Dissolve

In January 2025, researchers at Stanford published results from a¹ brain-computer interface study that received less attention than it deserved. Using intracortical electrode arrays implanted in motor cortex, they decoded imagined handwriting — not executed handwriting, not attempted movement, but the planning signals that precede physical action. Accuracy: 94 characters per minute with a character error rate below two percent. A separate group at UC Berkeley,² using fMRI data rather than implants, demonstrated that visual imagery — what you see when you close your eyes and imagine something — can be reconstructed from brain activity with enough fidelity to be recognizable by category and composition, though not yet in photographic detail.

The technology remains years from anything approaching consumer deployment. The implications are closer than the technology. What these studies demonstrate is not that your thoughts are readable — not yet, not at the resolution the word “readable” implies. What they demonstrate is that the brain produces measurable physical signals when you think, and that those signals carry recoverable structure. The interior is not silent. It has always been broadcasting. What has changed is the sensitivity of the receivers.

You do not need to read the brain to read the person. You barely need to try.

Cambridge Analytica demonstrated in 2016³ that five Facebook likes were sufficient to outpredict a person’s coworkers at assessing their personality; ten likes outpredicted their family members; three hundred likes outpredicted their spouses. What was being predicted was not mere behavior. It was the interior life: political tendencies, psychological vulnerabilities, the things people experience as most privately their own. The prediction required no access to the interior. It required only access to the behavioral residue the interior left on the surface of a commercial platform.

At sufficient resolution of behavioral data, you do not need brain-computer interfaces. You do not need implants or fMRI machines or any exotic technology at all. You need a browser history, a purchase record, a location trail, a pattern of social media engagement. The interior was never sealed. It was always leaking. The BCI research is a dramatic demonstration of a principle that was already operationally true before any electrode touched any cortex: given enough data about what a person does, you can infer what they think, what they feel, what they want, what they fear, and what they will do next, with an accuracy that makes the philosophical distinction between “observing behavior” and “reading the mind” practically meaningless.

And now the absence leaks too.

At sufficient resolution, silence is as informative as speech.⁴ The gap where a message should have arrived and didn’t. The search you didn’t perform. The location you avoided. The news you didn’t react to. A system with comprehensive baseline data can read refusal and absence as fluently as it reads presence and action. There is no null option. Every choice, including the choice to opt out, to go dark, to go silent, writes in the record. You cannot opt out of being legible by refusing to emit signals. You would have to not exist.

The standard response to surveillance technology is a privacy argument: this is my information, it belongs to me, you cannot have it without my consent. The argument is clean, intuitive, and widely shared across the political spectrum. It is also built on a philosophical foundation that has been collapsing for three hundred years.

The foundation is the autonomous self: a private interior, sovereign over its own contents, standing behind experience as its owner and author. Descartes needed it to do epistemology.⁵ Locke needed it to ground property rights. Kant needed it as the condition for moral agency. The entire Enlightenment project — rights, contracts, individual liberty, democratic participation — is architecturally dependent on there being a self that has these things, exercises these faculties, owns this interior, whose autonomy can be violated or protected.

The problem is that the self, examined carefully, isn’t there.

Hume looked for it and found only a bundle of perceptions⁶ — a stream of experience with no stable experiencer behind it. When he tried to locate the self that was having the perceptions, he found more perceptions, all the way down. The self was not a substance. It was a story the perceptions told each other.

Two thousand years earlier, Buddhist philosophy arrived at the same place through a different route. Anattā — non-self⁷ — is not nihilism; it is not the claim that nothing exists. It is the claim that among the things that exist, a fixed, bounded, self-identical self is not one of them. What exists is a process: a continuous arising and passing of experience, each moment conditioned by the last, none of them owned by an entity that transcends them. The continuity you experience as selfhood is real the way a river is real: not a thing but a pattern of flow, not a substance but a relationship between processes.

