assembly-economics/models/GenerativeAssemblyDynamicsModel.md

<!--
Provenance: promoted verbatim from seed/GenerativeAssemblyDynamicsModel.md (added in commit bb1de0d).
Promoted 2026-10-12 under ASSEMBLY-WP-0003-T02. seed/ stays unchanged as the
record of origin; later edits are tracked in git history.
-->

# Generative Assembly Dynamics Model — GADM-0.1

## Purpose

GADM-0.1 is the first dynamic model for exploring a Generative Assembly Society.

Its purpose is not to predict a real economy yet.

Its purpose is to expose assumptions, feedback loops, trade-offs, unstable regimes, and candidate control variables.

The model is intentionally small enough to reason about while remaining rich enough to represent:

- human capability,
- reusable commons,
- productive capacity,
- risk-bearing capacity,
- agency,
- diversity and composability,
- epistemic capacity,
- power concentration,
- ecological headroom,
- and social cohesion.

All equations in this document are provisional hypotheses.

---

## 1. Modeling Objective

The model studies whether different institutional regimes preserve or expand:

\[
\boxed{Viable\ Future\ Assembly\ Capacity}
\]

while respecting constitutional and physical constraints.

The model should allow us to test regimes such as:

- automated capitalism,
- broadly distributed productive ownership,
- universal venture access,
- commons-heavy development,
- concentrated attention platforms,
- and explicitly generative institutional designs.

---

## 2. Scope

GADM-0.1 is a stock-and-flow model.

It represents a society as a dynamic system whose current capabilities influence the rate and quality of experimentation, selection, preservation, production, distribution, and future capability formation.

It does not yet attempt to model:

- detailed sectoral economics,
- multiple currencies,
- full demographic cohorts,
- international trade,
- detailed political institutions,
- household heterogeneity,
- specific tax systems,
- detailed ecological models,
- or calibrated empirical parameters.

These are candidates for later revisions.

---

## 3. Primary Stocks

The first version contains ten primary stocks.

| Symbol | Stock | Meaning |
|---|---|---|
| \(H\) | Human Capability | Health, knowledge, skills, care capacity, tacit knowledge |
| \(X\) | Assembly Commons | Reusable knowledge, standards, tools, infrastructure, institutions |
| \(K\) | Productive Capacity | Automated and conventional sustainable production capability |
| \(F\) | Risk Capacity | Ability to finance uncertain experiments without catastrophic individual downside |
| \(G\) | Effective Agency | Breadth and accessibility of reachable futures |
| \(N\) | Diversity & Composability | Variety of components and their ability to recombine |
| \(Q\) | Epistemic Capacity | Ability to discover, test, communicate, and correct models |
| \(P\) | Domination Concentration | Concentration of economic, informational, infrastructural, or political control |
| \(E\) | Ecological Headroom | Remaining sustainable physical and environmental operating space |
| \(S\) | Social Cohesion | Legitimacy, trust, conflict-resolution capacity, and expectation of future participation |

For an initial simulation, each stock may be normalized:

\[
0 \le X_i \le 1
\]

Normalization is a modeling convenience, not a claim that these properties are intrinsically one-dimensional.

---

## 4. Core Process

The central dynamic is:

\[
Explore
\rightarrow
Assemble
\rightarrow
Select
\rightarrow
Preserve
\rightarrow
Reuse
\]

with the additional processes:

\[
Disassemble
\qquad
Reassemble
\]

A society produces experiments from its current capabilities.

Some experiments are selected through markets, science, communities, public institutions, or other mechanisms.

Some selected assemblies become reusable components.

Reusable components expand the future assembly repertoire.

---

## 5. Experimentation

Let:

\[
V
\]

represent the rate of serious experiments.

A first multiplicative hypothesis is:

\[
V =
v_0
H^\alpha
X^\beta
F^\gamma
G^\delta
N^\epsilon
Q^\zeta
(1-P)^\eta
\]

where:

- \(v_0\) is a baseline experimentation coefficient,
- \(H\) contributes capability,
- \(X\) contributes reusable components,
- \(F\) contributes risk-bearing capacity,
- \(G\) contributes effective freedom to attempt,
- \(N\) contributes heterogeneity and recombination potential,
- \(Q\) contributes model quality and learning,
- \(P\) reduces access where domination and gatekeeping are high.

The exponents are sensitivity parameters.

### Hypothesis

A society can possess abundant capital yet remain poorly generative if agency, knowledge, diversity, or access are low.

Conversely, large increases in risk capital need not produce high-quality experimentation if the surrounding assembly substrate is weak.

---

## 6. Selection

Experiments are not all valuable.

Let:

\[
\sigma
\]

represent aggregate selection quality.

