Abstract: It has been widely documented that retracted papers are often cited long after they have been retracted. Scientific beliefs, such as in the truth of positive findings that have failed to replicate, likewise tend to persist for a significant period after they have been refuted. In this work, we investigate the dynamics of retraction using a network model. We explore consequences of the following observation: it is typically more conversationally relevant to share new, positive findings than to share a retraction. I.e., there may be symmetries in the way that original claims and retractions spread socially. We explore how factors like the time delay before retraction, length of time with which individuals share novel news, and network structure influence this process.