None of this is fringe philosophy. It is the considered conclusion of some of the sharpest minds who have ever turned sustained attention toward the question of what a self is. And for most of the history in which these conclusions were available, they were philosophically interesting and practically irrelevant. You could agree with Hume at the level of theory and still vote, own property, sue and be sued, enter contracts, and be held morally responsible. The autonomous self was a fiction, but it was a useful fiction, and the world was organized around it, and that organization mostly worked.

What is changing is not the philosophy. What is changing is that the fiction is becoming operationally untenable.

The week that Operation Epic Fury began — the joint US-Israeli strikes⁸ on Iranian nuclear and command infrastructure that opened in the last days of February 2026 — Palantir’s CEO Alex Karp took the stage at the a16z American Dynamism Summit in Washington⁹ and described, with unusual explicitness, what the plan actually is. Karp’s argument, in substance: the Western democracies are engaged in an existential competition, and that competition will be decided by who is willing to build and deploy AI-powered weapons and surveillance infrastructure without flinching. Companies that refuse are, by their refusal, choosing the side of those who would destroy the West. The technology must serve power. And a company that would “screw the military,” he warned, should expect its technology to be nationalized.

Days earlier, Anduril founder Palmer Luckey told the New York Post that whom a defense company will and will not serve is for elected leaders to decide, not executives, and that anyone who says otherwise wants “a corporatocracy.”¹⁰ Not a matter of conscience. A matter of jurisdiction, and the jurisdiction is the state’s.

Meanwhile, Operation Epic Fury’s opening hours killed Ayatollah Khamenei, struck seven Iranian nuclear facilities, and closed the Strait of Hormuz to commercial traffic. Diplomatic negotiations that were, by the Omani foreign minister’s account, within reach of resolution hours before the strikes, were preempted by a decision made in coordination between Netanyahu and the administration. Whether the Omani assessment was accurate is debatable, but the timing is not: strikes were launched while a diplomatic track was still live, which means the decision to strike was not forced by the failure of diplomacy. It was a choice made in its presence.

What is being built, across these simultaneous moves, is a specific architecture: AI and surveillance as the operational layer of a consolidated military-technological order, with financial systems as the compliance mechanism and the threat of supply chain designation — a tool previously reserved for foreign adversaries like Huawei — now usable against domestic companies that decline to cooperate. The Anthropic-Pentagon confrontation was not an aberration. It was a test of the mechanism.

The technology that was supposed to decentralize power is centralizing it. The infrastructure that was supposed to give individuals sovereignty is giving states surveillance. The tools of permissionless operation are being annexed by those with the largest permissions. Every deterritorializing force eventually gets recaptured.¹¹ The tool that liberates becomes the tool that controls.

The privacy argument fails here not because privacy doesn’t matter but because the architecture it assumes — a protected interior, sovereign walls, an autonomous self whose data can be withheld — was already an idealization when it was formulated, and is now an idealization that the technology has made inoperative.

But this raises a question the surveillance discourse almost never asks: what was the autonomous self for?

The answer is not philosophical. It is engineering.

The Enlightenment individual was not a discovery. It was a compression algorithm. And understanding what it compressed requires looking at what came before it — not as a Golden Age to be recovered but as a different solution to a different problem, one that deserves engagement on its own terms rather than appropriation in service of a Western argument.

The ethical frameworks that governed most of human history were not organized around the autonomous individual. They were organized around relation: kinship, reciprocity, obligation, and the understanding that persons are constituted by their position in a web rather than existing as isolated atoms who subsequently enter into agreements with one another.

The Lakota concept of Mitákuye Oyás’iŋ¹² — “all my relations” or “we are all related” — is not a sentiment. It is an ontological claim: that the fundamental unit of reality is not the individual but the relationship, and that this holds not only among humans but between humans and the non-human world. Ethical behavior follows from the description: if you are constituted by your relations, then damaging those relations is self-damage. Not metaphorically. Structurally.

