No Country for Old Coders

Computed diagram: thirteen columns of dots, one per age band; the columns rise to a peak in the early thirties and fall away, and the four oldest, from sixty up, are short and magenta

I became a software developer at 54. How odd is that? In the United States one developer in fifteen is sixty or older, half the share in the workforce; they are scarce because few started, not because they leave.

I started working as a software developer in 2022, at 54, which makes me an odd case in a young trade. How odd? The question is simple: of all the people who write software for a living, how many are sixty or older? Most countries cannot answer it at that level of detail from published tables. The United States can, because its largest household survey, the American Community Survey, releases anonymised answers from more than three million people a year, each with an age and a job.[1] The job is coded from what people say they do; “software developers” is code 1021 in that scheme.[2]

In 2024 the survey counts 2.32 million employed software developers. Of them, about 155,000 were 60 or older: 6.7%, give or take 0.3 points.[1][3] Among all 169 million American workers, 14.5% were 60 or older. A sixty-year-old worker is about half as likely as the average worker to be a software developer.

6.7%of US software developers were 60 or older in 2024, about 155,000 people
14.5%of all US workers were 60 or older
37median age of software developers, against 41 for all workers
0.94developers aged 55–59 in 2019 still counted as developers five years on, per person; 0.85 for all workers
Every age

A young trade

The age profile of developers is a hill with its top at 30 to 34, where almost one developer in five sits. The workforce as a whole is closer to a plateau: roughly a tenth in each five-year band from 25 to 59, then a slope down.

Age of US software developers and of all US workers, 2024 Share of each group in five-year age bands. Developers peak at 30-34 (18.7%); all workers are spread evenly from 25 to 59. Aged 60 and over: developers 6.7%, all workers 14.5%. SHARE OF EACH GROUP, BY AGE, 2024 (%) 0 5 10 15 20 60 and over 16-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75+ Software developers All workers
Figure 1. Share of US software developers and of all US workers in each age band, 2024. Employed people aged 16 and over. Margins of error are under one point for every developer band. Source: American Community Survey 2024, public-use microdata, weighted.[1]

Dividing one share by the other shows where developers are over- and under-represented. They are over-represented from 25 to 44, at up to 1.6 times their weight in the workforce. From 45 the ratio falls steadily: 0.78 at 50–54, 0.59 at 60–64, 0.43 at 65–69 and about a tenth at 75 and over.

How over- or under-represented each age is among US software developers, 2024 Developers' share of an age band divided by all workers' share. Above 1 from 25 to 44 (highest 1.62 at 30-34); below 1 from 50; 0.59 at 60-64 and 0.11 at 75+. DEVELOPERS' SHARE / ALL WORKERS' SHARE, 2024 0 0.5 1 1.5 2 0.04 16-19 0.66 20-24 1.52 25-29 1.62 30-34 1.40 35-39 1.27 40-44 0.92 45-49 0.78 50-54 0.69 55-59 0.59 60-64 0.43 65-69 0.24 70-74 0.11 75+ 1 = same as all workers
Figure 2. Developers’ share of each age band divided by all workers’ share, 2024. A value of 1 means the band is as common among developers as among all workers. Source: as Figure 1.[1]

The bands at the edges are thin. Developers aged 70 and over are about 13,000 people, from fewer than two hundred survey answers; they hold the shape, not the decimals. At exactly 60 the survey has 282 developers, enough to say 1.15% of developers against 1.75% of all workers.

Leaving or never arriving

The same people, five years later

A snapshot cannot say why the older bands are thin. There are two simple stories. In one, developers leave the trade in their fifties, pushed out or moving into management. In the other, few people born in the early 1960s became developers in the first place, because in the 1980s the job was a fraction of its present size, and that small generation is simply getting older.

