Policymakers and scholars — both lawyers and economists — have long been pondering the optimal design of default rules. From the classic works on “mimicking” defaults for contracts and corporations to the modern rush to set “sticky” default rules to promote policies as diverse as organ donations, retirement savings, consumer protection, and data privacy, the optimal design of default rules has featured as a central regulatory challenge. The key element driving the design is opt-out costs — how to minimize them, or alternatively how to raise them to make the default sticky. Much of the literature has focused on “mechanical” opt-out costs — the effort people incur to select a non-default alternative. This focus is too narrow. A more important factor affecting opt-out is information — the knowledge people must acquire to make informed opt-out decisions. But, unlike high mechanical costs, high information costs need not make defaults stickier; they may instead make the defaults “slippery.” This counterintuitive claim is due to the phenomenon of uninformed opt-out, which we identify and characterize. Indeed, the importance of uninformed opt-out requires a reassessment of the conventional wisdom about Nudge and asymmetric or libertarian paternalism. We also show that different defaults provide different incentives to acquire the information necessary for informed opt-out. With the ballooning use of default rules as a policy tool, our information-costs theory provides valuable guidance to policymakers.