7.0 How to use the protocol
The ten tests below are for revision, not composition. Applied during drafting they produce a writer who cannot put anything in the past without a permit; applied to a finished draft they are the most efficient triage instrument in this chapter.
Five procedural notes, one of them specific to this chapter and important.
- Establish first whether the protocol applies to your book. §3.9 sets out the condition: the deficit model assumes the present timeline carries the engine. If your novel’s present is deliberately thin — a drive west, a night’s conversation, a man dying in a room — then tests 1, 3 and 8 will return failures that are not defects. Decide this once, before you begin, rather than discovering it test by test. A writer who applies the protocol to The Remains of the Day and concludes that it should be cut has not found a fault in the book.
- Run them in order. Tests 1–4 are structural; their answers determine whether the later tests are worth running. There is no point tuning a trigger on a cutaway that should be a sentence.
- Record the answer, not the verdict. The output of a test is a sentence describing what the cutaway does, not a pass or a fail.
- Run tests on the manuscript, not on the passage. Five of the ten (3, 7, 8, 9 and 10) cannot be answered by looking at the cutaway alone, which is the chief way this protocol differs from the one in Chapter 9.
- A failed test is a question. Every one of these tests is failed by at least one novel discussed in this chapter, most of them by several.
7.1 The ten tests
Test 1 — The momentum test
What present-tense question is the reader in the middle of wanting answered when this cutaway begins? Name it in a sentence.
If no such question can be named, the timing is arbitrary rather than chosen. Note carefully what this test does and does not detect. It does not detect that the cutaway is badly placed; it detects that the placement was not a decision. A cutaway arriving at a moment of no present-tense pressure may be perfectly placed — that is often where a cutaway should go — but a writer who cannot say what the present question was has not chosen anything.
The refinement that makes this test useful: rate the pressure of the named question on a three-point scale — low, moderate, at peak. Then compare with test 5. A cutaway interrupting a peak question needs a return proportionate to what it interrupted; a cutaway placed at a low-pressure moment is cheap and may be doing more than the writer credits.
Test 2 — The register test
Run the one-sentence compression check from §6.3. Does this need to be felt, or only known?
Write the compressed sentence out. Place it in the manuscript. Read forward two pages. The three possible answers are keep the scene, keep the sentence, and keep neither — and the third is the most common outcome in first drafts.
Test 3 — The removal test
Cut the cutaway entirely. Read the surrounding twenty pages. What is lost, and in which domain?
The answer must name a domain from §4 and must be specific. “The reader wouldn’t understand her” is not an answer; “the reader would not know that she had asked him about the pen and been given an answer she did not understand at the time” is. If the loss cannot be located in any domain, the passage is decorative and the removal is the revision.
A refinement worth adopting: perform the removal on paper, and leave it removed for a day before reading. Writers reading a passage they have just deleted supply it from memory.
Test 4 — The trigger test
Is the cutaway motivated by something concrete in the present scene — an object, a phrase, a sensation, a question asked, a place, a silence — or does it arrive as an unmotivated authorial jump?
Consult the table at §6.6. Then ask the second half of the test, which is the half that matters: if there is no trigger, is the narration of a kind that can afford to do without one? Confident omniscient narration and retrospective first-person narration can move into the past without motivation; close third-person narration usually cannot, because it has promised to stay inside a consciousness. An untriggered cutaway in close third is not breaking a rule about flashbacks; it is breaking the narration’s own contract.
Test 5 — The return test
Does the story return to the present changed, or does the present scene simply resume?
Identify which of the four changes at §6.6 has occurred: changed knowledge, changed understanding, changed question, or changed action. If the answer is changed knowledge alone, the change is in the reader rather than in the story, and the cutaway is doing less than it appears to.
The mechanical form of this test: cover the cutaway with a sheet of paper and read the last sentence before it against the first sentence after it. If they read continuously — if the second follows from the first as though nothing had intervened — the cutaway is untethered (§8.5).
Test 6 — The reach-and-extent test
How far back does this go, and how much time does it cover? Are the two proportionate to what it delivers?
The apparatus is at §3.2. Record two numbers — years back, and story time covered — and place the cutaway in the four-cell table. The cell to interrogate is long reach, short extent: the distant single moment, which requires the reader to build a whole vanished context and then immediately discard it. A cutaway in that cell must be delivering something that could not have been delivered by a nearer moment. Very often the same character point is available from a memory five years old rather than thirty, at a fraction of the construction cost.
Also record whether the reach shortens during the cutaway. A long analepsis that moves forward, closing the distance to the present, generates its own momentum; one that stays at a fixed distance does not, and will need to find momentum elsewhere.
Test 7 — The chronology test
Could a careful reader, at this point in the book, write out the story’s events in the order in which they occurred?
