1. Introduction: The Allocation Problem

1.1 An asymmetry between coverage and depth

Consider two manuscripts of identical length, both about a family of four after a death.

The first is told entirely through the eldest daughter. Everything the reader learns, they learn because she learned it — by being present, by being told, by inferring, by being lied to. Whole regions of the story are unavailable: the brother’s negotiations, the mother’s reasons, the father’s last months. The book is, in an obvious sense, badly informed. It is also, at the end of four hundred pages, in possession of one person so thoroughly that the reader can predict what she will notice in a room before she notices it.

The second is told through all four in rotation. Nothing is unavailable. The brother’s negotiations are dramatised; the mother’s reasons are supplied; the father speaks for himself in a section set before his death. The book is comprehensively informed. It is also, at the end of four hundred pages, in possession of four people to one quarter of that depth each — and, if the rotation was governed by fairness rather than by need, possibly to less, because a reader who has been moved four times has been asked four times to begin again.

Neither manuscript is better by construction. What separates them is not craft but allocation: a decision about how a finite quantity of reader attention and reader identification is to be distributed across a cast. That decision is made once, structurally, at the level of the whole book — and then honoured or violated at every scene boundary in it.

This is the allocation problem, and it is different in kind from the problem Chapter 5 (The Transition) solves. Chapter 5 asks how a reader is carried across a gap; it assumes the far side has already been chosen. This chapter asks who is standing there.

The asymmetry can be stated in its sharpest form, because it is what makes viewpoint a structural rather than a local decision:

• Coverage is additive. Each new viewpoint character adds the scenes that character can attend. Two viewpoints see roughly twice as much of the story as one.
• Depth is divisive. Each new viewpoint character divides the same finite quantity of pages, of remembered detail, of accumulated familiarity. Two viewpoints are known, on average, half as well as one.

A book cannot buy coverage without spending depth, and cannot preserve depth without accepting blindness. Every viewpoint architecture in §6.3 is a particular settlement of that trade, and the settlements are not interchangeable.

1.2 The claim in its strong form

Granting a character viewpoint is the largest single act of trust a novel asks of its reader — and the trust is drawn from a pool that does not grow when the cast does.

The metaphor of a budget will be used throughout, and — since §16 will argue against it — it is worth being exact about what it does and does not claim.

• It does claim that reader identification is a resource with an upper bound in any given book, and that the bound is set principally by length, by the reader’s working memory for persons, and by the number of times a reader can be asked to reorient before reorientation becomes the dominant experience of reading.
• It does claim that the distribution of that resource is a formal feature of the book, as analysable as its chapter lengths or its handling of time, and that it can be audited (§5.5, §7.1 test 5).
• It does not claim that fewer viewpoints are better. As I Lay Dying has fifteen narrators in a very short novel and is not diluted. The claim is about the relation between number and available page count, not about number.
• It does not claim that identification is the only thing a viewpoint delivers. A viewpoint can be granted for irony, for information, for rhythm, or — §4.6 — to demonstrate that no vantage suffices. Several of those uses require the reader not to identify.
• It does not claim that the pool is fixed by nature. It is fixed by the book’s own construction, and a book that constructs its cast well can enlarge what it can carry. This is the qualification §3.6 develops and §16 presses hardest.

The corresponding claim about the people who are not granted viewpoint is less obvious and, following Alex Woloch, more consequential:

The characters who receive no viewpoint are not merely absent from the allocation. They are shaped by it. A character seen only from outside, in a book whose other characters are seen from inside, is not neutral in the reader’s model; they are compressed, and this character compression produces a specific and legible distortion — the type, the eccentric, the villain, the saint.

That is the argument of §3.3, and it is what raises this chapter above a discussion of technique. The allocation of viewpoint is the allocation of interiority, and the allocation of interiority is, in a novel, very nearly the allocation of personhood.

1.3 Where this chapter sits in the book

Three chapters divide a single problem between them (Chapters 5 and 6 in Part II, The Joins, and this chapter in Part III), and each is incomplete without the others:

The division is not merely administrative. It corresponds to three separable failures, with three separable remedies. A reader who cannot tell whose head they have entered has suffered an anchoring failure, and the fix is a proper noun in the first clause (Chapter 5 §4.5). A reader who resents being pulled away has suffered a placement failure, and the fix is to move the break (Chapter 6). A reader who does not care whose head they have entered has suffered an allocation failure, and no amount of anchoring or moving the break will fix it; the character should probably not have viewpoint at all.

Four further dependencies are load-bearing here.

Chapter 1 (The Purposes of a Scene) supplies the domains model — the practice of asking what a unit of narrative is for, and requiring more than one answer. §4 is that model transposed from the scene to the viewpoint grant.

Chapter 14 (The Subplot) supplies the attention tax: the cost, in reader patience, of any material that pulls focus from the main line. The empathy budget is the attention tax narrowed to its most expensive form. A reader will tolerate a subplot that occasionally takes the page; being asked to become six different people is a larger request, made of a scarcer faculty. The two chapters share a single theoretical source in Woloch, and the book treats them as one argument applied at two scales.

Chapter 9 (The Opening Pages) supplies, at its §4.6, the formal contract: the requirement that a book declare its structural rules — including whether narration is single or plural — before the reader has invested in a single thread. That chapter states the requirement and gives the reason. This chapter owns the requirement in depth: §6.5 anatomises how a viewpoint change is signalled, §6.6 formalises the cost curve that makes early declaration cheap and late declaration expensive, and §7.1 test 6 converts it into a diagnostic. Chapter 9 says declare early. This chapter says what declaring consists of, what it costs not to, and — at §8.4 and §9.5 — when a book should decline the advice.

Chapter 10 (The Prologue) supplies the bounded grant in its most familiar form: one unit given to a consciousness that will not hold the reader again. §6.3, architecture 6, generalises it.