The models in this group describe how systems behave once they are running, and nearly all of them report the same surprise: the behaviour is not what the people inside the system expected. Well-meant fixes backfire, full capacity slows everything down, and choices made casually early on become impossible to reverse.
Feedback and emergence
The foundation is feedback. A Negative Feedback Loop counteracts change and holds a system steady; a Positive Feedback Loop amplifies it, so success breeds success and decline breeds decline. Emergence and Spontaneous Order describe the larger result: patterns that arise from many local interactions and belong to no single part. Simon's Near-Decomposability explains why complex systems that last tend to be built from loosely linked modules, each tightly organised inside.
The systems-thinking archetypes
A second cluster comes from the system-dynamics tradition of Jay Forrester, Peter Senge and Donella Meadows. Fixes That Fail describes the remedy that relieves a symptom quickly while worsening its cause. Shifting the Burden describes dependence on an outside intervener eroding a system's own capacity. Policy Resistance shows interventions absorbed when actors pull a system towards conflicting goals. Meadows's Leverage Points ranks where to intervene, arguing that changing a system's goals, rules and underlying assumptions matters far more than adjusting its parameters.
Flow, bottlenecks and history
The operations models are more precise. Kingman's Formula shows waiting time rising steeply as utilisation approaches capacity, which is why a service run at full stretch becomes slow for everyone. Goldratt's Theory of Constraints says throughput is set by one bottleneck, and effort spent elsewhere changes nothing.
Path Dependence and Lock-In add history. Early choices raise the cost of switching, so a standard or an institution can persist long after better alternatives appear. The Robustness–Fragility Tradeoff completes the picture: optimising for expected conditions makes a system brittle outside them.
Problems without solutions
Two models describe the limits of design itself. Rittel and Webber's Wicked Problems have no definitive formulation, no stopping rule and no test of whether a solution worked, which covers much of social policy. Christopher Alexander's Pattern Language offers a response: recurring problems have named, combinable solutions, which can be assembled at many scales without a master plan. The idea crossed from architecture into software, where design patterns became a standard vocabulary.
Where it applies
These models are the working tools of operations managers and policy analysts. Kingman's Formula argues against targets that push hospitals, airports or engineering teams to full utilisation. The archetypes give policymakers a way to anticipate backfire before it happens, and a vocabulary for explaining it afterwards. Path dependence explains why technology choices made in a hurry deserve more care than they usually get, since they may still be in place decades later.
Where it is argued over
The famous illustration of lock-in, the QWERTY keyboard, is disputed on the evidence that the layout is meaningfully worse than its rivals, although the mechanism is not in doubt. Spontaneous order carries political weight, since Hayek used it to argue for markets over planning and critics reply that many orders called spontaneous rest on deliberate rules. Meadows's ranking of leverage points is influential but was offered as a considered judgement rather than a tested finding.
Why it is worth knowing
The single most useful lesson here is that a system's behaviour often comes from its structure rather than from anyone's intentions. That lesson changes what people blame and what they try to fix. Instead of looking for the person who made the queue long or the policy fail, it points to the loop, the bottleneck or the early decision that made the outcome likely, which is usually where a lasting remedy lies.
The entries behind this essay: Model in Design & Systems. Every subject has its essays on the essays page.