The reason is, and always has been, to enable lazy police work. They won't have to do their jobs if facebook will just tell them who the criminals are.
Generally people want all the [other] criminals catching, but do not want all resources of the state used solely for policing. Optimisations then are preferred even if there is collateral harm.
Is it better to put boots on the ground, some of whom may die and who may cause collateral casualties in executing their mission, or a (hypothetically idealized) drone that kills only the target?
Still haven't decided if I have, or there even exists, a good answer to that.
A perfect drone wouldn't do that.
But on the other hand, a perfect drone has no cost of use, other than monetary. Which seems far too low a bar to set for the seriousness of taking a life.
I suppose a better question would be, why do you feel that in this particular situation - a murderer on rampage - you still feel that it is "deeply evil" to kill them in such a way that they "have no way to see it coming", to the point where you'd prefer other people to risk lives to do it in some other way?
Somehow the police managed to catch bad guys before mass surveillance existed. Maybe they should look at that?
This line of thinking has always struck me as extremely odd - as soon as current, presumably E2EE, methods of communication are tapped into for intelligence and law enforcement purposes, criminals can and will switch to projects that don't comply with the backdoor laws. It's a violation of privacy that only law-abiding citizens will be subject to, for everyone else it's optional.
You might catch a wave of them off guard in the beginning, but in the long term all you end up doing is surveilling innocent people and maybe catching lowest hanging fruit - the types of people who are already dumb enough to share their criminal activities on Facebook.