ALGORAND implements a dispute resolution system where any transaction (post) can only be disputed (flagged) once, and when a transaction is disputed, 1000 random users are notified and if they fail to respond within a set time frame (say, 48 hours) their option is forfeit and it goes to another random user. A transaction can only be canceled (removed) if a 60% consensus is reached by the randomized jury. If there is no 60% consensus (even if it's a 599 to 401 split) then the transaction remains unaffected. There is no penalty for simply being disputed, there is no "held pending trial" status. There is a small incentive paid to jury members for their participation and there is an added bonus for voting with the majority if there is a 60% majority (and the votes are hidden from all participants until voting is completed).

The core idea here is RANDOMIZED JURY.

Jury qualification can be based on any number of proposed factors, but should include 90% of active users in order to maintain neutrality and reflect a true "community consensus".

I've discussed this topic a few times and I am more than happy to entertain any and all critiques.