They couldve spent time to understand the problem domain and think more, if theyre not already busy fabricating a resume to sneak into The Actual Big Tech.
It probably made a lot of sense in certain contexts, and certain side effects weren't predicted, or it just has bugs that need fixing.
Presumably nothing malicious or stupid. But just ironing out the kinks.
"Ironing out the kinks" in production on live data is stupid.
Nothing has ever gone wrong in production in ways that were fine on dev?
You've never written code to improve one metric, only to discover in prod it made another metric worse that nobody had considered?
How about a little humility for the ways in which we're all fallible and usually just trying to do our best?