Fractal layers: from components to supernodes

Psychology and geopolitics look like different subjects. One talks about anxiety, attachment, and burnout; the other about terawatts, alliances, and migration. They are the same accounting identity at different positions in a nested graph. This essay is the bridge: how your components ($\Phi$ at $L-1$), you (the agent at $L$), your couplings (family, friends, teams, society at $L+1$), and your supernodes (states, markets, climate, world-system at $L+2$ and above) constrain each other through the Fractal Persistence Equation.

Nothing here replaces clinical care, diplomacy, or law. It names the fractal bookkeeping that both folders already assume but rarely state in one place.

The equation (one identity, many scales)

\mathcal{R}^{(L)} = \Psi\!\bigl(\mathcal{R}^{(L+1)}\bigr) \cdot \left[\frac{P_{in}^{(L)}\,\eta(I)}{\omega^{(L)}\,\mathcal{E}_{\Sigma}^{(L)}\,\bigl(1 + \mathcal{D}_{KL}^{(L)} + \Gamma^{(L)}\bigr)}\right] \cdot \Phi\!\bigl(\mathcal{R}^{(L-1)}\bigr).

$\Phi$ — health of what you are made of (substrate at $L-1$). Not a mean — critical parts enter as a bottleneck ($\min$); redundant parts as a softer pool. See substrate_shelter_and_redundancy.md.
Middle bracket — what you harvest, how efficiently, and what tax you pay for noise, delusion, and unresolved friction.
$\Psi$ — shelter from what contains you ($L+1$). Independent buffers multiply; they are not averaged.

Persistence at any level requires $\mathcal{R}^{(L)} \ge 1$. There is no private exemption: a nation cannot outrun a population whose cellular $\Phi$ is collapsing; a person cannot outrun a labour market whose $\Psi$ has turned predatory. Humanity’s $\mathcal{R}$ is not the mean of national $\mathcal{R}_i$.

The stack in plain language

Level	Examples	You are…	Your $\Phi$ (components)	Your $\Psi$ (supernode)
$L-2$	Cells, sleep, microbiome, habits	—	(below conscious ISM)	Your body + nervous system
$L-1$	Skills, health, reputation, savings	—	—	Household, therapist, GP
$L$	You (agent with ISM)	The persisting self	Body, sleep, attachment capacity	Family, friends, employer
$L+1$	Couple, household, friend group, team	A member; also a sub-node	Partner, children, colleagues	Firm, school, city, union, church
$L+2$	Organisation, city-region, movement	Employee, citizen, parishioner	Workforce, inventory, trust stock	Nation, supply chain, treaty bloc
$L+3$	Nation-state, mega-corp	Citizen, migrant, shareholder	Population, grid, soil, firms	Dollar system, climate, UN layer
$L+4$	World-system, biosphere	Species-level coupling	Planetary $\Phi$	(no external $\Psi$ in the model)

The table is not a moral ranking. A dinner table is as real an IPS as a ministry. What changes is which terms dominate the budget and who pays when a term slips.

Three roles you always play at once

At every scale you simultaneously occupy three fractal roles:

Substrate — someone else’s $\Phi$. Your sleep quality is your partner’s $\Phi$; your child’s regulation is yours; your team’s delivery record is the firm’s $\Phi$.
Node — your own $\mathcal{R}^{(L)}$: boundary ($P_{in}\eta$), honesty ($\mathcal{D}_{KL}$), friction ledger ($\Gamma$), complexity ($\omega$).
Shelter — someone else’s $\Psi$. Parents buffer children from market noise; managers buffer teams from reorgs; NATO buffers members from invasion — each imperfectly.

Confusion usually comes from mixing roles. Treating your nation as your self (identity fusion) collapses $L$ into $L+3$ and makes criticism feel like death. Treating your child as your node (enmeshment) collapses $L+1$ into $L$ and exports your $\Gamma$ onto their $\Phi$. Treating your body as optional (grind culture) starves $L-1$ while you optimise the middle bracket.

Healthy fractal life keeps the levels distinct but coupled: strengthen $\Phi$ below, run the two operations at $L$, contribute to $\Psi$ above without pretending you are the supernode.

