Work at One Scale at a Time

A writer sits down to a chapter and everything is open at once: whether the chapter should exist, the order of its sections, the strength of a paragraph, the rhythm of a sentence, the spelling of a name. A section moves, its first line is reworked, the premise is doubted, a comma is fixed. At the end of the evening the pages are different, and the writer can’t say whether they’re better, because no single question was ever asked.

The editors who prepared technical reports at the Jet Propulsion Laboratory gave this problem a formal answer. They defined nine types of editorial work, among them substantive, language, mechanical style, format and policy, and combined the types into five cumulative levels, in a document whose second edition appeared in 1980 under the title The Levels of Edit. A manuscript was assigned a level, so that editors, authors and managers could agree which work was to be done on it. The report’s authors say the concept improved communication among those three groups about the specific work each manuscript needed, and that it served as a tool for estimating and monitoring cost.

Revising alone, a writer has the same problem with nobody to disagree with, and the likely cause is that no one has decided which question the session is asking. At the scale of purpose the question is whether the chapter earns its place. Down at the sentence, it shrinks to whether this sentence sounds right after the one before it. Ask both at once and the vivid small problem tends to win, because it is easy to see and satisfying to fix, while the large one stays vague. A sentence labored over in the first hour also tends to resist the verdict, reached in the third, that its paragraph should go.

Practice 82 recommends an order for revision across the whole manuscript, from purpose downward. The unit here is smaller, the sitting. Whatever the order, a session is easier to run, and its results easier to judge, when it has one declared scale. You can still notice other things; you just don’t act on them. The scales don’t stay apart, either. A repeated local problem often has a structural cause, and a session at one scale will turn up evidence for another, so write the evidence down and let the response wait for its own session.

Hemingway, whose morning rereading Practice 81 describes, is a working counterexample. For a writer who drafts forward from yesterday’s pages, mixing scales is a feature, and the maxim is meant for the stages where evaluation, not momentum, is the task.

Practice. Take the next three revision sessions. Before each, write a line at the top of the working document naming the scale, the question and what is out of scope: “Structure. Does each section earn its place? Not wording.” In a short poem the line break, the sentence and the architecture are often one decision and the scales don’t separate cleanly, so there the useful declaration isn’t “sentences today” but a specific question, such as whether the turn arrives where it should. Keep a parking list beside the line, and whenever something at another scale catches your eye, add it and keep going. At the end of the session, sort the list by scale and assign each item to a later session. Afterward look at what changed in the manuscript, which should all belong to one scale, and at how long the parking list grew, which measures how often your attention strayed. If the parking list is longer than the list of changes, the manuscript is asking a question at a different scale, and the list shows which.