In 1985, the U.S. Army commissioned prominent psychologists to investigate the possibility of extending human capabilities using parapsychological techniques (Swets & Bjork, 1990). Influential members of the army were frustrated by the slow pace of advancements in human performance and believed that large gains could be made using methods outside of the mainstream. They believed that things like mental concentration and guided imagery could allow soldiers to walk through walls, view things remotely, and even kill adversaries by staring at them (Ronson, 2005). Not surprisingly, the panel of psychologists concluded that these ideas were without merit. In reviewing the psychologists’ work, Morrison (1988) observed, “Among the most difficult lessons in science is how not to deceive yourself” (p. 109).

In this article, I summarize and respond to key comments and criticisms offered in commentaries on Kraiger (2008). At issue is whether the proposed third-generation instructional model is fundamentally sound, represents a new or better approach, overemphasizes the importance of learner–learner interaction, and underemphasizes the role of the instructor. Responses to these concerns are offered, along with clarification that the major contribution of the third-generation model is at the point of training design.

Commentators expressed a wide variety of views on my evaluation of the state of assessment center (AC) research and practice. In this response, I first trace the evolution of the construct validity paradox “urban legend.” Next, I consider the commentators’ comments as they relate to (a) my recommendation to abandon dimension-based ACs in lieu of task- or role-based structures (b) my recommendation to discontinue design fix attempts toward making ACs conform to multitrait–multimethod construct validity criteria, and (c) considerations of construct validity and validation evidence. Finally, I offer some directions for future AC research and practice.