The debt model assumes that a reader leans forward into what happens next, and that the past interrupts the lean. For most plot-driven fiction this describes the reading well, and the tests built on it catch real faults. But it is a description of a kind of book, not of narrative, and the books it does not describe are not a margin. In The Remains of the Day the present is a drive west, a broken radiator and a walk up a hill, made thin so that the retrospect can carry the book. In Beloved the past is not visited but suffered, and the present at 124 is what interrupts it. In The Good Soldier and Absalom, Absalom! the reader’s question is not what happens next but what happened, and whether this teller can be trusted about it. In Proust the subject is the involuntary return itself. For these books forward motion is the interruption, and a writer who applied the tests to them would cut what they are for.
The practical consequence is a question to ask before any test is run: which timeline is this book about? If the answer is the present, the past must earn each departure, and the instruments in this chapter say how. If the answer is the past, the present has the burden instead. It must justify its claims on the reader’s attention as an occasion, a frame or a vantage, and a present that does not know it is a frame will drag in the way a bad flashback does. A great many uncertain drafts sit between the two answers, with a past the writer cares about more than the present that carries it, and for them the remedy is neither better flashbacks nor fewer, but a decision about which timeline is the story.
See also: the Reference coda.