Other thing that may take into consideration is the use of bots to gain trust, like someone receive positive trust from new participants and those same participants give trust between them.
That is an important thing to consider. My initial thoughts would be to limit it so one account can only leave one rating on another account, and have some minimum standard that much be reached before a new account is permitted to leave feedback. We can also add a Radon cost, to make botting financially unfeasible.
Thoughts?
Minimum account ages and one rating from one account to another would be good. However, users should be able to swap a previous rating from positive to negative and vice versa. If an account is marked as compromised, the trust influence of that account should automatically go to zero.