Renormalization & Self-Consistency

Recently, I'm struggling between two methods in solving a research problem. One is called renormalization, in laymen terms, it means to find a particular viewing angle so that one can make your problem 'looks' simplier. Renormalization involves forgetting some irrelevant details of the orginal problem, and of coz, 'irrelevant' is relative to your target research interest. The second method is called self-consistency, naively it means first, assume your problem belongs to a certain simple class of problems that can be solved in standard way, then see where it leads to. In the case where the conclusion is obviously not 'consisent' with the assumption, change certain parameters in the assumption until self-consistency is achieved. A few days ago, I realized these are also common methods for people to deal with real life siuations. Yet, one cannot go back and renew your decisions in the past, so in real life 'self-consistency' is not possible, but its cousin, 'self-deception' is popularly used. Yet, people who use the method of 'self-deception' always think they're just using the method 'self-consistency'! My boss wanted to stick to the method of self-consistency, and I want very much to follow the method of renormalization. If we really understand the problem well, we should be able to renormalize it! Maybe my boss also wants some 'self-deception' to easy the frustration in research !