Understanding Literary Editing—There Are Five Levels

And maybe more.

The idealized vision of writing and a writer’s life is a cruel joke, because writers are constantly reminded of their inability to produce ideal forms. Your instincts are mediocre—you’ve heard millions of words of talk, and everyone is a talker—until you’ve expended thousands of hours of effort to suppress your mode-seeking tendencies, and even once you achieve this, the skill you have acquired is still so exhausting that it cannot be sustainably applied for as many hours per day as you and society both expect you to be able to commit. Writing gets harder the better you get. The first version of a sentence, paragraph, essay, or story is often so inferior to what you intended to produce, you find it an embarrassment. You must revise.

The writing world has a paradoxical relationship with revision that is seldom discussed. Unpublished amateurs self-edit heavily—once a manuscript is submitted to a literary agent, it has often gone through rounds of beta reading and five or six revision passes—but the industry consensus is that they make little progress by doing so; effort cannot overcome a lack of talent, and anyone who relies on a query letter is presumed to be talentless. Published authors, whether in trade or as self-publishers, are encouraged to capitalize on momentum, which means little time can be spent on revision, as it is better to release imperfection while one is remembered than to polish but be forgotten. High-status “career authors” are told to eschew self-editing entirely, with the exception of structural work on direct advice from a literary agent or acquisitions editor; serious revision is considered a waste of time that could be better used for marketing. In other words, self-editing is seen by publishing as something unsuccessful writers do pointlessly—a low-status activity one quits once one’s time is worth something—and that successful writers no longer need to do. Trade publishing promises authors that all editorial needs will receive the attention of skilled professionals, who will work over the text in multiple passes, until perfection is achieved. In truth, the quality of others’ work is wildly variable, even in an elite publishing setting, and you will be unable to evaluate anyone else’s competency unless you learn how to do these jobs for yourself.

Revision is made difficult by the need to operate on the text at several levels; they compete for one’s attention. Writers are generally advised to start at high levels of abstraction and finish low; fixing a plot or character arc may introduce new spelling errors, but fixing spelling will usually not alter plot. This said, it is usually impossible to follow a strict high-to-low schedule, because low-level signals often indicate which high-level gambits have worked. In a large manuscript, different sections will sit at different stages of development, and often one must re-engage with the text, after some time away from it, to know which level of abstraction requires the most attention—sadly, you don’t get to know this in advance. If we could reduce editing to techne and predict how many stages there must be, writing well would be easier, but it is not that way. Still, in an effort to extract some order from chaos, I will argue that revision of literary text tends to exist at five levels: developmental, structural, tactical, line, and copy editing.

1: developmental editing—deciding what the project is, and what it is not

The obvious: You can’t edit until you have something to edit. The first stage of development is not editing; it is the first draft. Some authors are outliners who plan stories before they throw down words. Others are discovery writers (“pantsers”) who find the shape of the book as they write it. I’ll let you in on a secret: it doesn’t really matter. You will have, either way, a first draft full of missed opportunities, loose threads, flawed executions, and failed experiments. You’ll have information in the prose foreground that belongs in the world notes. You’ll make a spelling or grammar error every 300 words or so, because everyone does. You will make up the game’s rules as you play it. You’ll have to fix this stuff; all important writing requires significant rewriting. You can write a “vomit draft” with no outline and pay a slightly higher tax in later revision, or you can spend time upfront to build the story and have slightly less editing to do. If outlining motivates you, then outline; if discovery motivates you, then don’t. Your goal in the first draft is to stay motivated enough to finish.

Writing tends to expand, then contract, as you revise it. The first draft will be underweight. This is the phase in which “show, don’t tell” is good advice, and development is sometimes described as “the snowflake method” in reference to a fractal curve produced by iterative expansion. You swap out bland details in favor of evocative, interesting ones. You search for places where you took the easy way out, and try to do something hard and distinctive instead. You refine themes and story logic; you strengthen necessary connections, and prune what cannot be connected. You’ll expand what works and what matters; you’ll trim infrastructure until it is almost invisible; you’ll cut the diversions that didn’t lead anywhere.

When development turns into developmental editing, it’s often targeted. Acceptable subject matter, narrative style, and word count depend on one’s audience, and there are no absolutes. Authors seeking commercial success will make different choices from those aiming to win awards. Traditional publishers tend to target specific trends and genres—often, contrary to the author’s own desires—for season-specific commercial reasons or due to idiosyncratic career needs, while developmental editing toward artistic objectives—the mythologized Maxwell Perkins treatment—is nearly extinct. You’re on your own. The people pushing ideas—the literary agents with their “manuscript wish lists”—would be out there executing if they had any talent or taste.

Your first book will spend years in this phase. Your fourth draft will be superior to your first draft, but it will be closer to the first draft of a far better book than to an experienced author’s fourth draft.

