The increasing complexity of the network society is the result of a particular approach to design: that of mastery, control, ease of use and interconnectedness. The author analyzes this design approach for its negative and positive aspects, which he labels as "designing for dominance" and "designing for dialogue", respectively. Both of these aspects are inherent (and in constant opposition) in all designed learning objects and environments. He argues that the "dominant" aspect of design, as influenced by cognitive science, obscures or minimizes the possibility of learner dialogue in favor of empirical knowledge. Design for dominance eliminates the educative potential of chance and ambiguity in its design process, while it emphasizes complete control over the elements of production. Overly dominant designs are ill-suited to account for higher-order thinking skills, which can only result from more dialogical designs, in service of critical reflection, withholding judgment and productive ambiguity. The author suggests that a return to "relation" and "responsibility", as constituting design tenets in ubiquitous learning, would enable learners to learn the benefits of "being pulled up short" when confronted with challenging world views. He concludes his essay with the provocative suggestion that we may, in some cases, want a "worse" design if it exponentially increases the capacity for dialogue and that, in fact, there may be some learning problems that cannot be solved through the automatic striving for a "better" design.