Which makes me wonder if the FAANGs and others will go this route - basically robo-ban people from their services who meet certain criteria (or maybe no criteria at all - maybe some kind of neural network or something decides who should be allowed to use the service and who shouldn't).

Make the ban based on various obvious and non-obvious metrics. Like credit score, comments or whatnot made using the service, how they treated other people or products in their comments - as well as things like how they typed, or moved their mouse around, or other info from their browser (or even side-channel info if that can be done from a browser - which it probably can).

So once banned - even if they tried to go back with a completely different computer, home address, IP, etc - they'd still eventually get flagged as being banned in the past.

Then have a service that all of them use to make a centralized "blacklist".

It wouldn't surprise me if they don't already have this in testing in some manner.