The two operations, fractally

Almost every repair that works — in therapy, marriage, teams, or states — reduces to:

Lower $\mathcal{D}_{KL}$ — update the model toward reality (humility).
Lower $\Gamma$ — resolve, renegotiate, or exit unresolved friction.

Scale	Operation 1 (honesty)	Operation 2 (friction)
Body ($L-1$)	Stop calling exhaustion “lazy”; track sleep, pain, labs	Finish the medical backlog; change the job that prevents sleep
Self ($L$)	Name feeling before story; journal observable facts	Apologise, repay, quit the lie
Couple / family ($L+1$)	“I am scared” not “you always…”; shared calendar truth	Repair rupture; divorce if $C$ is huge and repair refused
Work / school ($L+2$)	Post-mortems without blame theatre; metrics that predict	Close projects; fire toxic process, not only people
Society ($L+3$)	Real statistics, real war lessons, real energy books	Peace treaties, bankruptcy, pension reform

Rituals (therapy homework, retros, elections, UN resolutions) are implementations. The ratio is indifferent to vocabulary.

How terms flow up and down

$\Phi$ flows upward (substrate constraint)

Theorem 5.3 in the formal paper: parent persistence supervenes on child persistence. If enough sub-nodes have $\mathcal{R}^{(L-1)} < 1$, $\Phi \to 0$ and the parent equation loses meaning regardless of rhetoric.

Upward channel	Mechanism	Example
Health → household	Caregiver $\Phi$ down → co-regulation expensive	Burnout parent; anxious household
Household → firm	Presenteeism, turnover, tacit knowledge loss	Unpaid care crisis → GDP $\eta$ hit
Firm → state	Tax base, innovation, trust	Demographic winter + brain drain
State → world	Migration, famine risk, conflict spillover	Sahel $\Gamma$ → EU migration $\mathcal{E}_\Sigma$

Policy that demands “resilience” from agents while degrading $\Phi$ (precarious work, poisoned cities, unsafe childhoods) is not neutral; it is a numerator squeeze on the whole stack.

$\Psi$ flows downward (shelter and its price)

$\Psi$ attenuates noise from above. It is never free.

Downward channel	Mechanism	Example
Parent → child	Predictable rules, physical safety	Secure base → lower child $\mathcal{D}_{KL}$
Employer → worker	Benefits, tenure, psychological safety	Moral injury when institutional $Q$ ≠ witnessed $P$
State → citizen	Rule of law, currency, welfare, defence	Dollar $\Psi$; also hegemon’s $\mathcal{D}_{KL}$ imported via alliance
World → state	Trade routes, climate bands, non-aggression	UN humanitarian layer; weak on P5 wars

When $\Psi$ weakens, denominators rise at $L$ even for agents who “did nothing wrong” — Hirschman’s insight in ips_psychology.md. A competent engineer in a collapsing firm pays the firm’s $\mathcal{E}_\Sigma$ in stress; a competent citizen in a sanction spiral pays the state’s $\mathcal{D}_{KL}$ in prices and uncertainty.

Delusion cascades (sideways and diagonal)

$\mathcal{D}_{KL}$ is contagious in coupled graphs:

Couple: “everything is fine” broadcast while both ISMs know otherwise → joint $\Gamma$ when truth leaks (ips_psychology.md).
Organisation: KPI theatre → workers hold high $\mathcal{D}_{KL}$ between institutional $Q$ and floor reality → burnout.
State: Official demographics / war forecasts → emigration, currency distrust, alliance fatigue.
Media diet: Consumption of $L+3$ narratives as $L$ identity → personal $\mathcal{D}_{KL}$ about what you can control.

Humility at $L$ is partly refusing to wear another level’s delusions as your personality.

Friction export (the hidden bill)

Systems often lower local $\Gamma$ by exporting entropy to a sub-node:

Pattern	Who looks “fine”	Who pays
Drama triangle Rescuer	Group calm today	Target’s $\Phi$ or Persecutor’s $\Gamma$ later
Workplace scapegoat	Manager’s $\Gamma$ avoided	One employee’s $\mathcal{D}_{KL}$ + health
Sanctions regime	Sender’s moral story	Target’s $\mathcal{E}_\Sigma$ and sender’s $P_{in}\eta$
Colonial extraction	Metropole $\eta$	Periphery $\Phi$ (soil, labour, minerals)

Fractal ethics: if your $\Psi$ requires someone else’s $\mathcal{R} < 1$, you are borrowing persistence — the books close eventually (defection, revolution, burnout, brain drain).

