Data Article · Culture
America Ran Out of Common Names
In 1950, a third of American boys got a top-ten name. Today it's one in eleven — and the names that do catch on burn out twice as fast. A century of Social Security records, read as a history of individualism.
Every year since 1910, the Social Security Administration has counted every first name given to five or more babies in each state. It is close to a perfect cultural record: near-universal, annual, and unpolluted by what people say they value. What it shows, over 116 years, is a country steadily abandoning the idea that a name should be ordinary.
In 1950, the ten most popular boys' names covered 35.2% of all boys born; the girls' top ten covered 25.6%. By 2025 both figures had fallen to around 9%. Measured as an effective number of names — the diversity index economists use for market concentration, applied to nurseries — boys went from the equivalent of 58 equally-common names to 347; girls from 89 to 429, still climbing.
The collapse of the common name, 1910–2025
Left: share of babies given a top-10 name. Right: effective number of names (inverse Herfindahl).
Data table (by decade)
The turn is precisely datable: boys' top-10 share peaks in the late 1940s and erodes slowly until 1990, then falls off a cliff. Girls led the whole way — female naming has been more diverse than male naming in every single year since 1910.
1 Look up any name
Aggregate indices hide the fun. The 2,163 names below are every name that ever reached 0.05% of same-sex births. Some are mountains (Linda's 1948 peak — 6.03% of all girls, the largest single-name year in the record, powered in part by a hit song). Some are mesas (Elizabeth, essentially flat for a century). Some are needles: Ayden, Nevaeh, Brittany — up and down within a generation.
A century of any name you like
Share of same-sex births, 1910–2025
Data table (selected name)
2 Fads burn faster now
Define a name's rise time as the years it takes to climb from a quarter of its eventual peak to the peak, and its fall time as the years to sink back below a quarter of peak. For names that peaked in the 1920s–40s, the median fall took 24–27 years — a name was a generation-scale commitment. For names peaking in the 1990s–2000s the median fall is 14–16 years and still shrinking. The rise has always been quick; what changed is how fast America moves on.
Name fads collapse twice as fast as they used to
Fall time (years from peak to 25% of peak) by year of peak · one dot per name · line = decade median
Data table (decade medians)
3 The one-way frontier
Names cross the gender line — almost always in one direction. Leslie, Ashley, Aubrey, Sydney: each began as a boys' name and became, within a few decades, overwhelmingly a girls' name. The reverse journey is nearly absent from the record. The pattern is so regular it has a sociological name — parents colonize "distinguished" male names for daughters, after which they become unavailable for sons.
Twelve names crossing (or guarding) the gender line
Share of the name's babies who are girls · midline = 50/50
Data table (decadal)
4 Watching a name conquer the country
Names diffuse through the states like any innovation. Jennifer ignites in the late 1960s — earliest and hottest on the coasts — saturates the whole country by 1972, and recedes everywhere in near-unison. Nevaeh ("heaven" backwards) starts in 2001 from nothing, after a musician named his daughter Nevaeh on MTV, and spreads inland with the geography reversed — strongest in the heartland, weakest on the coasts. Drag the slider.
How a name spreads, state by state
Share of same-sex births in each state · darker = more popular
Data table (selected year)
What unites these panels is a single long trend: naming has moved from a convention you follow to a statement you compose. The cost of that freedom is churn — the faster fads of Figure 3 — and its perimeter keeps expanding: the SSA now records about twice as many distinct names per year as it did in 1950, even as births have fallen.
Data & methods
- Source. Social Security Administration state-level baby-name files (public domain), 1910–2025, downloaded July 2026: 6,696,687 rows of (state, sex, year, name, count). Cross-validated against the BigQuery mirror of the same data for 1910–2021; decade totals agree within 0.7%.
- Suppression. Name-state-year cells with fewer than 5 babies are omitted by the SSA. "National" totals here are sums over states, so rare names are undercounted and every diversity measure is a lower bound — more so in small states and recent years (as naming spreads out, more names fall under the threshold).
- Early data. Social Security numbers were introduced in 1936; earlier years reflect retroactive registration and skew female. Levels before ~1937 are treated gently; within-era trends are fine.
- Definitions. Share = babies with the name ÷ same-sex births recorded that year. Effective number of names = 1/Σ(share²). Fad rise/fall times use a 25%-of-peak threshold; names whose fall is censored at 2025 are excluded from Figure 3, and peaks after 2007 are excluded entirely to avoid right-censoring bias.
- Unisex panel. Names with ≥2,000 babies of each sex all-time and ≥10,000 total; yearly points require ≥50 babies.