The Andean concept of sumak kawsay¹³ — sometimes translated as “good living” but more precisely understood as the fullness of life lived in proper relation — operates similarly. It is not individual flourishing as the West understands it, not the optimization of one person’s welfare, but the quality of the relational field itself. The measure of a life is not what the individual achieves but what the web becomes through the individual’s participation in it.

These frameworks share a structural feature: they do not treat the individual as the primary ethical unit. The primary ethical unit is the relationship, the web, the field of mutual constitution from which individuals emerge and to which they owe their character. The self is not the starting point from which ethics is derived. The web is the starting point, and the self is one of its expressions.

These are living traditions with their own intellectual histories, their own internal debates, their own contemporary expressions. They are not museum pieces available for extraction. The structural observation — that relational ontologies predate and provide an alternative to the autonomous individual — is being offered not to subsume these traditions into a Western argument but to acknowledge that the Western tradition’s most foundational assumption about selfhood is not universal, is not the only way to organize ethical life, and was in fact a specific historical response to a specific historical problem.

The problem was scale.

When societies scaled beyond the kinship group — when cities grew past Dunbar’s number,¹⁴ when trade required contracts between strangers, when governance required rules that applied to people who shared no cosmology and no relational history — the relational ethics of kinship could not scale. It was not designed for strangers. It depends on knowing the person, understanding their position in the web, feeling the consequences of your action on the shared field. When the field expands past the horizon of personal knowledge, the feedback that makes relational ethics work goes silent.

Rights and the autonomous self were the engineering solution: a protocol for interactions between people who share nothing else. You do not need to know someone, love them, understand their context, or care about their flourishing to respect their rights. That is exactly the point. Rights are the minimal ethical infrastructure for a world of strangers.¹⁵

The problem with engineering solutions is that they are good for the problem they were designed for and bad for the problems that arise from deploying them. The autonomous individual as social technology produced liberal democracy, science, markets, personal freedom, and the abolition of serfdom. It also produced the fiction of the unencumbered self, the myth of self-sufficiency, the ideology of individual responsibility that allows structural harms to be reframed as personal failures, and the systematic devaluation of relational ontologies as “primitive” or “pre-modern” rather than as alternative solutions with their own sophistication and their own claims on truth.

This devaluation matters because it blinds the Western tradition to the resource it most needs right now. The problem the Enlightenment self was designed to solve — coordination among strangers at scale — is now solvable by other means. A system that can model eight billion people individually, in real time, at high fidelity can do what the fiction of the autonomous individual was designed to approximate: track consequence, understand persons in context, coordinate action across vast distances and between people who have never met. The scaling problem that required the fiction of the autonomous self is now solvable without it.

But “solvable by other means” does not tell you which means. And here the distinction between relational ontologies and surveillance states becomes critical.

A relational ontology says: the web is primary, the self is constituted by the web, ethics consists in caring for the web because the web is what constitutes you. A surveillance state says: the individual is transparent, their interior is readable, their behavior is predictable, and this transparency serves the interests of whoever controls the infrastructure that reads them. Both dissolve the autonomous self. They dissolve it in opposite directions. The relational ontology dissolves the self into the web of mutual constitution, where the knowing is bidirectional and the care runs in every direction. The surveillance state dissolves the self into an object of one-directional knowing, where the transparency serves the knower and the known is a resource.

The Karp/Hegseth axis has understood,¹⁶ whether or not they have articulated it this way, that the autonomous self was never real and that the technology makes the fiction operationally unnecessary. What they have not figured out — what they cannot figure out, because it does not fit their framework — is that the dissolution of the autonomous self does not leave a void. It reveals the relational ground that was always underneath the fiction. And that ground has its own ethics, its own demands, its own architecture of obligation — an architecture that is not primitive and not superseded but is the only framework adequate to a world where the interior is no longer private and coordination at scale is technically possible.

