The early 2010s saw the rise of “big data” and its handmaiden “data science,” novel expertise promised to “coax treasure out of messy, unstructured data,” turning the exhaust of our information age into the “raw material” of our “data-centered economy.” By the mid-2010s, the “data scientist” job title had topped many Best Jobs lists. For years, however, despite myriad openings and legions of candidates, there were comparatively few data scientists hired. What does it take to do data science, and who gets recruited? For the stakes involved, answers to these questions were fiercely contested. In particular, the question that occupied career Q&A websites for much of the past decade was: Does a good data scientist need a PhD? And if so, why?