Cutting Your Darlings: A Working Novelist's Revision Process
In June of last year, three weeks before I sent my second novel to agents, I deleted a character. Not a paragraph. Not a scene. A person -- forty-one scenes across three drafts, a younger brother named Desmond who had lived in that manuscript for two years before I admitted he did not belong in it. He had a job, a marriage falling apart on schedule, a running argument with the protagonist about their dead father's boat. None of it touched the plot. I knew this in draft two. I did the cut in draft four, at eleven at night, with a glass of wine I did not finish, and I felt something closer to grief than relief.
That gap -- between knowing a cut is right and actually making it -- is the part nobody writes advice about. Everyone quotes "kill your darlings" like it settles the matter. It does not. It names the problem and then walks away, leaving you alone with a character you built out of real hours and no instructions for what happens next. I have come to believe the phrase has done more harm than good, not because the advice is wrong, but because it frames cutting as a test of nerve instead of what it actually is: a mechanical process with steps, most of which have nothing to do with courage.
The Man Who Never Actually Said It
Here is a fact that surprises most writers when they hear it: William Faulkner almost certainly never said "kill your darlings." The line belongs to Sir Arthur Quiller-Couch, a Cambridge professor of rhetoric who used it in a 1916 lecture later published as On the Art of Writing. His actual instruction was "murder your darlings," and he was not talking about characters or subplots at all. He was talking about a writer's private inventory of fine phrases -- the sentences you are proudest of, the ones you reach for because they are clever rather than because the piece needs them. Cut those, he said, "whole-heartedly," before you send the manuscript anywhere.
Somewhere across a century of repetition, the advice migrated from sentences to scenes to whole characters, and the attribution drifted from an obscure rhetoric don to the most quotable American novelist of the twentieth century, because a stern piece of advice sounds better coming from Faulkner than from a professor nobody has heard of. I think the migration matters. Quiller-Couch was diagnosing a specific vanity -- writing to be admired rather than to be understood. What most novelists actually struggle with is bigger and less vain than that. Desmond was not a beautiful sentence I was showing off. He was two years of real narrative labor that happened to be structurally load-bearing to nothing.
Stephen King gets closer to the size of the actual problem in On Writing, where he writes that you should "kill your darlings, kill your darlings, even when it breaks your egocentric little scribbler's heart, kill your darlings." I like this version better because King is honest about the emotional stakes instead of pretending the whole exercise is about taste. But even King leaves you at the threshold. He tells you the door needs opening. He does not tell you how to tell which door.
What Actually Counts as a Darling
Most of what writers agonize over cutting is not a darling in any useful sense. A clumsy paragraph that just needs a rewrite is not a darling. A scene that is badly placed but structurally necessary is not a darling, it is a scene that needs to move. A darling, in the sense that actually costs you something to remove, has one specific property: you are defending it for a reason that has nothing to do with whether the story needs it.
I have watched myself and other writers defend darlings on exactly three grounds, and none of the three is "the story requires this."
The first is difficulty. Desmond's dead-father's-boat argument took me four drafts to get the rhythm right on. It is a genuinely well-built scene. That is precisely why it was dangerous -- the effort I had sunk into it made me confuse craft quality with narrative necessity. The second is cleverness, the Quiller-Couch problem in its original form: a line, a structural trick, a piece of foreshadowing so elegant you built scenes around it that the plot did not otherwise need. The third is grief for the version of the book you thought you were writing. Sometimes a subplot survives three drafts not because it works but because cutting it means admitting the novel you are actually writing is not the one you set out to write, and that admission is its own kind of loss, separate from whether Desmond himself was ever good on the page.
None of these are moral failings. They are just not evidence. The test for a darling is not "do I love this" -- most good writing is loved by its author, that is not disqualifying. The test is whether the reasons you can give for keeping it survive being said out loud to someone who has not read the manuscript.
The Three Questions Before Any Cut
I now run every suspected darling through three questions before I touch it, in this order, and I do not let myself skip ahead.
Does removing this break causality anywhere else in the book? Not "will I miss it" -- does any later scene depend on information, an object, or a decision that only exists because of this material? Trace it forward, chapter by chapter, and write down every dependency you find. Desmond's boat argument turned out to have exactly one dependency: a single line in chapter twenty-two where the protagonist mentions "what Des said." I rewrote that line in about ninety seconds.
Does any other character's behavior only make sense because of this? A subplot can feel load-bearing because it explains why a character acts a certain way, even when the plot itself never touches it. If cutting Desmond meant the protagonist's guilt in the third act suddenly had no source, that is a real dependency, and the fix is not "keep the subplot," it is "find the guilt a cheaper source."
Would a reader who has never seen this draft notice its absence? This is the hardest question because you cannot answer it as the person who wrote the material. I sent the Desmond-free draft to two readers who had never seen an earlier version. Neither one asked where the protagonist's brother was. That silence was the actual answer, more reliable than anything I could reason my way into at my desk.
Three questions, asked in order, and by the third one most darlings have already answered themselves. The nerve required drops sharply once the decision is a checklist item instead of a referendum on your own resolve.
