JesseMarkowitzandClaude Opus 5
e5617b86ba
The pipeline's embed-and-cluster step is dead, and this commit holds both the evidence for that and the step proposed to replace it. Predicaments. Scenes are re-described as "what the person is up against", with no names, jobs or places, then embedded and clustered (redescribe.py, topic_words.py, topic_share.py). The pilot chose qwen3:14b over 3b by reading both side by side. Two defects the pilot exposed are fixed: split.py missed titles in quotes and a contents subtitle after a dash, so three stories had been merged into their neighbours, and strip_names.py read New York place names as people. The corrected corpus is probe/v2 (97 stories, 839 scenes); carry_summaries.py reuses the 829 unchanged v1 summaries. Topic share fell from 20% to 13% at k=60, short of the pre-registered 10%. Hand references. Three corpora were read scene by scene and written up by hand, under the same prompt rules the local models get, as a baseline to judge them against: O. Henry (probe/v2/claude, 839 scenes, 20 situations), Wharton's Descent of Man (probe/wharton, 262 scenes, 16 groups) and Jacobs's The Lady of the Barge (probe/jacobs, 157 scenes, 19 groups). Each has its own README and a readable page. No inference was used for any of them. Catalogue. probe/catalogue maps every hand group in the three references onto 36 situation entries, with an answer key per corpus and one recurrence rule applied to all three. classify.py assigns a scene one entry or none, leave-one-corpus- out; score.py checks it against the key, with a self-test on random labels. Why clustering is out: hand-written predicaments, embedded and clustered exactly as the model's were, agree with the hand grouping at ARI 0.05 — no better than the 14B text's 0.07. Better rewriting cannot rescue it. Embeddings cannot even shortlist: the hand label is the nearest entry 13% of the time and in the top 8 half the time. The classification runs are not here. The dev and test runs are pre-registered in probe/catalogue/README.md with the bar set beforehand, and are blocked on the inference host, whose GPU has fallen off the PCIe bus three times. The 30-scene partial output in out/ is not a result. Review page. The situation review is now a browser page rather than JSON edited by hand (review_page.py, review_page_logic.cjs with Node tests, format schema v2). It has never been rendered in a real browser. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014BygvsUXV9eU6oHkTCkKZ1
30 lines
2.9 KiB
JSON
30 lines
2.9 KiB
JSON
{
|
|
"about": "Reference setting for a pack built from Edith Wharton, The Descent of Man and Other Stories (1904). Written by hand (Claude) as a comparison target for the story-to-pack tool. Labels only: the stat spine is the fixed one in DESIGN.md, and no number is chosen here.",
|
|
"player_role_suggestion": "You are new to money and to the people who have it, in a city where everything is forgiven except being talked about. (A suggestion only: DESIGN.md makes the one-line player role the designer's to write.)",
|
|
"stat_labels": {
|
|
"money": "Money",
|
|
"owed": "Obligations (what you have accepted and will be asked to repay)",
|
|
"standing": "Standing (what people say about you)",
|
|
"skills.tact": "Tact: smoothing things over, saying the acceptable thing",
|
|
"skills.candour": "Candour: saying the true thing and bearing the cost",
|
|
"relationship.confides_in_you": "Confides in you",
|
|
"relationship.obliged_to_you": "Obliged to you",
|
|
"relationship.resents_you": "Resents you"
|
|
},
|
|
"why_these_labels": "Across the ten stories, the thing people spend is reputation, and the two ways of spending it are the two skills: smoothing over (the wife in 'The Other Two', the publisher in 'The Descent of Man', Mrs. Vervain as trained by Thursdale) and plain speaking at a cost (Hope Fenno, Julia Westall at the end, the mother in 'The Quicksand'). Almost nobody borrows money in these stories; what they incur instead is an obligation to someone, so the debt slot is labelled that way.",
|
|
"roles": [
|
|
{"token":"{spouse}","role":"The spouse","from_scenes":"stories 0, 1, 2, 3, 4, 5"},
|
|
{"token":"{former}","role":"The one from before: a former spouse or old intimate who still has a claim","from_scenes":"stories 1, 5, 7"},
|
|
{"token":"{rival}","role":"The rival for the person you want","from_scenes":"stories 1, 5, 6"},
|
|
{"token":"{patron}","role":"The elder with power over you: a relation, a publisher, a senior partner","from_scenes":"stories 0, 1, 2"},
|
|
{"token":"{ward}","role":"The young person in your charge, or your child","from_scenes":"stories 4, 8"},
|
|
{"token":"{employer}","role":"The person you work for, in their private trouble","from_scenes":"stories 1, 3, 6"},
|
|
{"token":"{confidant}","role":"The candid friend or cousin who tells you what you would rather not hear","from_scenes":"stories 2, 7, 8"}
|
|
],
|
|
"stages": [
|
|
{"id":"outside","title":"On sufferance","standing_rule":"The protagonist is dependent, new, in service or hard up: they need something from the people around them.","example_scenes":[88, 89, 90, 91, 174, 241, 242]},
|
|
{"id":"inside","title":"Established","standing_rule":"The protagonist is married, known and comfortable: their trouble is what their position costs to keep.","example_scenes":[12, 29, 58, 145, 212]}
|
|
],
|
|
"stage_assignment_note": "Assigned by the protagonist's standing in each group's member scenes, not by the story. Groups whose members fall on both sides are marked 'any'."
|
|
}
|