7. Diagnostic Method — the ten tests

7.0 How to use the protocol

The tests below are for revision, not composition. Applied during drafting they produce the beat-shaped inertness described at §4.0; applied to a completed sequence they are the fastest triage instrument in this chapter.

Four procedural notes.

• Run them in order, and on a pair, not a scene. Tests 1–5 are structural, and their failures invalidate the rest: there is no point auditing a dilemma in a sequence whose disaster has not occurred.
• Record the answer, not the verdict. The output is a sentence describing what the text does. “The goal is find out why the locks were changed, and it descends from the previous decision” is useful. “Passes test 2” is not.
• A failed test is a question. Every test in this protocol is failed, deliberately and successfully, by at least one novel discussed in this chapter. The test’s function is to make a choice visible so that it can be confirmed as a choice.
• Test the events, not the connective tissue. Per §3.7, causality can be asserted in transitions rather than dramatised in beats. Strip the connectives — and so, because of this, which meant that — before running tests 5 and 6.

7.1 The ten tests

Test 1 — The goal test

State, in one sentence, what the point-of-view character wants at the start of this unit.

If the sentence will not come, the unit has no engine, whatever else is happening in it. The sentence must contain a verb of pursuit and a specific object: she wants Dov to say why the locks were changed, not she wants answers. If the only available sentence is book-sized, the scene goal is missing (§4.1).

Test 2 — The descent test

Does that goal descend from the decision that closed the previous sequel — and can I point to the sentence?

This is the goal test’s other half and catches a different fault. A goal may be perfectly stated and structurally free. Trace the line: which decision produced it? If the answer is “the plot required it”, the free goal is present (§8.1). Legitimate exceptions exist — a goal may descend from another character’s decision, or from an event in a parallel chain — but each should be nameable.

Test 3 — The disaster test

Does the unit end worse, or “yes, but”, rather than in a clean costless win?

State the terminal condition in one sentence containing the word but or the word worse. If neither will fit, ask the follow-up that matters: what is the next unit’s goal supposed to be built from? A costless success is diagnosed not by the satisfaction it gives but by the vacuum that follows it.

Test 4 — The cost test

What did the character pay, and in what currency?

Name the currency: time, money, safety, information, an ally, a self-image, a possibility. If no currency can be named, the disaster is decorative. This test catches the very common case in which something dramatic happens and nothing is actually spent — the shouting match after which both parties hold the positions they held before.

Test 5 — The chain test

Trace backwards three links. Is each caused by the one before it, or merely followed by it?

Write the three links as bare events, with the connectives removed. Then ask of each pair whether the second would have happened, in the circumstances of this story, without the first. This is Forster’s of grief test applied serially (§3.1), and the point at which the answer becomes “yes” is the point at which the chain has become a list. Two or three consecutive “yes” answers mark an episodic passage — which may be correct, and which should be a choice.

Test 6 — The “and so / but” test

Reverse-outline the sequence using only three connectives: and then, but, and therefore. Count the “and then”s.

E. M. Forster’s king-and-queen formulation (§3.1) is the classical statement of the principle; the working heuristic in its modern form — that beats should be joined by but or therefore and never by and then — circulates widely in screen and craft culture, most famously as a formulation of Trey Parker and Matt Stone’s, and it should be marked as popular craft lore rather than theory. It is nonetheless a good instrument, for a mechanical reason: it forces the writer to write the connective explicitly, which is the one thing prose lets you avoid.

Interpretation: a run of three or more “and then”s marks a stretch of the draft in which the chain has failed. A sequence in which every link is “therefore” and none is “but” is a different fault — a plot with no resistance in it, in which everything the character intends occurs.

Test 7 — The dilemma test

Are both options genuinely costly, and can I state each cost in one sentence?

Write the two costs before looking at the text. If one sentence is easy and the other is vague, the dilemma is false (§8.4). A useful supplementary question: would a reasonable, sympathetic reader disagree with the character’s choice? If nobody could disagree, no dilemma was weighed.

Test 8 — The criterion test

At which decision in this book would the protagonist’s earlier self have chosen differently — and what changed?

This is the falsifiable form of the Hinge (§4.7). It asks for a coordinate, not an interpretation. Locate the decision, then name the rule that changed. If no such decision exists anywhere in the manuscript, the book has a working plot and no arc, and the escalation treadmill (§8.9) is the likely reader complaint. If several exist in close succession, the arc is being spent too fast.

Test 9 — The reversibility test

Could the character abandon this decision between chapters at no cost, and would anyone notice?

If yes, it is a preference rather than a decision, and the sequel has not terminated. The strongest decisions foreclose something: a door locked, a letter torn, a call not returned for a fourth time. The test is not that characters must be trapped, but that a decision with no effect on the option set is structurally indistinguishable from an intention.

Test 10 — The beat-weight audit

Does the page-space given to each beat track its weight in the story, or is the manuscript applying a default?

Run the procedure at §5.5 over ten consecutive pairs and read the result against the interpretation table there. This is the slowest test in the protocol and the one that most reliably finds problems the writer cannot see, because the default register is invisible from inside.

7.2 Reading the results