7. Diagnostic Method

7.0 How to use the protocol

The ten tests below are for revision, not composition. Applied during drafting they produce paralysis of a particular kind — the writer stops before every scene to ask whose head it belongs in, which is the surest way to write nobody’s. Applied to a completed draft, they are the most efficient structural triage in this chapter.

Four procedural notes.

Run them in order, and stop when one fails badly. Tests 1 to 3 are about whether a grant should exist at all; tests 4 to 7 are about the shape of the allocation; tests 8 to 10 are about execution and consequence. There is no point tuning the knowledge boundary of a vantage that should not have been granted.

Run them on grants, not on scenes. The unit under test is a viewpoint character across the whole manuscript. A test applied to a single chapter will return a local answer to a structural question.

Record the answer, not the verdict. The output of each test is a sentence describing what the book does. “Dov’s vantage serves sympathy and contrast, and appears once, at 78%” is useful. “Fails test 7” is not.

A failed test is a question. Every test below is failed by at least one novel discussed in this chapter, several of them by Middlemarch. The test’s function is to make a choice visible so that it can be confirmed as a choice.

7.1 The ten tests

Test 1 — The necessity test

Could an existing vantage deliver this material by report, inference, reassignment, document or ellipsis?

Procedure. For each scene held by the vantage under test, name the cheapest instrument that could carry its content, and write the substitute in one sentence. Then read the substitutes as a set.

Interpretation. If most of the scenes translate without loss, the grant is an access grant that should be reconsidered. The important refinement is that the test asks about loss, not about information: report can always convey a fact. What report cannot do is dramatise a decision as it is taken, render a misapprehension from inside, or make a silence feel like a choice rather than an absence. If the loss is one of those three, the necessity test is passed.

Test 2 — The domain test

Which of the six domains does this specific grant serve?

Procedure. For each vantage, list the domains from §4, and beside each write the specific textual evidence — the scene, the passage, the effect. Domains asserted without evidence do not count.

Interpretation. A grant serving only access is under-employed and should be converted to a cheaper instrument or given more to do. A grant serving two or more, at least one of which is not access, is earning its seat. A grant serving none — which happens more often than writers expect, particularly with viewpoint characters inherited from an early draft — should be removed.

A useful secondary output: tabulate the domains across all vantages. A book whose every grant serves access and pacing has a delivery mechanism rather than a design; a book whose every grant serves contrast may be an argument with characters attached.

Test 3 — The investment test

Has this character earned enough presence and interiority to justify holding the reader’s head, or do they read as a plot vehicle with a name?

Procedure. Before the character’s first viewpoint section, count their appearances, and write down what the reader knows of their wants — not their function, not their relation to the protagonist, but what they are trying to get.

Interpretation. A vantage whose recipient had no legible want before the grant will read as an appointment. The remedy is usually not to add interiority inside the grant but to add presence before it: the character should have been on the page, wanting something visibly, for some time. This is the “cold start” charge at §6.6, and it is paid before the grant rather than during it.

Test 4 — The distinguishability test

Swap the contents of two viewpoint sections between their holders. Could a reader tell?

Procedure. Take a passage of two hundred words from vantage A and a passage of two hundred from vantage B. Change only the proper nouns required to make each grammatically consistent with the other’s situation. Give both to a reader who knows the book.

Interpretation. This is the most unforgiving test in the protocol and the most diagnostic. If the passages survive transplantation, the book has two cameras and one consciousness — the redundant vantage (§8.5). The repair is at the four levels of §5.3: what each notices, what each is willing to appraise and in what words, what each cannot know, and what each calls things. Note that the test is about mind, not about voice: two vantages may be perfectly distinguishable in idiom and still fail, if the idiom is the only difference.

A weaker but faster version, usable alone: cover the chapter headings of a multi-viewpoint manuscript and read the first paragraph of each section. Count how many you can assign correctly.

Test 5 — The page-share audit

What proportion of the book does each vantage hold, and what shape does the distribution make?

Procedure. §5.5 in full. At minimum: word count per vantage as a percentage, number of sections, onset page, longest gap.

Interpretation. Per §5.5’s table. The two findings that most often force a structural decision are approximate equality across five or more vantages (an ensemble governed by fairness rather than by need) and a long gap (a vantage the reader must re-enter as a stranger). Neither is automatically wrong; both must be defended rather than inherited.

The audit also answers a question writers cannot answer introspectively: which character is the book actually about? The answer is very often not the one on the synopsis.

Test 6 — The first-grant timing test

On what page does the reader learn that this book has more than one vantage?

Procedure. Find the page. Then find the page at which the reader’s investment in the first vantage becomes substantial — in practice, the end of the first sustained scene in which that vantage wants something and is resisted.

