Computing Careers 2026

What the evidence says, and what nobody knows

Publishing about computing careers at scale costs money, and a great deal of that money belongs to people selling something into the field: courses, placements, bootcamps, subscriptions, consultancies. That is not a conspiracy and it is not a measurement either. It is an incentive, and it shapes what gets said.

In April 2024 a United States regulator found one of those schools advertising 71% to 86% job placement while its own internal metrics ran closer to 50%, on agreements capped at $30,000: how schools count job placement.

This book was written without any of that behind it. Forty-six computing careers, scored against the same ten measures, and every number followed by where it came from, when it was collected and how hard you can lean on it. Where the evidence was too weak to count as measured, the figure is marked where it is printed, and where nothing was found at any grade the box stays empty. No career in the book clears the standard on both pay and time to enter.

The two halves

The first half is method: twenty chapters on reading a labour market, telling a measured number from a marketing one, and checking a training provider before you give them money.

Six of those chapters are free to read in full, with no email box in the way. That is Part I, the part carrying the grading scale, and none of it is specific to computing or goes stale.

The second half is reference. Forty-six careers, each scored on pay, competition, how you get in, what it costs to get in, what it costs to stay in, exposure to AI, geography, whether it can genuinely be done remotely, other disruption, and what employers actually screen for when they hire.

The first half is written to be read once; the second is written to be returned to. If you are holding a job advertisement and cannot work out what the title really means, the index of 452 job titles will tell you which career it belongs to.

Every number carries a grade

Figures in this book are graded from 1 to 5, and the grade is printed beside the figure rather than buried in a note on methodology at the back.

A grade 1 source is a public body that can compel an answer and has nothing to sell you — a tax authority, a national statistics office. A grade 5 source is a company describing the thing it sells. Salary comparison sites land at grade 4, and Chapter 5 spends a while on why you should not make decisions with them.

The grading gets applied rather than announced, which means it keeps producing results that are awkward for the book itself:

• Where a career has no pay figure meeting the standard, the score is a dash. The book does not estimate to fill a column.
• Where a figure is sound but measures the wrong population, it carries a dagger, and the reason sits next to it rather than in an endnote.
• The chapter explaining the scoring prints a table showing how many careers clear the bar on every measure, which is a chastening thing to publish about your own book.

Leaving that out would have made the book look more authoritative than it is. Every source it used is published, free, with its grade attached, so you can go and disagree with the reasoning.

What it does not know

There is an appendix, a long one, whose only purpose is to list what the book could not find out.

How long it actually takes to enter a computing career is the figure the book most often could not find: none of the 46 careers has a measurement of it that meets the standard. Two countries publish something close, for graduates of a subject rather than for a named job. Both were found by searching in Polish and Italian, and the book says what they can and cannot tell you. Pay data outside the United States and Western Europe runs from thin to non-existent, and there is a chapter on why that is unlikely to change soon.

The whole register is published before you buy anything. Read what is missing, then decide whether the rest is worth your money.

Who it is for

• Graduates who cannot get a first job and want to know whether it is them or the market
• Students choosing subjects, and the parents being sold expensive courses alongside them
• People considering a career change who need to know what the move actually costs
• Anyone stuck in support, testing or admin work looking for the route out
• Readers in India, Nigeria, the Philippines and elsewhere, where the entry routes, the money and the published evidence are all different

There are two country annexes, and a third covering six more markets, because “get a remote job” means something quite different depending on how your money gets home and what happens to it on the way.

Who it is not for

If you want to be told that the right bootcamp will change your life, this is the wrong book. It names doors that are already crowded and are still being sold as opportunities, and it says which certificates have stopped being worth holding.

It also prices them over ten years rather than at the exam fee. What a certificate costs to hold is decided by its renewal rule, and the renewal rule is almost never on the page that shows the fee: across 18 credentials marketed as broadly equivalent, ten years of holding one runs from $165 to $2,225. The working is here: what a tech certification costs to hold for ten years.

That is one of the findings published in full — single questions settled out of the book’s own sources, free to read, with the grade of every figure beside it and a note on what the answer does not cover.

What you get, and how it arrives

• EPUB — the accessible edition, for phones, e-readers and any reading app.
• PDF — the fixed 6×9 edition, for reading on a screen or printing.
• The data — every source, every figure, every gap, as CSV and JSON.

Both formats come with one purchase.

The EPUB meets EPUB Accessibility 1.1 at WCAG 2.1 AA, verified with the DAISY Ace checker. If you read with assistive technology, that was in the specification from the first draft.

A download link is emailed to the address you use at checkout, and the same link stays in the payment provider’s customer portal, so you can fetch the files again later without asking anyone. One warning worth giving in advance: the page you land on immediately after paying shows your invoice, not the book. It looks like something has failed. Nothing has. The files are in the email.

Price

$19, and it stays there. One price, no launch discount and no countdown. A limited-time price that never quite runs out is just a price with a story attached, and this book spends twenty chapters teaching people to notice that sort of thing.

$19 is the ceiling, not the price. It is a very different amount of money depending on where you are, and a good part of this book is written for readers in markets where it is real money, so the checkout adjusts downward based on where you are buying from. That happens automatically. You do not have to ask for it and you will not be asked to prove anything.

If it is still out of reach after that, there is one more route, and it is not a discount code.

If the price is the obstacle

Write and ask, and the book is yours. Two things are asked in return, and neither is a means test.

Read Part I first. It is free, it runs to six chapters, and it is where the method is set out. If it does not prove useful to you, the rest will not either, and you will be out nothing but an afternoon.

Then write a real message. Nothing here will ask what you earn, ask you to document it, or ask you to describe your circumstances: that question is the single reason schemes like this go unused by the people they were built for. Tell us what you are trying to decide and where you are reading from. A career question, in your own words, a few sentences long is plenty.

So it is a test of effort rather than of poverty, and that is deliberate. Anyone who has read six chapters and then written that message is a reader. Write to the address on the corrections page. Asking is not a favour, and it will not be treated as one.

What your copy includes

Every correction to this edition, for as long as corrections are made.

If a figure in this book turns out to be wrong, being told so is part of what you already paid for, and it is not a separate product.

The line: anything that makes the book more accurate is included. Anything that makes it bigger is a new edition. A source revising its figure, a newer year of the same table, a chapter rewritten because it was wrong — all yours. New countries and new careers are a new book.

There is no schedule attached to any of that, deliberately. A correction ships when it is found, not on a calendar.

A correction is not an upsell. This book is sold on the claim that its figures are checkable. Charging you to be told that one of them was wrong would put the evidence behind a second payment.