escapekey: «The Black Swan Factory» — каббалистическое Древо Жизни как карта глобальной системы управления

The Black Swan Factory Source: https://escapekey.substack.com/p/the-black-swan-factory

Between 2011 and 2016, Gabriel Burstein and Constantin Virgil Negoita published a series of papers in journals like Kybernetes, Systems and Applied Mathematics. Using category theory, algebraic topology and cybernetics, they mapped out a control hierarchy for what they called ‘knowledge, finance, and humanistic systems’.

They named this Kabbalah System Theory because the structure they’d formalised had been sitting in the Zohar — a 700-year-old Kabbalistic text — where it’s known as the Tree of Life: eleven nodes arranged in a hierarchy and linked by paths, with an abstract objective at the top, the physical world at the bottom, and everything in between serving as the machinery that converts one into the other.

That model was originally designed to describe the structure of creation. What follows is a description of the structure that’s now being built around you.

The top of the tree is will. In the original text it’s called Keter — the crown, the first intention that everything else flows from. Hermann Cohen, the most important philosopher of the neo-Kantian tradition, gave this its modern form in his Ethik des reinen Willens, or Ethics of Pure Will. In Cohen’s framework, ethics isn’t derived from anything prior and doesn’t serve a purpose beyond itself — the will is the purpose.

Spinoza provides the mechanism. His idea of God, or Nature — Deus sive Natura — erases the gap between human will and the natural world, so intention becomes indistinguishable from the way things are. Sustainability, planetary health, the common good — these aren’t presented as political ideas to be debated but as facts to be accepted, like the climate.

You can’t argue against them any more than you can argue against the climate itself. Pure will, dressed up as natural law, doesn’t need a vote. Its authority depends on one thing: that nobody recognises the premise as a premise. The moment people see ‘sustainability’ as an installed axiom rather than a discovered truth, everything that follows loses its authority.

Under the ethic sit two nodes that together define how the system classifies things. The first — Chokmah — records who everyone is. Every person and entity is tagged and filed: digital IDs, ISO categories, national registers, legal identifiers. If you take part in the system, the system can see you.

The second — Binah — records what everyone is allowed to do; credentials. This is worked out through risk ratings, regulatory frameworks, climate scenario models like the NGFS, and accreditation systems like GLOBAC — the new international body that decides which organisations are qualified to certify, test, inspect and validate.

Identity and capability together give the system a full list of what exists and what it’s allowed to do.

Between this list and the working layers sits a hidden node, Da’at, usually translated as knowledge. It doesn’t appear on most diagrams. It pulls the identities, capabilities and ethic together, and turns them into the rules the rest of the system follows. The BIS Innovation Hub’s Project Mandala is Da’at in action: it turns each country’s regulatory requirements into a shared digital format and compiles different national frameworks into one machine-readable language the system can process.

Whatever Da’at recognises becomes a variable in the system. Whatever it doesn’t recognise doesn’t exist. The system doesn’t control what you do. It decides which tools you’re allowed to do it with.

The middle of the tree is where judgement happens. Three nodes form a group that acts as a sorting office.

The first is Chesed — traditionally loving-kindness, but in practice it’s surveillance. Constant monitoring: SDG indicators, satellite observation, health data, ecosystem metrics, transaction records.

The system watches everything, and it’s dressed up as generosity — not the system spying on you, but you being ‘transparent to the system’. ESG reporting is voluntary disclosure. Digital identity is financial inclusion. Health data sharing is community responsibility. You open yourself up completely, because holding back is what bad people do and, surely, you have nothing to hide.

Opposite Chesed sits Gevurah — severity. That’s audit. Real-time compliance checking against the compiled rules: BIS supervisory technology, Project AISE, automated regulatory review.

Everything disclosed through Chesed gets measured against what Da’at has decided the rules are.

Between them sits Tipheret — traditionally beauty, harmony or balance. In this governance structure, that’s the clearinghouse, eventually to become artificial intelligence. It receives the surveillance stream and the audit demands, balances the two, and produces a ruling. The AI doesn’t set the standards and doesn’t collect the data; it sits between those functions and outputs a judgment.

This clearinghouse role is being moved from human ‘stakeholders’ into machines, producing rulings that are continuous, automated, and not open to appeal in any form the ruled would recognise.