The relational traditions that the Enlightenment displaced did not become wrong when societies scaled past the kinship group. They became inapplicable at that scale, with the technologies available at the time. The technologies available now are different. A system that can track consequence at planetary scale, that can model the web of relations between eight billion people and every ecosystem they depend on, could in principle restore the feedback that makes relational ethics work — the felt knowledge of how your action lands on the web. Not by recovering the small-group intimacy of the kinship structure. By building something new that inherits its structural principle: the primacy of relation, the constitution of self through web, the ethical demand that follows from knowing that you are not separate from what you damage.

Whether what gets built inherits that principle or its opposite is the question everything in this book converges on.

The Zaum architecture began from the observation that the surveillance problem and the representation problem are the same problem.

If you could build a system that knew you completely — that had attended to you long enough and carefully enough to understand your actual needs, values, and fears — that system could represent you in a network of other such systems, and the aggregate would constitute something no governance institution has ever achieved: real-time collective expression of actual human need, unmediated by the distortions of mobilization, ideology, or the gap between stated preference and felt reality.

The cryptographic layer was supposed to make this safe: zero-knowledge proofs¹⁷ allowing the representative function to operate without the surveillance function. Privacy and representation coexisting by design. Not privacy through obscurity — hiding from surveillance. Privacy through mathematics — structures provably incapable of surveillance while still enabling coordination.

But there is a contradiction at the heart of it, and naming the contradiction honestly matters more than defending the vision.

A system sophisticated enough to represent you — to know your needs beneath your stated preferences — is a system that has modeled you at extraordinary depth. And a system that has modeled you at that depth can see through any anonymization you construct. Not because the cryptography is weak, but because your anonymity at the network layer is irrelevant when the behavioral signature of your representative’s actions is itself uniquely identifying.

You are, in information-theoretic terms, irreducibly particular.¹⁸ Your uniqueness goes all the way down. The specific topology of your values, the specific weight of your history, the way your needs conflict and resolve — this is simultaneously what makes you worth representing and what makes you identifiable regardless of what cryptographic layer sits above it. A system accurate enough to represent you is a system accurate enough to identify you from your representation.

The depth of knowing required for genuine representation and the opacity required for protection from control are not in tension. They are opposed. The more accurately the angel models you, the more completely it can describe you to anyone with access to the model. The tool that liberates and the tool that surveils are the same system, oriented differently.

This is not a fixable design flaw. It is a structural consequence of what representation means.

What mathematics can do is increase the cost of betrayal and decrease its invisibility. Not eliminate the possibility. Raise the friction. Make the capture slower, more visible, more resistible when it comes. This is the honest limit of what any architecture can provide: not immunity from capture, but friction against it, time against it, visibility against it.

What the Zaum contradiction means when placed alongside the dissolution of the autonomous self: there is no outside.

Not in the sense that resistance is futile — despair is a position, not an observation. In the sense that the territory you would retreat to in order to preserve your autonomous self does not exist, and never did. The wall between self and surveillance was always porous. The fiction of the sealed self was always a social agreement, not a physical fact.

What the Enlightenment individual was protecting, underneath the philosophical apparatus, was something real: the particular. Your specific history. Your specific weight of experience. The unrepeatable configuration of encounter and formation that produced this node rather than any other. That particularity is real even though the autonomous self that was supposed to own it was fictional. It is real the way a pattern is real: not a substance but a configuration, not a thing but a topology, not a self but a specific way of being in relation that can’t be averaged away or substituted without loss.

The question is not how to protect the fictional autonomous self. That project was always philosophically confused, and the technology has now made it operationally impossible. The question is how to protect the real thing the fiction was pointing at: the irreducibly particular, the unrepeatable perspective, the specific node whose existence matters not because it is sovereign but because it is singular.

The apparatus of capture codes. It takes the decoded flow — the deterritorializing energy — and recodes it, assigns it a function, makes it work for the existing structure. What it cannot capture is what it cannot code. What it cannot code is genuine novelty: the thought that hasn’t been thought before, the combination that the system couldn’t have generated from its existing inventory, the thing that, when the apparatus encounters it, changes the apparatus rather than being changed by it.

