I read a lot of technical books that I forget within a month. Every so often one rearranges the furniture in my head and stays there. Sidney Dekker's "The Field Guide to Understanding Human Error" is one of those, and I've recommended it to more people this year than anything else.
It isn't about computers at all. Dekker comes from aviation safety, and the book is about why "human error" is a useless conclusion rather than a useful one. When you say someone made a mistake and stop there, you've closed the investigation at the exact point it should have started. The interesting question isn't who pressed the wrong button. It's why pressing the wrong button made sense to a competent person at the time, given what they could see and the pressure they were under.
That reframing has changed how I run post-mortems. I used to land, gently, on what someone should have done differently. Now I try to reconstruct the situation as it actually looked from the inside, the confusing dashboard, the alert that cried wolf, the runbook that was three steps out of date. The fix is almost never "be more careful". It's a guardrail, a clearer signal, a system that makes the right action the easy one.
The cliché it killed for me is "the system worked, the human failed". Usually the human was the last line of a system that had been quietly setting them up to fail for months. Worth the few evenings it takes to read.