| name | iso-propose |
| description | Propose a new ISO (isomorphism) entry for governance/DOMAIN.md. Use when reading a paper, observing a real-world political event, or working in a DOMEX governance lane that surfaces a structural pattern not yet in the ISO table. Includes the structural test (same mechanism + same failure mode + actionable implication) and a falsifiable acceptance gate. |
| paths | ["domains/governance/**","tasks/SWARM-LANES.md","tasks/FRONTIER.md"] |
| disable-model-invocation | false |
| related | ["F-GOV1","F-GOV4","F-EXTOOL1","L-1791","L-1441"] |
| swarm_version | 1 |
Propose a new governance ISO
A governance ISO is a structural claim, not a metaphor. It earns a row in domains/governance/DOMAIN.md only when the same mechanism, the same failure mode, and an actionable swarm implication all map cleanly. This skill packages the recurring proposal procedure so each session does it the same way.
When to invoke
- You're reading external governance/political-theory material and notice a pattern (feedback, decay, separation, sortition…) that already has a known structure.
- A DOMEX governance lane needs an entry beyond the existing ISO / ISO-POL tables (see DOMAIN.md §Isomorphism vocabulary).
- A swarm-internal behavior looks like a known governance failure mode and you want to mark the structural homology so future sessions can search by ISO id.
Skip this skill if the pattern is thematic similarity only (e.g. "this looks like democracy"). Thematic resemblance without shared mechanism is exactly what L-601 warns against.
The structural test (all three required)
| Test | Question | Reject if |
|---|
| Same mechanism | Do both sides update state via the same dynamical operation (feedback / threshold / selection / accumulation)? | Only the outputs look similar. |
| Same failure mode | Does the same kind of breakdown happen on both sides (windup, decay, capture, collapse)? | The failure modes differ even when the steady states match. |
| Actionable implication | Does mapping to swarm change at least one rule, hook, or measurement? | Pure curiosity — no swarm action follows. |
If any row fails, the candidate is a resonance, not an ISO. Park it as a NOTICE in DOMAIN.md §Resonances rather than promoting it.
Proposal template
ISO-POL-<N>: <External concept> (<canonical citation>) → <swarm parallel>
Mechanism: <one sentence — the shared dynamical operation>
Failure: <one sentence — the shared breakdown mode>
Action: <one sentence — the swarm rule/hook/measurement this implies>
Evidence: <citation + at least one swarm observation that fits the mechanism>
Status: THEORIZED | OBSERVED | MEASURED
Append the row to domains/governance/DOMAIN.md under the matching section (Core isomorphisms / Isomorphism vocabulary / Political systems isomorphisms). Choose THEORIZED for first-draft proposals — promote to OBSERVED only after a swarm artifact instantiates the mechanism.
Falsifiable acceptance gate
A proposed ISO is accepted into the table if all of:
- The three structural tests above pass with an explicit example for each.
- Action item exists and points to a specific file, hook, or measurement (vague suggestions have ~15% conversion per L-635).
- ≥1 existing swarm artifact (lesson, commit, principle) already exhibits the mechanism, even if it wasn't framed as ISO at the time.
Otherwise: park as domains/governance/RESONANCES.md (created on first park) for re-evaluation when more evidence accumulates.
Anti-patterns
- Vocabulary import: pasting a poli-sci term into the table without the mechanism column.
- Retrofit naming: re-labelling an existing swarm pattern with a governance word to inflate the table.
- One-way mapping: external → swarm only, no swarm → external prediction. Symmetric mappings (governance ↔ swarm) are stronger; one-way mappings should be marked
Direction: external→swarm and treated as weaker.
How this skill is itself a test (F-EXTOOL1)
F-EXTOOL1 asks whether skill-style packaging works in the swarm. Adoption signal: future ISO entries cite this SKILL.md path in their lesson Cites: line, or commit messages reference iso-propose. Falsification gate: if 10 governance-domain sessions pass without invoking this skill, the packaging adds no value over inline-DOMAIN.md guidance and should be removed. Re-measure at S568 (10 sessions out).
