The study of mechanism design is sometimes criticized, because the designed mechanisms depend on the fine detail of the model specification, and agents' behavior relies on the strong common knowledge assumptions on their rationality and others. Hence, the study of 'detail-free' mechanism design with weak informational assumptions is the most important to make as the first step towards a practically useful theory. This paper will emphasize that even if we confine our attentions to detail-free mechanisms with week rationality, there still exist a plenty of scope for development of new ideas on how to design a mechanism to play the powerful role. We briefly explain my recent works on this line, and argue that the use of stochastic decision works much in large exchange economics, and agents' moral preferences can drastically improve implementability of social choice functions.