Ex 24.6 I think it’s easy to see why program inspections are successful, because peer review is a cornerstone of checking new discoveries and building quality knowledge. Simpler than that big idea is just my personal experience going back and looking at my own code for errors, especially after I sleep on it. I haven’t spent as much time checking other programmers’ code, but I would probably be able to find even more errors just because I’m less familiar with what they wrote and therefore have fewer blind spots and bad habits.
I would think that program inspections would not catch system errors that emerge when the program is plugged into a complex network of interactions. There’s often no way to predict the exponentially complex interactions in a whole system or especially a system of systems just by looking at the program code. Potentially taking a step back and reviewing system architecture and such via UML would help, but often such systemic errors can only be found through actual testing.