The Levees Were Already Broken

Their findings were stark. The levees protecting New Orleans didn't fail because Katrina was unpredictable. They failed because they were inadequately designed, improperly maintained, and subject to fifty years of incremental decisions that prioritized cost over resilience. The storm exposed a system that was already broken. It just needed sufficient pressure to show where.

This is the kind of finding that should be obvious but somehow never does. We persist in thinking of disasters as events — as moments of acute failure, lightning strikes, acts of God. The IPET report was 6,000 pages of evidence that disaster is mostly slow. It accumulates in budget meetings and engineering memos and decisions to build a floodwall to a standard that was technically defensible but practically inadequate.

The universe charged compound interest. August 2005 was the settlement date.

i · the audit of a system that was already failing

The IPET found multiple concurrent failures across the New Orleans flood protection system. Floodwalls were designed with inadequate safety margins — built to survive water levels that proved too low given Katrina's actual storm surge. Foundation materials had been incorrectly assessed. Sections of the London Avenue and 17th Street canals gave way not from overtopping but from structural failure beneath the waterline — the kind of failure that happens when the engineering math was wrong to begin with, or when the ground wasn't what the models assumed.

The Industrial Canal breached catastrophically, sending a wall of water into the Lower Ninth Ward. The Mississippi River Gulf Outlet — a navigation canal cut straight through the natural barriers between the Gulf and the city's interior — channeled storm surge like a funnel into the heart of New Orleans. Over three hundred breach and failure points were documented across the system.

What made the IPET report landmark wasn't just the technical inventory — it was the honest accounting of how a protection system erodes over decades. Budget pressures produce design compromises. Maintenance gets deferred. Inspections surface warnings that are documented, filed, and not acted upon. Each individual decision is defensible in isolation; the chain of those decisions, viewed across fifty years, is a portrait of a system in steady drift.

The levees weren't destroyed by Katrina. They were revealed by it.

This distinction matters enormously. If the levees failed because the storm was unprecedented, the lesson is probabilistic: build for the 500-year event, not the 100-year event. If the levees failed because the system was already compromised by accumulated neglect, the lesson is different: the thing that kills you is usually the drift you stopped measuring.

The IPET report is clear about which lesson applies. Katrina was severe, but it was within the range of storms the protection system should have been able to handle — if the system had been what it was supposed to be.

ii · what fifty years of good enough looks like when the bill arrives

The Corps of Engineers and its partner agencies didn't set out to build an inadequate system. The engineers who designed the original levee network were working within the technological and budgetary constraints of their era. The problem is that systems age, storms intensify, land subsides, and the margin between adequate and catastrophic narrows steadily when no one is actively closing it.

New Orleans sits in a geologically hostile location. Much of the city is below sea level and continues to subside — some areas by several feet since the original levee network was built. The surrounding wetlands that historically absorbed storm surge had been substantially degraded by decades of canal dredging for oil and gas extraction. Every acre of wetland lost was a yard of surge reduction removed from the system's natural buffer. The engineered protection system was asked to do more work precisely as the conditions it operated in became more hostile.

This is the structural pattern that IPET documented in engineering terms: when a system's load increases and its resilience decreases simultaneously, the gap between them is where failure lives. Widen that gap slowly enough and no single inspection catches it. Let it persist long enough and you get August 29, 2005.

The human cost: 1,833 people dead. Hundreds of thousands displaced. Entire neighborhoods erased. The Lower Ninth Ward, flooded by the Industrial Canal breach, became a national shorthand for systemic failure and abandonment — the image of people on rooftops, of a city drowning while the mechanisms designed to prevent it took time to mobilize.

New Orleans had been warned. Levee engineers had written memos. Journalists had published disaster scenario reporting as far back as 2002. FEMA had conducted a major exercise — Hurricane Pam — just fourteen months before Katrina, modeling nearly exactly what happened. The knowledge existed in abundance. The institutional will to act on it didn't accumulate the way the risk did.

iii · what the invoice teaches

Twenty years after the IPET report, the rebuilt system protecting New Orleans is substantially stronger. The Hurricane and Storm Damage Risk Reduction System — constructed at a cost of approximately $14.5 billion — represents the most significant flood protection infrastructure investment in American history. Floodwalls are taller. Pump stations are larger. The system is rated to survive a hundred-year storm.

This is genuinely good news, and it deserves acknowledgment before the uncomfortable question: why did it require 1,833 deaths to produce that investment?

The answer is not mysterious. Systems that protect against low-frequency, high-consequence events are structurally difficult to fund through normal political processes. The storm that hasn't happened yet has no lobby. The thousand people who won't die in a hurricane if you build the wall correctly don't know who they are and can't organize around that fact. Meanwhile, the money that would fund prevention is visible, immediate, and subject to competing claims from problems that are already happening.

This isn't a Louisiana problem. It's a universal feature of how humans manage systemic risk. We underfund prevention and overfund response because prevention is invisible and response is dramatic. The IPET report is one of thousands of such documents making this argument for one infrastructure system or another. The pattern recurs: accumulate deferred maintenance, receive catastrophic correction, build back better, begin accumulating again.

The IPET report also documented something harder to quantify than engineering specs: the cost of distributed accountability. The levee system was overseen by a complex web of federal, state, and local authorities, each with partial responsibility and none with full ownership. In systems designed this way, failure travels along the boundaries between institutions indefinitely — warnings passed from one body to another, risks acknowledged but belonging to someone else's budget, maintenance deferred because the next actor in the chain is responsible for it.

No single decision broke the levees. Every decision participated.

What the IPET report reads as, twenty years on, is less a technical manual and more a case study in what misalignment with physical reality costs over time. The levees weren't aligned with the storms they needed to survive. The funding structures weren't aligned with the slow-building risk. The maintenance regimes weren't aligned with the actual condition of the infrastructure. The accountability structures weren't aligned with the scale of what could go wrong. Every layer drifted. Every drift compounded.

Katrina didn't break the levees. The levees broke themselves, over fifty years, one small misalignment at a time. The hurricane arrived at the precise moment when all those misalignments finally added up to something that would give.

The 6,000-page report is available in full. The lessons are not complicated. Whether they compound differently than the risk did is the question the next fifty years will answer.