It is a toy. The odds behind it are not.
Spin the globe and the page picks one country. The fun part is watching the globe fly somewhere you have never thought about. The part worth explaining is why that particular somewhere came up — because the answer is not "at random" in the way people usually mean it.
Most "where would you be born" toys pick a country off a list, so every country is equally likely and you land in Tuvalu about as often as you land in India. That tells you nothing. This one asks a narrower question:
Of all the babies being born in the world this year, pick exactly one at random. Where was that baby born?
So a country's chance is its share of the world's births. India and Nigeria come up constantly. Iceland almost never does. That lopsidedness is the whole point — it is what the world actually looks like, and it is not what most people expect before they roll a few times.
The By population switch asks a different question with the same machinery: pick one person alive in 2026 rather than one baby born in 2026. The two give noticeably different answers, and the gap between them is roughly the gap between where the world lives now and where it is heading.
Every figure comes from the UN World Population Prospects 2024, the medium variant, for the year 2026. Two columns per country: projected population and projected annual births.
| Source | UN World Population Prospects 2024, medium variant |
|---|---|
| Year | 2026 |
| Entries | 237 countries and areas — every one the UN lists |
| Units | Thousands, as published |
That last row of the table matters more than it looks. Because all 237 entries are in the draw, the weights add up to the whole world. There is no "other" bucket and no rounding-off of small places, so every birth the UN counts has somewhere to land on this globe. An earlier version carried only the 196 largest and covered about 99.9% of births — which sounds close enough until you notice it meant the remaining 0.1% of people simply could not be born.
The medium variant is a projection, not a measurement. 2026 has not finished happening. Treat the numbers as the UN's central estimate rather than a count.
Both columns for all 237 rows are published here as data.json, which is the file the globe actually draws from — there is no second copy. If you would rather read them than roll for them, the full odds table ranks every country and area by its share of the world's births.
The draw is a weighted pick. Line up all 237 countries end to end, each taking a stretch as wide as its births, then throw a dart at the line.
let random = secureRandom() * totalBirths
for (const item of data) {
random -= item.births
if (random < 0) return item
}secureRandom() reads from crypto.getRandomValues(), not Math.random(). Nothing here needs cryptographic randomness — it is a globe toy. But Math.random() makes no promises about its distribution, and the one claim this page makes is that the odds are honest. It would be a strange thing to be sloppy about.
Each roll is independent. The globe has no memory, no streak protection, and no nudge toward interesting answers. If it sends you to India three times in a row, that is simply what a 1-in-6 chance does sometimes.
The globe is an orthographic projection rendered to canvas with d3, using the Natural Earth 110m basemap. That basemap is built for speed, and at that scale it drops countries too small to draw — Singapore, Bahrain, Kosovo, Tuvalu and others.
They are still in the draw, so they would have been picked and then silently not shown. Two things fill the gap:
Loading the full 50m map instead would cost roughly eight times as much per frame, which is enough to kill the auto-rotation on a phone. Two small files were cheaper than a slow globe.
Nothing you do here is stored. There are no cookies, no accounts, and no ads. Anonymous usage counts are collected on a self-hosted analytics server — the details are in the privacy policy.