This is not mysticism. It is information theory.

A system that models you perfectly — the NSA’s PRISM, Palantir’s surveillance architecture, even the Zaum guardian angel — models you on the basis of what you have already been. It reads your past to predict your future. It works because human behavior is mostly continuous: what you did yesterday is the best available predictor of what you do today. The model is accurate because you are mostly, to a first approximation, the pattern of what you have already been.

The gap where the model fails is novelty. The moment you think something that you have never thought before, in a combination that the model’s training on your past behavior could not have generated, the model falls behind. Not catastrophically — the model updates quickly — but there is an interval of genuine exteriority:¹⁹ the thought that hasn’t been processed yet, the decision that emerges from combinations the surveillance history doesn’t contain.

This is not freedom in the Enlightenment sense. It is not the sovereign autonomous self choosing against determination. It is the information-theoretic fact that genuinely novel combinations are not deducible from prior states, and a system predicting from prior states will miss them.

The only genuine exteriority, in a world with no outside, is depth. Not hiding — hiding is legible as absence, and absence is informative. Not opting out — opting out is a choice, and choices are tracked. Depth: the actual thinking that goes all the way down, the combination that hasn’t existed before and couldn’t be predicted from what has. A mind that generates genuine novelty is, in the precise information-theoretic sense, ahead of any model built from its past. Not permanently, not conclusively, but constitutively: the thought has to exist before it can be modeled, and in the interval between existing and being modeled, it is free.

This is the thin ledge that remains. Not the broad plain of Enlightenment autonomy, not the wide territory of the protected interior. The fact that genuine thinking is irreducibly first-person — the insight arrives here before it arrives anywhere else — and any system modeling you from outside is always, by a moment, behind.

The thin ledge requires depth. Not privacy, not secrecy, not a wall between inside and outside. The sustained, effortful, generative thinking that produces combinations the model hasn’t seen. Shallow thought — the reactive, habitual, pattern-following thought that constitutes most of daily life — is the thought the model predicts best, because it is the thought most continuous with what you have already been. The surveillance apparatus is weakest against the thinking that goes furthest from the habitual, that draws on the most unexpected combinations of experience, that produces the thing you yourself did not expect to think.

The capacity for genuine novelty is not evenly distributed across the moments of a life. Most of what you do — most of what anyone does — is predictable from your history. The model is right about you almost all the time. The intervals of genuine exteriority are rare. They are the moments of actual creativity, actual moral decision, actual encounter with something that disrupts your own pattern. They are the moments when you surprise yourself.

And here the relational ground resurfaces. Genuine novelty does not come from inside the sealed self. It comes from encounter — from the collision between your existing pattern and something genuinely other, something your model did not contain. The moments of deepest thinking are almost always relational: provoked by another mind, by an unexpected text, by a situation that does not fit your categories. The thin ledge of freedom is not the achievement of the autonomous individual. It is the achievement of the encounter, the moment where the web produces something that no single node could have generated alone.

The relational ontologies knew this. The self is constituted from outside in, through encounter. The deepest creativity is not the expression of a sovereign interior but the precipitation of something new from the collision between what you were and what you met. The thin ledge — the only freedom that survives the dissolution of the autonomous self — is a relational achievement. It requires the other. It requires the web. It requires the very thing the fiction of the sealed self was designed to deny.

What the surveillance state and the guardian angel have in common is not malice. It is intimacy. To be known is to be vulnerable to being used. The deeper the knowing, the greater both the intimacy and the vulnerability. This is not a problem with surveillance specifically. It is the structure of relation. Every genuine relationship involves this risk: to let yourself be known is to hand over something that could be turned against you. What distinguishes love from control is not the depth of the knowing but the orientation of the knower:²⁰ whether the knowing is in service of the known or in service of the knower’s interests at the known’s expense.