This migration follows from a principle W. Ross Ashby established in the 1950s: the Law of Requisite Variety. A controller must have at least as much variety in its responses as the system it’s trying to control. Once the nexus frameworks declare health, climate, finance, food, water and security interdependent — once ‘holism’ becomes the governing principle — the variety of the system explodes beyond what humans can process.

No committee can handle the interactions between climate scenarios, financial risk models, health surveillance streams and ecosystem metrics at once, and no parliament can deliberate at the speed the data arrives.

This holistic framing makes AI structurally necessary: only a computational system has enough variety to match the complexity of what the architecture calls ‘Spaceship Earth’. Every new nexus declaration — every claim that domain A determines outcomes in domain B — increases the system’s variety, which increases the demand for computational management, which justifies moving the clearinghouse from humans to machines.

The bottom of the tree is where the ruling gets carried out. Three more nodes form the behavioural group.

Netzach — persistence or endurance — is governance. It’s the conditional logic that determines what happens next, the policy frameworks that translate a judgment into a set of requirements. Where Tipheret produces a ruling, Netzach determines what follows from it.

Hod — splendour or form — is enforcement. FATF grey-listing, Basel capital requirements applied through national regulators, procurement conditionality and sanctions compliance are the formal mechanisms that make the ruling stick.

Where Netzach sets the logic, Hod provides the institutional weight.

Yesod — foundation — is settlement. Conditional CBDCs, programmable money and the three-party lock documented in the BIS’s Project Rosalind — payer, payee, and a set of conditions that must be satisfied before value moves — mean the transaction either clears or it doesn’t.

Yesod has always been the node through which everything above is channelled into manifestation, the last stage before contact with physical reality. In this architecture, it’s the moment the money moves — or refuses to.

At the base of the tree sits Malkuth — the kingdom, the physical world. The outcome lands here: the individual experiences the consequence inside the total human ecosystem, the integrated environment in which they live, work, transact and are governed.

The traditional tree has three pillars.

The right pillar — Chokmah, Chesed, Netzach — is expansion.

The left — Binah, Gevurah, Hod — is constraint.

The middle carries the synthesis at every level: the ethic, the compilation, the AI ruling, the settlement, the outcome.

This creates an asymmetry that runs from top to bottom. The right pillar is the complete picture that each unit offers to the level above it. Chokmah: here’s who I am. Chesed: here’s what I’ve been doing. Netzach: here’s how I’m governing myself. Identity, activity and self-regulation — the full account a subordinate unit sends upward.

The left pillar is what comes back down. Binah: here’s what you’re allowed to do. Gevurah: here’s whether you’ve complied. Hod: here’s what happens if you haven’t. Capability boundaries, judgment and enforcement — the authority the parent level exercises over the one below.

The right pillar goes up, the left comes down, and the middle resolves each pair into a single output: Da’at compiles, Tipheret rules, Yesod settles. You self-regulate. They enforce. The money decides.

Because the structure is recursive, this same interface repeats at every scale. A commercial bank presents its identity, transaction data and self-governance to the central bank. A country presents its credentials, indicators and policy compliance to the UN system. An individual presents their digital ID, transaction history and behavioural record to the settlement layer.

At every level, you give, they constrain, and the AI or the clearinghouse in the middle harmonises the two and calls it balance.

The system doesn’t tell anyone what to do. The options are narrowed before the choice appears. A transaction that fails the compliance checks doesn’t go through. An asset that misses the risk threshold becomes unbankable. A country that doesn’t meet reconstruction conditions doesn’t get funding. The choice is always free, but the options were fixed in advance.

This runs on input constraint — the architecture’s central principle. The ethic decides what matters, the standards decide what the system can see, the surveillance collects it, the audit checks it, the AI rules on it, the governance logic conditions it, the enforcement backs it, the money executes it. And the individual receives the outcome — inside a system whose parameters were set before they were born, by processes they were never invited into.

Any control system has three stages — standard, clear, settle. A basketball game has rules, referees and a scoreboard. That part is obvious, but this is more specific: complete hierarchical governance systems for complex human domains — comtemporary global governance — don’t just settle on three stages. They settle on this exact eleven-node topology — with pillar asymmetry, a hidden compilation node between the cognitive and evaluative layers, recursive holonic scaling, and the particular relationships between nodes described above.

