River embankments are one of the oldest and most widely used structural measures for flood control in South Asia, credited with protecting lives, crops, and infrastructure across flood-prone river basins in India. This paper examines the dual character of embankments: their demonstrated socio-economic value in stabilising agriculture, infrastructure, and settlement patterns along rivers such as the Ghaggar, Yamuna, and Kosi, set against a growing body of hydro-geomorphic evidence, most notably the 2008 Kosi embankment breach in Bihar, showing that embankments can, over time, convert manageable flood risk into infrequent but catastrophic risk. The paper then moves from this qualitative case-based critique to a quantitative treatment of how and why embankments physically fail during overtopping, presenting the shear-stress-based erosion model used to predict soil and road-surface erosion rates, including the empirically derived erosion-rate equations for cohesive and noncohesive soils. Read together, the socio-economic case for embankments and the engineering case for their eventual failure point toward the same conclusion: embankments are not a permanent structural fix, but an infrastructure class whose value depends on active monitoring, material selection, and reinforcement over their operational life.