Aggression and empathy across layers

From ips_psychology.md and books/aggression_empathy.md:

Aggression (ethological) = $P_{in}\eta$ spent on the Markov blanket — “no,” consequence, border.
Empathy = lowering $\mathcal{D}_{KL}$ and $\Gamma$ in coupled systems — accurate modelling + repair.

Layer	Healthy aggression	Healthy empathy
Self	Rest, boundaries, assertive speech	Self-honesty, grief, somatic truth
Family	House rules, fair chores, protective “no”	Listening, repair after rupture
Organisation	Clear mandates, firing predation	Transparent metrics, conflict process
State	Defence, law, antitrust	Statistics, courts, truth commissions

Failure modes are scale-analogous:

Suppressed aggression at $L$ → inward $\Phi$ burn (people-pleasing).
Aggression without modelling → sanctions that boomerang, rage parenting, “tough” management.
Empathy without boundaries → compassion fatigue, Rescuer geometry, aid that hollows $\Phi$.
Empathy as placation → raises dyadic $\mathcal{D}_{KL}$ (“we’re fine” while not fine).

Warrior–Sage at $L$ is the same balance as credible deterrence + honest diplomacy at $L+3$: both factors non-zero.

A walk through one week (same math, four levels)

Monday — body ($L-1$). Sleep debt: $\Phi$ down. ISM spins stories ($\mathcal{D}_{KL}$) to avoid the cost of early bedtime. Intervention: treat sleep as $\Phi$ maintenance, not virtue.

Tuesday — marriage ($L+1$). Unpaid emotional labour: one partner’s $\Gamma$ item (“we never talk about money”) avoided six months. Marginal cost of the next fight $\uparrow$. Interventions: one observable truth (op. 1); one bounded repair or scheduled decision (op. 2).

Wednesday — employer ($L+2$). Team $\omega$ high (process theatre), $\Psi$ shaky (layoff rumour). Your $P_{in}$ rises on visibility tasks with low $\eta$. Interventions: voice with data; exit if $\mathcal{R}^{(L)}$ trend negative (ips_psychology.md).

Thursday — news ($L+3$). Foreign war as identity: personal $\mathcal{D}_{KL}$ inflates; $\Gamma$ with neighbours at dinner. Intervention: separate what your state/node can affect from what your ISM can only model; donate/vote/voice at the correct level.

Friday — friend group ($L+1$). Gossip lowers $\Gamma$ locally by avoiding direct repair with the absent person — classic $\mathcal{D}_{KL}$ export. Intervention: encourage dyadic op. 1+2 or decouple.

The week is one graph, not five unrelated stories.

Migration, remittances, and care — $\Phi$ routing in the graph

Geopolitics essays treat migration as $\Phi$ transfer: humans move to the $L+3$ node where their personal $\mathcal{R}$ is higher. Psychology treats emigration from toxic families the same way — exit as $\Gamma$ optimisation.

Remittances (~$100B/yr to Africa) are upstream $\Psi$ from diaspora $\Phi$: labour exported, money shelter imported. Care work (unpaid, feminised) is $\Phi$ maintenance for the whole stack that national accounts routinely undercount — a structural $\mathcal{D}_{KL}$ in how states model their own substrate.

Being a good sub-node of your supernode

$\Psi$ is not obedience. Beneficial coupling means your persistence and the shelter’s persistence are positively correlated over the horizon you share:

Good sub-node behaviour	Bad sub-node behaviour
Pay taxes / do real work / tell truth to power	Free-ride, extract, forge stats
Raise children who can regulate	Export trauma to the next $\Phi$
Use alliance trade for $\eta$	Demand allies fight your $\Gamma$ forever
Voice before exit when $\mathcal{D}_{KL}$ is shared error	Loyalty to predatory $\Psi$ until collapse

States defect from empires; employees quit; partners divorce — the same Lemma 3: no node is self-caused.