Cutting Is Not the Same as Deleting
The single change that made me capable of cutting Desmond at all was one I had already built into my process for an unrelated reason: a Scraps folder, sitting quietly outside the manuscript, that catches everything I remove instead of destroying it. Cutting a character out of a novel and deleting a character from existence are two different operations with two very different emotional costs, and conflating them is, I think, the real reason "kill your darlings" reads as brutal instead of practical.
I moved all forty-one Desmond scenes into Scraps the night I cut him. I have opened that particular folder exactly once since, out of curiosity, eight months later, to reread the boat argument. I did not put it back. The point was never that I would need it. The point was that cutting stopped being a referendum on whether the work had value and became a question of whether it belonged in this book, which is a much smaller and much more answerable question. Faulkner-who-did-not-say-it and Quiller-Couch-who-did agree on one thing across a century: murder, not exile. I would rather exile. Exile is reversible in theory, and being reversible in theory is apparently all a writer's nervous system needs in order to stop clutching.
What a Cut Broadway Song Taught Me About My Own Draft
Stephen Sondheim understood this trade better than most novelists I have read, which is a strange thing to say about a composer, but the mechanism is identical. For the original 1970 production of Company, he wrote a song called "Marry Me a Little" as Bobby's Act One closer -- a genuinely good song, technically accomplished, that revealed more of Bobby's interior fear of commitment than anything else in the show. It was cut before opening night, because showing that much vulnerability that early undercut the ambiguity the rest of the show needed Bobby to hold. The song did not fail. The placement failed. Ten years later Sondheim let it become the title track of an entirely different revue built from cut material, and productions of Company have since reinstated it in other slots where the earlier vulnerability does not cost the show anything.
What strikes me about that story is how little it resembles the myth of the ruthless artist murdering his own children. Sondheim did not decide the song was bad. He decided it was in the wrong place, in the wrong show, at the wrong moment, and he was specific enough about why to find it a second life a decade later. That is Scraps-folder logic, executed decades before I ever needed a folder of my own, because the logic is not mine. It belongs to anyone who has ever had to separate the quality of a piece of work from its fitness for the container it currently occupies. A scene can be good and wrong at the same time. Learning to hold both facts without one canceling the other is most of what this whole skill is.
Annie Dillard would push back on me here, and I want to be honest about that rather than pretend the tension does not exist. In The Writing Life she argues almost the opposite position -- "do not hoard what seems good for a later place in the book, or for another book; give it, give it all, give it now." I take her point as aimed at a different failure: the writer who saves their best material for some imagined future project and never actually uses their good ideas on the page in front of them. That is not what a Scraps folder is for. Mine is not a savings account I am building toward some future novel starring Desmond. Of the forty-one scenes I moved out, I expect to use zero of them again. The folder's job was never storage. Its job was permission -- letting me make the cut today by removing the false choice between "keep it" and "destroy it forever," even when the honest long-run outcome is closer to Dillard's position than mine.
The Checklist That Replaces Nerve With Method
Once the character-level cuts are made, the same discipline scales down to the sentence level, which is where Quiller-Couch's original advice actually lived. A gorgeous sentence that exists to be admired rather than to move the reader forward is a smaller version of the exact same problem Desmond was, and it deserves the exact same three questions, compressed: does the plot depend on this clause, does a character's next move depend on it, would a cold reader notice its absence. Doing this pass by feel, sentence after sentence, for a 90,000-word manuscript is how good writers burn out halfway through revision and stop trusting their own judgment. A line-level revision checklist exists precisely to turn that fatigue into a repeatable sequence -- weak verbs, redundant modifiers, the sentences you are proudest of, checked in the same order every time, so the fortieth decision of the day is not weaker than the fourth.
I ran the Desmond-free manuscript through exactly that kind of pass afterward and found six more sentences I had written specifically to service a subplot that no longer existed -- callbacks to a boat that had sailed, so to speak, along with the brother who owned it. None of them were hard to cut. The hard cutting had already happened. What was left was housekeeping, and housekeeping does not require nerve. It requires a list.
When Not Cutting Is the Actual Mistake
I want to end on the failure mode nobody warns you about, because it is more common than the one everyone talks about. Writers do not usually fail by cutting too aggressively. They fail by treating "kill your darlings" as a moral hurdle they have already cleared once -- I cut my prologue, I am the kind of writer who can do this -- and then stopping, leaving three more Desmonds standing in a manuscript because the one act of courage felt sufficient. Cutting is not a trait you acquire. It is a question you have to keep asking, scene by scene, all the way to the last pass, and the questions get easier exactly to the degree that you stop treating them as tests of character and start treating them as a checklist with three items on it.
The wine I did not finish the night I cut Desmond is still, in some sense, the most honest response to that scene I have had. Not triumph. Not relief. Just the recognition that something real had been made and then correctly set aside, the way you'd set aside a chair with a broken leg you built yourself -- proud of the joinery, unable to pretend the thing stands up.
A version history that keeps every cut scene retrievable and a project structure with room for a proper Scraps folder are exactly what Plotiar is built around, so cutting a chapter never has to mean losing it. Free to start.