Most proposals for new technological approaches fail, and that is reasonable. Despite that, most of the technological proposals arising from basic science are promoted unhesitantly, with little attention to critical appraisal, even little opportunity for the presentation of criticism. We discuss several case histories related to devices intended to displace the transistor in computer logic. Our list includes devices using control of quantum mechanically coherent electron transmission, devices operating at a molecular level, and devices using nonlinear electromagnetic interaction. Neural networks are placed in a different category; something seems to be coming out of this field after several decades of effort. © 1990.