Disentangling Perpetrators from Victims
When someone is accused of moral wrongdoing, they have the right to procedural protection and a fair hearing, and should not be presumed guilty of the act without cause.
When someone has been morally wronged, they have a right to compensation and restitution, and should not be assumed to be lying about the wrong in absence of specific evidence to that effect (e.g., if I say I’ve been robbed, the first instinct of the police shouldn’t be “insurance fraud!” unless there is specific evidence pointing in that direction).
The right of the perpetrators to a fair hearing and due process, and the right of the victim to compensation and restitution, are in tension when they are joined as part of one proceeding. There is a dis-juncture between them — making sure the proceedings are fair probably means some “guilty” will go free, for example. If the only remedy for the victim is punishing the wrongdoer, or if the victim is only seen as deserving of remedy in conjunction with a punishable perpetrator, then this conflict is unavoidable. However, in most cases, one can compensate victims in ways that do not require punishing the perpetrator, and so victims should be seen as deserving compensation via proceedings that lie outside the proceedings used to determine whether the accused can be seen as guilty. In these situations, we should recognize paired obligations: to provide remedy and restitution to the victims so as to erase, as far as possible, the unjust harm they faced, as well as to not label any accused perpetrator as guilty of wrongdoing without proof. Recognizing this fact is crucial in constructing all ethical proceedings — racial and otherwise — in which there is a identifiable victim class which may not be able to muster evidence to “convict” their oppressors sufficient to satisfy our duties to the accused. A moral system grappling with these issues is only adequate when it recognizes and is responsive to both the “perpetrator” (accused) perspective and the “victim” perspective.
Far more analysis at the above-link.