There are three examples in the article: removing /usr (!!), running "hdfs dfs -rm /data" and "terraform destroy". Even with basic knowledge, these are clearly destructive operations - and to top it all, being run in Production. What strikes me as odd is that in all the cases, there is no change control. And by that I don't necessarily mean a full-on ServiceNow-type change process; I mean there was no "review". Even if you are a small firm, it is not difficult to ensure that you ask atleast one colleague to review any destructive step in Production prior to executing. Ignoring to do so is inviting disaster.