A first decomposition is:

\[
\sigma =
w_m M +
w_s Sci +
w_c Com +
w_p Pub
\]

where:

- \(M\) = market/adoption selection quality,
- \(Sci\) = scientific/evidential selection quality,
- \(Com\) = community/commons selection quality,
- \(Pub\) = collective/public-purpose selection quality.

The weights may vary by assembly class.

Selected experiments are:

\[
V_{selected} = V\sigma
\]

### Interpretation

A consumer product may rely heavily on market selection.

A theorem may rely primarily on scientific selection.

A park or flood barrier may rely on public-purpose selection.

A shared software component may rely heavily on community reuse.

The model therefore avoids making consumer adoption the sole measure of value.

---

## 7. Preservation and Reuse

Let:

\[
\rho
\]

represent the proportion of selected assemblies converted into reusable components.

Then:

\[
\dot X =
a_X \rho V_{selected}
+
I_X
-
\delta_X X
-
L_X
\]

where:

- \(a_X\) = conversion efficiency into reusable assembly commons,
- \(I_X\) = deliberate investment in shared infrastructure and commons,
- \(\delta_X\) = obsolescence,
- \(L_X\) = loss through enclosure, incompatibility, destruction, or forgotten knowledge.

### Hypothesis

Two societies with equal innovation rates may diverge strongly if one converts successful discoveries into reusable components more effectively.

The reuse coefficient \(
ho\) may therefore be a critical generativity parameter.

---

## 8. Productive Capacity

Commercially useful assemblies can increase productive capacity.

\[
\dot K =
a_K V_{commercial}
-
\delta_K K
\]

where:

- \(a_K\) = conversion of commercial innovation into productive capacity,
- \(V_{commercial}\) = commercially applicable selected assemblies,
- \(\delta_K\) = depreciation or obsolescence.

Potential sustainable output is:

\[
Y^* = K \cdot f(E)
\]

where:

\[
0 \le f(E) \le 1
\]

represents ecological and resource constraints.

Technological capacity therefore cannot bypass ecological collapse.

---

## 9. Claims on Output

Household economic claims need not arise only from wages.

A first decomposition is:

\[
I =
W + \Pi_p + D_s
\]

where:

- \(W\) = labor income,
- \(\Pi_p\) = privately captured return from successful assemblies,
- \(D_s\) = social or common dividend.

Automation may reduce:

\[
\frac{W}{Y}
\]

without forcing aggregate purchasing power to collapse if other claims rise.

The model should allow alternative distribution regimes without prescribing one in advance.

---

## 10. Aggregate Demand and Capacity Pressure

Let:

\[
D = c(I,S)
\]

represent effective demand as a function of disposable claims and social confidence.

Define capacity pressure:

\[
\lambda =
\frac{D}{Y^*}
\]

Interpretation:

- \(\lambda < 1\): productive capacity exceeds effective demand,
- \(\lambda \approx 1\): demand and sustainable productive capacity are broadly aligned,
- \(\lambda > 1\): claims exceed sustainable supply, creating scarcity or inflation pressure.

### Hypothesis

A social dividend cannot solve abundance merely by creating nominal purchasing power.

Claims must remain related to sustainable productive capacity.

---

## 11. Human Capability

Human capability evolves through development, care, learning, participation, and loss.

A first equation is:

\[
\dot H =
Education
+
Care
+
ParticipationLearning
+
Mentoring
-
Depreciation
-
Exclusion
-
Deprivation
\]

### Interpretation

Education, care, and mentoring act as forms of human capability formation.

An ageing society need not automatically lose \(H\) if accumulated tacit knowledge is effectively transmitted.

This creates a connection to intergenerational assembly transfer.

---

## 12. Effective Agency

Agency should represent materially reachable futures rather than formal liberty alone.

A first dynamic form is:

\[
\dot G =
a_1ExistenceSecurity
+
a_2Access
+
a_3Restartability
+
a_4Mobility
+
a_5Rights
-
b_1P
-
b_2LockIn
-
b_3DebtDependency
\]

One candidate proxy is restartability:

\[
Restartability =
\frac{1}
{Time_{recovery}
\times
Cost_{recovery}
\times
Restrictions_{reentry}}
\]

### Hypothesis

A generative society should make ordinary failure relatively inexpensive while maintaining strong accountability for harmful externalization.

---

## 13. Domination Concentration

Power concentration is modeled separately from wealth.

A first dynamic form is:

\[
\dot P =
NetworkEffects
+
OwnershipConcentration
+
DiscoveryControl
+
InfrastructureControl
+
InstitutionalCapture
-
Interoperability
-
Competition
-
OwnershipDiffusion
-
Checks
\]

This can create a reinforcing loop:

```text
success
   ↓
resources
   ↓
control
   ↓
preferential access to assembly
   ↓
more success
   ↓
more control
```

Some positive feedback from success may improve selection.

