I feel like I've experienced similar situations with post mortem and sprint retrospective processes. Sometimes shit goes wrong in a sort of unique way and people go too far with wanting to have the outcome always be "we will do X to make sure Y never happens again", even though X is more costly than Y in terms of engineering effort and business value. So I think this is also applicable to software design.