Skip to content
Elin Marrow
Back to Blog
Craft

On writing work you have actually done

Novels are bad at jobs. A case for research first, metaphor a long way second, and for letting the work be boring where it is boring.

Elin Marrow
Elin Marrow
2026-05-14 · 7 min read
On writing work you have actually done

Fiction is unusually bad at work. Characters have jobs the way they have eye colour — as an attribute, mentioned once, with no bearing on how they spend their days. And when a novel does go into a job, it tends to go in reverent, with the job standing in for something else.

Two failure modes

Romance. The pilot boat as metaphor for the crossing between life and death. The reader can hear the machinery from three rooms away.

Condescension. Work described from the outside by somebody impressed with themselves for describing it. You can tell because the vocabulary is right and the rhythm is wrong: the terms are correct and the sentences do not sound like anybody who does the thing.

The test I use

Would somebody who does this job read the paragraph and wince?

That is the whole test, and it is not about accuracy. You can be entirely accurate and still wince-inducing, because you have made the job significant in a way the person doing it never would. A drainage clerk does not experience the levy roll as a meditation on obligation. A drainage clerk experiences the levy roll as 288 letters that have to go out by the 14th.

Let it be boring where it is boring

The most useful thing I took from eleven years of clerking is that most work is repetitive and that the repetition is not tragic. It is just the shape of the day. Fiction reaches for the exceptional shift, the storm, the accident — and in doing so it misses that the accident is legible only against several thousand days where nothing happened.

So I will spend half a page on a routine that goes correctly. Not to be austere about it, but because the fourth time a character checks the same gauge, the reader starts checking it too, and then when it reads wrong you have not had to tell anybody it matters.

Where the research goes

Interviews, not books, wherever possible. Ask somebody what they did yesterday, not what their job is like — the second question gets you a summary and the first gets you the texture. Ask what goes wrong most often, which is always a smaller and more specific thing than an outsider would guess.

And then leave nine-tenths of it out.

Share this article

The books Get the letter