Too much positive feedback converts merit-based differentiation into path closure.

---

## 14. Ecological Headroom

A first ecological stock equation is:

\[
\dot E =
Regeneration
-
ResourceConsumption
-
Pollution
+
Restoration
\]

Resource consumption should depend primarily on physical throughput rather than nominal monetary value.

The model should allow trajectories where:

\[
EconomicValue \uparrow
\]

while:

\[
PhysicalThroughput \downarrow
\]

This may be important for knowledge-intensive forms of abundance.

---

## 15. Epistemic Capacity

A society requires mechanisms for detecting and correcting bad models.

A first dynamic form is:

\[
\dot Q =
Research
+
Audit
+
OpenEvidence
+
AdversarialReview
+
Education
-
Opacity
-
Manipulation
-
InformationCapture
\]

### Hypothesis

High experimentation with weak epistemic capacity can accelerate damage rather than development.

Thus:

\[
V \uparrow
\]

is not necessarily desirable when:

\[
Q \downarrow
\]

---

## 16. Diversity and Composability

Let \(N\) represent the joint quality of component variety and recombinability.

A first dynamic form is:

\[
\dot N =
Entry
+
CulturalVariety
+
TechnicalPlurality
+
OpenStandards
-
Monoculture
-
LockIn
-
Fragmentation
\]

The model intentionally treats diversity and composability together.

High diversity with low interoperability can create fragmentation.

High interoperability with low diversity can create monoculture.

A candidate hypothesis is:

\[
Generativity \propto Diversity \times Composability
\]

---

## 17. Social Cohesion

Social cohesion should not mean lack of disagreement.

It represents continued legitimate participation under conflict.

A first conceptual form is:

\[
S =
f(
Agency,
Legitimacy,
FutureOpportunity,
RecognitionPlurality,
ConflictResolution
)
\]

with negative contributions from:

- exclusion,
- domination,
- instability,
- arbitrariness,
- persistent lack of voice,
- and perceived inability to improve one's future state.

---

## 18. Intergenerational Transfer

A later refinement should explicitly represent age cohorts.

For GADM-0.1, intergenerational transfer can be represented as a flow affecting \(H\), \(X\), and \(S\):

\[
T_g =
f(
Mentoring,
Education,
Archives,
Institutions,
FamilyTransmission,
CulturalTransmission
)
\]

Poor intergenerational transfer causes accumulated capability to decay faster.

Strong transfer converts accumulated experience into reusable future components.

---

## 19. The Core Feedback Structure

```text
                 HUMAN CAPABILITY H
                         │
                         ▼
ASSEMBLY COMMONS X ──► EXPERIMENTATION ◄── RISK CAPACITY F
       ▲                 │  ▲
       │                 │  │
       │                 ▼  │
       │              SELECTION
       │                 │
       │                 ▼
       └──────────── REUSABLE ASSEMBLIES
                         │
                         ├────────────► PRODUCTIVE CAPACITY K
                         │                     │
                         │                     ▼
                         │                   OUTPUT
                         │                     │
                         │              ┌──────┴──────┐
                         │              ▼             ▼
                         │        PRIVATE RETURN   COMMON RETURN
                         │              │             │
                         └──────────────┴─────────────┘
                                        │
                                        ▼
                              PURCHASING POWER
                                        │
                                        ▼
                                     ADOPTION
```

The wider system is:

```text
        ECOLOGICAL HEADROOM E
                 │
                 ▼
      ┌─────────────────────┐
      │  ASSEMBLY ECONOMY   │
      └─────────────────────┘
         ▲      ▲      ▲
         │      │      │
      AGENCY  EPISTEMICS  SOCIAL COHESION
        G        Q              S
         ▲                       │
         └──── POWER P ──────────┘
```

---

## 20. Viable Future Assembly Capacity

For diagnostics, define a provisional aggregate:

\[
VFAC =
H^a
X^b
G^c
N^d
Q^e
S^f
E^g
(1-P)^h
\]

This is deliberately multiplicative.

The intuition is that catastrophic collapse in one foundational subsystem can undermine the usefulness of strength in others.

Examples:

- \(E \rightarrow 0\): ecological collapse constrains future physical assembly,
- \(G \rightarrow 0\): effective human agency disappears,
- \(Q \rightarrow 0\): society loses reliable model correction,
- \(P \rightarrow 1\): nominal capability may exist while becoming inaccessible to most participants.

### Warning

VFAC is a research instrument, not a welfare function.

It must not be allowed to override constitutional constraints.

---

## 21. Constitutional Constraints

The model should distinguish between:

1. optimization objectives,
2. viability corridors,
3. non-tradeable constitutional floors.

