docs(linkedin-studio): S30 hardening log entry — discrete-% magnitude-scrub + S31 multiplier catalog

Appends the S30 row to the hardening journal (FIXED, 20 edits/7 files, commit 1bab1d6) and
catalogs the surfaced S31 multiplier-scrub class (~50 unsourced "Nx" instances across ~15 files,
with the SSOT-backed-vs-unsourced and coefficient-vs-frequency distinctions the next pass must make).
New entry only — prior entries untouched.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016qgzo6rxthw7KuxHjn5vyE

docs/hardening/log.md

@@ -2201,3 +2201,61 @@ an honest record. The only FORM A left anywhere is the one kept ironic quote
 magnitude-scrub.
 
 ---
+
+### S30 magnitude-scrub (discrete-% class) — unsourced reach/engagement penalties → SSOT
+
+**CLASS: magnitude-scrub / hardening** (NOT re-hardening — surgical SSOT-reconciliation of discrete
+percentage penalties/declines asserted as fact with no primary source in the SSOT
+`references/algorithm-signals-reference.md`; same discipline as S27/S28: read-and-show → grep-confirm →
+re-grep final). Re-scoped from S27 Bucket D + S28 Bucket D. Predicate: the SSOT encodes ordering + two
+officially-named signals + directional/sourced estimates — never hard coefficients; links "correlate
+with lower reach" (~38%, contested band), topic alignment is a real input with **no discrete off-topic
+figure**, AI content carries a **~45% correlational engagement gap (medium)**; no hashtag / post-length /
+posting-frequency / engagement-bait figure exists.
+
+**TEST (tool-grounded).** Plugin-wide grep for the discrete-% class (penalty/decline/underperformance/
+engagement-%). The plan's stored catalog had drifted and was incomplete — re-grep surfaced same-class
+siblings not listed (`linkedin-growth-playbook:225` 32% post-length, `:435` 25% twice/24h, `:828` 25%
+multiple/3h, `:158` a SECOND 47-50% decline; `engagement-coach:455` AI-comment table row) + one adjacent
+multiplier (`:221` 1.17x). All surfaced and operator-approved before edit. The first broad sweep silently
+failed on a zsh `--include=*.md` glob (must quote `"*.md"`); re-run quoted.
+
+**HARDEN (20 edits, 7 files).** Reconciliation vocabulary 1:1 with S27/S28.
+- `linkedin-formats.md` (5): :6 "47-50% decline" + :7 "15%→31% feed share" → directional; :176 "AI
+  comments reduce reach 30% / engagement 55%" → "~45% less engagement (correlational), no separate reach
+  figure"; :231/:279 "5+ hashtags = -68%" → "diminishing returns; no discrete figure".
+- `linkedin-growth-playbook` (6): :158 (47-50% twin of formats:6), :166 (-68% hashtags), :224/:225
+  (25%/32% post-length), :435/:828 (25% posting-frequency) → directional. **:221 1.17x multiplier folded
+  in per operator approval (adjacent); the rest of the multiplier class → S31.**
+- `glossary.md` (2): :91 bait -30-50% + :235 topic-gap -15-25% → "correlate with lower reach; no discrete
+  figure".
+- `engagement-coach.md` (2): :195 55% + :455 -30%/-55% → "~45% less engagement (correlational), actively
+  suppressed".
+- `post-feedback-monitor.md` (1): :330 -25%/post → "tends to split your own audience".
+- `ab-testing-framework.md` (1): :66 -68% → "no discrete figure".
+- `poll-strategy-guide.md` (2): :20 / :205 poll-overuse penalty → "declining effectiveness (directional)".
+
+**KEPT INTACT.** Operator-locked: engagement-pod + AI-slop "suppressed" framing (SSOT high-confidence,
+`engagement-coach:456`, glossary pod def). Different class / no magnitude: `firsthour:112` (no number),
+`poll:206` / `poll:3` (qualitative), `growth:567` (conversion rate, not reach), `formats:268` (list item).
+
+**VERIFY.**
+- Re-grep discrete-% penalty/decline class across the 7 files → **NONE**.
+- Leave-items intact (firsthour:112, poll:206/:3, growth:567, engagement-pod) → confirmed.
+- `git diff --stat` = **7 files, +20/−20**.
+- `bash scripts/test-runner.sh` → `Passed: 81 · Failed: 0 · Warnings: 0`, **exit 0**; counts **29/19/26/6**
+  + v0.5.0 unchanged.
+- Disposition: **FIXED** (20 edits, 7 files). Commit `55de831` (local; push held — `open/` is a public
+  namespace, STATE.md must not go there; all parked).
+
+**SURFACED → S31 (multiplier-scrub, new cataloged pass — operator: run everything, multiple sessions).**
+The unsourced **"Nx" reach/format multiplier** class — ~50 instances across ~15 files (`poll:7` 1.64x,
+`growth:175/205/234/247-250` 1.6x/1.4x/1.64x, `:493` 5.6x, `:636` 40x, `collaborations-guide` 2-24x,
+`report:249` 3.4x, `articles-strategy-guide:28-29` 2-3x/10x, `video-strategy-guide:308`, `ab-testing:49`
+1.17x-1.6x, …). S31 MUST distinguish (a) SSOT-backed engagement-order multipliers that STAY ("save ≈ 5x a
+like", "quality comment ≈ 2x a like" — SSOT medium) and (b) frequency counts that are NOT coefficients
+("3-5x/week", "10-15x daily"). Also catalog the descriptive-% sub-class: suppression-severity bands
+(`analyze:230`, `troubleshooting-guide:180` "50-70% drop") + engagement-rate benchmark (`growth:198` "11%
+engagement rate") + `report:215` "content format multipliers" reference.
+
+---