This is a marvelous collection of failures from NASA who takes their failures seriously. Such documentation of failures is akin to Knuth’s recommendation to document bugs.

IBM kept statistics on OS bugs. There were circumstances where fixing a bug introduced 1.2 bugs on the average. I think this was due to ambiguous specifications that led some programmers to expect one thing and others to expect the opposite. One of these programmers would complain to IBM who would change the OS behavior and thus break the other code. The 370 hardware specs were famously unambiguous.