In the original Kabbalah, each sefirah contains a complete Tree of Life — the pattern nests inside itself at every level. Burstein and Negoita formalised this as a holonic architecture where each node runs the same cognitive-evaluative-behavioural loop, from top to bottom. The BIS runs it, as does a national central bank, a commercial bank, even an individual CBDC transaction.

Global standards cascade into national implementation, into institutional compliance, into point-of-sale settlement — and at every level, the same template governs the processing. Each unit runs autonomously on the same grammar. No central coordination’s needed, because the grammar’s the coordination. And there’s no level where you step outside the architecture. The individual transaction’s subject to the same structural logic as the UN Emergency Platform.

The KST papers came out between 2011 and 2016. The BIS projects that instantiate the same topology — Mandala, Rosalind, AISE, Aurora, Agorá — were specified between 2022 and 2025; the topology was formalised before the institutions that now implement it were designed, and those institutions arrived at structurally identical solutions without citing the formalisation.

The template comes from the Zohar, a text seven centuries old, and its intellectual lineage runs through Hess, Ashlag, Laitman and Burstein. But the architecture’s built by institutions of every nationality and staffed by people of every background — the BIS, the UN, ISO, national central banks, FATF, the Basel Committee.

Anyone solving this problem reaches the same solution. Reducing the argument to ethnicity is avoiding the architecture itself, and that’s the only part that matters.

In the acknowledgements of three of the papers, including the first one from 2011, Burstein thanks Michael Laitman — a student of Baruch Ashlag, whose father Yehuda Ashlag wrote the Sulam commentary on the Zohar and proposed what he called ‘altruistic communism’: a system that doesn’t abolish egoism but channels it through institutions toward collective outcomes.

This principle matches what Moses Hess described in the 1840s — egoism and love, processed through a clearing mechanism — and matches what the architecture implements in practice. The intellectual line runs from the Zohar through Ashlag’s social theory, through Laitman, into a formal systems framework published years before the BIS unified ledger, conditional CBDCs, or AI supervisory systems were designed.

Burstein and Negoita didn’t design the infrastructure. They mapped out the control grammar. The infrastructure was built by institutions that arrived at the same structure independently — because the control problem demands this topology.

The architecture doesn’t need to hide what it’s doing because nobody involved can see the whole thing. The BIS engineer working on Mandala understands the compilation layer but not the full eleven-node map, just as the compliance officer applying Basel risk weights knows the regulation but not the climate scenario assumptions behind it. The loan officer receiving an AI recommendation knows the customer but not the upstream parameters that shaped the decision, while the customer only sees the outcome.

The complete picture lives in the topology itself — and it’s too abstract, too spread out and too nested for any human mind to hold. The concealment isn’t a design choice. It’s a natural result of the recursive structure. Nobody needs to keep a secret, because the structure is the secret.

Da’at and Malkuth are the two gateways in and out of the machine’s abstract world. Da’at turns reality into data the system can read. The middle layers judge and act on that data. Malkuth turns the result back into real life. Everything between those two points runs in abstraction.

Whoever controls what Da’at recognises — whoever sets the encoding — controls everything downstream without touching the operational layers directly.

In a normal democracy, the legislature writes the rules, the judiciary judges them, and the executive enforces them. All three answer to the same voters. That shared accountability’s what makes democracy legitimate: the governed can replace the governors.

What’s happened over the better part of a century is that this unity’s been broken apart by the architecture’s own recursive nature.

The cognitive layer split first. Because each part of the tree contains a complete copy of the whole, the top thinking layer folded inward into three smaller layers: one that thinks, one that judges, and one that acts.

After 1945, the topmost thinking layer — the part that decides what the system believes — got claimed by the international ‘rules-based order’: Bretton Woods, the UN system, GATT, the Basel accords, ISO. These set the supranational standards that fix the premises everything else follows.

Elected representatives were pushed down to the judging layer: they still pass laws, but they now only assess and adapt standards whose premises were set above them. They’re still in the room, but they’re no longer at the head of the table. National administration became the doing layer, executing domestically what was decided supranationally. From outside, it still looks like one function — standard-setting. From inside, the power has shifted to the part that writes the standards the other parts implement. Democracy wasn’t replaced; it got reclassified internally.

