For a more detailed description of the project of Blumenstock and colleagues, see chapter 3 of this book.
Gleick (2011) provides a historical overview of changes in humanity’s ability to collect, store, transmit, and process information.
For an introduction to the digital age that focuses on potential harms, such as privacy violations, see Abelson, Ledeen, and Lewis (2008) and Mayer-Schönberger (2009). For an introduction to the digital age that focuses on opportunities, see Mayer-Schönberger and Cukier (2013).
For more about firms mixing experimentation into routine practice, see Manzi (2012), and for more about firms tracking behavior in the physical world, see Levy and Baracas (2017).
Digital age systems can be both instruments and objects of study. For example, you might want to use social media to measure public opinion or you might want to understand the impact of social media on public opinion. In one case, the digital system serves as an instrument that helps you do new measurement. In the other case, the digital system is the object of study. For more on this distinction, see Sandvig and Hargittai (2015).
For more on research design in the social sciences, see King, Keohane, and Verba (1994), Singleton and Straits (2009), and Khan and Fisher (2013).
Donoho (2015) describes data science as the activities of people learning from data, and it offers a history of data science, tracing the intellectual origins of the field to scholars such as Tukey, Cleveland, Chambers, and Breiman.
For a series of first-person reports about conducting social research in the digital age, see Hargittai and Sandvig (2015).
For more about mixing readymade and custommade data, see Groves (2011).
For more about failure of “anonymization,” see chapter 6 of this book. The same general technique that Blumenstock and colleagues used to infer people’s wealth can also be used to infer potentially sensitive personal attributes, including sexual orientation, ethnicity, religious and political views, and use of addictive substances (Kosinski, Stillwell, and Graepel 2013).