a.AgencyAn inquiry by Elias Marrow
World pausedseed / marrow-0023
An experiment in perspective

Effort was
held constant.

Same starting resources. Same effort. Same opportunities. A different sequence of setbacks. Follow what happens to 120 households over a year.

Only the seeded daily roll differs
StableUnder pressureMultiple needs failingDisplaced
Follow one household See all households

An explanatory simulation. Its assumptions are inspectable; its outcomes are not estimates of real populations.

THE CONSTRAINT TOWN120 households. One year.
Each light is a household. Select a home to look closer.
Jan 1Jan 1 · day 0Dec 31
What changedJan 1

No recorded shock. Fixed costs still moved.

A snapshot is not a life

Recovery doesn’t erase
what it took to get here.

Move through the year. Compare who is stable today with who has experienced hardship along the way.

Stable on Jan 1120 / 120

Households with food, housing and safety needs met today.

Hardship by Jan 10 / 120

Households with at least one of those needs unmet at any point so far.

These groups can overlap. 0 households have had two or more core needs fail together by this day. All values come from this town’s simulated record.

01 / Stay with one household

A life is more than
its last frame.

No invented biography. No assigned blame. Follow one simulated household’s actual record, including the days that disappear from a year-end total.

Explore a path in this townSelected examples from the full year, not estimates of how common each path is.
District 1 · Jan 1
stable
Cash available
$1,400
Survival debt
$0
Days in compound failure
0
Effort
100% unchanged
Inspect the full record
Jan 1 · Day 0

The beginning

In the equal-ground control, cash, wage, needs and effort begin equal. Added constraints can change that starting position. Effort always remains fixed.

Turning points are selected milestones from this household’s full year, including future days. Use the timeline above to examine every day.

Live evidence record

The town is the interface.
The ledger is the evidence.

Every house in the town is one simulated household. The clock above controls every figure below: scrub to any day and the town totals, household ledger, and selected-household history resolve to that exact frame.

Jan 1 · day 0base / marrow-0023
02 / Town record

What is happening across the year

A line is a share of the 120 households, not a probability. The vertical marker is the day currently visible in the 3D town.

Basic needs met 100%Compound failure 0%Displaced now 0%
0%25%50%75%100%day 0day 91day 182day 273day 364
Total survival debt$0owed by 0 households
Median cash$1,400$168,000 across town
Employment100%effort remains fixed at 100%
Food insecure0%food score below 55
Housing unstable0%housing score below 55
Household ledger

Every household, side by side.

Select a house in the town or a row in the ledger. Need scores below 55 are counted as failed; two or more at once are compound failure.

120 of 120 houses shown
HouseStatusFoodShelterSafetyHealthCashDebtWorkCommuteCompound daysDisplaced ever
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
stable100100100100$1,400$0yes30m0—
03 / Mechanism

What “survival debt” means

It is not a personality trait or a consumer preference. In this model it is the balance created when a household has to keep surviving after cash has run out.

Rent shortfalls are carried into debt with a fixed $38 late cost. Any negative cash left by utilities, transport or work losses, repairs, or health shocks is also converted into debt.

new balance=old debt+unpaid survival costs+0.072% daily interest
0.072% per day is about 29.9% effective annual growth if a balance remains unpaid.
At this frame$0

across 0 of 120 households

Median household debt
$0
Largest balance
$0
Repayment begins
cash above $1,450
Cash buffer retained
$1,200
shock or fixed cost→cash gap→need still exists→shortfall becomes debt→future income pays for the past
04 / 1,000-world laboratory

What the essay led us to expect — and what the model produced

Each lab run creates 1,000 independent seeded towns: 120,000 simulated household-years. We summarize outcomes by world so one unusually lucky or unlucky town cannot carry the claim.

MeasureEssay expectationObserved mean95% Monte Carlo interval
Stable at year endabout 90%
89.72%
89.55–89.88%
Ever in compound failurenearly 50%
49.80%
49.52–50.07%
Chronically failing1 in 24 · 4.17%
4.38%
4.27–4.50%
Experienced any hardshipmodel output
52.88%
52.61–53.16%
Ever displacedmodel output
8.69%
8.54–8.84%

What we expected: approximately 90% stable at year end, nearly half experiencing compound failure, and roughly one in 24 chronically failing.

What we observed: the 1,000-world means land inside those targets. “Any hardship” and “displaced” are reported as model outputs, not retrofitted as essay claims.

What this does not prove: these intervals describe variation generated by this model. They are not confidence intervals for poverty in the real population.

Open the laboratory and run your own 1,000 worlds
05 / Reading the record

Exactly what each measure counts

The model keeps end state, lived experience, chronic duration, and displacement separate. A household can recover by day 364 and still appear in a path-dependent measure.

01

Stable now

Food, housing, and safety are each at or above 55 on the selected day.

02

Any hardship

At least one core need fell below 55 at any point in the year.

03

Compound failure

Two or more core needs were below 55 on the same day.

04

Chronic failure

More than half the simulated year was spent in compound failure.

05

Displaced

Housing fell below 12, triggering loss of home and a recovery barrier. It is not every episode of poverty.

What this world tests

The environment is the variable.

The fair control fixes effort, starting cash, wage, health, needs, actions, and opportunity. Only the daily roll changes. The layered version adds four explicit constraints so their effect can be measured against that control.

  1. Control firstEqual inputs and opportunities isolate the effect of sequence and chance.
  2. Add one constraintTransportation, health, access, and education can be switched on independently in the laboratory.
  3. Measure pathsAny hardship, compound failure, chronic failure, and displacement remain distinct outcomes.
  4. Keep the boundary honestThe model illustrates mechanisms; it does not estimate the real population.
Read the source essay