[megameter]

J Paul Morrison "Flow-based Programming", while not really about documenting historical methods, got my head thinking about programming in terms of the pre-CPU paradigms of data processing with unit record machines, since that is what this style draws on most strongly.

Along the same lines, while not a book, videos and documents about debugging electromechanical pinballs and modular analog synthesizers provide great reference points for what digital systems could look like or strive to emulate, and physical engineering, woodworking and construction decenter the computer as the whole of the process while still being very logical and systematic. At the top level it's usually finding the right structural metaphor that dogs practical programming and gets it into a loop of endless data collection and reprocessing, so it helps to have other things in mind and try to discern ultimate aims.