The evaluative clearinghouse layer is now moving into AI. The institutional path’s been built over decades: from the G77 and the New International Economic Order through the IFDA’s ‘Third System’, then Blair’s Third Way, then Reinicke’s ‘trisectoral networks’ adopted as UN policy in 2000 — the public-private partnership model for social good originally proposed by Julius Wolf and Eduard Bernstein in the 1890s, now codified in ISO 37000 as the international governance benchmark, and it operates as continuous, automated assessment against pre-set standards with no human discretion and no way to appeal in context.

The behavioural layer’s following into programmable money. Settlement’s becoming conditional execution — funds released or withheld based on parameters checked at the transaction level.

Each separation removed one more function from democratic reach. The current phase — AI evaluation and programmable settlement — completes the circuit.

The split isn’t only structural — it’s about speed.

Standards tend to change on decadal timescales: treaties, accords and framework agreements negotiated over years. But evaluation runs in real time, with AI assessment continuous and instantaneous. Execution is faster still — a conditional transaction clears or fails in milliseconds. Democracy moves at the speed of debate, while the system moves at the speed of settlement.

By the time a standard becomes visible to public debate, it’s already encoded; by the time a ruling is issued, it’s already executed; and by the time anyone notices, it’s already irreversible.

This speed gap doesn’t just leave people out of the process — it flips the causal order entirely. Democracy needs perception, then deliberation, then standard-setting, then evaluation, then execution. This system, however, delivers standard, then encoding, then evaluation, then execution, then perception.

People experience the outcome of a process they were supposed to sit upstream of. They don’t participate late — they participate after everything’s already done.

The ratchet has a companion mechanism: the fait accompli. The architecture doesn’t propose standards for debate. It builds the infrastructure first and presents adoption as adaptation to reality.

The unified ledger isn’t a proposal — it’s already being piloted.

The conditional CBDC isn’t a policy option — it’s technical infrastructure being tested.

The AI supervisory system isn’t a recommendation — it went operational in early 2025.

By the time the democratic body is consulted, the only question is how to adapt to what already exists. Reversal would require demolishing infrastructure that’s already absorbed capital, trained personnel and integrated with trading partners.

The feedback loop seals the system shut. Every failure at the bottom — every crisis, shock or model miss — feeds back into the surveillance layer as a demand for more data, higher resolution and expanded monitoring.

The assumption is always that the model couldn’t see enough, never that its assumptions were wrong. Climate scenarios need more granularity. Financial risk models need transaction-level data. Health surveillance needs genomic resolution. Every failure of the system becomes the argument for expanding the system.

The system is unfalsifiable.

The architecture doesn’t need manufactured crises — it metabolises whatever crises occur. A pandemic, a war, a financial crash or a climate event — all produce the same expansion. That makes it more robust than any conspiratorial reading suggests, and more dangerous, because no single actor needs to orchestrate anything for the ratchet to turn.

The architecture also expands through the simulations that justify pre-planned responses — the stress test that requires more data sharing, the scenario that demands more surveillance, the model that mandates more integration.

The UN Emergency Platform, adopted in the September 2024 Pact for the Future, institutionalises this ratchet. Its trigger is the ‘complex global shock’ — defined by what cascading consequences the models predict.

The models generating these predictions are ‘black boxes’ whose internal logic can’t be explained by their designers, nor audited by the public. Surveillance data feeds the model, the model generates a prediction, the prediction crosses a threshold, and the threshold triggers the emergency. Whoever calibrated the threshold and designed the model has pre-determined when the emergency will be declared.

The usual name for this process is ‘adaptive management’ — the claim that the system learns from its failures and adjusts. In proper adaptive management, the model adapts to reality. In this architecture, however, reality is forced to adapt to the model. When the output diverges from the prediction, the system doesn’t revise its assumptions — it tightens enforcement. We saw this during Covid-19.

Together these make up anticipatory governance: complex global shocks as the trigger, ‘black box’ models as the prediction engine, and adaptive management as the justification.

It’s pre-emptive authority drawn from opaque models, exercised before the event happens, then legitimised afterwards by the claim that the intervention prevented the predicted outcome.

