Subglacial bedforms are a range of landforms (10-105m long) shaped mostly in glacial sediments and are abundant on ice-sheet beds. Numerous explanations for their generation, especially of drumlins, have been proposed. Rather than viewing them as resulting from erosion or deposition directly by ice, Smalley and Unwin (1968) proposed that both might occur simultaneously if, beneath the glacier, a layer of lubricating sediment existed and ‘flowed’, conducting geomorphic work (erosion, deposition, shaping) of its own. This idea appears to have steered the field, leading perhaps to final resolution of the centuries-old problem of drumlin genesis. Here I trace how the idea evolved, leading to the conclusion that subglacial bedforms are emergent phenomena arising from self-organization in the coupled flow of ice, sediment and water. The ubiquity and patterning of bedforms appears to be well explained by a naturally arising flow instability in the subglacial system, and which has been demonstrated by analytical and numerical modelling. A problem for the instability theory is an apparent mismatch between predicted and observed sedimentological properties of bedforms. A distinction is made between emergent drumlins, drumlin clones and obstacle drumlins which helps explain some apparent contradictions, and a conceptual framework is erected that might form a basis for confronting the theory with the wealth of observational evidence that exists.