assembly-economics/models/GADM-Sectors-Scope.md

GADM Sectors and Goods — Scoping Note v0.1

Status: scoping only. Nothing here is implemented.

GADM up to 0.4 has a single production sector, one undifferentiated output Y*, one ecological headroom stock E, and selection quality σ as a fixed-weight mix of channels. GADM-0.1 § 26 plans four extensions. This note fixes their scope and order, and names the questions each is meant to answer, so they do not become structure for its own sake.

---

1. Rival versus non-rival goods (first)

Question. The program's core hypothesis rests on reuse of non-rival assemblies (commons X), while ecological limits bind rival, physical output. GADM currently mixes both in Y*. How much of the generative effect comes from the non-rival share?

Minimal structure.

• Split output into a rival part Y_r (physical throughput, draws on E) and a non-rival part Y_n (information, knowledge, software, organism services: near-zero marginal cost, feeds X).
• Information organisms K_Ω produce mainly Y_n. K_phys produces mainly Y_r.
• Demand and claims B apply to both. Only Y_r enters the ecological draw.

Tests. Throughput decoupling (GADM-0.1 § 14): can VFAC rise while physical throughput falls? Does the generative regime's ecological cost (the lowest E in GADM-0.2) shrink when its growth is mostly non-rival?

2. Ecological resource classes (second)

Question. Is the single E stock hiding trade-offs between renewable flows, non-renewable stocks and sinks?

Minimal structure. Three classes: renewable flows (regenerate), depletable stocks (no regeneration, recycling slows depletion) and sinks (pollution absorption capacity). Each has its own floor, following the planetary-boundaries logic: no aggregation across classes.

Tests. Which class binds first in each scenario? Does compute-intensive organism growth (e2) bind energy and sinks rather than materials?

3. Multiple sectors (third, only if 1–2 need it)

Question. Are there results that depend on sector composition, such as care versus production, or infrastructure versus consumer goods?

Minimal structure. At most four sectors: production (rival), information and organism services (non-rival), care (labour- and attention-intensive, links to cohort care need), infrastructure (feeds X and the infrastructure channel). Shares of K and of labour by sector.

Scope guard. Only add sectors if a concrete question from §§ 1–2 or from the cohort model needs them. A full input–output model is out of scope.

4. Institutional selection mechanisms (fourth)

Question. σ weights market, science, community, public and organism selection with fixed weights. Do outcomes depend on who sets those weights, and can institutions shift them over time?

Minimal structure. Make the selection weights a slow state variable, pushed by domination channels (for example, discovery control raises the organism/market weight) and by governance levers (public investment raises the public weight).

Tests. Does domination capture selection itself (a meta-governance failure, WS-11)? Does a plural selection mix make the generative regime more stable?

---

Order and dependencies

1. Rival vs. non-rival goods, because it tests the core hypothesis.
2. Ecological classes, because it builds on the Y_r / Y_n split.
3. Sectors, only if needed.
4. Endogenous selection weights, which is independent and can go in parallel with 2.

Every extension follows the practices established so far: a separate module so earlier versions stay reproducible; calibration to the previous baseline; a robustness check of the key readings; an entry in ParameterGrounding.md.