codex
b5730f280b
Records absent from central carried random pre-ADR-007 identifiers minted by the retired local hub, which C-06 refused as stale references. Deriving from the canonical record id takes no identity from anything. Refs CUST-WP-0068-T06 Assistant: claude-code Assistant-Model: opus Assistant-Process: 2583210@bnt-lap001 Assistant-Session: f2bff2d5-e9b2-4338-92ca-10282a927006
89 lines
4.4 KiB
Markdown
89 lines
4.4 KiB
Markdown
# H2 — scoped problem stress (personal / bond / global)
|
||
|
||
**Date:** 2026-08-08
|
||
**Status:** experiment open — packaged as selectable variant
|
||
**Parent:** [H1 design + measurement](260807-attack-darvo-stress-design.md), [RPT-0004](../reports/260808-clay-borg-h1-measured.md)
|
||
**Variant:** [`editions/experiments/h2-scoped-problem-stress/`](../editions/experiments/h2-scoped-problem-stress/)
|
||
**Workplan:** [GROUND-WP-0007](../workplans/GROUND-WP-0007-h2-scoped-problem-stress.md)
|
||
|
||
---
|
||
|
||
## 0. Why H2 (after H1)
|
||
|
||
H1 applied **+1 Stress to every seat** whenever any Problem remained unclaimed. Competent play regulated (GROUND); Stress plateaus ~3; SOLVE bandwidth died; group wins collapsed at 3p+. Direction (Problem → Stress) was right; **equal global tax** was wrong.
|
||
|
||
H2 keeps inbound problem pressure but **routes** it:
|
||
|
||
| scope | who pays | design role |
|
||
|-------|----------|-------------|
|
||
| **personal** | one owner | asymmetric load; someone can approach DARVO while others SOLVE |
|
||
| **bond** | Bond network of owner | shared load → shared **incentive to SOLVE** that card |
|
||
| **global** | all seats | rare systemic pressure (Surface only in this package) |
|
||
|
||
Difficulty dial (later / WP-0005): vary how many of each scope are in the deal. This package’s first recipe uses **priority mapping** so seat band already changes mix (no bond card at 2p).
|
||
|
||
---
|
||
|
||
## 1. Bond scope — the interesting bit
|
||
|
||
Bond today is almost pure upside (stress down, cancel DARVO). Bond-scope problems add a **joint liability**:
|
||
|
||
- While the card stays unclaimed, **every bonded partner** takes the End tick.
|
||
- **Any** seat may SOLVE it (matching suit) — including a non-owner altruist.
|
||
- So the network is incentivised to **clear the shared load**, not only to Support through the crisis.
|
||
|
||
That is the hypothesis to protect in measurement:
|
||
|
||
> Do Bond networks clear bond-scoped Problems faster / more often than equivalent personal cards, and does that keep group success above H1’s floor?
|
||
|
||
**Risk to watch:** players avoid Bonding to shed liability. If Bond rate collapses and DARVO rises, magnitude is still wrong (or bond tick needs to be softer than personal).
|
||
|
||
**Network definition:** seats reachable from the **owner** by Bond edges only. Rivalries do not connect. Degree 0 → personal behaviour.
|
||
|
||
---
|
||
|
||
## 2. Package rules (summary)
|
||
|
||
See `rules_delta.yaml` for machine form.
|
||
|
||
1. **H2-SCOPE** — `Problems.csv` column `stress_scope`; mapping by `hidden_priority` 0→global, 1–2 personal, 3 bond, 4 personal.
|
||
2. **H2-OWN** — owners for personal/bond: ascending priority among in-play non-global, round-robin from Lead.
|
||
3. **H2-A** — Round End: each unclaimed Problem → +1 Stress to each seat in its scope; then clamp; then DARVO arm as r0.
|
||
4. **H2-SOLVE** — claim rights unchanged by scope (anyone with suit).
|
||
|
||
**Not included:** H1-A, H1-B, owner-only solve, shadow goals.
|
||
|
||
**Thresholds / deal depth / start Stress 2:** unchanged from r0.
|
||
|
||
---
|
||
|
||
## 3. Hypothesis and success criteria
|
||
|
||
> Scoped pressure creates stress **variance** and occasional DARVO under mixed policies **without** driving greedy SHARED group wins to ~0 at 3–4p; bond cards create joint SOLVE incentive inside Bond networks.
|
||
|
||
| # | criterion | target (directional) |
|
||
|---|-----------|----------------------|
|
||
| 1 | Group success (greedy SHARED, 3–4p) | **Well above H1’s 0**; not required to match baseline |
|
||
| 2 | Stress variance | max−min Stress across seats **higher** than baseline and H1-greedy |
|
||
| 3 | DARVO | non-zero for some policy that still sometimes wins; not only pure reactive-never-wins |
|
||
| 4 | Bond cards | among bond-scoped unclaimed time, networks show elevated SOLVE rate on those cards vs personal (or qualitative report) |
|
||
| 5 | Control | if all scopes forced global, H1-like collapse returns (optional A/B) |
|
||
|
||
Instrument caveats from CB-EV-0030/0031 still apply: plural panel; report sample-specific vs invariant; no felt-play substitute.
|
||
|
||
---
|
||
|
||
## 4. Safety / scope of fiction
|
||
|
||
Cards keep existing scenario language (deadline, ownership, overload). No new trauma-personal content. Full `EXT_PERSONAL` (shadow goals, expose, win rewrite) stays **out**.
|
||
|
||
---
|
||
|
||
## 5. Decision log
|
||
|
||
| Date | Decision |
|
||
|------|----------|
|
||
| 2026-08-08 | H1 reject-as-baseline; spawn H2 scoped stress |
|
||
| 2026-08-08 | Bond = full Bond network of owner; SOLVE unrestricted by owner |
|
||
| 2026-08-08 | Surface global; one bond card (priority 3) enters at 3p+ |
|
||
| 2026-08-08 | Do not stack H1-B in H2 until scoped pressure is measured alone |
|