Treat the Outline as a Hypothesis

Practice. Take your outline, or a quick list of your units if you have no outline, and spend twenty minutes marking each element as one of three things. Fixed means something outside the draft requires it: an external constraint, a documented fact, a promise already made to a reader or collaborator, or an element the draft has already tested and confirmed. Probable means you expect it to hold and have a reason. Exploratory means you are guessing and want the draft to tell you. Beside every fixed item, write the reason in a few words, and demote any item you cannot give one. If everything is fixed you have written a contract, and if nothing is, you have a mood. Start drafting at the exploratory items, where the plan has the least to say.

George R. R. Martin, asked often whether his novels would end the way the television adaptation did, took up the question in a 2022 blog post. An architect, he wrote, could give a short, simple reply, but he is much more of a gardener. The final destinations and the big set pieces have been in his head for years. The details are another matter, he said, and the story sometimes shifts under him as he writes.

An outline is full of predictions like his destinations: that if the confrontation falls here, the reader will feel the double betrayal; that three examples will carry the second claim; that the reader needs the definition before the objection. Drafting is the experiment, and when the draft disagrees, ask which prediction failed. Maybe the second example contradicts the third, which means the claim as stated is wrong and the argument has just taught its author something. Predictions can also come true, and a confirmed plan was still a hypothesis while the draft was testing it.

A writer who holds the plan as a contract treats any surprise as a defect in herself, and may produce scenes she no longer believes because the outline says the scene exists. One who drops the outline at the first friction keeps no record of what the friction showed. Each has thrown away information. Treated as data, a deviation can be examined; treated as a failure, it only gives you someone to blame.

Some plans are rightly stiff. A textbook with chapters assigned to four authors or a legal brief with a word limit needs a structure that holds still enough to coordinate. There a change has to be negotiated with the people who depend on the plan, and recorded.