2: structural editing—building the form

Publishing tends to combine developmental, structural, and tactical editing under one banner. My view is that they’re separate. Developmental editing encourages expansion, because the project is still mostly undone. The decision to add—to turn the love interest into a real character with backstory and independent motivations—is often correct at that stage. Structural editing begins when the story feels full; it requires saying no to good ideas. Pacing, proportionality, and cause-and-effect logic become major concerns. Plot issues are debugged. Characters are removed from scenes where they add nothing. Scenes that start too early or too late are fixed.

At this point, you’ll start to find the work’s natural writing style. Are you writing with heavy interiority, or taking a McCarthy-esque landscape approach? Can a year pass between two chapters, or does all time have to be accounted for? Does your dialogue prefer blunt naturalism or literary flourish? You can use a three-act structure or a five-act structure or something else entirely, but you should know what it is by this point. Pacing has no absolutes, but consistency is mandatory; the story should speed up as it approaches the climax, but not become so fast it turns omissive or reads as an easy way out (e.g., a deus ex machina).

The developmental phase answers the question of what your book is; the structural phase is the one in which you decide how it should work. Writing can always break “the rules,” but it is almost never a good idea for a story to show indifference toward the rules it sets for itself. Development settles the question of what kind of story should exist; structural editing defines how it should work and what each piece should do. And the next step is not yet language—line and copy editing—but execution; the author must get the tactics right.

3: tactical editing—solving the puzzles

Tactical editing tends to mix most readily with the other phases. Structure sets objectives and determines what tactics are possible, so a tactical desire or insight might require upward changes, while line editing itself tends to have tactical implications. As already discussed, these phases do not always separate clearly, and this is most true in tactical editing. One needs to make sure the writing gets where it needs to go on time.

There are “hard” and “soft” tactical concerns. In a medieval story, a character who is in Paris on Friday cannot be in London on Saturday; that’s hard tactics. Soft tactical concerns are aesthetic—scenes followed by sequels, avoidance of repetition—but equally important. Some authors do separate tactical passes for dialogue, pacing, and description, but I’ve never been able to separate concerns so cleanly, and there is an inherent danger in stretching this phase out for too long, which is that one can drift. Consistency here is essential. A story that is deliberate, descriptive, and maximalist cannot jump to thriller pacing or introduce rapid travel in its final act; similarly, one that is tightly plotted should not become lyrical and introspective as it heads toward its climax.

The long middle of the revision process is mostly tactical in nature. The author knows who the characters are, how they speak, what they will do and what they will not do… and must remove old elements that no longer fit. Chapters are shaped and checked for relative size and span. Transitions between movements are refined, so plotting feels inevitable rather than contrived. Sentences that obviously only do one thing are replaced by those that do (or seem to do) several.

Tactical editing’s objectives are genre-specific, so hired editors are often ill-equipped to do it. Traditional publishing ignores this field of concern entirely. Unlike developmental editing, it cannot be sales-driven, because no novel can be pitched on the strength of its tactics. Unlike copy editing, it does not lead to negative reviews and embarrassing pull quotes if it is skipped. Thus, it shares with line editing the distinction of being artistically relevant but commercially invisible, and therefore it is ignored by traditional publishing.

4: line editing—the sentence of the turn

Line editing is exhausting. You get about 90 minutes of this per day; two or three hours, if you’re lucky. You’re permuting sentences and tweaking words, trying to try on an exponential number of combinations, but we all know that it’s intractable to search the entire space, so you’re finished when it looks like you’ve given every scene, paragraph, and sentence its best execution. How do you know if you’re done? You don’t. At some point, fatigue sets in and your judgment deteriorates; you will find errors no one else sees and make “improvements” that don’t need to be made, while missing things others would spot.

Such as: When do you combine sentences? When do you cut details for speed or musicality of prose? Do you split a 200-word paragraph, or trim it down to 120, or leave it as-is because it’s the exceptional case that holds? Do you deploy the rule of three, or violate it, or fulfill it in some silly way to make a point? When do you pull a word that four people out of a hundred will recognize, and when do you use more common language—at the cost of imprecision or word count? When do repeated words smell like death, when do they hold positive rhetorical value, and when are they benignly invisible? Commas; semicolons… if you know you know. And then the hardest one: How do you tell the difference between a true cliché and an idiom that we deal with because it has become ordinary language and disappears harmlessly? The list goes on.

An author who nails 60 percent of the truly close line calls will read as sophisticated. One who hits 70 will exceed, in writing mechanics, all but a few dozen authors in traditional publishing at any given time. And 80 might be unachievable, because some of these calls are so subjective, opinions flip by the hour. What tortures us as literary writers is that we want 100.

This is probably the lowest level we can do for ourselves at book length, the reason being that it is never done. We have the skills copy editing requires—with time and emotional distance, we can polish our own copy to a professional standard. The problem is that we never do, not at 50,000+ words, because we will always find line-level issues, even if none exist. And then you are line editing, not copy editing; as you are not fully attentive to copy-level issues, you’ll introduce new errors at the lower level.

