This chapter by Oscar H. Gandy reads like a cigarette warning label that does not yet exist: “We are caught in the matrix of information we provide in our everyday use of the Internet, this ‘transaction-generated information’ (TGI) is used in an unregulated fashion, and the goals of private companies will win out over the pubic good. Be careful.” But more than this, much like the impact of cigarettes where the combination of lack of health resources with unhealthy behaviors accumulate, the already disadvantaged may be even more severely affected by potentially predatory use of personal information. These conclusions may seem to the uninitiated an unjustifiably dystopian picture of systems that use the information we leave when we surf the web to ‘personalize’ targeted information. But what some may see as beneficial and convenient (“Hey, Google knows what I’m looking for!”) may have serious consequences for disadvantaged segments of the Internet-using population.
In his chapter, Gandy outlines several issues arising from the use of what he calls ‘advanced analytics’ including problems associated with identification and classification, private interests and social networking sites (SNS), agents and issues of trust, and discrimination and cumulative disadvantage. He finishes with a discussion of how we might regulate the use of TGI in such a way so as to better protect our most vulnerable populations.
In his first section on identification and classification, Gandy describes the “data mining” process of deriving knowledge from patterns and relationships provided by TGI and explains that the use of data mining has “expanded quite dramatically” since September 11, 2001—but his focus for the chapter is on market segmentation and targeting. In this subset of data mining, he cautions that research has shown that some forms of TGI “might be characterized as personally identifiable information (PII) even though it may not include the user’s name or email address,” stressing potential problems with surveillance and privacy issues (131). However, he does note that not all use of TGI for segmentation and targeting lead to “commercial exploitation, marketplace discrimination, and cumulative disadvantage” (133). Here, Gandy makes the distinction between ‘corrupt segmentation’ where the content and vehicle for providing information to audiences segmented by race, class, gender, and sexual orientation are determined by the potential for profit rather than ‘authentic segmentation’ in which the “scale and intensity of identifiable need” of users takes precedence (134).
He then moves briefly to social network sites (SNS) and the extent to which these sites have been able to meet identifiable needs such as a the necessity for increased political participation among disenfranchised groups. In this section, Gandy describes Dara Byrne’s (2007) work on the cultivation of a politically active Black identity on sites like BlackPlanet.com (which is, importantly, Black-targeted but not Black-owned), finding little evidence that SNS activity promotes political engagement. Gandy stresses that private companies and marketing typically wins out over other goals.
The third section of Gandy’s chapter discusses agents and issues of trust, including ‘infomediaries’ and personalized search. Infomedaries are companies that gather information provided by Internet ad services for the purpose of classification. This process typically occurs without the knowledge or consent of users, yet companies create detailed profiles. Gandy explains that companies treat personal information as a “commodity that can be packaged and sold” and assume that users will get something in return, but may not (136). For example, Gandy cites the Viswanathan et al (2005) finding that portal users may pay higher amounts in spite of a lower “product fit” (137). And a less savvy consumers are likely to be more vulnerable. The same story is true of personalized searches where profit motive rules; Gandy says that companies are non-neutral entities. He argues,
Those who pay the piper are more likely to be governed by concerns about the needs and interests of their owners and investors than they are about the needs and interests of their customers or clients (139).
In other words, these infomediaries are not to be trusted.
In the fourth section Gandy describes discrimination and cumulative disadvantage in the context of the use of advanced analytics. I believe is the crux of Gandy’s argument and wanted it to more explicitly anchor the chapter. His point is simple, but important: The most disadvantaged will be the losers in this scenario. Those with lower levels of education, limitations on technological access, lower levels of skill, and other points of disadvantage will be less likely to benefit from the Internet, particularly if private companies to focus Internet-based resources on more advantaged populations, and are more likely to be susceptible to predatory action. But that’s not the only area where disadvantaged populations are at risk. Connecting the case of the Internet to the larger body of research on cumulative disadvantage, Gandy explains how advanced analytics, targeted marketing, and lack of legal protections impacts many facets of life. For example, Gandy points out that the “target[ed] marketing of predatory loans to minority communities was at the heart of a rapidly spreading cancer that placed the global economy at risk” (140).
In the final section, Gandy suggests some ways that this problem can be addressed. Although he advocates formal restrictions of and laws against marketplace discrimination, he does not see that as an area in which change will easily occur because of “the wake of massive investments in the management of public opinion and its representation to members of Congress” (141). This, he says, largely leaves the responsibility for protection to the individual in a manner that ‘blames the victim’ who are typically unaware that there is even a cause for concern. He notes that in the interim, we will have to rely on the technological advocacy work of such groups as Howe and Nissenbaum (2009) who created a Firefox extension that “helps protect web searchers from surveillance and data-profiling by search engines…[through] noise and obfuscation… [A]ctual web searches…are essentially hidden in plain view” (as described in their website, http://cs.nyu.edu/trackmenot/). Although Gandy says it is likely marketers and search engines will develop “countermeasures” he hopes (as do I) that they “attract an army of open-source imitators who will help them improve this defensive technology” (142).
Overall, Gandy’s chapter on the Internet-based use of advanced analytics adds to an established body of work on quantification and the process of making groups ‘legible’ (or illegible) by the state and, often, private companies (e.g., Igo 2007, Espeland and Sauder 2007). Even more importantly, it adds to a growing body of work on the role of technology in cumulative disadvantage. For example, a study by Calder, Lleras-Muney, and Vogl (2008) on health inequality found that the negative effects of educational disadvantage on health could be largely explained by a lower ability to “process new information and take advantage of new technologies” (35). The myriad types of ever-growing inequalities (health, political, economic, etc.) need to be even more central in our discussion of the impact of new technologies, as do conversations about what we can do about it.
Calder, D., A. Lleras-Muney, and T. Vogl. 2008. "Socio-economic Status and Health: Dimensions and Mechanisms, Working Paper 14333." Cambridge, MA: National Bureau of Economic Research (NBER).
Espeland, Wendy and Michael Sauder. 2007. “Rankings and Reactivity: How Public Measures Recreate Social Worlds” American Journal of Sociology, 113 (1): 1-40.
Igo, Sarah. 2007. The Averaged American: Surveys, Citizens, and the Making of a Mass Public. Cambridge, MA: Harvard University Press.