Universe 25
In 1968, ethologist John B. Calhoun built a paradise for mice at NIMH, Maryland. A 4.5-foot metal cube with unlimited food, unlimited water, zero predators, zero disease. The perfect conditions for life.
He introduced 8 mice — 4 males, 4 females. The population doubled every 55 days. By day 560, there were 2,200 mice. The space could hold 4,000.
Then something broke. A generation emerged that Calhoun called the “beautiful ones” — mice that refused to mate, refused to fight, spent their time obsessively grooming. They were physically perfect and behaviorally dead.
The last birth occurred on day 920. The population spiraled to zero. On May 23, 1973, the final mouse died.
“I shall largely speak of mice, but my thoughts are on man.”— John B. Calhoun
Calhoun ran it 25 times.
Every time: extinction.
Experiment Timeline
8 mice introduced
Population doubles every 55 days
Peak population (capacity: 4,000)
"Beautiful ones" emerge
Last birth recorded
Final mouse dies
Modern Parallel
The Global Picture
A world map colored by fertility rate reveals a stark pattern: the more modernized a country, the fewer children it produces.
Loading world map...
The Classifier
Three logistic regression models were tested. The simplest one — just 3 features — performed best. Adding more variables made it worse.
| Model | Features | LOO-CV Accuracy | Misclassified |
|---|---|---|---|
| Structural (6) | urban, speed, edu, internet, WLFP, GDP | 75.8% | 15 |
| + Meaning (7) | above + fight_for_country | 75.8% | 15 |
| Minimal (3) | urban, edu, internet | 79% | 13 |
Modernization Threshold Test
Countries where urbanization > 65% AND internet penetration > 70%
Of 41 modernized countries, 35 have collapsed below replacement. The 6 survivors:
P(Collapse) vs Actual TFR
Israel stands out as the largest anomaly — high predicted probability of collapse, yet the highest TFR among developed nations.
Feature Explorer
Select any two variables to see how they correlate across 62 countries. Israel is always highlighted — it defies every trend line.
Nine Countries, Nine Stories
Each country tells a different chapter of the same story.
South Korea
Seoul: 0.55$270B in incentives. Zero effect. 63% of income goes to mortgage. The lowest fertility rate in recorded human history.
Japan
11% would fightThe lowest willingness to fight for country globally. Calhoun's "beautiful ones" — below replacement since 1974. 50 years and counting.
Iran
7.0 → 1.7 → 2.08Fastest decline ever recorded. The Islamic Revolution couldn't stop it. Slight bounce during economic crisis — stress sometimes raises fertility.
Ukraine
War = accelerantGDP collapsed, millions fled. War drove fertility lower. The exact opposite response to Israel's war baby boom. Same stimulus, opposite effect.
Sweden
$52K GDP, 480 days leaveThe perfect welfare state. Universal healthcare, subsidized childcare, 480 days parental leave. Still below replacement. Welfare can't fix aspiration.
Morocco
Collapsed at $3,800 GDPA smartphone in Casablanca delivers the same aspiration as a penthouse in Stockholm. You don't need to be rich to stop having children.
Egypt
5 governorates = 45% of birthsA split country. Rose from 3.0 to 3.5 between 2006–2014, then fell again. Rural Egypt and urban Cairo are different civilizations.
Nigeria
Lagos 3.6, Sokoto 7.2The full model in one country. Southern Nigeria is modernizing and fertility is falling. Northern Nigeria looks like 1960. The entire thesis in miniature.
Israel
P(collapse) = 87%The anomaly. 93% urban, 90% internet, $42K GDP, 13 years education. Should have collapsed decades ago. It didn't. We don't know why.
Dead Ends
Three popular explanations for fertility decline. Each one fails the data.
"Fight for Country"
Willingness to defend doesn't predict fertility
Countries at both extremes of patriotic willingness have collapsed. The variable adds noise, not signal.
"Welfare State"
Money and benefits can't buy babies
Nordic countries prove that even perfect welfare states cannot maintain replacement fertility. Bangladesh proves poverty doesn't protect either.
"Religion"
Faith doesn't prevent collapse
Every major religion has populations below replacement. Modernization overpowers religious pronatalism in every case tested.
The Israel Anomaly
The model predicts Israel should have collapsed. It didn't. This is the most important data point in the dataset.
Model Prediction
87%
P(Collapse)
Actual TFR
2.90
Well above replacement
Sub-Population Breakdown
2024 War Baby Boom
Following the October 7th attack and war, Israel experienced a measurable baby boom — the opposite of Ukraine's response to war.
Sept 2023
14,878
births
Sept 2024
15,968
births
Change
+7.3%
increase
Same Stimulus, Opposite Response
October 7 attack → existential war
TFR: 3.06 → 3.19
↑ Fertility increased during war
Russian invasion → existential war
TFR: 1.16 → 0.90
↓ Fertility collapsed further
The Druze Puzzle
The most important finding in this research. A population that should have the highest fertility in Israel — by every theory — has among the lowest.
Druze Fertility Trajectory
From 7.9 children per woman to 1.66 — a collapse steeper than any modernized nation.
Every Protective Factor Present
By every mainstream theory, the Druze should have the highest fertility in Israel.
Result: Below replacement, shrinking 0.9% annually
Population: 148,000 → declining
“If every theory says this population should have the highest fertility in Israel, and it has among the lowest — every theory is wrong.”
What Comes Next
Countries approaching the modernization threshold — and likely to cross it within 10-15 years.
High Confidence (by 2035)
Already at replacement
At the boundary
Already collapsed
Already collapsed
Urban areas already below
Already collapsed
Moderate Confidence (by 2040)
Rapid urbanization
High internet, urban
99% internet, fast decline
Entering Danger Zone (2035-2040)
Lagos already at 3.6
Urban Kenya declining fast
High internet for Africa
Calhoun ran it 25 times.
Every time: extinction.
The human experiment is running once.
62 data points say we're on the same curve.
1 data point says we might not have to be.
We don't yet understand what makes that one different.