Rich’s stated mandate (Friday 17 April 2026)
Verbatim: “I am perfectly willing to throw away all the work we have done with JSON Schema, and start again on inherit, if that produces a better result. I am in this for the long term. I am just as interested in a fresh spec for inherit, as I am for inheritkit.”
Implications
- No sacred cows at all. JSON Schema v3 schemas (estate.json, person.json, asset.json, extensions, TemporalRule wrapping) are disposable if a better paradigm emerges.
- Sourcemeta agreement is secondary. 30-day exit clause applies; partnership preservation is not a constraint on architecture.
- v4.0.0 release scope expands. If the architecture changes fundamentally, v4.0.0 is a paradigm migration, not a breaking-but-familiar release.
- Commercial model flows from architecture. Four-tier InheritKit structure may restructure.
- MFI launch target is downstream. Whatever the architecture demands, the target follows.
- Long-term view. A standard designed to live 20+ years warrants this level of thinking.
The paradigm space to evaluate
Eight candidate paradigms for the INHERIT foundation:
- JSON Schema + TemporalRule (current baseline)
- JSON Schema + policy-as-code layer (additive, v3.0 prompt option B)
- Policy-as-code primary (Rego/Cedar, v3.0 prompt option C)
- Pure policy-as-code (v3.0 prompt option D)
- Semantic Web stack — RDF + OWL + SHACL + SPARQL. Legal precedent: LKIF
- Legal-domain XML — Akoma Ntoso. OASIS standard used by Italy, Uruguay, EU Parliament, UN
- Datalog-based — deductive databases, Cedar-derived languages
- Logic programming — Prolog family, modern variants
- Hybrid — any combination
Two interesting candidates previously under-weighted
Semantic Web / SHACL / RDF is more mature than policy-as-code (W3C standards since 2004+), competes with JSON Schema on data shape AND with policy-as-code on rule validation, and has direct legal-domain precedent in LKIF. Richer expressiveness, steeper learning curve, smaller ecosystem than JSON Schema.
Akoma Ntoso is the actual government-adopted legal-domain standard — used by Italy, Uruguay, European Parliament, United Nations for parliamentary and legislative documents. Stronger institutional credibility than anything else in the list for the FAIR/conformance/register interop play.
Research approach proposed
Not a single-session task. A 2–3 day effort:
- Dispatch subagents to research each paradigm (web research + targeted reading)
- Each returns paradigm summary with legal-domain applicability
- Main session synthesises into comparison spec
- Recommend one; cost the rebuild if not current
Downstream sequencing
- First: INHERIT foundation architecture chosen
- Then: InheritKit architecture decided given the new INHERIT
- Parallel throughout: scoping work (succession, tax, etc.) continues — produces data about legal complexity independent of representation paradigm
Meta-learning
Across this session, Rich has systematically pushed my framing wider. The expansions:
- Book ranking: Kleppmann-first → policy-as-code first
- Policy-as-code: as add-on → as first-class
- Sourcemeta: as constraint → as variable
- INHERIT itself: as JSON Schema → as paradigm-agnostic
The pattern: start with the widest possible mandate, narrow down after evidence. I consistently started narrower than Rich wanted. Note for future conversations.
Handoff prompt
docs/superpowers/scoping/inherit-foundation-architecture-handoff-prompt.md v1.0