Interpretation. If the second is earlier than the first, the book has trained the reader to expect single narration and then contradicted the training. This is Chapter 9 §4.6’s formal contract, and the expected symptom is resistance rather than confusion: the reader understands perfectly and resents it.

The remedy is rarely to move the second vantage’s material. It is to declare the possibility earlier — a short section, a heading convention, a part division — so that the eventual shift is a fulfilment rather than a surprise.

Test 7 — The late-grant test

If this vantage arrives after the book’s opening movement, does the payoff cover the disruption?

Procedure. Name the three charges from §6.6 — formal surprise, attention withdrawal, investment cold start — and estimate each. Then name the domain payoff and ask whether it could have been delivered by report.

Interpretation. Late grants pass when three conditions hold together: the recipient has been present and observed from outside for a long time; the grant is bounded; and the material dramatises a decision rather than delivering information. They fail when the recipient is new, the grant is open-ended, or the content is expository.

A specific and common failure: a late grant introduced because the plot has become impossible to convey from existing vantages. That is a plot problem presenting as a viewpoint problem, and adding a vantage treats the symptom.

Test 8 — The knowledge-boundary test

Does this vantage know only what it could know?

Procedure. Mark every sentence in a viewpoint section that reports a fact the focaliser could not have. Include: other characters’ thoughts and motives; events elsewhere; physical descriptions of the focaliser’s own face and expression; information the focaliser learns later in the book; and — the subtlest — judgements the character is not capable of forming.

Interpretation. A handful of marks is normal and, in most architectures, invisible to readers. A steady scattering is the omniscient slip (§8.6) and indicates that the book’s architecture is not what its author believes. A cluster around plot-critical moments is the vantage that knows too much (§8.7), and it is usually the writer solving an information problem at the reader’s expense.

The productive follow-up is not to delete the marked sentences but to ask what the vantage would have had instead: an inference, a misreading, a wrong guess. Those are more interesting than the facts they replace, which is why the test so often improves a chapter it was meant only to correct.

Test 9 — The scene-integrity test

Within a single continuous scene, does the narration stay with one consciousness — and if it does not, is the book’s architecture one that permits movement?

Procedure. Take five continuous scenes at random. Mark each sentence with the initial of the mind whose perceptions or judgements it reports. Where a scene shows more than one initial, ask the second question.

Interpretation. The two-part structure of this test is deliberate and is the practical form of §2.3 and §8.1. Mixed initials in a book whose architecture is single fixed (1), single mobile (2), alternating dual (3) or rotating ensemble (4) constitute head-hopping: the narration is leaking outside the convention it has established. Mixed initials in a mobile omniscient architecture (10) are the architecture operating.

The diagnostic question is therefore never did the narration move? but has this book established the right to move, and does it exercise it consistently or only when convenient? A narration that moves at moments of plot necessity and stays put elsewhere has not established a right; it has found a loophole, and readers detect the difference.

Test 10 — The character-compression test

What has happened to the characters who receive no viewpoint?

Procedure. List every character with a name and more than two scenes who holds no vantage. For each, record: (a) how many scenes they appear in; (b) whether their motives are ever explained; (c) whether the explanation comes from them, from a viewpoint character’s inference, or from the narrator; (d) whether the reader could state what they want; and (e) whether they have acquired a repeated identifying tag — a gesture, a phrase, a physical feature — that stands in for characterisation.

Interpretation. This is Woloch’s test (§3.3), and it is the only one in the protocol that looks away from the material under revision. Item (e) is the key. A repeated tag is the visible mark of character compression: it is what a person becomes when a book needs them to be recognisable and has no room to make them knowable.

Character compression is not a fault. Every novel compresses its minor characters, and must. What the test asks is whether the character compression is distributed as the book’s argument requires. Three findings to watch for:

• The unexamined antagonist. The only character whose motives are never available is the one the plot needs to be wrong. The book has arranged for its own verdict.
• The instrumental relative. A parent, a spouse or a child who appears in many scenes, is never explained, and exists to make a viewpoint character feel something.
• The compressed group. A whole class, profession, ethnicity or generation present in the book without a single interior. This is the finding with the largest consequences and the one most often invisible to the author, because absence is not salient.

The output of the test is not a demand for more viewpoints. It is a demand that the writer be able to say why these people are outside, and be willing to own the argument the arrangement makes.

7.2 Reading the results

A warning about test 5. When a page-share audit reveals that a beloved secondary vantage holds four per cent of the book, the instinct is to expand it. This is almost always the wrong direction. Four per cent is what the book has decided that character is worth; the productive question is whether those pages are the right four per cent, and whether they should be one bounded, unforgettable section rather than six thin ones.