Why the word exists, how it decays, and why development is growing its own version of it right now.
I used the word "manual" in a post and heard about it. Several people, independently, making versions of the same argument: nobody says manual development, or manual sales, or manual product management. So why testing?
I sat with that longer than I expected to. Here is where I ended up, and it is not quite where I started.
Testing got the word because the work got divided. One person designs what to check, another runs it. Once a machine could do the running, the part left over needed a name, and "manual" was the word lying around.
Development never got divided that way. The thinking and the artifact were the same object — you could not hand the typing to a machine and keep the reasoning. So there was nothing left over and nothing to name.
A tester I respect pushed back on exactly this, and the pushback was fair. He argued that the thinking and the doing were never actually separable in testing either — that typing is simply how understanding becomes code, and the same is true of a test. The division was invented by people who did not understand the discipline, on a false premise, and the industry propagated the mistake.
I think we were using "split" to mean two different things, and both are true. He means the work was never genuinely separable. I mean organisations divided it anyway — one team writing cases, another executing them, frequently at different pay bands. The premise was false and it got operationalised regardless. The word is what came out of the operationalising.
I was treating this as a vocabulary problem. The word is inaccurate, so pick a better one. Several people have tried: core testing, functional, exploratory, experiential.
Every replacement inherits the same flaw, and one of the people proposing an alternative spotted it in his own suggestion: it keeps the division and relabels the sides. Whatever you call the two halves, one of them ends up looking like the modern one.
But the sharper objection came from someone describing what happens to the word over time, and it is the thing I had not thought through.
First the word is just a label. Then it becomes a question — why are you not using AI for that? Then it becomes a line in a budget, read by someone who has never watched the work being done.
That is the mechanism. The damage is not done on the day the word is coined. It is done years later, in a meeting you are not in, by someone who wants the work to look cheap. A vendor deck promising to save a trillion manual days does not have to argue that judgement is worthless. It only has to make one side of the division easy to price.
Testing did not lose that argument on the merits. Nobody sat down and decided that judgement mattered less than tooling. It just became easier to put a number on one than the other.
Here is what makes this urgent rather than academic.
Since models started producing code, developers have begun saying "hand-written" and "AI-generated." I have caught myself saying it in a code review without thinking about it. Same division, forming in real time, in a discipline that was immune to this two years ago.
If it follows the path testing took, the label will not stay on the code. Labels move to people. Someone gets described as a hand-written developer in a job posting, it reads as a downgrade, and nobody can quite say when that happened.
Testing had a twenty year head start on this argument and still has not fixed its own vocabulary. That is not encouraging.
Honestly: I still use the word, and I want to be straight about why rather than pretend otherwise.
The people I write for have "Manual Test Engineer" printed on an offer letter. They type "manual QA to automation" into a search bar at eleven at night. If I use more accurate language, the person who most needs the guide never finds it. So I use their words to reach them, and then spend the piece arguing that the judgement they already have is the valuable half and the tooling is the smaller one.
Whether that is the right trade is genuinely arguable. One hiring manager told me he treats the old terminology as a red flag in a candidate — which I understand, and which also places the cost on the tester rather than on the organisation that wrote the job description. The person most likely to say "manual tester" is the one whose contract says it.
What I have stopped believing is that a better adjective is the fix. The word is a symptom. The division is the disease, and the pricing is what makes it fatal.
The most useful thing in any of these conversations was not a proposed replacement term. It was an automation lead saying he goes to bat for the core QA team when leadership is considering cuts. That does more than the vocabulary ever will.
If you are making the manual-to-automation move yourself — and yes, I am using the word — the 90-day roadmap is the practical version of this. Or find out where you would land in an interview.