The irreducible epistemic atoms underlying the curriculum. 4,828 atoms across 8 types and 2 molecules
Design pilots with baseline measurements collected before implementation and control conditions to distinguish intervention effects from external factors.
Select pilot contexts that are representative and moderately challenging rather than ideal or enthusiastic, to generate generalizable learning.
Run pilots for at least three full cycles of the process being changed to distinguish initial learning effects from steady-state performance.
Investigate pilot surprises as valuable signals of system dynamics missing from the original analysis rather than dismissing them as anomalies.
Treat resistance to change as diagnostic information about system properties rather than opposition to overcome.
Test systemic changes by removing the attention, personnel, and resources that accompanied the change initiative — genuine system changes persist while attention-dependent improvements revert.
Implement structural changes before behavioral training to ensure the environment supports rather than contradicts the behavior being taught.
Test pattern validity by making falsifiable predictions about future instances before they occur, because patterns that survive prediction testing on unseen data are structurally different from patterns that only fit historical data.
Assign decision authority to the lowest level with sufficient information and capability (subsidiarity), escalating only when decisions exceed that level's scope, because distributed decision-making produces speed, quality, and capability development benefits that outweigh occasional reversible mistakes.
Define explicit escalation criteria specifying when decisions should move to higher levels rather than relying on default escalation, because ambiguous authority produces rational escalation behavior that creates organizational bottlenecks.
Optimize organizational processes by minimizing queue time, handoffs, rework, and non-value-adding steps rather than increasing worker effort.
Design processes from the customer's desired outcome backward rather than from internal functional responsibilities forward.
Build error detection and quality verification into each process step rather than inspecting for defects at process completion.
Make process execution the default path and workarounds the exception requiring active override.
Deploy technology to enable fundamentally different processes rather than to automate existing dysfunctional processes.
Design technical systems and social systems jointly rather than sequentially to avoid suboptimal adaptation.
Specify minimum necessary constraints when implementing technology and leave remaining design to users.
Align formal metrics, informal recognition, and promotion criteria with changed behaviors to prevent incentive-driven reversion.
Start new compounding habits at absurdly small scales that require minimal motivation, because consistency over time produces exponentially greater returns than intensity without consistency.
Remove structural, political, resource, and attention obstacles that only organizational authority can address.
Build organizational capability for systemic change as a permanent competency rather than treating each transformation as a one-time project.
Create sufficient internal system variety to match environmental complexity and enable adaptive responses.
Distribute decision authority to the organizational level where the relevant information naturally resides, not where formal authority currently sits.
Transition decision authority gradually through progressive phases of transparency, consultation, bounded authority, and full distribution, with each phase building the infrastructure needed for the next.