on incumbents to behave, more ways to say no.

Seeds don’t require collapse to be valuable. They just require someone to plant them.

The seeds exist. They’re being planted now, in small communities and open repositories and cooperative structures that most people haven’t heard of. Most of these seeds may never need to be the primary system. Most of them will simply make the existing systems more honest, by providing alternatives that keep the bunkers from becoming the only option.

Non-Participation and Building

There’s a story that Wangari Maathai, the Kenyan environmentalist and Nobel laureate, used to tell.

A great forest is consumed by fire. All the animals come out and stand transfixed, watching it burn. They feel overwhelmed. Powerless. Except for one small hummingbird. It flies to the nearest stream, takes a single drop of water in its beak, flies back and drops it on the fire. Then returns for another drop. And another.

The other animals, much larger, watch this. The elephant, with its trunk that could carry so much more water, stands there doing nothing. Finally they say to the hummingbird: What do you think you’re doing? You’re too small. The fire is too big. Your wings are too little. You can only bring one drop at a time.

The hummingbird doesn’t stop. Without wasting a moment, it turns to them and says: I am doing the best I can.

That’s not a strategy for defeating fire. It’s a statement about integrity. About what you do when the scale of the problem exceeds your capacity to solve it. You do what you can. You don’t wait for the elephants to move first.

The most powerful thing you can do starts with what you stop doing.

Non-participation is not withdrawal. It’s recognition. Every engagement with extraction systems feeds them. Attention is not neutral. Attention is fuel. Your outrage is monetized. Your engagement trains the algorithm. Your participation, even in opposition, strengthens what you oppose.

Recognize when your prediction engine is being targeted. The manufactured urgency that demands immediate response. The outrage cycle that keeps you clicking. The content designed to bypass your prefrontal cortex and speak directly to your amygdala. When you feel the pull, the compulsion to react and share and engage, pause. Legitimate action survives a 48-hour wait. Manufactured urgency does not.

Verify before amplifying. The cost of checking is small. The cost of spreading corruption is large. Every time you share without verifying, you become part of the machine-speed attack. Be the friction that manipulation cannot afford.

Slow down when urgency is manufactured. Real emergencies don’t need emotional acceleration. They present facts and trust you to respond. Manufactured emergencies need your emotions overriding your judgment. The tempo itself is diagnostic.

But non-participation alone is not enough. The minnow doesn’t just avoid poison. It needs clean water. We have to build the environments that make honesty possible.

This is the constructive work: building environments where the constraints are structural, not policy. Where you can’t claim more than capacity because the system won’t let you, not because a rule says you shouldn’t. Social structures that distribute rather than concentrate. Institutions designed for collective action rather than collective extraction. Businesses where skin in the game is mandatory. Communities where the participants maintain the commons because the commons maintains them.

In the high desert of New Mexico, farmers have maintained acequia irrigation systems for four centuries. The parciantes, the members, govern themselves. They share the water according to established rights. They share the labor of maintaining the ditches each spring. The physics of gravity and water flow determines who can take and when. No one enforces the rules from outside. The water enforces itself.

This is what seed logic looks like in practice. It’s old. It works. And it’s waiting to be applied to information, to finance, to AI.

Stop feeding what feeds on you. Plant what survives the fire.

What We Carry

When I was a child, my grandmother cared for me in the afternoons. We’d sit together on her front porch, the kind with that green patio carpet everyone had in the nineties, the front of the house covered in big leaves from a vine that had been climbing the brick for years.

My grandparent’s yard was magnificent, luscious, and brilliantly green. Colors popping from every bed, including the lavender bush that pulls in the bees all summer long.

In the fall, the marigold flowers would dry up, leaving behind small pods full of tiny seeds. My grandmother showed me how to pluck the old flower off, pinch and roll the pod between my fingers, and spread the seeds into the small flower bed by the front door.

She didn’t explain the botany. She didn’t lecture about plant reproduction or seasonal cycles. She just showed me what to do. How to put my love and hands on the inevitable cycle. How to carry what the flower had made into the next year.

In a gift economy, the measure is not what you accumulate. It’s what you pass on.

What’s your seed story? What are you carrying into next year, and who will you plant seeds with?

The fire may come, or it may not. But the seeds are worth planting either way.

Chris Perkins writes from Las Cruces in Doña Ana County, New Mexico, where he lives with his wife Nayomi, their two daughters, and three dogs. He is the founder of NMCITRA (New Mexico Cyber Intelligence & Threat Response Alliance) and the architect of the Kinetic Trust Protocol. His work bridges communities and technology through data literacy and data storytelling, empowering people to see the systems that shape their lives.

This is the third of four essays. The previous, “The Abundance of Trust,” named the layer beneath coordination. The next, “Surviving Adolescence,” asks how a civilization answers the question the aliens already knew.