If the shock arrives, the model was right and more surveillance is needed; if it doesn’t, the intervention worked and more surveillance is justified anyway. There’s no way to prove the system wrong, and it expands either way.

Stafford Beer, building on Ashby’s earlier work in Design for a Brain, drew a distinction between two kinds of stability. A homeostatic system corrects deviations from its reference — like a thermostat keeping a room at twenty degrees. An ultrastable system can change the reference itself when the environment demands it — like a thermostat that learns it should heat to fifteen instead. Ashby’s original homeostat achieved ultrastability through a double feedback loop: the first corrected deviations, and when that loop failed repeatedly, a second randomly reconfigured the system’s own internal connections until a new stable configuration emerged. The system didn’t just tune parameters — it changed its own wiring.

The architecture described here has the first loop in abundance — it can adjust the carbon budget, recalibrate the risk model, retrain the AI endlessly. It doesn’t have the second loop. It can’t reconfigure its own structure or question whether its parameters should exist, because Keter is presented as self-evident.

The ethic sits above the feedback loop, not inside it. And Ashby’s Law of Requisite Variety explains why expanding surveillance is the wrong response to model failure: you can’t match a complex adaptive system’s variety by watching harder. You match it by being able to change yourself.

If the architecture dropped the alignment narrative entirely — if it stopped pretending and governed through pure operational necessity — the right pillar would collapse. The system depends on people volunteering their identity, their activity reports, their self-regulation. That self-generated transparency is what the surveillance layer processes. Kill the narrative that compliance serves a shared purpose, a ‘global ethic’, and the self-presentation stops. People still comply — the transaction still clears — but they stop volunteering the behavioural data the system needs to maintain its model.

The system goes blind at exactly the layer where it needs the most granular information. The alignment narrative isn’t rhetorical decoration — it’s the incentive structure that generates the data stream the architecture runs on.

A system that can self-correct but can’t change its own wiring keeps correcting itself within its own logic for as long as that logic matches reality. When reality drifts too far from what the model assumes, the system doesn’t adapt — it breaks.

Combined with the speed mismatch — moving too fast for democracy to intervene, but setting its benchmarks too slowly for democracy to revise them — the conclusion is plain. The architecture, once closed, can’t be reformed from within: the reference model sits above democratic reach, while the operating speed sits below democratic perception.

The two structural defects reinforce each other. The system can’t reform because it can’t fail — the unfalsifiable loop means no outcome is ever read as evidence against the model. And it can’t fail because it can’t reform — without ultrastability there’s no mechanism to act on disconfirming evidence even if it were admitted.

Each property makes the other inescapable.

The architecture doesn't just face black swans — it manufactures them.

Arthur Koestler called it the ghost in the machine: the pathology built into the structure by the structure itself. Every crisis absorbed by expanding surveillance pushes more of reality into the space the system can't encode. What isn’t encoded doesn’t disappear — it piles up. The system’s own expansion creates the conditions for the disruption it can’t process.

The more complete the encoding, the larger the blind spot left behind, and the more catastrophic the eventual collision with what’s hiding in it.

A system with gaps has slack — it absorbs unencoded shocks continuously, in small doses. A system approaching total encoding has no slack at all, so every encounter with something it can’t classify is a genuine surprise, and the architecture has no mechanism for handling genuine surprises. It can only expand the encoding, which requires the surprise to be encodable. When it isn’t, the system doesn’t adapt — it shatters.

The disruption that breaks the architecture won’t be a routine crisis — the system feeds on those and grows fatter each time. But as it reaches for total closure, the crises it manufactures grow larger too. It absorbs them at first, building more encoding capacity with each shock. Eventually the black swan outgrows even that appetite. The final encounter isn’t with a bigger data point but with an unencodable one: meaning as distinct from classification, purpose as distinct from function, and qualitative experience as distinct from metric.

Da’at can classify a person’s identity, capabilities and compliance status, but it can’t classify their reasons, their loyalties, or their sense of what matters — except by reducing them to variables, which means missing them entirely. When the unencodable arrives at sufficient scale, the system doesn’t adapt — it explodes.

The system can create stand-ins for these things — community resilience scores, sentiment indices, wellbeing metrics — but when people recognise them for what they are, the gap doesn’t close; it becomes official. Every new metric is another admission that the system can’t capture the real thing. The encoding fails not because it’s missing data, but because people feel it’s synthetic.