There are also, at this level, a lot of rules that midwit literary agents push; unskilled writers often take them seriously. “Omit adverbs.” Sorry, but no. You should audit all adverbs, because they can be very inelegant—the very stays, midwits; go back to your Manhattan lunches—but if you are a serious self-editor then, guess what, you’ll be auditing all the other words, too. You should study how language works and be aware of its general principles, but there’s no reason to take “agent rules” to heart, because literary agents will never read you unless you have the social status to get read and, even if they do, they will never read closely enough to notice what you are doing at the line level. Scared of your writing looking like the rest of the slush pile? Relax. No one will ever know if it does or does not. You can cut all or none or some of the adverbs in your manuscript and, in that time, the six-figure deal you ought to have earned for all that effort will be given to an illiterate with 1,800,000 Instagram followers and butt cheeks made out of siliconized rubber.

You don’t line edit because it will be rewarded. It probably won’t be. You do it because, dammit, words matter.

5: copy editing—completeness and production values

All writers make mistakes. We repeat words, break grammatical rules unintentionally, and leave sentences hanging because we got distracted by other things. This might be why it’s so dangerous to internalize “agent rules” like the supposed prohibition against adverbs; the cognitive load of trying to follow them tends to produce inattention upward and downward. Every constraint added to writing consumes our finite attention; some of these constraints are artistically beneficial or necessary, but those driven by the tastes of literary agents will never be worth what they cost.

Readers have a justified expectation of correct execution on matters of spelling, punctuation, grammar, and stylistic consistency. A number of these rules are habit or protocol; it isn’t objectively wrong to write May 7th instead of May 7, but everyone who’s taken a course in copy editing will find the former to be unprofessional and circle it, and quite a few readers will notice inconsistency at a subconscious level. No reader wants to buy a book that wasn’t taken seriously at every stage; in this light, production values really do matter. Books are judged on their covers. Interior design matters. Justifiably, readers hate having to second-guess the words in front of them, so every spelling or grammar error has a cost.

Copy editing is—as far as publishing is concerned, and even in the minds of most writers—the least interesting kind of editing. I have a background in logic and computer science, which leads naturally to linguistics, so I enjoy this stuff, but I would agree that it is inessential for most authors to know the details of comma placement and orthography when there are paid professionals who specialize in it. And yet it is, in terms of establishing and retaining readers’ trust, the most important.

LLMs are excellent at short-form copy editing. Where they still fail—they are getting better, but I’d still rather have an attentive human—is at long-form copy editing, which pertains to consistency both in story facts (e.g., a character who is fifteen in 1980 should not be forty in 2000) and stylistic decisions. Furthermore, LLMs should never replace humans at a nonfiction copyeditor’s most important job, which is fact checking—the hallucination problem is, it should be obvious, an absolute dealbreaker here.

6: … and then there is chaos

To take this five-level model of editing as absolute truth is dangerous, as this grappling with the writer’s inability to easily produce ideal forms cannot be used as the source of a new one. In my experience, it is impossible to know the degree to which a previous pass has succeeded until one has taken time away from the work, and the level of abstraction at which one must work is not uniform while working over a manuscript. As I said, some line fixes have tactical implications, while tactical considerations can expose serious structural flaws. Writers learn over time how to prevent “emergency” upward revisions, but this issue cannot be removed fully. Also, it is impractical, for most writing, to do five or six editing passes; concerns must sometimes be combined.

The truth remains less-than-ideal. We do not will ideal forms into existence; we iterate toward them. The product is exceptional writing that looks easy and obvious, and even we fall prey to the disease of false triviality—the tendency to believe something is embarrassingly easy after achieving it—while simultaneously being daunted by the threat of having to follow our own achievements without repetition. If we could devise an orderly process to overcome all this and to produce excellent work on time, we would not need one in the first place.

Given the class implications, this might be the oddest metaphor ever, but golf comes to mind. Please ignore your justified disgust toward the sorts of people who stereotypically play it; I am going somewhere with this. Getting a tiny ball into a tiny hole over a 500-yard distance in four or five strokes seems like an impossible task. I’m not able to do it; I’ve played before, and I’m terrible. The skills are developed over time; the golfer drives for distance, and putts for precision, and achieves with apparent effortlessness what ought to be extraordinarily difficult. Writing and self-editing contain the same paradoxical combination. The prose reads easily on first glance, as if it came naturally, as a stand-up comic’s rehearsed jokes also seem improvised. And then deeper study shows an apparent pattern of sustained best execution—a writer who reliably picks the best option in an astronomical search space. And both of these impressions obscure the banal fact of the task being seriously (but not infeasibly) difficult.

Serious writing cannot be timed; convergence always takes longer than one expects. It cannot be industrialized; trade publishing has tried, and the result is writing that moves copies based on trends but that no one takes seriously as writing. We can only move gradually toward understanding; in this light, I believe the model I have given above is a contribution.