Thinking Big Data in Geography: New Regimes, New Research ed. by Jim Thatcher, Josef Eckert, and Andrew Shears (review)

With the advent of "Big Data" penetrating ever more deeply into the social sciences, this book sets out to explore how the discipline of geography contributes to data science. Big Data presents geography, as well as other disciplines, with a multitude of epistemological, political, technical, and affective questions. Mark Graham's chapter, the last in the book, effectively summarizes some of these questions, particularly the crucial role of digital labor and its often exploitive working conditions.

Graham's chapter provides a gentle pushback against the largely critical tone of most of the chapters ahead of it. As he points out, despite repeated calls throughout the book for hybrid qualitative–quantitative work, and the abiding value of small data, the time has probably arrived when geographers need to be contributing empirically. Jin-Kyu Jung and Jung Yeop Shin's analysis of about 800,000 geocoded tweets in the Seattle area (Chapter 5) attempts to identify spatial patterns of the sentiment expressed about two referenda concerning the legalization of marijuana and marriage equality. Although hardly Big Data (especially given their 3 percent sample), the results show an interesting disparity between public sentiment and voting outcomes; many more people opposed legalization of both marijuana and marriage equality in the voting booth than on Twitter. Emily Fekete (Chapter 7) correlates posts on the Foursquare site with urban demographics by census tract. It would be interesting to repeat her empirical work using Cambridge Analytica–style psychographics (not the business practices), or personal information that scales to the population.

The book is organized into five main parts plus a conclusion. Rob Kitchin and Tracey Lauriault (Chapter 1) note that Big Data today goes well beyond the original "three V's" of volume, velocity, and variety; they add seven other characteristics. The most important is the relational nature of data, which becomes most powerful in the presence of other data. Their own critique centers around how "raw data are always already cooked," listing four concerns, including "anticipatory governance"—in the form of, say, predictive policing (14)—warning against the potential for misrepresentation and harmful decisions when it is based on citizens' data shadows.

The chapters by David O'Sullivan (Chapter 2), Ryan Burns (Chapter 11), and Christopher D. Weidermann, Jennifer N. Swift, and Karen Kemp (Chapter 6) also discuss this potential for harm. As O'Sullivan observes, interpretive dangers lurk within processes that are overly mechanistic or that ascribe static attributes to people without accounting for systemic changes or differences at different scales. Burns focuses on the opportunity for geographers to contribute to digital humanitarianism and pays respects to those who do so contribute. He could have provided more discussion about machine learning for object recognition from imagery (or "GeoAI"), such as Microsoft's recent demonstration, which identified 128 million building footprints in the United States by algorithm. Weidermann, Swift, and Kemp report on a fascinating experiment that revealed users' "geosocial footprint" or their geographical traces across social media. Exposure to these data increases awareness of oversharing insecurities, pointing to how increased transparency can help to shape user behavior.

Britta Ricker (Chapter 4) addresses the theme of reflexivity, or how individuals relate to their own data. As she points out, because data are often unrepresentative of those on the social margins, we need a better understanding of data across the entire stack or assemblage of processes (much of it material in nature). Much of this analysis will be qualitative. Renee Sieber and Matthew Tenney (Chapter 3) make a strong, detailed case for hybrid analyses that move across scales and between big, small, fast, and slow data—a case in which geographers might be able to provide insight, without being accused of special pleading.

Two chapters focus on the urban scene. Jessa Lingel (Chapter 8) provides four short vignettes that highlight the difference between reading the city and...