The present paper addresses the problem of literary history as a problem of data comprehensiveness and selection, seeking not to resolve the impossibility of literary historical narrative, but to reframe it through a computational perspective. Our focus is on the Quan Tang shi 全唐詩 (Complete Tang poetry), the massive comprehensive anthology of Tang poetry that was produced at the height of the Qing dynasty (1644–1912). The sheer quantity of Tang poetry preserved in the Quan Tang shi (over 50,000 poems and poem fragments) exceeds the human-scale perspectives of close reading. To make sense of the corpus as a whole, we will show how two related forms of distant reading — topic modeling and divergence measures — allow us to reframe and rethink these literary historical questions and provide a new perspective on what it means to read Tang poetry.