Early on, experts warned about serious flaws in the system. In 1984 Wilson Shaffer, a storm-surge modeler at the National Weather Service, told the corps that the Standard Project Hurricane, the hypothetical storm against which engineers tested their levee designs, was too small to represent the true threat. Stronger storms—such as the Category 5 Hurricane Camille, which slammed into Mississippi four years after Betsy—could easily overtop the system and flood the city, Shaffer said. "There are no high areas near the city that wouldn't flood in extreme cases," he wrote. "High ground is several tens of miles away. Evacuation routes are limited. … Imagine, if you can, the massive destruction and loss of life."

The corps rejected these warnings. Protection against these "rare" events, the corps deemed, would be "prohibitively expensive," a conclusion seconded by the Orleans Levee District, the local flood-protection authority. The corps also dismissed another, longer range threat, summed up in a graph made by John S. Hoffman of the Environmental Protection Agency: rising sea level because of global warming. By 2100, Hoffman projected, the sea could rise "at least two feet (0.6 meters), with a more likely amount as 3.7 feet (1.1 meters). A rise as high as 12.6 feet (3.8 meters) cannot be ruled out." The corps responded: "Given the uncertainty of projections of sea level changes, an attempt to accommodate such changes in the design of the project … would represent a very poor use of funds."

"Locals wanted the cheapest possible protection system," says Oliver Houck. "But it wasn't cheap, it was just badly built."

The floodwalls along the city's major drainage canals were a classic example of the shortcomings. The corps didn't want to build most of them. Initially it planned to block storm surge with giant barriers across the eastern inlets of Lake Pontchartrain, beef up the levees along the southern lakeshore, and erect massive floodgates to keep high water out of the canals. Environmental groups, concerned about impacts on the lake and its wetlands, blocked the plan in court. The corps dropped the barriers and switched to a system that would rely on higher lake levees and floodgates. State and local officials—who were required to pick up nearly a third of the ballooning tab—balked at the cost of the gates. They also feared that closing the gates could actually cause flooding, as rainwater piled up in the canals. City leaders pushed instead for floodwalls along the canals. The groups remained at loggerheads until 1992, when Congress passed a water resources act that forced the corps to do it the city's way.

Foundation problems plagued the levees and floodwalls from day one. A contractor building the 17th Street Canal floodwalls in the mid-1990s actually tried to sue the corps for more money as the mucky soils drove up costs. The underlying sheet piles—steel panels driven into the ground to form a barrier—were shifting and pushing the concrete walls on top out of line.

Katrina, alas, exposed these weak underpinnings. When the storm drove floodwaters to within four feet (1.2 meters) of the top, the walls deflected backward, opening a crack at their base. Water poured in, found a thin layer of clay as slick as jelly, and forced nearly 450 feet (137 meters) of levee into Orleans Parish. On the London Avenue Canal, sandy soils led to similar blowouts. Floodwall failure let in nearly 80 percent of the water that flooded the central part of the city. "Just ten million dollars more spent on sampling and foundation investigation, and the system wouldn't have failed," says engineer J. David Rogers, who investigated the breaches with a team from the University of California, Berkeley. "It didn't come within a country mile of the design load."