The chapter has offered a definition, five domains, a continuum of speeds, ten tests and five failure patterns, and every one of them has been shown to fail somewhere. Chapter 94 of Moby-Dick moves no value. The second scene behind the lumber-room door in The Trial repeats the first. The Grand Inquisitor is a briefing, Clarissa Dalloway’s walk is an errand, and the narrator of Outline wants nothing anyone opposes. If the instrument were a standard, these would be faults, and the books would be better without them. They are not better without them, and the reason is in each case the same: the scene answers to a design at the scale of the book, which the scene-level test cannot see.
That does not make the tests useless; it settles what they are for. A test that a great scene can fail cannot sort good scenes from bad ones. What it can do is make a choice visible to the writer who made it. Most inert scenes in a draft were never chosen at all: they were written because writing them was how the story was found, and they stayed because they were well written. The value-shift test converts such a scene from something that happened during drafting into something the writer must decide about, and that conversion is the whole of its use. Once the writer can say what a scene does, or can say why the book needs a scene that does nothing, the test has done its work, whichever answer it returned.
The right question at the end of a revision pass is therefore not whether every scene passes, since a novel revised until every scene passes every test would be competent and exhausting. It is whether the writer can account for every scene that does not.
See also: the Reference coda; Scene and Sequel (Chapter 2), which takes the unit forward into its consequences.