Part one ended with an inventory: the list of places you have to visit to write one page of your own book. This part is about the bill.
It is a strange bill, because it never arrives. No session ends with a line item saying eleven minutes lost to reassembly. The cost is real and the invoice is not, which is why writers can pay it for an entire manuscript without once deciding it was worth paying.
So let me itemise it. Three line items, in ascending order of how much they matter.
Scatter charges you three times: minutes at the start of every session, continuity across the whole book, and - the expensive one - the consistency of your own voice.
Line item one: the re-entry cost
Every writing session starts with reassembly. Not writing - reassembly. Opening the draft, opening the outline to remember where you were going, opening the notes to reload the details, scrolling back through yesterday's pages to find the tone.
Call it ten minutes. Sometimes it is three, sometimes it is twenty, and it is longer when the gap since the last session was longer, which is the worst possible property for it to have. The writers with the least time - the ones fitting a book around a job, writing in forty-minute windows - pay the largest proportional tax. A ten-minute re-entry against a three-hour Sunday is a rounding error. Against a forty-minute weekday window it is a quarter of the session, gone before a word.
This is the line item that gets the most attention, and it is the least important of the three. Minutes are recoverable. You can find more of them. What follows cannot be recovered by working harder.
Line item two: the continuity cost
Here is the mechanism, and it is worth being precise about it, because the usual framing - writers should keep better notes - misses where the failure actually occurs.
You are mid-scene. It is going well. You need a detail you established earlier: the brother's age, the name of the street, whether the ritual needs blood or only intent. The detail exists. You wrote it down. It is in the apparatus, which is in another app.
You now make a decision, in about a second and a half, and you make it with the momentum of a good scene on the line. Almost always you guess and keep writing. This is not a failure of discipline. It is correct. Momentum in a draft is more valuable than a single accurate detail, and an experienced writer knows it.
The trouble is that the guess is silent. It does not announce itself. It goes into the draft looking exactly like the details you were sure of, and three chapters later you write consistently with the guess rather than the original, because the guess is now what you remember. The error does not stay a single error. It propagates.
- Details established early and needed late - the ones furthest from working memory.
- Anything you invented (place names, rules, invented vocabulary), because there is no real-world fact to fall back on.
- Relationships and ages, which are the details most often stated obliquely rather than written down.
- Rules you set for your own world, which are the most expensive to break because a reader who catches one stops trusting the rest.
The bill for this one arrives late and in a lump: a revision pass that is not craft work but repair work, hunting inconsistencies you introduced under momentum months earlier. I have written about the way manuscripts stall at predictable points; a good share of what makes the late zones feel unclimbable is accumulated repair debt of exactly this kind.
Line item three: the voice cost
This is the one worth the whole piece.
Your voice on the page is not a decision you make once. It is a thing you stay in, sentence by sentence, and staying in it depends on being able to hear it. That is why writers reread the last page before starting a new one: not for the plot, for the sound.
Now put the book in eleven places. What can you hear? You can hear the last page of the file you have open. You cannot easily hear chapter one from chapter twenty-two. You cannot scan the whole book for the rhythm you started with, because there is no whole book to scan - there are fragments, in different apps, in different fonts, some of them in a boards tool that renders your prose in a sidebar.
A voice you cannot hear across the whole book is a voice you can only maintain from memory. Memory drifts.
And so it drifts - not dramatically, never in a way you would catch in a single session. Chapter twenty-two is written in the voice of a writer who has read chapter twenty-one. Chapter one was written by someone else, months ago, and you cannot hear them from here. This is a slower cousin of the flattening I wrote about in The Beige Drift, and it has nothing to do with what tools you use to draft. It is purely a function of what you can hear from where you are standing.
Why word count hides all three
Every one of these costs is invisible to the metric most writers track.
A session where you paid ten minutes of re-entry, guessed at two details, and drifted slightly out of voice can produce a perfectly respectable eight hundred words. The counter goes up. The bar fills. By the only measure on your dashboard, it was a good day. The costs land later - in a revision that takes three months instead of one, in a continuity pass, in the strange dispiriting experience of reading your own first act and not recognising the writer.
That gap between the metric and the reality is why this bill goes unnoticed for entire books. Word count measures production. It cannot measure coherence, and coherence is what scatter erodes.
Open your first chapter and your most recent one side by side and read both aloud. Not for quality - for sound. Sentence length, rhythm, how much you explain, how often you use a comma where you could use a full stop. If they sound like two writers, that is the voice cost, and it is measurable in ninety seconds.
The counter-argument, taken seriously
Plenty of good books were written scattered. Plenty were written on index cards and in six notebooks, and the finished work shows none of it. So the honest version of this argument is not that scatter prevents books. It obviously does not.
What scatter does is make finishing more expensive - in reassembly minutes, in repair work, in the willpower needed to keep a voice you cannot hear. Writers with a great deal of time, or a very forgiving book, can absorb that. Writers working in forty-minute windows around a job, on a book with an invented world and a large cast, mostly cannot. The cost is real for everyone; the question is whether your circumstances let you absorb it.
And there is a version of this that is genuinely no problem at all: a short book, held whole in your head, written in one long stretch. If that is you, this series is not for you, and you should go and write it.
What actually fixes it
Note what does not fix it. Better notes do not - the notes are already good, they are just elsewhere. More discipline does not - the guess under momentum was the right call. A stricter filing system does not, because the cost is not in the filing, it is in the distance between where you are writing and where the answer lives.
What fixes it is reducing that distance to nothing. Part three is the working method for doing that: how to consolidate a scattered book without losing a week to the migration, and what becomes possible in editing once the whole manuscript is in one place - the passes that simply cannot be run on a book that lives in eleven.
