[samhw]

Yeah, this 'blameless' ethos has definitely trickled down from FAANG to decently-sized decently-reputed places I've worked at - and certainly to #EngTwitter.

I think it's a bit over-applied in some cases. Does it not commit you to the theorem that every process can be made so perfect as to be completely invulnerable to one human being making a mistake? (At least, in the form exemplified by the common tweets to the effect that "your processes are to blame for $incident, not your interns/engineers/etc".)

Even if you required two-person auth for every single thing, two people will make a mistake now and then, and in reality - due to our being social animals - the two probabilities are not truly independent.

I just don't see how this is feasible in reality. A more realistic principle feels like: "people will infrequently make mistakes, and that's of course natural and human and forgiveable, but far fewer incidents should be vulnerable to human error than currently are".

[assbuttbuttass]

I of course agree that mistakes are inevitable. That being said, the point of blameless culture is not to make a process invulnerable to mistakes. Instead during a post-mortem, we look at how to prevent that particular incident from happening again.