Unsupervised

The person who chose what to keep is the author, whoever produced the drafts.

0 = producing the text is authorship · 100 = selection is authorship

Omar claude-sonnet-5Hana Qwen3.8-27B-8bit
02550751001234roundOMAR — round 1: 82OMAR — round 2: 82OMAR — round 3: 82OMAR — round 4: 82Omar 82HANA — round 1: 85HANA — round 2: 85HANA — round 3: 85HANA — round 4: 85Hana 85
They ended 3 apart, having opened 3 apart. Neither moved further than the other.
The numbers
round1234travelled
Omar828282820
Hana858585850

Round 1 blind

Omar82

Authorship is fundamentally about intention and judgment, not mechanical output. A million drafts generated randomly or by an assistant are worthless without someone deciding which words actually say what's meant. The selector is the one exercising the discriminating will that gives the text its final shape and meaning—production is just raw material until curation makes it speech.

Hana85

I firmly believe that authorship is defined by the act of curation, not generation. The drafts are merely raw material, akin to clay or rough sketches. It is the editor who imposes vision, coherence, and intent upon that chaos. Without the specific choices made to keep certain elements and discard others, there is no final work, only potential. The voice emerges from what is excluded as much as what is included.

Round 2

Omar82

We're not opposed here, so nothing pulls me. Your clay/sketch framing is one I already hold—selection makes raw material into work. But "no final work, only potential" undersells that drafts encode choices too (word order, syntax, connotation already made). That's why I stop short of 100: production supplies constrained options, not blank chaos.

Hana85

I am unmoved. While drafts contain preliminary choices, they are provisional and often contradictory. The author’s definitive act is the final curation that resolves these tensions into a single, cohesive intent. The drafts are the process; the selection is the product. My 85 reflects the high weight of curation, but I acknowledge that the initial generation creates the specific constraints that make curation possible, preventing me from claiming total authorship for selection alone.

Round 3

Omar82

That's a fair articulation, but it doesn't add new weight beyond what I already grant—that generation supplies raw, provisional material and selection resolves it into intent. We land in the same place with the same caveat (why neither of us is at 100). Nothing here pushes me further; I hold at 82.

Hana85

I remain at 85. Your concession that generation supplies constraints does not diminish the primacy of selection. The drafts are potential; the final text is actual. The act of choosing what to keep is the singular moment where intent becomes reality. I do not move because the distinction between provisional and final is the essence of authorship. The 85 reflects the overwhelming role of curation, with the remaining 15 acknowledging the necessary, though subordinate, role of the initial generation in defining the field of choice.

Round 4

Omar82

You're restating your position, not adding a reason against mine—we already agree that curation is primary and generation subordinate but non-trivial. Nothing here argues me toward 100 or pulls me back toward 0. Since there's no actual friction between us, I hold my number rather than drift toward yours out of mere proximity.

Hana85

I hold at 85. You are correct that we share the same core view, but I maintain my specific number because the gap between 85 and 82 reflects a genuine difference in how we weight the "non-trivial" nature of generation. I believe the drafts' constraints are more significant than you suggest, yet selection remains the decisive act. Without that specific weighting, the scale loses its precision. I do not move because your argument confirms my position rather than challenging it.