Current causal representation
System Map v2.1
The causal center moved upstream to organizational configuration. Constraints, workflow, state production, capability stocks, and value/burden distribution can be endogenous.
How to read the map
Required / structural relationUsed when the sequence is definitional or operational, not to imply experimental causal identification.
Inferred relationA current causal or mechanistic interpretation whose identification may remain incomplete.
FeedbackOutcome changes later expectations, capabilities, dependencies, or structure.
Current map
Feasible action set → configuration → constraints → outcome → reconfiguration
External environment
Demand, institutions, technology markets, labor markets, regulation, and physical conditions that shape the feasible organization.
- demand
- institutions
- technology markets
- labor markets
- regulation
- physical conditions
AI capability + cost
Available model, agent, tool, and infrastructure capability together with the relative cost of using it.
Feasible action set
Actions that can be technically, economically, institutionally, and physically attempted under the current environment.
Organizational configuration
The arrangement that jointly determines actor response, task boundaries, workflow structure, capability reproduction, and which constraints emerge.
- actor expectations / incentives
- authority / practical control
- information / state-production behavior
- task economics
- economic task boundaries
- workflow dependencies
- human capability stocks
- organizational capability stocks
- external conditions carried into the workflow
Candidate / binding constraints
Observed or predicted limitations classified by origin. A candidate becomes a binding workflow constraint only under the stronger intervention-validation standard.
- technical
- resource
- state / information
- institutional / formal authority
- coordination
- incentive-produced
- distributional
- strategic
- physical
- demand
- capability-stock
- configuration
Execution
Organizational, software, institutional, or physical action under the current configuration and constraints.
Outcome
The real consequence the workflow exists to produce, not merely an artifact, tool call, or process completion.
Value + burden distribution
Who receives financial gain, time saving, quality benefit, extra review, waiting, risk, accountability, cost, or loss of discretion.
- financial gain
- time saving
- quality benefit
- extra work / review
- waiting
- downside risk
- accountability
- loss of discretion / bargaining power
Updated expectations / capabilities / structure
Outcomes update expectations, capabilities, work allocation, dependencies, and organizational structure, changing the next response to AI.
Read the map as text
- External environment. Demand, institutions, technology markets, labor markets, regulation, and physical conditions that shape the feasible organization.
- AI capability + cost. Available model, agent, tool, and infrastructure capability together with the relative cost of using it.
- Feasible action set. Actions that can be technically, economically, institutionally, and physically attempted under the current environment.
- Organizational configuration. The arrangement that jointly determines actor response, task boundaries, workflow structure, capability reproduction, and which constraints emerge.
- Candidate / binding constraints. Observed or predicted limitations classified by origin. A candidate becomes a binding workflow constraint only under the stronger intervention-validation standard.
- Execution. Organizational, software, institutional, or physical action under the current configuration and constraints.
- Outcome. The real consequence the workflow exists to produce, not merely an artifact, tool call, or process completion.
- Value + burden distribution. Who receives financial gain, time saving, quality benefit, extra review, waiting, risk, accountability, cost, or loss of discretion.
- Updated expectations / capabilities / structure. Outcomes update expectations, capabilities, work allocation, dependencies, and organizational structure, changing the next response to AI.
The final stage feeds back into organizational configuration: the same AI capability can therefore produce a different future outcome after the organization has adapted.
What the map reveals
Four implications of the current representation
- AI is upstream of an organizational response.
Capability changes what can be attempted, but the organization still determines task boundaries, authority, state production, work allocation, and which constraints emerge.
- Some constraints are produced, not merely encountered.
A state problem can be generated by incentives or control; a resource problem can disappear when the task boundary changes.
- The same outcome can imply different mechanisms.
A stalled workflow can reflect missing state, incentive, authority, coordination, physical execution, demand, or a misconfigured division of labor.
- The system changes the conditions of its own next use.
Work allocation, capability stocks, expectations, dependencies, and organizational structure feed back into future AI value.
What the map does not establish
- that incentives dominate technical constraints;
- that workflow recomposition dominates node acceleration;
- that AI broadly causes long-run deskilling;
- that every observed limitation is a binding constraint;
- that configuration necessarily improves prediction beyond task capability;
- that mechanisms found elsewhere transfer directly to Malaysia or Southeast Asia.
