Successful interventions reorganise systems around their own outputs. The gains are real, the new baseline becomes normal, and the conditions that made the original system resilient are quietly dismantled in the process. See Infrastructure Without Culture for the related pattern of what happens when the governance needed to sustain an intervention is never built. See The Loop of Reinvented Knowledge for what happens to the traditional practices that the intervention displaces.

A successful intervention increases output, the gains become the new baseline, and the system loses the capacity to function without continued external inputs — by the time the trap becomes visible, the exit options have already closed.

The coastal polders of Bangladesh are the clearest illustration. Before the embankments, communities had developed a finely adapted system: seasonal flooding fertilised the land with nutrient-rich sediment, community-managed tidal channels regulated salinity, and traditional crop varieties were matched to the rhythms of the delta. The system was low-productivity by modern measures. It was also resilient — self-renewing, locally governed, and not dependent on inputs it could not generate itself.

The embankments changed all of this. Agricultural yields rose dramatically. Farmers shifted from one crop per year to two or three. Food security improved. By any conventional measure, the intervention was a success.

But the success reorganised the system around its own continuation. The traditional water management knowledge that had sustained the delta for generations became irrelevant and was lost. The natural sediment flows that had maintained soil fertility were interrupted. The community governance structures that had managed tidal channels atrophied. Farmers became dependent on chemical fertilisers to replace what floods had once provided for free, on motorised pumps to replace what tidal rhythms had once regulated, and on external maintenance budgets to sustain embankments that the natural system had never needed.

When the deterioration came — drainage congestion, salinity intrusion, soil exhaustion, land subsidence — the communities had no fallback. The system they had lived in before the intervention was gone. The knowledge needed to manage it was gone. The only available response was more intervention: more pumping, more chemicals, more external maintenance. The trap had closed.

The pattern is not unique to Bangladesh. It appears wherever a successful intervention reorganises a system around its outputs: irrigation infrastructure that eliminates traditional water sharing arrangements, flood protection that enables development in areas that were previously naturally managed, groundwater schemes that displace rainfall-fed agriculture and then deplete the aquifer. In each case the success is real. The dependency it creates is also real. And the two arrive together, so gradually that the moment of no return passes without being noticed.

What makes the trap a trap rather than simply a trade-off is the asymmetry of the exit. The gains from the intervention are captured quickly and distributed widely. The costs accumulate slowly and fall unevenly — on the poorest, on the most exposed, on those with the least capacity to adapt. And by the time the costs become undeniable, the landscape has been so thoroughly reorganised around the intervention's continuation that reversing course requires dismantling not just infrastructure but livelihoods, institutions, and the expectations of an entire generation.

Before an intervention is implemented, map what the existing system does that the intervention will not replace — the ecological functions, the governance arrangements, the knowledge, the informal resilience. Build the continuation of those functions into the design, not as an afterthought but as a condition of success. Monitor not just output but the system's capacity to function without continued external inputs. When that capacity is declining, treat it as an early warning, not a footnote.

Linked patterns: Infrastructure Without Culture — the governance and maintenance culture that sustains an intervention must be built alongside it, not assumed. The Loop of Reinvented Knowledge — the traditional practices displaced by the intervention will be rediscovered elsewhere at enormous cost. The Tragedy of the Rational Actor — individual farmers and planners making rational decisions collectively produce the dependency the trap depends on. Protection Enables Exposure — the related pattern where successful protection reorganises risk rather than reducing it.