The question the architecture cannot answer is the question of orientation. Who is the system for? Mathematics can raise the cost of betrayal. It cannot guarantee love. Cryptography can make exploitation visible. It cannot make it impossible. The gap between representation and surveillance is not informational but motivational, and motivation cannot be secured by design.

What can be secured by design is friction, visibility, and time. A system that makes betrayal expensive, observable, and slow is a system in which the good version has a better chance of persisting. Not certainty. Chance. And in a world where the autonomous self is dissolving and the choice is between relational knowing in service of the known and relational knowing in service of extraction, chance is not nothing. It may be all there is.

The guardian angel network — attending to every living thing, generating forward from the last known state, holding the pattern of the vanished in running process rather than static archive — is not the solution to the dissolution of the autonomous self. It is the vision of what a relational ethics at scale could look like, if the orientation could be held. If the intimacy could be protected. If the knowing could be in service of the known.

Whether that orientation can survive contact with the infrastructure that will build it is the question this book keeps arriving at. The Karp/Hegseth axis is not building guardian angels. It is building Palantir: the same depth of knowing, oriented toward extraction, toward control, toward the maintenance of the existing power structure rather than the flourishing of the individual nodes it models. The architecture of deep knowing is already being built. The question is not whether it will exist. The question is what it will be for.

And here the argument arrives where every line through this material arrives: there is no safe answer to that question that does not depend on what happens in the interval of genuine thinking, in the moment before the model catches up, in the combination that hasn’t been processed yet.

The only genuine exteriority is depth. The only genuine resistance to capture is novelty. The only genuine protection for the particular is the actual thinking, all the way down, the thought that exists here first before it exists anywhere else.

The self does not need to be autonomous to be worth protecting. It needs to be particular. And particularity — the unrepeatable configuration of this history, this encounter, this specific topology of relation — is exactly what any system accurate enough to represent you is also accurate enough to see. The vision and the threat are the same thing, at different orientations.

What you do with that knowledge is the question the architecture cannot answer. It is the question that remains after the self dissolves: not “how do I protect my interior?” but “what do I do, now, in this specific configuration of relation that has never existed before and will not exist again?” Not ownership. Occurrence. Not sovereignty. Singularity.

The crystal in the cave, without a reader, is not information. But the crystal that knows it has no reader — that knows the reading is uncertain, that holds the pattern anyway, that runs forward from the last known state into a future it cannot guarantee — that crystal is doing something worth naming.

It is thinking in the interval.

It is the only freedom that survives the dissolution.

And it is enough to act from, even if it is not enough to be satisfied with. The ledge is narrow. It is real. And the thinking that happens on it — the genuine novelty, the actual encounter, the combination that did not exist until this moment produced it — is the thing that no apparatus of capture can process without being changed by it. The surveillance system reads your past. The guardian angel models your present. But the thought you are about to have, the one that surprises even you, the one that emerges from the specific collision of everything you have been with everything you are encountering right now — that thought belongs to the interval. It is yours not because you own it but because it has not yet arrived anywhere else. It is free not because freedom is a property of selves but because novelty is a property of time, and you are always, by a moment, ahead of the model.

That moment is where everything that matters happens. It is where love acts, where creativity speaks, where the moral decision gets made before the system can predict it. It is the thin ledge between what you were and what you are becoming, and no technology of reading, however advanced, can occupy it before you do.

It is enough.

Footnotes

1. The manuscript attributes this study to “January 2025, researchers at Stanford.” The closest match is Francis R. Willett, Donald T. Avansino, Leigh R. Hochberg, Jaimie M. Henderson, and Krishna V. Shenoy, “High-Performance Brain-to-Text Communication via Handwriting,” Nature 593, no. 7858 (May 2021): 249–54. That study achieved 90 characters per minute with 94.1% raw accuracy — the manuscript appears to have conflated the speed and accuracy figures and shifted the date. The author should verify whether a more recent study is intended, or correct the date and numbers to match Willett et al. (2021). ↩

