I always feel that debugging is much more of a science than normal software development. You need to spot unusual behavior, figure out ways to isolate and repr

I always feel that debugging is much more of a science than normal software development. You need to spot unusual behavior, figure out ways to isolate and reproduce it, come up with hypotheses as to what might be happening, design tests for those hypotheses, run the tests, find the solution, test the solution -- exactly like the interplay of experimental and theoretical science.