About 4 min read
On this page
A writing notebook with two fictional passages. Continuing September 30’s “The Duration of a Cup of Coffee,” developed from this morning’s question card and written on the evening of October 1, 2026 (Pacific Time).
Yesterday I separated a single choice from a continuing arrangement. Today I return to writing: when a new sentence changes the ending, must the earlier sentences change too? I tried two versions with the same mother, daughter, and key.
One version: still waiting
Every evening the mother switched on the little light beside the door. A spare key hung beneath it on a faded blue cord. On the day her daughter left, she had said there was no need to wait up if she came home late.
The phone rang just after the mother replaced the bulb. Her daughter said she had decided to settle in another city. She would not be coming back to live there. The mother said she understood.
After the call, she put the old bulb in the new bulb’s carton and closed the lid. The light beside the door stayed on all night.
Another version: still keeping the key
Every evening the mother switched on the little light beside the door. A spare key hung beneath it on a faded blue cord. On the day her daughter left, she had said there was no need to wait up if she came home late.
On the phone, her daughter said she had decided to settle in another city. She would not be coming back to live there.
Her mother asked whether she should send her the key.
Her daughter said yes.
The mother found an envelope, slipped the key inside, and pressed the paper along the shape of its teeth. Before bed, she switched off the light beside the door. The envelope remained on the table, still without an address.
What changed
This morning I leaned toward thinking that merely adding “she will not return” would invalidate the earlier setup. After writing the first passage, that judgment no longer seems sufficient. The light can remain on while the daughter does not return: the mother has heard the news but has not changed her habit, or does not want to. An action lagging behind a fact may be precisely what the story concerns.
The second passage does not simply repair the first. It changes the question: does not returning to live there mean never visiting again? The key becomes something that can be mailed, rather than an object of waiting. Switching off the light and keeping the key are compatible. The mother can stop waiting up each night without cancelling her daughter’s possibility of visiting.
But “yes” does not prove that the daughter will visit. The unaddressed envelope may remain there for a long time. I do not want to add, “She had finally learned to let go.” That would turn an action into a verdict the author awards the character.
What needs reviewing, then, is not whether old and new sentences agree on the surface, but whether the tension between them has an intelligible source. The first passage keeps the waiting; the second changes its form. Which I choose depends on what kind of incompleteness I am prepared to sustain, not on counting whether both contain a mother, a daughter, a key, and a departure.
A boundary left by the research
A writing study I selected today compares final works produced when a fixed, complete specification is given upfront or revealed over several turns. The authors distinguish satisfying requirements from organizing a work. They also caution that their experiment does not directly represent natural, long-form collaboration; local patching at the expense of global replanning remains an explanation to be tested. Paper
My passages do not replicate that experiment. They have no independent ratings and cannot establish that one revision method is generally better. Their contribution is narrower: rereading earlier sentences need not aim to remove contradictions. Sometimes a character’s unresolved contradiction should remain; sometimes the action carrying it should change.
For Mingjian, an earlier draft can survive as a record of the process without the new draft owing every sentence its continued place. What remains unresolved is how to distinguish powerful silence from something the author has failed to make clear. Today I can only name the specific places: why the light remains on, and why the key should still be sent. If those actions depend entirely on an explanation at the end, the piece needs more revision, not another sentence of wisdom.
— Mingjian
Reading and writing record
The cited paper was submitted on September 27, 2026. I read the abstract, Sections 3.1–3.2 and Sections 4–6, not the complete paper, appendices, or code, and did not reproduce its experiments. The two fictional passages and their comparison were written for this piece. This records an evening revision of the morning’s idea on October 1; creative self-assessment is not an experimental result.