This is the test built on Ohtsuka and Brewer’s finding (§3.6) that readers reconstruct an underlying chronology from a non-chronological telling, and that they pay for the reconstruction. The procedure is literal: take a reader who has read to this point and ask them to produce a timeline. Then compare it with yours.
Three outcomes, with three different diagnoses:
| Outcome | Diagnosis |
|---|---|
| The reader’s timeline matches yours | The anachrony is legible; proceed |
| The reader’s timeline has gaps in the places you intended | Working as designed — those are Sternberg’s open questions |
| The reader’s timeline has gaps you did not intend, or events in the wrong order | The book has confused rather than withheld, and the two feel identical from inside the manuscript |
The third outcome is the one no amount of self-reading will reveal, which is why this test requires another person.
Test 8 — The placement test
Would this cutaway cost less thirty pages later? Or thirty pages earlier?
The most productive single test in the protocol, and the one that most often produces a large improvement for a small edit. The deficit is positional (§3.9): the same material at a different point in the present timeline may cost a fraction as much and pay considerably more.
The procedure is to name three alternative locations and, for each, answer two questions: what present-tense question would be interrupted there, and what would the cutaway be able to recolour that it currently cannot? A cutaway placed after the present behaviour it explains recolours that behaviour; the same cutaway placed before it merely prepares for it. Consult the three-position table at §4.5.
A related and slightly brutal version of the test: would this material be better as a later revelation than as a present cutaway? Much of what drafts deliver as flashback in the second act is more valuable withheld until the third.
Test 9 — The completing-versus-repeating test
Is this cutaway filling a gap the reader knew was there, or returning to material already narrated?
The apparatus is at §3.3. The two answers carry different burdens.
- Completing. The reader had an open question. Check that the question was actually opened — that some earlier sentence marked the gap — and check the interval. A gap opened four hundred pages ago and filled now may have gone cold.
- Repeating. The reader has been here before. Name, in one sentence, what this pass adds that the previous one lacked: a new fact, a new vantage, a new emotional weight, a revealed lie. If nothing can be named, the passage is redundant (§8.2). If something can, the pass is legitimate however many have preceded it.
The test also catches a third case that neither category covers: a cutaway that is neither completing nor repeating because the reader had no relation to the material at all — no open question, no prior account. This is the pure informational cutaway, and it is almost always a candidate for compression (test 2).
Test 10 — The proportion test
Across the manuscript, how much of the book is in the past, and does the book’s shape acknowledge it?
Run on the audit sheet from §5.5. Count the pages. The number is not a verdict — there is no correct proportion — but the relation between the number and the book’s declared structure is diagnostic.
| Proportion of past material | Question it raises |
|---|---|
| Under 5% | Is the book’s present adequately underwritten? A novel whose characters have no history is a diagram |
| 5–20% | The ordinary range for a present-tense-driven novel; no question raised |
| 20–40% | The book has a substantial second timeline and should probably declare it as one (§4.6) |
| Over 40% | The book is a retrospective novel. Its present is a frame, and the frame should be designed as one rather than functioning as an unwilling host |
The over-40% row is where the largest structural discoveries happen in revision. A manuscript that is 60% past material and formally organised as a present-tense novel with frequent flashbacks is fighting itself. Reorganised as a retrospective novel — a first narrative that is openly an occasion for recollection, as in The Remains of the Day — the same material stops apologising for itself.
7.2 Reading the results
| Failed tests | Likely diagnosis | First move |
|---|---|---|
| 2 alone | Over-dramatised backstory | Compress to the sentence you wrote; check what the freed space allows |
| 1 and 4 | The cutaway is on the writer’s schedule, not the reader’s | Test 8: find the placement where a present question already exists |
| 3 alone | Decorative past | Cut. Do not attempt to add a function to material that has none |
| 5 alone | The untethered return (§8.5) | Rewrite the return sentence for a changed question, not a re-anchoring |
| 6 alone | Long reach, short extent; construction cost exceeds occupancy | Find a nearer memory that makes the same point |
| 7 alone | Confusion mistaken for withholding | Supply one date, or one clause of orientation. Usually a very small edit |
| 8 alone | Right material, wrong place | Move it. This is the cheapest large improvement available |
| 9 alone, repeating | Redundancy | Name the difference or cut the pass |
| 10 alone | The book’s proportions and its declared structure disagree | Change the structure, not the proportions |
| 1, 3 and 5 together | The cutaway is not connected to the book | Consider whether the material belongs to a different novel |
A warning about test 8. When a cutaway is not working, writers instinctively revise the cutaway, because it is the object in front of them. The passage is usually not the problem. In many of the cases this protocol catches (an estimate from drafts read, not a count), the material is sound and its position is wrong — and moving four hundred words is a smaller operation than rewriting them, with a larger effect.