Write What Breaks, Not What Works

A generator that is running well is nearly impossible to describe past its first sentence. It hums, it produces light, the fridge stays cold. There is nothing else to say about it, because working correctly is not a source of information, it is simply the absence of anything worth reporting. A generator that is dying, on the other hand, tells you its whole history in one stutter: the specific pitch of a slipping belt, the colour of smoke that means the oil was never changed, the moment it cuts and the exact half second before the inverter catches it. My father could name the make and the fault of a failing generator from two streets away, by ear, and he could not have told you a single distinguishing fact about one that was simply working.

I think most invented systems in fiction are described the way you would describe a healthy generator, which is to say barely, and mistaken for worldbuilding anyway. A magic system's rules, a government's structure, a family's unspoken order, these get a paragraph of specification early on and then run quietly in the background for the rest of the book, doing exactly what they were designed to do, which means they have nothing left to tell the reader that the specification did not already say.

The failure state description

Here is the rule I use now. An invented system earns page space through a scene where it is failing, or has recently failed, not through a scene where it is merely operating as intended. If a system in a manuscript never fails on the page, its description should be cut down to the minimum needed to move the plot, because a reader learns nothing extra from watching something perform its own specification correctly.

Consider a governing council with a fixed procedure for passing law: three members, majority rule, the presiding elder breaks a tie. Written as ordinary function, this is a paragraph of civics that could sit in an appendix. Written as a failure, the same rule tells you where power actually lives in this world. Give the council a law that contradicts a precedent nobody remembered existed, watched by a room that assumed today's vote would be routine. The presiding elder, faced with a genuine tie and a precedent that favours the losing side, rules against the precedent instead, on grounds she invents in the moment and nobody present is positioned to challenge. That single failure, one broken tie, tells the reader more about who actually holds power in this council than a chapter of working procedure ever could, because working procedure only shows you the rule. Failure shows you who the rule bends for.

Why this feels risky to say

I refuse hardship as atmosphere, and I have argued in other pieces that a failing system without a repair ceiling is just magic pretending to be structure, that competence and maintenance matter and should be shown working when they are earned. This is not a contradiction of that, though it will read as one to anyone who took the earlier pieces as an argument for showing systems succeed. A system's competence is proven in one sentence: the light stayed on, the trial proceeded, the bridge held the caravan. That sentence is owed to the reader, once, so failure has a baseline to depart from. What is not owed is the paragraph, or the chapter, spent admiring the system continuing to do what it was already established to do.

The obvious objection

You could say a manuscript built entirely from breakdown is exhausting and eventually reads as despair rather than craft, that readers need scenes of ordinary function to believe a world is stable enough to be worth caring about, and that constant failure is its own kind of dishonesty, no more true to how systems actually behave than constant competence is. I agree, and I do not mean zero working scenes. I mean that once a system's ordinary operation has been established economically, in the space of a sentence, further page space devoted to watching it work should stop, and the page space that would have gone to more demonstrated competence should go toward its next failure instead, because that is where the information actually is.

What the generator taught me that the manual never did

Nobody in our house ever asked my father to describe how the generator worked when it worked. We asked him, every single time it failed, what that particular stutter meant, because the failure was the only part of the machine that ever told us anything we did not already know. I have come to think every system I invent for a page deserves the same asymmetry: a single sentence of function, owed and paid, and then silence, until the day it breaks, because that is the only day it has anything left to say.