@tedu It's always a careful play. Without a doubt, they understand this well. They understand the Streisand effect, but they also control the platform, what surfaces, what people see etc.
So my educated guess is they estimate how effective a ban would be. I wouldn't be surprised if they use ML/AI or something. They already do for curating interests.
Besides, they have a plethora of methods to censor. Shadowbanning, non-chronological manipulated timelines, priming, psychological warning info boxes, link bans, topic bans, temporary bans, intentionally ambiguous ToS.