The two stories can be told apart by following the same birth years. The survey’s occupation codes have been the same since 2018,[4] so the people aged 55–59 in 2019 can be compared with those aged 60–64 in 2024. They are different respondents, not the same people re-interviewed; the comparison is between two samples of one generation.[5]

The same birth cohorts in 2019 and 2024: software developers and all workers Number employed in 2024 at ages five years older, divided by the number in 2019. 20-24 to 25-29: developers 3.55, all 1.19; 25-29 to 30-34: developers 1.55, all 1.05; 30-34 to 35-39: developers 1.26, all 1.06; 35-39 to 40-44: developers 1.25, all 1.06; 40-44 to 45-49: developers 1.01, all 1.02; 45-49 to 50-54: developers 1.04, all 0.99; 50-54 to 55-59: developers 1.06, all 0.93; 55-59 to 60-64: developers 0.94, all 0.85; 60-64 to 65-69: developers 0.54, all 0.56; 65-69 to 70-74: developers 0.42, all 0.52. 2024 COUNT / 2019 COUNT, SAME BIRTH YEARS 0 0.5 1 1.5 2 ↑ 3.55 20-24 → 25-29 25-29 → 30-34 30-34 → 35-39 35-39 → 40-44 40-44 → 45-49 45-49 → 50-54 50-54 → 55-59 55-59 → 60-64 60-64 → 65-69 65-69 → 70-74 Software developers All workers
Figure 3. For each group of birth years, the number employed in 2024 divided by the number employed in 2019, five years younger. Above 1, the group gained people; below 1, it lost them. Source: American Community Survey 2019 and 2024, public-use microdata, weighted.[5][1]

Young developers multiply: the cohort aged 20–24 in 2019 was three and a half times larger as developers five years later, as graduates took their first jobs. Into the forties every cohort grows, faster than the workforce does. From 45 to 59 developers hold their numbers better than workers in general: the cohort aged 50–54 in 2019 was 6% larger in 2024, while all workers of that age lost 7%. That cohort is mine. Its 6% is about 7,000 people, net, roughly one in a hundred of the developers added across all ages in those five years: late arrivals are rare, but they exist. Between 55–59 and 60–64 developers lost 6%, all workers 15%. After 60 both lose about half in five years: that is retirement, at the ages traced in Long Retirement. Only after 65 do developers leave a little faster, and there the numbers are small.

Portugal

Younger still at home

Portugal publishes nothing as fine as the American survey. The 2021 Census counted everyone, but its published tables cross age only with broad occupation groups and with the industry people work in.[6] The nearest thing to software developers is the professionals working in firms whose business is computer programming and consultancy: 52,160 people. The measure misses developers employed by banks, the State or manufacturers, who are probably older, and it includes professionals in those firms who do not write code.

Age of software professionals and of all workers in Portugal, Census 2021 Share of each group in five-year age bands. Professionals in computer programming and consultancy firms peak at 25-29 (23.5%); aged 60 and over: 1.0%, against 10.1% of all workers. PORTUGAL, SHARE OF EACH GROUP, BY AGE, 2021 (%) 0 5 10 15 20 25 60 and over 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75+ Professionals in programming firms All workers
Figure 4. Share of professionals (CPP 2010 major group 2) working in computer programming and consultancy firms (CAE 62), and of all workers, in each age band, Portugal, 2021. A census count, not a sample. Source: INE, Censos 2021, table 6.33.[6]

By that measure 1.0% of them were 60 or older in 2021, 536 people; among all Portuguese workers the share was 10.1%. Where an American sixty-year-old is about half as likely as the average worker to be a developer, a Portuguese one is about a tenth as likely. More than half of these professionals were under 35, against a quarter of all workers, and the band I was in that year, 50–54, held 3.8% of them. The census is a count, so the doubt here lies in the definition, not in chance; and it is a single year, so it cannot follow a generation as the American data can.

The European labour survey points the same way. It splits ICT specialists only into under and over 35, and 44% of Portugal’s were under 35 in 2024, against 37% across the European Union.[7] And it is growing faster than the American count of developers: ICT professionals in Portugal went from 82,500 in 2019 to 150,400 in 2024, up 82%, against 40%.[8] A trade that nearly doubles in five years is filled mostly by the young, and its sixty-year-olds are, for now, the few who were there before.

