polymath-society 0.2.4

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (45) hide show
  1. package/LICENSE +52 -0
  2. package/README.md +311 -0
  3. package/dist/DATA-MAP.md +109 -0
  4. package/dist/cli.js +24136 -0
  5. package/dist/engine/closing-rubric.md +50 -0
  6. package/dist/engine/exp-allfacets.js +16832 -0
  7. package/dist/engine/exp-person.js +16922 -0
  8. package/dist/engine/exp-pipeline.js +16700 -0
  9. package/dist/engine/feel-seen-rubric.md +192 -0
  10. package/dist/engine/ingest-export.js +1423 -0
  11. package/dist/engine/peak-demos.js +16752 -0
  12. package/dist/engine/person-dimension-summary.js +16744 -0
  13. package/dist/engine/person-facet-lines.js +16784 -0
  14. package/dist/engine/person-headline.js +16733 -0
  15. package/dist/engine/person-report.js +16845 -0
  16. package/dist/engine/person-self-image.js +16689 -0
  17. package/dist/engine/public-report-guidelines.md +80 -0
  18. package/dist/engine/public-report.js +17284 -0
  19. package/dist/engine/run-analysis.js +16035 -0
  20. package/dist/engine/run-tagger.js +16092 -0
  21. package/dist/engine/top-companies.md +41 -0
  22. package/dist/engine/writing-well.md +56 -0
  23. package/dist/index.js +22021 -0
  24. package/dist/pipeline/TECHNIQUES.md +406 -0
  25. package/dist/pipeline/WRITING.md +48 -0
  26. package/dist/pipeline/coding-agglomerate.js +15876 -0
  27. package/dist/pipeline/coding-aggregate.js +440 -0
  28. package/dist/pipeline/coding-build.js +1168 -0
  29. package/dist/pipeline/coding-coaching.js +16893 -0
  30. package/dist/pipeline/coding-day-digest.js +15869 -0
  31. package/dist/pipeline/coding-delegation.js +16292 -0
  32. package/dist/pipeline/coding-expertise.js +15925 -0
  33. package/dist/pipeline/coding-flow.js +313 -0
  34. package/dist/pipeline/coding-frontier-detail.js +15878 -0
  35. package/dist/pipeline/coding-frontier.js +51 -0
  36. package/dist/pipeline/coding-gap-dist.js +293 -0
  37. package/dist/pipeline/coding-gap.js +17108 -0
  38. package/dist/pipeline/coding-grade.js +16195 -0
  39. package/dist/pipeline/coding-nutshell.js +15725 -0
  40. package/dist/pipeline/coding-projects.js +104 -0
  41. package/dist/pipeline/coding-walkthrough.js +15947 -0
  42. package/dist/web/app.js +12024 -0
  43. package/dist/web/index.html +1 -0
  44. package/dist/web/styles.css +1 -0
  45. package/package.json +61 -0
@@ -0,0 +1,80 @@
1
+ # Public report — phrasing guidelines
2
+
3
+ How to phrase the outward-facing public report (the artifact a person shows to
4
+ others). Companion to `feel-seen-rubric.md` (that one is about being SEEN; this
5
+ one is about getting the framing right so nothing reads worse than the truth).
6
+
7
+ ---
8
+
9
+ ## Failure modes — frame by ABSOLUTE rarity, NEVER a ratio
10
+
11
+ This is the rule that matters most, because a weakness phrased wrong does real
12
+ damage to how the person is seen.
13
+
14
+ 1. **State how often it actually arose, as an absolute count across his WHOLE
15
+ history — never a ratio.** "Twice, ever" — NOT "2 of 9 conversations." A ratio
16
+ reads as a *rate*: "2 of 9" sounds like it happens ~1 in 5 times (frequent). The
17
+ truth is it happened **two times, total**. That difference is everything — it is
18
+ what stops a rare slip from reading as a standing character flaw.
19
+ - GOOD: "twice in his entire history", "a handful of times", "three moments
20
+ total", "only once".
21
+ - BANNED: "2 of 9", "in 30% of cases", "2 out of every N", any fraction or rate.
22
+ - The grader's denominators (e.g. "seen in 2 of 9 graded conversations") are an
23
+ internal artifact — the 9 is just how many were scored on that facet, not a
24
+ meaningful base rate. NEVER surface the denominator. Use the numerator as an
25
+ absolute count.
26
+
27
+ 2. **Every failure mode carries a `frequency`** that does the reframing work, e.g.
28
+ "twice, ever — both under direct attack on his venture" · "a handful of times,
29
+ only on relationship calls" · "three moments, all competitive" · "his one
30
+ genuinely recurring one — but only on low-stakes things".
31
+
32
+ 3. **Bound it to its context.** A weakness that only fires in one narrow situation
33
+ is not a general trait — say where it fires: "only under direct attack on his
34
+ venture", "only when his conviction is low", "only on low-stakes sends". The
35
+ bound is as important as the count.
36
+
37
+ 4. **Don't let the body imply a habit.** For something that happened a few times,
38
+ write "On the two times it surfaced, he reframed the critic first" — NOT "He
39
+ reflexively reframes critics." Avoid frequency-implying words (recurs, reflex,
40
+ always, tends to, habitually) unless it genuinely recurs.
41
+
42
+ 5. **Be honest both ways.** If one truly recurs, say "recurring" — but still bound
43
+ it. Don't inflate a one-off into a pattern, and don't bury a real recurring one.
44
+
45
+ 6. **Ground every frequency in the real occurrence count** the grader recorded
46
+ (absolute instances across the corpus). Never invent a number.
47
+
48
+ ---
49
+
50
+ ## Drives — the person's own account leads; the behavioral read never does
51
+
52
+ A person's stated drive carries FIRST-PERSON AUTHORITY. In the public report:
53
+
54
+ - **The PRIMARY driver is what the person themselves says drives them**, taken from
55
+ their own words (the self-image read's "what they say drives them" bucket),
56
+ phrased in the report's voice and ideally carrying one short verbatim quote.
57
+ - **Revealed-preference / behavioral drives may appear only as SECONDARY mentions**
58
+ — "he'd tell you X; the record also shows Y" — clearly subordinate, second or
59
+ third, never as the primary, never overriding their account.
60
+ - Never present an analyst's behavioral verdict ("what actually drives him is …")
61
+ as the person's drive, and never list an emotional need (to be witnessed, to be
62
+ loved) as a drive or want.
63
+
64
+ ---
65
+
66
+ ## The rest of the public report (brief — see the generator + code for detail)
67
+
68
+ - **Percentiles are deterministic, never written by the model.** Each spike/trait
69
+ carries a `facet` tag (a key from the score catalog); the percentile is computed
70
+ from that facet's real graded score. A number with no backing facet is dropped.
71
+ - **Trait labels = the facet's real name in the detailed report** ("agency:initiative"
72
+ → "Bias to action"), not editorial inventions. Per-spike-deduped.
73
+ - **Bodies are pithy** (one sentence, the claim). **Examples are the deep layer**:
74
+ 3–6 ordered bullets — what happened, in order, with real verbatim quotes and the
75
+ different instances — drawn from the graded demonstrations.
76
+ - **Third person throughout** ("he / his"), for others to read. Convert the
77
+ second-person demonstration steps. Leave his verbatim quotes exactly as said.
78
+ - **Obfuscate private individuals' names** (use a role: "a close friend"). Keep
79
+ specific verifiable facts about him. Public figures named for stature are fine.
80
+ - **No slop / no jargon / no role-label praise.** Plain verbs, concrete specifics.