They could make a warning/inquiry first, allowing a dialogue to explain a murky case before damage is done by a 'unilateral' enforcement action. (Pretty sure that's what grandparent post meant.)
Just take a look at the number of spammers on Twitter, or (recently) Instagram. It's not even possible to open up a dialogue for every single instance of possible spam.
Dude, you are missing completely the point. There's a vast difference between automatically dealing with spam in Twitter and shutting down a paying customer with legitimate (non-drone account) followers.
And this is not about spam, it's about TOS violation, itself much more subjective than just spam.
I suspect you've never been on the other side of this equation.