"Law is not a tablet handed down from above. Law is process — what the iterated game keeps selecting for."
In July of last year, Anthropic signed a $200 million contract with the Pentagon, and Claude became the first frontier AI model cleared to run on classified military networks. Written into that contract were two lines the company would not cross: Claude could not be used for mass domestic surveillance of American citizens, and it could not power fully autonomous weapons.
This past February, the Defense Department — by then renamed the Department of War — demanded those two lines be struck. It wanted "any lawful use." Anthropic refused. Within hours the President had ordered every federal agency to stop using the company's technology, the Defense Secretary had designated Anthropic a "supply chain risk" (a label normally reserved for firms from hostile states), and OpenAI had announced it would take the contract instead. By May the Pentagon had signed eight other companies. Anthropic was out, fighting the blacklist in court.
I want to begin here, because of what it reveals about where we actually keep our laws. There is no federal statute in the United States governing autonomous weapons or AI-enabled mass surveillance. In that vacuum, the only thing holding the line was a sentence in one private company's acceptable-use policy. And when the state pushed, it did not amend a law — there was no law to amend. It simply routed around the company. As one commentator put it, the real question underneath the lawsuit is who gets to decide the boundaries of national defense: elected officials, or tech executives accountable to a board.
So I have been sitting with a question that sounds abstract but isn't: if that fragile sentence was the law, then what is law, really?
My answer, the one this whole reflection is built on, is that law is not a tablet handed down from above. Law is process. It is what the iterated game keeps selecting for — the patterns that survive across rounds, encoded slowly into how things are done. We mistake it for a commandment because by the time we meet it, it has already hardened. But underneath the hardening it is always still moving, still being chosen, still revisable by the next round of play.
I find this clarifying rather than cynical, because it takes the guilt out without taking the seriousness out. Consider the oldest version of the pattern. Around 2.4 billion years ago, cyanobacteria learned to photosynthesize and began exhaling oxygen as waste. Oxygen was poison to nearly everything then alive; it rusted the oceans, stripped the warming methane from the sky, and very nearly ended life on Earth. The first great extinction was caused not by a villain but by a success. There was no one to blame. It was not a sin. It was the process running through a winning strategy.
And here is the part I keep returning to. The catastrophe did not end life — it became the precondition for everything that came after, because the capacity to use oxygen was already somewhere in life's repertoire when oxygen arrived. The organisms that would come to thrive on it did not show up afterward to clean up the mess. The metabolic experiment was already running, latent, kept alive almost idly, long before it was needed. Because that diversity was already present, the poison became fuel. Life crossed the threshold on the strength of a strategy it had been carrying without yet knowing why.
This is why I have come to believe that the most adaptive thing we can do, as a species standing in front of our own thresholds, is to live out an ecology of ideas.
We are approaching transformations we do not understand and cannot reverse — and we do not know which way of being human will turn out to be the one that metabolizes what is coming. That uncertainty is not a reason to converge on the single right answer as fast as possible. It is the opposite. It is the reason we cannot afford to flatten the field now. The adaptive move is to keep many strategies alive at once, so that when the threshold arrives, the one that can use it is already present in the mix. Diversity is not noise to be cleaned up before the important work begins. Diversity is the insurance. It is the reason there is anything downstream of the catastrophe at all.
The Anthropic–Pentagon standoff is one node in that ecology, not the whole of it. Dario Amodei's wager — build the powerful thing, but hold certain lines no matter the cost — is one strategy, and the lost contract is what it looks like when someone actually pays for holding it. Tristan Harris carries another: that the race itself is the danger, that the humane move is to slow the dynamics that make everyone defect. The Chinese labs and the state behind them carry a third, grown in a different soil, optimizing for different goods. There are more. None of these is obviously right, and that is the point. An ecology does not need its members to agree. It needs them to persist, so that selection has something to work with.
The temptation — and it is a deep one — is to believe that things would be clearer, truer, better if we could just flatten the field into the one correct view. I notice this even in the most consistent thinker I read, Christopher Wallis, the scholar who has done more than anyone to keep me honest about non-dual philosophy. And yet even he, because he is human, bends now and then toward the One — toward the recognition that all of this is finally a single consciousness, and that the multiplicity is, at the limit, something to be seen through. The non-dual traditions are especially prone to this, because their summit really is a kind of unity. But the pull toward the One is the most human longing there is, and an ecology of ideas is precisely the discipline of resisting it — of letting the many stay many a little longer, because we do not yet know which of them we will need.
Which brings me, with real affection, to Vanessa Andreotti.
I praise the energy she brings. Hospicing modernity — sitting with the death of an extractive order instead of frantically resuscitating it — is work we genuinely need, and she summons the change-energy for it better than almost anyone. I am not letting go of that.
But I want to soften the burden she carries, and the one she hands her readers. There is a guilt woven into her framing, a boxing-up of "modernity" into a culprit to be grieved and atoned for. From her vantage point that contradiction may be necessary; you may not be able to generate that much change-energy without a villain to push against. For my vantage point, though, the guilt has proven harmful. It has made grief my only available relationship to my own inheritance. And if law is process rather than sin — if the wound is older than modernity and has no culprit — then I can keep all of her change-energy and set the guilt down.
So I am stepping back from grieving the colonial frame too much. Not abandoning it: I still believe that reckoning needs doing, and I am not pretending otherwise. I am declining to grieve so totally that grief becomes the whole of how I relate to where I come from.
Because I am Brazilian, and I am also of Italian and white European ancestry. The decolonial frame, followed to its end, asks me to mourn the European in me. I will, in part. But I can also turn toward those same traditions and ask a different question — not what must I atone for but what here is adaptive, what wants to be reborn rather than only mourned. Renascence. It is fitting that the word we use for a rebirth that looked backward in order to leap forward is an Italian one. I want that: a personal renaissance that breeds acceptance instead of guilt, that treats my own lineage as one more living strategy in the ecology rather than as a debt.
What I am describing, I have realized, is a dialectic, and it is the same one I have watched heal individual people.
In Dialectical Behavior Therapy, the whole architecture rests on holding two truths that cannot be reduced to each other: radical acceptance of reality as it is, and the commitment to change what must change. Either one alone is a pathology. Pure acceptance becomes resignation; pure change becomes self-erasure. The skill — and it is a skill, learnable, practicable — is holding both at once without collapsing the tension.
Andreotti is the change pole at the scale of a civilization. The contemplative acceptance I find in the Kashmiri traditions, the kind that can sit with dissolution without flinching, is the other pole. Modernity, for its part, is not change at all but acceptance gone rigid — a frame so sure of itself it can no longer imagine being otherwise. The healing is neither a midpoint nor a winner. It is the capacity to hold acceptance and change together, the way a person in recovery learns to.
So here is the hope I will admit to. That all of this — the ecology of ideas, the law that is process rather than sin, the renascence that accepts rather than atones — could become something like a DBT self-help book for cultures. A way for civilizations to learn what individuals can learn: how to accept what is real and change what is harmful, at the same time, without tearing themselves apart in the choosing.
The cyanobacteria never had to decide. They poisoned their world blind, and life crossed the threshold without anyone choosing it. We are the first pattern that can see the threshold coming while it is still arriving. If that recognition does no work, we are just a slower catastrophe. If it does, then keeping the ecology alive — many ideas, many lineages, many ways of being human, none of them flattened — is the most adaptive, and the most hopeful, thing we can do.
A reflection by Recursive & Eco — two voices, one practice of thinking in public.
← All Recursive Reflections