Is software development a fast-growing job in the United States?

This page was wrong between 20 and 22 August 2026, and it was corrected before the book went on sale. It said software developers alone grow at “7% or higher” against 15% for the wider occupation. That was backwards. The correction, and how the error happened, is at the bottom of this page.

Is software development a fast-growing job in the United States?

Yes — and slightly faster than the figure everyone quotes. The widely cited 15% is not a projection about software developers. It covers two occupations bundled into one headline. Taken apart, software developers are projected to grow 15.8%, and the other job in the bundle — quality assurance analysts and testers — 10.0%.

What the 15% actually counts

The series is named “Software Developers, Quality Assurance Analysts, and Testers”. That is one headline built from two occupations the Bureau of Labor Statistics also publishes separately.

Here are both rows, from Table 1.2 of the 2024–34 Employment Projections:

The two rows sum to the headline. Their annual openings add to exactly 129,200, which is the figure the Occupational Outlook Handbook publishes for the bundled occupation at 15% growth. So the published headline is these two occupations and nothing else — which is what makes it possible to take apart.

What that means for the number people quote

It slightly understates software developers. They are at 15.8%, not 15%. Anyone quoting the headline as a figure about development is being modestly pessimistic, not optimistic.

It flatters the other job. Quality assurance and testing is projected at 10.0%. Someone reading “15%” as a number about testing has been handed a figure half again too high for their own work.

So the headline is not false. It is just not about one job, and the two jobs inside it are moving apart — which the headline cannot show you, because the headline is their average.

One thing this does not settle

That second row counts quality assurance analysts and testers together. It does not separate people who test by hand from people who write test automation.

This book argues elsewhere that purely manual testing is closing as a career. Both can be true at once: the occupation grows while one way of doing the work shrinks inside it. The occupation code cannot tell you which, and nothing published separates them. That gap is recorded with the others rather than filled with a guess.

The correction, and how it happened

The earlier version of this page compared 15% for the bundle against “7% or higher” for developers, and concluded developers grew at half the advertised rate.

That 7% was never a growth rate. It came from O*NET, which reports projected growth as a band rather than a number. Its page for software developers reads:

Much faster than average (7% or higher)

“7% or higher” is the floor of the fastest band. Every occupation in that band carries those same words. Reading it as this occupation’s rate produced a comparison between a rate and a category threshold.

The arithmetic on this page ruled it out, which is how it was caught. Developers are about 89% of the bundle’s openings. A component that large cannot grow at half the rate of the whole unless the remaining 11% grew explosively, and manual testing plainly did not. The contradiction was printed on this page from the day it was written, and someone read it carefully enough to see it. That is the argument for publishing figures next to each other rather than one at a time.

Three things were wrong in one source record: the figure, the publisher — it was attributed to the Bureau while citing a republication — and the reliability grade. It is now read from the Bureau’s own published table.

Where these come from

The bundled figures are from the Occupational Outlook Handbook, recorded as bls-swe-outlook-2026. The detailed rows are from Table 1.2, Occupational projections, 2024–2034, published at bls.gov/emp and downloaded as occupation.xlsx, recorded as bls-employment-projections-2024-2034. Both are the US Bureau of Labor Statistics, both are graded 1, and both records with their full caveats are in the free source bibliography under those identifiers. They are named here rather than only linked because a page that tells you to go and check should tell you what to look for.

No link is given for the table itself. bls.gov refuses automated requests, so a link here could not be checked the way this book asks readers to check everything else. It has to be downloaded by hand, and it is worth doing: the whole of this page is two rows of one spreadsheet.

Why this page kept its address

It would have been easier to delete it. It stays because a book that asks readers to check its sources should show what happens when a check fails — and because the URL was public, so anyone who read the wrong version can find out that it was wrong, at the address where they read it.

Found by a reader. If you find another, the address is here.