[overu589]

A multi pronged approach may be necessary to achieve your ambition.

- identity. One aspect is establishing an identity. Once an identity is established a probability would be maintained.

- continuity. Always assume someone would convert a human validated identity into a {bot}. Maintain the probability through workflows. Include other presumed humans in the workflow for affirming the continuity of human identities.

- the grind. It seems you anticipate a certain amount of grind a human would go through for vetting, and from this a probability of human would be maintained.

- AI determiners. Are you ruling out the use of agents in rating identities? Algorithms for catching trending behaviors may funnel suspicions to qualified humans.

- accept everything. Reject nothing well formed. If the suspicion probability falls past a threshold simply black hole everything everywhere related to the account. Require a human to battle with admins over account restorations. An access controlled layer of notes and attributes could keep track of all of this.

Inevitably a human feedback loop may accomplish something with time and devotion.

Good luck with that one!