Methods and apparatus associate a computed trust level to avatars that interact with one another in a simulated environment. The avatars may represent legitimate users of the virtual world or spammers. System monitoring of each avatar provides ability to recognize potential spammers and create an alternate indication of the spammers. A user index may be used to store data describing attributes of each avatar for analysis using programs stored in memory.