Empathy is something I'm always exercising. It would be unwise of me not to, especially after considering how emotionally volatile humans are.
And no, not strictly by trial and error. I watch others' trials to formulate rational, seemingly-logical conclusions. Then, I start my trial, and if there is error or some abstract miscalculation, then I revise my approaches for future trials. I examine as much as I can before any trial, in other words. Labeling my methods simply "trial and error" makes it seem like I'm inductively acting without any regard for observable patterns.