You can already see how this buildup feels.

People meet the system’s requirements while finding those requirements meaningless. They file the ESG report, present their digital identity, accept the conditional transaction. They do what the system asks because exclusion from the settlement layer is worse than going along. But they don’t believe in it. They don’t experience the ethic at the top as their own purpose — they experience it as imposed, a set of parameters they navigate rather than a goal they pursue.

The system can’t see this gap. It monitors compliance, which is a behavioural variable, but it doesn’t monitor alignment, which is a first-person variable that can’t be encoded by nature. Tipheret (the clearinghouse) can measure whether someone filed the report, but it can’t measure whether they were genuine about it. The system sees compliance rates improving and concludes it’s working. The absence of alignment is invisible because alignment isn’t in the grammar.

This gap doesn’t exist only in the public. It runs through the system’s own machinery. The loan officer processes the AI recommendation without questioning it. The compliance analyst applies the taxonomy without endorsing it. The central banker calibrates the risk model without believing the scenario. They comply because their careers depend on it, not because the architecture’s premises are their own. At its operational core the system runs on executants, not believers, and that hollowness is invisible to the system for the same reason the public’s misalignment is invisible: the system monitors behaviour, not belief.

The gap widens as the encoding expands. As the system demands more compliance across more domains, the distance between what people do and why they do it grows. The compliance becomes thinner — sustained not by belief but by the absence of alternatives. Thin compliance is brittle compliance. It holds until the cost of maintaining it exceeds the cost of breaking it, and then it doesn’t gradually erode. It snaps.

This has happened before. Soviet central planning measured production targets, resource allocations and whether plans were met, but the numbers were often lies — everyone had an incentive to cheat on reporting, so the metrics showed success while reality was sharply different. The system didn’t measure the quality of experience, the meaning of work, or the purposes people pursued outside the plan. What wasn’t measured piled up as informal economies, underground networks and people quietly losing belief. The system saw plan fulfilment improving and concluded it was working. When it collapsed, it wasn’t because production had stopped — it was still producing — but because the gap between the system’s model and lived reality grew too large for the model to hold. The metrics looked fine but the system was dead on the inside.

Automation might remove the incentive to cheat on reporting, but that doesn’t close the gap — it shifts it. When the data becomes accurate, human reason and lived reality drift even further from what the system can encode. The machine gets better at measuring while understanding less about what it’s measuring.

This architecture is further along that path than Soviet planning ever reached, because it’s a more complete system than any previously attempted. It’s also more brittle, because completeness removes the buffers that previous systems relied on to absorb shocks they couldn’t process.

The informal economy that cushioned the Soviet system has no equivalent in a world of programmable money and universal digital identity. The pressure has nowhere to go except through the system — which has no way to process it.

The final black swan event shatters the entire system.

The template describing this entire architecture was formalised as peer-reviewed control theory between 2011 and 2016, and preserved in the Zohar for seven centuries before that. Whether the convergence reflects conscious design or the structural constraints of the control problem itself is a different matter — the architecture functions identically either way. What matters isn’t who drew the topology but that it’s being built.

What the architecture does answer is where the individual sits. If the cognitive layer’s pre-set institutionally, the evaluative layer’s algorithmic AI, and the behavioural layer’s programmable money, then democratic agency has moved from the centre of the system to its terminal node. The human subject isn’t the operator any more — they’re what the system operates on, and the politician claiming to represent you does little more than choose how to implement standards which cannot be challenged.

The Mark of the Beast governs the individual within the system, The Gaia Hypothesis governs the planet around it, and the Total Human Ecosystem integrates the two. Humanity, as a managed specie, allegedly balanced with nature — whatever it takes.

But the architecture isn’t yet complete.

The compilation layer’s in phase two. The AI supervisory system launched in 2025. The unified settlement platform’s in active exploration. Every component described here’s at some stage of design, pilot or deployment — none of it is speculative. The trajectory’s documented, and the window between partial installation and full closure’s measurable in project phases, not decades.

The architecture requires your data, your compliance, and your belief. Two of those can be compelled. The third is why it isn't finished yet.

To display that this is more than hypothetical, the next essay walks through three examples — three people passing through all eleven gates on the same Monday morning.