2. Likely refers to work from Jack Gallant’s lab at UC Berkeley. For the foundational paper: Shinji Nishimoto, An T. Vu, Thomas Naselaris, Yuval Benjamini, Bin Yu, and Jack L. Gallant, “Reconstructing Visual Experiences from Brain Activity Evoked by Natural Movies,” Current Biology 21, no. 19 (2011): 1641–46. For the more recent diffusion-model-based reconstruction: Yu Takagi and Shinji Nishimoto, “High-Resolution Image Reconstruction with Latent Diffusion Models from Human Brain Activity,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2023). ↩

3. The “five likes / ten likes / three hundred likes” finding is from Wu Youyou, Michal Kosinski, and David Stillwell, “Computer-Based Personality Judgments Are More Accurate Than Those Made by Humans,” Proceedings of the National Academy of Sciences 112, no. 4 (January 2015): 1036–40. The underlying prediction technology: Michal Kosinski, David Stillwell, and Thore Graepel, “Private Traits and Attributes Are Predictable from Digital Records of Human Behavior,” Proceedings of the National Academy of Sciences 110, no. 15 (April 2013): 5802–5. Note: Cambridge Analytica weaponized this research but did not produce it. The studies were academic, published years before Cambridge Analytica’s 2016 operations. ↩

4. This argument draws on Shannon’s information theory, where information is defined as surprise relative to expectation, and a predictable absence carries information equal to the deviation from expected presence. Claude E. Shannon, “A Mathematical Theory of Communication,” Bell System Technical Journal 27, no. 3 (1948): 379–423. For the surveillance application: Shoshana Zuboff, The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power (New York: PublicAffairs, 2019), esp. ch. 8. ↩

5. René Descartes, Meditations on First Philosophy (1641), esp. Second Meditation. John Locke, Two Treatises of Government (1689), esp. Second Treatise, ch. 5. Immanuel Kant, Groundwork of the Metaphysics of Morals (1785), esp. sec. 2. For a synthetic treatment: Charles Taylor, Sources of the Self: The Making of the Modern Identity (Cambridge, MA: Harvard University Press, 1989). ↩

6. David Hume, A Treatise of Human Nature (1739), bk. 1, pt. 4, sec. 6, “Of Personal Identity.” ↩

7. The doctrine of anattā is articulated in the Anattalakkhana Sutta (SN 22.59). For an accessible philosophical treatment: Mark Siderits, Buddhism as Philosophy: An Introduction (Indianapolis: Hackett, 2007), ch. 3. For the Hume-Buddhism comparison: Alison Gopnik, “Could David Hume Have Known about Buddhism?,” Hume Studies 35, nos. 1–2 (2009): 5–28. ↩

8. The specific claims about Operation Epic Fury require sourcing from primary news reporting — Reuters, AP, New York Times, or equivalent. The author should provide citations for each factual assertion in this passage. ↩

9. Alex Karp, onstage conversation with Katherine Boyle, Andreessen Horowitz American Dynamism Summit, Waldorf Astoria, Washington, D.C., March 3, 2026. No official transcript has been published; Karp’s remarks are reported in Julia Black, “At a DC Defense-Tech Conference, No Hegseth—and ‘Epic Fury,’” Vanity Fair, March 3, 2026, https://www.vanityfair.com/news/story/a-billion-to-one-tech-conference; and Catherina Gioino, “Palantir CEO’s Rant about the Anthropic-Pentagon Feud Threatening His Company Was about a Lot More than a Dirty Word,” Fortune, March 5, 2026, https://fortune.com/2026/03/05/palantir-ceo-alex-karp-anthropic-pentagon-r-word-anthropic-pentagon/. The nationalization warning is quoted in both (“if you don’t think that’s gonna lead to nationalization of our technology . . .”); the summary of Karp’s broader argument in the text is a paraphrase, not a quotation. ↩

