Coastal erosion occurs along many of the world’s shorelines, but is of particular concern to small islands. In response, island shorelines are often armoured through engineered structures such as seawalls, rip-raps, dikes or similar. Such hard protection is rarely successful in island contexts; rather than stop erosion, hard protection often simply redistributes and sometimes exacerbates shoreline erosion, and fails to protect property, land and food production. Such hard-protection structures frequently collapse shortly after construction. Why then does hard protection remain widespread? We argue that diffusion, specifically learning and emulation, helps explain the spread of hard protection. Island communities learn from the apparent effectiveness of hard protection in some contexts, and consider it an appropriate and accepted response to coastal erosion, regardless of its actual consequences (emulation). Case studies from the Pacific (Fiji, Federated States of Micronesia) and Indian Ocean (Maldives, Mauritius) show that both mechanisms are at work in small island states (SIDS) and may be mutually reinforcing. The case studies underline the role of the specific context in the understanding of not only why a specific adaptation measure is introduced, but also how it may change and evolve over time—and accordingly, when and how communities can move to more effective and sustainable alternatives to hard protection.