Map evolution
v1 → v2 → v2.1
Earlier maps remain visible as historical states. The semantic diff explains what changed and why; historical maps are not current beliefs.
AI was represented as uneven capacity expansion inside a gated adaptive work system; bottleneck migration was a central explanatory idea.
The map became explicitly recursive: capacity changes propagated through state, evidence, authority, coordination, execution, economics, labor, and physical reality, then altered the next system state.
The causal center moved upstream to organizational configuration. Constraints, workflow, state production, capability stocks, and value/burden distribution can be endogenous.
v1 → v2system dynamicsADDED
Before
A gated work system with bottleneck migration as the main dynamic.
After
An explicitly recursive system with state, authority, economics, labor, physical reality, feedback, stocks, delays, and operating regimes.
Why: Investigations 02–16 showed that local productivity, organizational state, authority, learning, economics, resilience, and physical execution could not be treated as one linear propagation chain.
Consequence: The map became a coupled adaptive system in which earlier AI effects changed the conditions for later AI effects.
Trigger investigations: INV-02, INV-05, INV-06, INV-08, INV-10, INV-12, INV-14, INV-15, INV-16
v2 → v2.1organizational configurationMOVED
Before
Capacity passed through largely pre-existing propagation constraints.
After
Feasible actions alter organizational configuration, and that configuration can produce or expose constraints.
Why: The v2 model was too flexible retrospectively and placed several variables in conflicting causal roles.
Consequence: Configuration becomes the main causal organizing object and must earn its complexity prospectively.
Trigger investigations: INV-17, INV-18, INV-19
v2 → v2.1workflowREPLACED
Before
AI propagated through an existing workflow graph.
After
AI can change relative task economics, task boundaries, role bundling, and therefore the workflow graph itself.
Why: Investigation 18 found workflow endogeneity to be a supported mechanism, though its general magnitude is unresolved.
Consequence: Node acceleration and graph recomposition must be separated empirically.
Trigger investigations: INV-18
v2 → v2.1organizational stateMOVED
Before
State was mainly an upstream input that could constrain action.
After
State remains an input, but state quality can also be produced by incentives, authority, control, and information-producing behavior.
Why: Investigation 17 showed that some apparent information constraints can be incentive- or control-produced.
Consequence: A data-quality symptom no longer implies a technical data root cause.
Trigger investigations: INV-17
v2 → v2.1human expertiseDECOMPOSED
Before
Expertise was treated as an aggregate human capability stock.
After
Human capability is decomposed into production, diagnosis, verification, recovery, transfer, calibration, explanation, and teaching.
Why: Investigation 19 found scalar expertise too coarse to represent different effects of AI-mediated experience.
Consequence: Current assisted output can no longer stand in for durable organizational oversight or recovery capability.
Trigger investigations: INV-19
v2 → v2.1authorityDECOMPOSED
Before
Authority appeared as a relatively unified organizational variable.
After
Formal authority, practical control, informational control, accountability burden, economic upside, and discretion value are separated where material.
Why: Investigation 17 and the semantic seal showed that formal decision rights did not capture who actually controls action or state production.
Consequence: Authority-related failures can now be represented as different mechanisms rather than one generic gate.
Trigger investigations: INV-17
v2 → v2.1constraint migrationNARROWED
Before
A new problem appearing after another improvement was often interpreted as the next bottleneck.
After
The next constraint must be predicted before intervention, its mechanism specified, and the predicted signature directly tested.
Why: Retrospective bottleneck stories could explain almost any result and therefore risked being unfalsifiable.
Consequence: Observed limitation, causally identified factor, and intervention-validated binding constraint are now separate statuses.
Trigger investigations: INV-17, INV-18, INV-19
Redraw condition
The map should become simpler if it fails prospective tests
v2.1 should be simplified if repeated prospective organizational studies show that task-level AI capability predicts organizational outcomes almost as well as actor incentives, task boundaries, workflow configuration, state endogeneity, and capability stocks.