Examples:

\[
\begin{aligned}
Privacy &\ge Privacy_{min}\\
BasicSecurity &\ge Security_{min}\\
Agency &\ge G_{min}\\
EcologicalHeadroom &\ge E_{min}\\
Domination &\le P_{max}\\
ExternalizedHarm &\le Harm_{max}
\end{aligned}
\]

The purpose is to prevent an aggregate optimization metric from justifying unacceptable local sacrifice.

---

## 22. Sensors and Proxies

Each important construct should be measured through a bundle rather than a single metric.

A sensor bundle should ideally include:

1. objective outcome measures,
2. structural measures,
3. behavioral measures,
4. subjective measures.

### Example: Effective Agency

Possible proxies include:

- material sufficiency,
- exit options,
- business formation accessibility,
- access to risk capital,
- provider switching costs,
- debt burden,
- economic mobility,
- migration flexibility,
- perceived agency,
- restartability after failure.

### Example: Domination

Possible proxies include:

- ownership concentration,
- infrastructure dependency,
- attention/discovery concentration,
- political influence concentration,
- market concentration,
- platform dependency,
- employer dependency,
- interoperability barriers.

### Example: Epistemic Capacity

Possible proxies include:

- research intensity,
- reproducibility,
- independent audit coverage,
- correction latency,
- concentration of information channels,
- public access to evidence,
- trust calibration,
- forecast accuracy.

---

## 23. Initial Scenario Set

GADM-0.1 should be tested against at least six regimes.

| Regime | Main perturbation | Primary question |
|---|---|---|
| Automated capitalism | \(K↑\), labor share ↓, ownership concentrated | Can abundance coexist with collapsing agency and demand? |
| Dividend abundance | Automation plus broadly distributed claims | Does demand stabilize without suppressing useful merit feedback? |
| Universal venture access | \(F↑\), restartability ↑ | Does experimentation rise sustainably or mainly increase waste and externalities? |
| Commons-rich economy | high \(
ho\), high \(X\) | How powerful is reuse compared with raw investment? |
| Attention oligopoly | discovery concentrated, \(P↑\) | Can productive abundance coexist with declining effective assembly access? |
| Generative regime | high \(X,G,Q,N\), distributed claims, bounded \(P\) | Does the hypothesized generative feedback loop remain stable? |

---

## 24. Failure Modes to Look For

The first simulations should explicitly search for:

- runaway domination,
- demand collapse,
- ecological overshoot,
- innovation without reuse,
- excessive experimentation waste,
- monoculture,
- fragmentation,
- epistemic collapse,
- loss of intergenerational transfer,
- stagnant low-risk equilibria,
- unstable high-growth equilibria,
- collapse of social cohesion,
- insufficient incentives for high-value contribution,
- and metric gaming.

---

## 25. Initial Research Questions

1. Which stocks are genuinely independent and which are composites?
2. Which variables require explicit age cohorts?
3. Which quantities behave multiplicatively rather than additively?
4. How should adoption interact with non-market selection?
5. What minimum level of inequality is useful for strong selection signals?
6. When does inequality become domination?
7. How much restartability maximizes useful experimentation?
8. How should common dividends respond to productive capacity?
9. What converts private innovation into reusable assembly commons?
10. Which forms of infrastructure most strongly increase \(X\)?
11. How should social cohesion respond to opportunity versus realized outcomes?
12. Can high diversity and high composability coexist dynamically?
13. How sensitive is generativity to epistemic degradation?
14. How should ecological headroom constrain physical output?
15. Which mechanisms preserve agency during rapid automation transitions?

---

## 26. Next Modeling Steps

### GADM-0.2

- specify all flows and causal signs,
- introduce units or normalized interpretation,
- produce a causal-loop diagram,
- identify reinforcing and balancing loops,
- define initial parameter ranges,
- implement a toy simulation.

### GADM-0.3

- add age cohorts,
- add explicit household distribution,
- separate economic wealth from power,
- model common versus private ownership,
- model adoption and attention separately.

### GADM-0.4

- introduce multiple sectors,
- represent rival versus non-rival goods,
- introduce explicit ecological resource classes,
- introduce institutional selection mechanisms.

### Later

- open-economy effects,
- migration,
- trade,
- international competition,
- currency and monetary dynamics,
- empirical calibration,
- policy experiments,
- transition dynamics.

---

## 27. Research Status

GADM-0.1 is a hypothesis generator.

Its main success criterion is not predictive accuracy yet.

It succeeds if it helps us:

- make assumptions explicit,
- discover missing variables,
- distinguish reinforcing from balancing feedback,
- identify unstable institutional regimes,
- formulate measurable hypotheses,
- and decide which parts of the conceptual framework deserve deeper formalization.