This is a very interesting idea.
Now the reality would be, all these bots and bad actors would still be interacting with you, but you just wouldn't see it by default. Only if you actively add them to your whitelist.
It may be the most free-speech friendly way of dealing with the issue that I have seen presented so far.
As a father of three daughters, I would love to see this idea built.
Discovering who to follow would become even more important under this scheme, though. Maybe we need a recommended follows function that uses WoT to help discover people you might want to add to your whitelist.