Service providers are often reluctant to support anonymous access, because this makes it hard to deal with misbehaving users. Anonymous blacklisting and reputation systems can help prevent misbehaving users from causing more damage. However, by the time the user is blocked or has lost reputation, most of the damage has already been done. To help the service provider to recover from abuse by malicious anonymous users, we propose the vote-to-link system. In the vote-to-link system, moderators (rather than a single trusted third party) can cast votes on a user's action if they deem it to be bad. After enough moderators have voted on the action, the service provider can use these votes to link all the actions by the same user within a limited time frame and thus recover from these actions. All the user's actions in other time frames, however, remain unlinkable. To protect the voting moderators from retaliation, we also propose a (less efficient) variant that allows moderators to vote anonymously. We implemented and evaluated both variants to show that they are practical. In particular, we believe this system is suitable to combat malicious Wikipedia editing.