Suffice it to say that any researcher interested in determining just what made the front page in the nineteenth century, as my talk is titled, would have a difficult time knowing where to start. This broad array of genre and topics, however, also illustrates the challenges related to classifying these texts in a meaningful way. On the page of a newspaper and to a human reader, those genres seem obvious, but how can we map those human understandings of textual difference across millions of clusters? What happens when machine models can’t match human models? These questions and associated challenges, the ways I attempted to overcome them, and how I learned that overcoming the challenges might not be the point at all, will make up the bulk of my talk today.