The grandmother who won’t let anyone borrow is the family’s insurance policy.
Three generations sit at a dinner table. The grandmother survived a famine. The father survived a recession. The child has grown up in prosperity. Each carries a different model of what risks are real, what reserves are necessary and what counts as ruin, and when they decide something together, those calibrations collide. The grandmother stops the child from leveraging into foolishness. The child stops the grandmother from refusing all risk long after safety has returned.
Most people would describe what happens at that table in moral terms: duty, love, tradition, or, from the other side, obligation and constraint. Why, then, do families outlast individuals in hard times, and why do civilizations that erode them keep discovering, too late, what they were for? The answer is in risk rather than morality.
You are not a population
The ergodicity insight is this: you are not a population. You are a single individual traversing a time series, and ruin is permanent. The ensemble average says a strategy produces superior expected returns, but for any one person that strategy includes the possibility of hitting an absorbing state, zero, from which no recovery is possible. You cannot change the fact that you traverse the time series alone. You can only diversify your exposures so that no single event destroys everything.
The family is the oldest human solution to that problem, and it diversifies across time rather than across assets. The grandmother stores experiential capital: the calibration of what fails under what conditions and which comforts are worth sacrificing. It cannot be handed over as data, because it lives in the shape of her anxiety, her instinct toward reserve, her refusal to borrow. The father stores economic and relational capital: assets, a network of trusted people, knowledge of how the current world works. The child stores adaptive capital, the flexibility to absorb new technologies and social forms without the friction of updating old models. The child is the family’s bet on the future.
No single generation holds all three, so the pool of their different time-series experiences is more ergodic than any member alone. It also runs on credit. Parents invest in children on an expectation, not a contract, of care in old age; grandparents transmit what they know on the expectation that it will be received and used. Atomization is a default on that intergenerational credit, and like other credit collapses it is non-ergodic: the relationship cannot be rebuilt at the same cost.
Why totalitarianism attacks the family first
The hedge protects against more than famine. Arendt documented that totalitarian movements specifically destroy family bonds, and the dinner table shows why. The family is a competing epistemological unit, with its own shared history and its own interpretation of events. The grandmother who says “this reminds me of 1932” is a prediction error that ideology cannot process. Her model was built from direct experience in a different era, so she generates updates by comparison, which is exactly the kind of bottom-up signal managed ideology needs to suppress.
The family is also a competing loyalty. It provides security, meaning and belonging, the things a totalizing movement needs to monopolize, and it has one thing the movement can never replicate: a shared history that does not depend on ideological conformity. So in authoritarian systems atomization is a deliberate precondition for control rather than an accident of modernity. The isolated individual, with no family memory and no alternative community, is the ideal subject. Arendt’s warning follows: the terror comes second. The loneliness — the loss of the epistemological community that the family provided — comes first.
The bet nobody notices making
The puer aeternus refuses multigenerational commitment: no children, no single partner, no place that would make him legible to a community across time. He calls this freedom and optionality. The ergodicity frame shows the hidden wager. He is reasoning from the ensemble (“most people get through life without a catastrophe that needs a family to absorb it”), which is correct as an average. For an individual traversing the time series, it is Russian roulette. The hedge is not needed on most pulls, but the pulls where it is needed are exactly the ones from which no one, unhedged, recovers.
Whole generations make the same bet without deciding to. Every cohort that grows up without a famine, a war or a disruption large enough to need the hedge sees it as overhead rather than insurance. The grandmothers who carried the anxiety of 1932 die, and their calibration goes with them.
None of this means defending one form of family. “Family values” and “chosen families are equally valid” both argue about the wrong thing, because the family’s value is not in its specific form but in its function as a multigenerational temporal hedge. The question is not “nuclear or chosen?” but “what institution provides the pooling of different time horizons, the transmission of experiential capital across generations, and the epistemological independence from any single ideology that prevents managed hallucination?” Any institution that provides these functions is doing what the family does. Any that does not is leaving its members unhedged.
That is the hard part, though. Twentieth-century communes tried to build the hedge without kinship, and most lasted less than a generation. The grandmother’s warning works because her calibration was formed before the current consensus existed, and a chosen family usually forms inside one. The erosion of multigenerational hedging is a civilizational bet that the time series will stay smooth enough for unhedged individuals to survive it. The bet has run long enough to look prudent. It has not yet met a disruption large enough to show what it costs.
Threads to Pull
- Emmanuel Todd, The Explanation of Ideology — maps family structures onto ideologies, suggesting families shape the ideological immune system of whole civilizations.
- Joseph Henrich on cumulative cultural evolution — most of what humans know lives across generations, not in any one head, and the family is a primary channel for transmitting it.