[variaga]

Look up "normalization of deviance".

Perversely, a bunch of near-disasters can reduce people's concern and make them less likely to demand fixes because "it did that last time too and everything turned out okay" is a powerful rationalization.

[kunwon1]

A good real-world example of the consequences of this normalization is British Airways flight 5390 [1]

  This problem extended far beyond this one individual, who was merely a symptom. The entire Birmingham maintenance facility, and perhaps British Airways more broadly, had a singular focus on “getting the job done.” If doing the work by the book took longer and jeopardized schedules, then doing the work by the book was discouraged. The shift manager who used the wrong bolts stated in an interview that if he sought out the instructions or used the official parts catalogue on every task, then he would never “get the job done,” as though this was a totally normal and reasonable attitude with which to approach aircraft maintenance. This attitude was in fact normalized on a high level by supervisors who rewarded the employees who most consistently kept planes on schedule. That a serious incident would result from such a culture was inevitable. The shift manager believed it to be reasonable to just “put on whatever bolts came off” and make a quick judgment call about what kind of bolts they were — not because he was personally deficient, but because he had been trained into a culture that didn’t consider this a flagrant safety violation.

[1] https://admiralcloudberg.medium.com/the-near-crash-of-britis...

[tialaramex]

Very few industries are safe enough to actually capture the "That could have been bad" events, that's what ASRS https://asrs.arc.nasa.gov/ does for the Aviation industry (there are equivalent agencies in various other wealthy countries e.g. CHIRP in the UK)

In the absence of a proper means to report "That could have been bad" as you say it can cause normalization. But it's understandable that you don't implement something like ASRS when you haven't solved most of your "That was bad" problems. If you regularly have CI failures due to the code not even compiling, "We need more unit tests" isn't top of the list of your problems.

[toss1]

Yup.

Meanwhile, smart organizations have decades-ago stopped tracking (primarily) "Time-Lost Work Accidents" and replaced that with tracking "Close Calls".

I've seen prominent signs for "N Days Since a Time Lost Accident", and more recently "X Days Since a Close Call".

Sadly, it is so obvious that this CEO clown was doing everything possible to avoid experienced people ("not as inspiring to hire 50yo white guys as hiring young upstarts") so he could overrule any safety or redundancy concerns, firing people as soon as they raised things like "this porthole window is only rated to 1500m and we're going to 4000m", using cheap scrap scaffolding as ballast, and completely ignoring any kind of redundancy in case something went wrong. He seems to have gotten a just end, but his deceived customers didn't deserve that.