I look at stories like this and a key moment of failure that is obvious to anyone that has ever deployed code is don't make a change and roll it out to 100% of all devices/servers immediately. Feels like there is just some basic things missing from folks brains that gradual release and validation of the impacted cohort isn't a built in instinct for us.
yeah, they should have absolutely piloted this approach, look at the results, re-avaluated or fixed things, try again, before making it the absolute new policy
