How it works — 5 of 7

Numbers under the hood

Two layers run quietly underneath the actors and committees covered so far: a population of cohorts carrying each country's public mood forward, and a numeric network carrying its economy forward. Both are deterministic once a tick's inputs are set, and both are explicit about what is real data and what is calibration.

Every tick, the platform doesn't track one national mood for a country — it tracks a small, fixed set of 8 cohort archetypes: a mix of urban and rural, wealthy and poor, public-sector and privately employed, each carrying its own running level of grievance and loyalty, weighted differently for every country.

Economic conditions and any repression in a given tick nudge each cohort's grievance and loyalty a little, capped so no single tick can swing either one too far. Once a cohort's own grievance clears its threshold, its members start acting on their own, without waiting for anyone else; as more of the population mobilizes, the bar for the rest to join drops, so a cascade can build from ordinary, individually rational decisions rather than an assumed coordination. That climb has an explicit ceiling, too: some share of any population never mobilizes no matter how bad things get, and the model says so outright instead of letting the number drift toward a false 100%.

What that pressure does next depends on the kind of government in charge. In a democracy, sustained grievance can trip an electoral constraint that blocks certain actions outright — repression and capital controls among them — a hard institutional check that either fires or it doesn't, even though the pressure driving it is continuous. In an autocracy or hybrid regime that electoral channel doesn't exist, so the same pressure instead shows up as a continuous strain on the narrower circle of elites and enforcers the regime actually depends on, penalizing a ruler's own destabilizing moves once that strain passes its own threshold.

Underneath a country's headline indicators — growth, inflation, unemployment, and the rest — sits a numeric substrate that evolves by fixed difference equations each tick, mean-reverting toward that country's own baseline rather than an LLM's guess at what the economy "should" do. Layered on top, a handful of largely independent, additive channels connect one economy to others': trade transmission carries a growth shock to trading partners, an energy channel passes commodity swings through with a deliberately capped effect on growth, and a currency-and-debt block tracks reserves and exchange-rate pressure — government default risk stays only a scenario variable, never a full interbank settlement network, since a government is not a bank. None of the coefficients or country-level weights behind any of this are published here: much of the economic and governance data this layer draws on carries licensing terms that permit using it, not republishing the specific figures.

That last point is worth being direct about, because it happened here. An earlier version of the democracy-blocking rule above had the right idea but the wrong direction: under sustained grievance, the computed pressure moved further from the blocking threshold instead of closer to it, so the block almost never fired the way it was meant to. Once found, the formula was rewritten, not quietly retuned, and the earlier, wrong version stayed documented rather than deleted — publishing how a system works means publishing its mistakes too, the subject of the next page.

flowchart TD
    accTitle: Population pressure and institutional gates
    accDescr: Economic conditions and repression change cohort grievance and loyalty, threshold cascades create mobilization, and regime type selects an electoral constraint or selectorate strain.
    A[Cohort archetypes carry grievance and loyalty, per country] --> B[Economic conditions and repression nudge both, tick by tick]
    B --> C[Threshold cascade: mobilization spreads as more of the population joins]
    C --> D{Institutional gate}
    D -->|Democracy| E[Electoral constraint can block select actions outright]
    D -->|Autocracy or hybrid| F[Selectorate strain penalizes destabilizing moves instead]