It's hard to argue against the concepts of self-describing data, contextual interfaces, and richer metadata for content that eventually will make up the Semantic Web. However, it is easy to imagine semantic environments suffering from the same challenges that many content management system implementations and the Web itself suffer from: the preoccupation with data could easily leave us drowning in it. We focus on approaches being explored to promote feedback and user involvement for the maintenance of semantic representations, to ensure they remain useful and current.