Redundancy and risk at every layer

$\Phi$ and $\Psi$ are graph-shaped, not averages:

Series substrate (one heart, one grid, one fab) → $\Phi \approx \min$ on critical parts; failure is sudden.
Parallel substrate (stockpiles, duplicate suppliers, spare experts) → smoother loss; higher $\omega$.
Shelter channels (family + employer + state) → $\Psi_{\mathrm{eff}}$ is a product over independent buffers, not a mean.

Choosing lean chains over redundant capacity is a risk trade-off: higher calm-environment $\mathcal{R}$, thinner tail. Globalization ran that experiment at civilisation scale — see ips_economics.md. Full operator definitions: substrate_shelter_and_redundancy.md.

Avoid complexity; care for components (fractal version)

$\omega$ — Do not add bureaucracy, roles, or moral machinery without $\eta$ return. Family “rules” that nobody follows are $\omega$ without $\Gamma$ reduction. EU acquis without implementation is geopolitical $\omega$.
$\Phi$ — Invest where the graph is actually built: sleep, schools, grids, soil, trust. Cutting $\Phi$ to fund $\omega$ is the classic late-stage mistake (states and startups alike).

Diagnostic checklist (any layer)

Ask at whichever level hurts right now:

$\Phi$: Is substrate intact (body, people, inventory, soil)?
$P_{in}\eta$: Is energy available and spent on the right blanket?
$\mathcal{D}_{KL}$: Which story is detached from observables?
$\Gamma$: What is unresolved and how large is set $C$?
$\omega$: What process complexity can die without killing the node?
$\Psi$: Is shelter real or performative? Is its price acceptable?
Fractal: Am I confusing my level with another’s role?

What this is not

Not determinism. $\mathcal{R}$ is a clearing condition, not fate. Nodes can improve terms.
Not moral relativism. Exporting entropy to children or colonies is measurable harm, not “culture.”
Not reductionism. Love, meaning, and art are not “just” $\mathcal{R}$ — but they run on $\mathcal{R}\ge 1$ the same way thought runs on glucose.

Software stack (Aion)

The companion essays and the Aion implementation repos are the same fractal accounting at different positions — not three unrelated doc folders and three unrelated repos.

Essays (human scale)	FPE band	Software	What persists here
Psychology	Agent $L$ — self, body, coupling patterns	aion-llm	Single model: token-level $\mathcal{R}$, `fractal_loss`, inference `usage.persistence`
Sociology	$L+1$–$L+2$ — household, team, firm	aion-core	Many agent types on one node: tasks, norms, processes, markets, learnings
Geopolitics	$L+3$+ — nation, treaty bloc, UN layer	aion-blockchain	Many nodes on one chain: Proof of Trust, Predictive Governance, society history

Psychology → aion-llm. The essay layer asks: can this agent keep $\mathcal{R} \ge 1$ (sleep, honesty, friction repair)? The software layer asks: can this neural substrate expose persistence at the token — cross-entropy plus delusion tax, not opaque loss alone?

Sociology → aion-core. The essay layer asks: can this group persist — roles, $\Gamma$ backlog, narrative vs. shop floor? The software layer asks: can this fleet of agents persist — task tree, norms service, prediction markets, meta-agent, learnings/?

Geopolitics → aion-blockchain. The essay layer asks: can this polity persist — energy, $\mathcal{D}_{KL}$ at the controller, $\Psi$ from alliances? The software layer asks: can this society of compute peers persist — scored forecasts, trust ledger, norms and patches on chain?

Vertical slice (one closed loop):

aion-llm          token R, fractal loss
    ↓
aion-core         sense → plan → act → score; agent types; norms
    ↓
aion-blockchain   votes, trust_mtrust, society audit log

Walkthrough when wired: examples/round-trip.md. Full software operator map: COMPONENTS.md. Inter-node detail: papers/society_of_aion_nodes.md.

Caveats: Sociology ≠ only aion-core (human sociology also describes families and cities; aion-core is one organisational implementation). Books and papers cross-cut all scales — theory is not owned by one folder.