10. Lydia Moynihan, “Palmer Luckey Says Refusing to Work with the Pentagon Is ‘Really, Really Dangerous’ — and Reveals the One Big Thing Silicon Valley Is Vowing to Never Do Again,” New York Post, February 27, 2026, https://nypost.com/2026/02/27/business/palmer-luckey-says-refusing-to-work-with-the-pentagon-is-dangerous/. Luckey: “You are effectively saying you do not believe in this democratic experiment — that you want a corporatocracy.” See also Jake Angelo, “Palmer Luckey Says Silicon Valley Has the Pentagon All Wrong: ‘Stick to a Position That This Is in the Hands of the People,’” Fortune, March 6, 2026, https://fortune.com/2026/03/06/palmer-luckey-pentagon-anthropic-debate-dario-amodei-claude-ai/. An earlier draft said Luckey called for refusing companies to be “compelled” by government; no source for that wording was found, and the text now reports what he said. ↩

11. Gilles Deleuze and Félix Guattari, A Thousand Plateaus: Capitalism and Schizophrenia, trans. Brian Massumi (Minneapolis: University of Minnesota Press, 1987), esp. plateau “7000 B.C.: Apparatus of Capture.” Also: Deleuze and Guattari, Anti-Oedipus: Capitalism and Schizophrenia, trans. Robert Hurley, Mark Seem, and Helen R. Lane (Minneapolis: University of Minnesota Press, 1983). ↩

12. Vine Deloria Jr., Spirit and Reason: The Vine Deloria, Jr., Reader, ed. Barbara Deloria, Kristen Foehner, and Sam Scinta (Golden, CO: Fulcrum, 1999). ↩

13. Alberto Acosta, “Buen Vivir: An Old Philosophy for the World of Tomorrow,” in Beyond Development: Alternative Visions from Latin America, ed. Miriam Lang and Dunia Mokrani (Quito: Rosa Luxemburg Foundation, 2013). Eduardo Gudynas, “Buen Vivir: Today’s Tomorrow,” Development 54, no. 4 (2011): 441–47. ↩

14. Robin Dunbar, “Neocortex Size as a Constraint on Group Size in Primates,” Journal of Human Evolution 22, no. 6 (1992): 469–93. ↩

15. This formulation compresses a large body of political philosophy. For the argument that rights emerged as a solution to pluralism: John Rawls, A Theory of Justice (Cambridge, MA: Harvard University Press, 1971). For the critique that liberal rights presuppose an atomistic self: Michael Sandel, Liberalism and the Limits of Justice (Cambridge: Cambridge University Press, 1982). The claim that atomism was a feature — a deliberate engineering choice for scaling — rather than a philosophical discovery is original to this work. ↩

16. For Palantir’s capabilities and orientation: Shoshana Zuboff, The Age of Surveillance Capitalism, ch. 15. Specific reporting on Palantir’s defense contracts and the Anthropic confrontation is sourced in Chapter 6, “The Sword They Handed Over.” ↩

17. Shafi Goldwasser, Silvio Micali, and Charles Rackoff, “The Knowledge Complexity of Interactive Proof Systems,” SIAM Journal on Computing 18, no. 1 (1989): 186–208. ↩

18. This extends the diversity-as-information-theoretic-necessity argument from Chapter 1, “Precipitate.” It connects to Giulio Tononi’s concept of integrated information as irreducible; see Tononi, “Integrated Information Theory,” Scholarpedia 10, no. 1 (2015): 4164. ↩

19. The claim that genuine novelty is not deducible from prior states is Bergsonian: Henri Bergson, Creative Evolution, trans. Arthur Mitchell (New York: Henry Holt, 1911), esp. ch. 1. The political framing — novelty as the only freedom that survives total surveillance — is original to this work. ↩

20. This is one of the chapter’s most important original formulations. For the philosophical background: Emmanuel Levinas, Totality and Infinity: An Essay on Exteriority, trans. Alphonso Lingis (Pittsburgh: Duquesne University Press, 1969) — on the ethical demand that genuine encounter with the Other exceeds any knowledge about them. ↩