Are you familiar with the book "On the Art and Craft of Doing Science"? It makes some good points for normalizing discussing dead-ends, failed experiments, etc. with a "CV of failures" and has good examples of how hypotheses and methods evolve before publications