The Generator Test for Worldbuilding

Every compound in Lagos with any ambition owns a generator, and every generator owns a man who knows it more exactly than the manufacturer does. The machine is not the infrastructure. The man is.

I did not understand this until I watched Mr. Eze from three doors down spend forty minutes with our neighbour's generator, listening to it the way a doctor listens to a chest, before he found the fuel line had been fouled by a bad delivery from the filling station on Adeniran Ogunsanya. He did not open a manual. He had one number in his head, the amount of diesel that generator should burn in an hour when it was healthy, and the machine was burning less, which meant something upstream was starving it. That number, the expected burn rate, is the whole of his expertise. Everyone else on the street just knows there is light or there is not.

This is the test I now run on any impressive object in a draft, mine or someone else's. I call it the generator test, and it is one question: does this thing have an expected burn rate, and does anyone in the story know it?

What the question actually checks

It is not a question about mechanism in the technical sense. I am not asking whether the floating city's antigravity is explained. I am asking whether anyone inside the story treats the floating city as a system with a normal operating state, so that when it deviates, someone notices before the reader has to be told directly. A generator running low does not announce itself. It changes pitch. Someone who has spent a year listening to it hears the change before the lights do anything.

Run the test on your own draft. Take the single most impressive invented thing in it, the artifact, the institution, the working method of the villain's order. Ask: who in this story would notice if it were running at eighty percent instead of a hundred, and how would they know, and what would they do about it before it failed completely. If your honest answer is nobody would notice, the thing is decoration. It can still be lovely decoration. It is not load-bearing, and you should stop asking it to carry plot weight it cannot support.

A failed generator test, corrected

An early chapter of mine had a witch's coven maintaining a ward around a border town, described once, gloriously, in the opening pages, and never checked on again until it dramatically failed in chapter twelve. That failure read as arbitrary to my first reader, and she was right, because nobody in the story had an expected burn rate for the ward. I fixed it by giving the coven's youngest member, Ijeoma, a Tuesday ritual: she walks the boundary stones every week and each one should hum at a pitch she has memorised. In chapter nine, one stone hums a third low. She reports it. Nobody acts on the report for three chapters, because the elder in charge is distracted by a funeral, and that inaction, not the ward's failure itself, is what causes chapter twelve. The mechanism did not need explaining. The habit of checking did, and once it existed, the failure had a history instead of a coincidence.

The honest objection

Sceptics of this approach will say that constant monitoring kills suspense, that a ward which hums a warning three chapters early has told the reader the ending. This is a real risk and I do not think it is fully answerable. Foreshadowing that is too legible is a genuine failure mode, and I have cut Ijeoma's report down twice in revision because early drafts had her escalate too loudly, too fast, and the reader stopped worrying about anything except when the elder would finally look up.

The fix was not removing the warning. It was making the response to the warning fallible in an ordinary way rather than a plot-convenient way. The elder does not ignore the report because the plot needs a disaster. She ignores it because she is grieving, which is a reason a reader recognises from their own life, and only in hindsight does grief look like negligence. The generator test does not require your systems to fail loudly. It requires someone to be capable of noticing quietly, and it requires you to have a reason, independent of plot convenience, that they did not act on what they noticed.

Why this beats the usual advice

Most guidance about worldbuilding tells you to be consistent, which is true and useless, because consistency is a property you check after the fact, not a thing you can do while drafting a scene. The generator test is a thing you can do while drafting a scene, because it is a single question aimed at a single object, answerable in under a minute. It converts an abstract virtue into a specific person with a specific number in their head, listening to a specific machine. Mr. Eze retired two years ago. The compound hired a younger man who uses an app that logs the fuel gauge automatically, and the older residents complain that outages last longer now, because the app tells you the tank is low after it is already low, and Mr. Eze could tell you it was going low. The instrument changed. The requirement, that someone in the system be listening before the failure, did not.