What it shows

Scarce because few started

On these numbers the shortage of sixty-year-old developers is mostly a matter of when people arrived, not of an early exit. The trade grew so fast that each generation of developers is larger than the one before it, and the one now reaching sixty was small. As the bulge in the thirties and forties ages, the share over 60 should rise. It has started: from 6.1% in 2019 to 6.7% in 2024, while the ratio to the workforce barely moved (0.45 to 0.46).

Three limits are worth stating. First, a net figure hides gross flows: a cohort that holds its size may be losing older developers and gaining others, from abroad or from other jobs, and the survey cannot tell these apart. That leaves room for age discrimination in hiring and firing, which these numbers neither show nor rule out. Second, the job is what people report. A senior programmer whose title is “architect” or “engineering manager” may be coded as a manager, which would make older developers look rarer than they are. Third, the occupation grew from 1.65 to 2.32 million in five years, 40%. That growth is what lets mid-career cohorts gain people at all, and some of it may come from how free-text answers were coded in each year rather than from new jobs.

On method and tools

This note was written collaboratively with Claude Opus 5.5 (Anthropic) on 19 November 2026: Luis Matos Ferreira’s question and editorial direction; the machine’s analysis and drafting. The American figures come from the Census Bureau’s public-use microdata for 2019 and 2024, downloaded unchanged and tabulated by the script scripts/dev_ages.py: employed people aged 16 and over, occupation code 1021, person weights, and 90% margins of error from the 80 replicate weights by the method in the Bureau’s accuracy notes. The Portuguese figures come from INE’s Census 2021 table 6.33, downloaded unchanged and read by the same script; the choice of proxy (professionals in programming and consultancy firms) is ours. The Eurostat figures were read from its database on 19 November 2026. The cover is computed from the same numbers by scripts/old_coders_cover.py.

Authored by: Luis Matos Ferreira — Physicist, Developer, Writer

Related essays on this blog
  1. Long Retirement — the exit age from work, falling for thirty years and rising again since about 2000.
  2. The Load-Bearing Volunteer — the software the world runs on, and the few people who maintain it.
Sources
  1. U.S. Census Bureau, American Community Survey 2024, 1-year Public Use Microdata Sample, person files (csv_pus.zip), released 2025; downloaded 19 November 2026.
  2. U.S. Census Bureau, 2024 ACS PUMS Data Dictionary: OCCP 1021 “Software Developers”, ESR employment status, AGEP age.
  3. U.S. Census Bureau, American Community Survey 2024 1-year PUMS: Accuracy of the data: standard errors from the 80 replicate weights (successive difference replicate method).
  4. U.S. Census Bureau, 2019 ACS PUMS Data Dictionary: the same occupation code 1021, “Software Developers”.
  5. U.S. Census Bureau, American Community Survey 2019, 1-year Public Use Microdata Sample, person files (csv_pus.zip); downloaded 19 November 2026.
  6. INE, Censos 2021, quadros de apuramento, table 6.33, “População residente empregada segundo o grupo etário por ramo de atividade económica, profissão e sexo” (Q633.zip), October 2023; downloaded 19 November 2026.
  7. Eurostat, Employed ICT specialists by age (isoc_sks_itspa), 2024, percentages; read 19 November 2026.
  8. Eurostat, Employed persons by detailed occupation (ISCO-08 two digit level) (lfsa_egai2d), ISCO 25, Portugal, all ages, 2019 and 2024; read 19 November 2026.

Comentários

Mensagens populares deste blogue

Le Grand Raid des Pyrénées

ITRA Performance Index - Everything You Always Wanted to Know But Were Afraid to Ask

Pequenos conselhos práticos - Solo Duro

Provas Insanas - Westfield Sydney to Melbourne Ultramarathon 1983

How Trust Becomes Access

UTAX - Ultra Trail Aldeias do Xisto - 2014