# Experiment H2 — scoped problem stress (personal / bond / global)

| | |
|---|---|
| **variant_id** | `h2-scoped-problem-stress` |
| **base** | `ground-darvo-r0` (not stacked on H1) |
| **status** | experimental |
| **catalog** | [`../../catalog.yaml`](../../catalog.yaml) |
| **design note** | [`../../../history/260808-h2-scoped-problem-stress.md`](../../../history/260808-h2-scoped-problem-stress.md) |
| **workplan** | [`../../../workplans/GROUND-WP-0007-h2-scoped-problem-stress.md`](../../../workplans/GROUND-WP-0007-h2-scoped-problem-stress.md) |
| **control** | `h1-problem-stress` (reject-as-baseline) |

## Hypothesis

If unclaimed Problems apply End stress only to their **stress_scope**, with mostly personal, one bond at 3p+, and rare global (Surface), then some seats can climb toward DARVO **without** taxing the whole table into zero group wins — and **bond-scoped** cards give the Bond network a shared reason to SOLVE.

## Scopes

| stress_scope | End +1 Stress while unclaimed goes to… |
|--------------|----------------------------------------|
| **global** | every seat |
| **personal** | assigned **owner** only |
| **bond** | owner + every seat in the owner’s **Bond network** (Bond edges only). No Bonds → behaves as personal |

**SOLVE:** any seat with matching suit may claim any Problem (owner need not solve). Clearing a bond card is a network public good — that is intentional.

## Assignment in this package

By `hidden_priority` in `Problems.csv`:

| priority | scope | in play at |
|---------:|-------|------------|
| 0 Surface | global | all seats |
| 1 | personal | 2p+ |
| 2 | personal | 2p+ |
| 3 | **bond** | 3p+ only |
| 4 | personal | 5p+ only |

Owners: non-global in-play Problems, ascending priority, round-robin clockwise from **Lead**.

## Bond incentive (design intent)

Bond is still the regulation tool (−2 Stress, cancel DARVO). Bond-scope pressure **spreads** cost across the network so partners are incentivised to clear *that* card together rather than leave one seat to melt alone. Measure whether networks SOLVE bond cards faster or avoid Bonding.

## Not in H2

- H1 flat global +1 for “any open problem”
- H1-B ATTACK self-soothe
- Owner-only SOLVE, shadow goals, competence track

## Simulation

Select `variant_id: h2-scoped-problem-stress`. Implement `rules_delta.yaml` (H2-SCOPE, H2-OWN, H2-A, H2-SOLVE). Compare to `ground-darvo-r0` and optionally `h1-problem-stress`.