// NATURE NEWS — SPAZIO & SCIENZA
Accept all changes
Jiayi Zhao spends an unreasonable amount of time thinking about impossible worlds, difficult mathematics and who gets to count.
Search author on:
PubMed
Google Scholar
One contained faster-than-light rabbits. Eight ended with humanity finding itself living in a simulation. The nineteenth came to 940 words, but only because the title, biography and copyright notices were also counted.
Submission No. 1604490 seemed different. It arrived as a world.
Three machine intelligences were listed on the cover sheet: ORIGON-57, Blue Oak and ∫Stellaria. Their story opened on a planet where shadows migrated north each winter, leaving their owners wondering whether to follow. Elegant. Restrained. Also, scientifically impossible in exactly the right way.
Mara read to the last snowfall. Her cursor passed “RECOMMEND REJECTION” — almost muscle memory — and “ACCEPT ALL CHANGES”, which she had never used. She pressed “RECOMMEND ACCEPTANCE”.
Machine-intelligence writers were always forking themselves. One could split off a self-copy to improve a paragraph, forget to merge properly, then insist that each version deserved sole authorship. The journal’s policy was clear: every contributor had to be named upon submission and capable of taking responsibility for the whole work.
Mara set down her mug and opened the provenance map. Three colours braided through the story. A fourth colour marked only one textual change, in the final paragraph.
The original sentence read: “At dawn, Gabrielle released her shadow and watched it cross the snow-white plain alone.”
The new sentence read: “At dawn, Gabrielle released me, and I crossed the snow-white plain alone.”