Other Mental Models · OM-36
A problem's true cause often isn't any single failing component, but the structure of relationships and feedback loops connecting the components together — fixing the part that appears broken frequently fails, because the actual dysfunction lives in the interactions, not in any one piece.
An approach to understanding complex problems that focuses on the relationships, feedback loops, and structural interactions among a system's components, rather than analyzing components in isolation — treating a system's behavior as an emergent property of its overall structure, not merely the sum of its individual parts examined separately.
Formalized as a distinct discipline through the field of system dynamics, pioneered by Jay Forrester at MIT beginning in the 1950s, and popularized more broadly for general audiences by Donella Meadows and Peter Senge, notably in Senge's widely read 1990 book The Fifth Discipline.
The Mechanism
The dysfunction usually lives in the structure and feedback loops, not in any single broken part
Forrester's original system dynamics modeling of industrial supply chains found that inventory oscillations and shortages, often blamed on individual poor forecasting decisions by specific managers, were frequently better explained by the overall structure of information delays and feedback loops connecting suppliers, manufacturers, and retailers — replacing individual managers rarely fixed the oscillation, because the actual cause lived in the structural relationships between the parts, not in any single manager's decision-making.
01 · IT SPECIFICALLY LOOKS FOR FEEDBACK LOOPS, DELAYS, AND STRUCTURAL RELATIONSHIPS AS THE PRIMARY UNIT OF ANALYSIS
This is a genuinely different analytical unit than examining individual components
Rather than asking 'which specific component is malfunctioning,' systems thinking asks 'what pattern of relationships, feedback loops, and time delays among the components produces this observed behavior' — a structurally different question that frequently reveals causes invisible to an analysis that only examines components one at a time, since the actual dysfunction may exist entirely in the interactions rather than in any single part.
02 · IT EXPLAINS WHY REPLACING OR FIXING AN INDIVIDUAL 'BROKEN' COMPONENT OFTEN FAILS TO RESOLVE A PERSISTENT PROBLEM
A well-documented and costly failure pattern in organizational troubleshooting
When an organization repeatedly replaces individual people, tools, or process steps in response to a persistent problem, without success, systems thinking suggests the actual cause may be structural — a feedback loop, an information delay, a misaligned incentive connecting multiple parts of the system — meaning the problem will simply recur with the new component in place, since replacing a part does nothing to address a dysfunction that lives in the relationships among parts.
03 · IT FREQUENTLY REVEALS THAT AN INTUITIVE, LOCALLY-OBVIOUS INTERVENTION MAKES THE OVERALL SYSTEM WORSE, NOT BETTER
Interventions that look correct at the component level can backfire at the system level
Because a system's overall behavior emerges from feedback loops and interactions rather than from any single component in isolation, an intervention that appears obviously correct when evaluated only at the level of a single component (speeding up one specific step in a process) can worsen overall system performance if it interacts poorly with other feedback loops elsewhere in the structure — a genuinely counterintuitive but well-documented pattern across supply chains, organizations, and ecosystems.
Where It Fails / Inversion
Where it fails / inversion
Not every problem requires a full systems-level analysis — many problems genuinely are caused by a single, identifiable, isolated component failure, and insisting on an elaborate systems-level analysis for a simple, localized problem can waste time and obscure an otherwise straightforward fix; systems thinking is most valuable specifically for persistent, recurring problems that resist repeated attempts at component-level fixes.
How To Use It
Worked example · diagnosing a persistent bottleneck despite repeated individual fixes
A team that has repeatedly replaced or retrained the specific individual or tool blamed for a persistent process bottleneck, without lasting success, should consider mapping the actual structure of information flow, feedback loops, and time delays connecting the different steps and people involved in the process — the true cause may be a structural feature (a delayed feedback signal, a misaligned incentive elsewhere in the system) that will keep producing the same bottleneck regardless of which individual component is swapped out.
How to use it
When a problem persists despite repeated attempts to fix the specific component that seems to be at fault, consider mapping the broader structure of relationships, feedback loops, and information delays connecting the system's parts — the actual dysfunction frequently lives in that structure rather than in any single replaceable component, which is why component-level fixes so often fail to produce a lasting resolution.
See Also