In our conversation, John and I discuss his work on Primer Science, a tool that harvests content uploaded to arxiv, sorts it into natural topics using unsupervised learning, then gives relevant summaries of the activity happening in different innovation areas.