We Are Data: Algorithms and the Making of Our Digital Selves
Cheney-Lippold, John. We Are Data: Algorithms and the Making of Our Digital Selves. United States, NYU Press, 2017.Chapter 0: Introduction
This book "is about how algorithms assemble, and control, our datafied selves and our algorithmic futures," including how data categories about us speak on our behalf (pg. xiii). Pasquale calls this a "black box society," where algorithms have determined the reality of the world often without our knowledge. Our identities, in terms of data identities, have become proprietary to the companies surveilling our activities, even while their predictive outputs determine our presents and our futures. Quoting Bowker and Star, Cheney-Lippold states that "the process of classification itself is a demarcation of power, an organization of knowledge and life that frames the conditions of possibilities of those who are classified" (pg. 7). Categorical structures are "a structuring of the world on terms favorable to the classifier," particularly for market or state purposes (pg. 7). Actual identity is meaningless - what is important is that you act, particularly in some marketable sense, like the category you have been classified as.Oscar Gandy supposed that our identity is formed through interaction with others, so when identity is formed without conscious interaction, Cheney-Lippold proposes we are neither free to develop nor do we know how. The categories operationalized and generated by predictive models are conservative in that they are based on historical data, and therefore they are reifying assessments and decisions of the past. Cheney-Lippold calls these algorithmic caricatures, which are not at all real or reflective of real identity, "measurable types." Data both highlights and marginalizes difference. Only some types of data are made legible to machines (like white skin). Further, our algorithmic identities are not stable; they are subject to change based on our physical conditions (did you buy a lamp so facial recognition now works?) and our digital journeys (becoming interested in a new hobby leading to a change in gender classification). Terranova calls this "network power," where individuals are decomposed into data and are integrated into dividuals; categories of identity, like race and gender (macrostates), then become connected to an array of algorithmic meaning (microstates).
The purpose of this book is to show how data is used to produce new versions of the world, which may be nothing like the physical world. Besides understanding how identities are encoded into measurable types, Cheney-Lippold also focuses on how these classifications reconfigure power.
Chapter Overviews
Chapter 1: Categorization: Making Data Useful focuses on how patterns of data are made meaningful and useful to computers. Categories like 'man' are not social or politically meaningfully; they could be called anything. They are simply data objects that sum up a set of observed patterns believed to be correlated with the social and political identity of man. He repurposes Weber and Goffman's notions of the "ideal type" to define measure type as: a data template of different datafied elements which construct a new interpretation of the world. These templates are used to assign users an algorithmic identification that compares streams of new data to prior data.Chapter 2: Control: Algorithm is Gonna Get You focuses on how our data is used as a form of control (soft biopolitics), through the lens of Deleuze. He discusses how even models with explicit error percentages are viewed as true generalizations of the world. He states that the employment of algorithms which are wrong proves "the swindle of big data's theoryless objectivity. What is useful is what is decided to be useful" (pg. 148). Control here is not about subsuming one's will so much as framing one's world and conditions the possibilities of your life.
Chapter 3: Subjectivity: Who Do They Think You Are? is focused on how soft-biopolitical categories structure our possible actions online and in real life (datafied subject relations). Cheney-Lippold opened by comparing Google's and the NSA's use of information. While Google packages up data for "knowledge, order, and advertising," the NSA does so for "knowledge, order, and state power" (pg. 153). He argues that the enabling constraints of our algorithmic identities "position us within a unique kind of subject relation" (pg. 154). Taking from Mary Flanagan, this is labeled a "combined" subject position "in which virtual systems allow for a 'multiple positioning of the user' that pairs a user's corporeal body with [their] avatar's digital body" (pg. 154). He calls the logics behind such subjectivities relational, as we are located and relocating according to our data patterns. We are not made subject but made "as if" subject. As he has already argued, this is because data are representations that are not actually subjects. The NSA, for example, "created an entirely new conception of citizen and thus actively reframed how we can thank of ourselves, as users, as subject to state power online" (pg. 157). He labels this mode of citizenship "jus algoritmi ... a formal citizenship ... that is not theorized as ordained and stable" (pg. 157). This differs from jus sanguinis (to be a citizen is that one's parents were citizens) and jus soli (to be a citizen is one is born within the borders of a state). In discussing Althusser's theory of interprellation, he writes that algorithms rarely hail people directly; rather, they structure conditions of possibilities and access, and so making a subject is more indirect.
Chapter 4: Privacy: Wanted Dead or Alive focuses on the history of the concept of privacy and how privacy might be translated to an algorithmic world. He says that we need to develop a dividual privacy that extends beyond individual bodies. Cheney-Lippold recommends Nissenbaum's "obfuscation" for overwhelming algorithms with data to make them more inaccurate and less able to track us.
