Algorithmic Life

Calculative Devices in the Age of Big Data

Louise Amoore, Volha Piotukh

• Publisher: Routledge
• Year: 2015
• ISBN:
  9781317527374
• Edition: 1
• Topic: Politica e relazioni internazionali
• Subtopic: Politica

This book critically explores forms and techniques of calculation that emerge with digital computation, and their implications. The contributors demonstrate that digital calculative devices matter beyond their specific functions as they progressively shape, transform and govern all areas of our life. In particular, it addresses such questions as:

• How does the drive to make sense of, and productively use, large amounts of diverse data, inform the development of new calculative devices, logics and techniques?
• How do these devices, logics and techniques affect our capacity to decide and to act?
• How do mundane elements of our physical and virtual existence become data to be analysed and rearranged in complex ensembles of people and things?
• In what ways are conventional notions of public and private, individual and population, certainty and probability, rule and exception transformed and what are the consequences?
• How does the search for 'hidden' connections and patterns change our understanding of social relations and associative life?
• Do contemporary modes of calculation produce new thresholds of calculability and computability, allowing for the improbable or the merely possible to be embraced and acted upon?
• As contemporary approaches to governing uncertain futures seek to anticipate future events, how are calculation and decision engaged anew?

Drawing together different strands of cutting-edge research that is both theoretically sophisticated and empirically rich, this book makes an important contribution to several areas of scholarship, including the emerging social science field of software studies, and will be a vital resource for students and scholars alike.

Part I
Algorithmic life

1
The Public and its Algorithms

Comparing and experimenting with calculated publics

Andreas Birkbak and Hjalmar Bang Carlsen

Introduction

One of the fascinating promises of the World Wide Web (web) is that it seems to hold the key to ordering its own ‘messiness’. This is how Sergey Brin and Larry Page (1998) thought of their Google search engine: helping web users navigate the overwhelming amount of webpages by exploiting the fact that these sites refer to each other by means of hyperlinks, and interpreting this as a recommendation. Based on this logic, every page can be assigned a rank, making it possible to generate hierarchies in the form of Google search results, thereby “bringing order to the web” (Page et al., 1999). To achieve this, web algorithms, such as Google’s PageRank, draw on methodologies found within the social sciences (Rieder, 2012). Sociological methodologies especially, and social science methodologies more generally, seem to be entering into a new relation with digital platforms that re-appropriate research methods to create hierarchies of relevance. And it is not only ‘our methods’ that are being redistributed and re-appropriated digitally (Marres, 2012c), but also political philosophies, that is, methods for envisioning a just social order, as will be argued below.

The idea that the solution to navigational difficulties posed by the web is to be found in the web itself has also been taken up within social research. One prominent example is Bruno Latour’s MACOSPOL project (MApping Controversies in Science for POLitics), where the disorientation associated with the web is to be turned into an opportunity for mapping issues:

Why mapping? It is possible we think, that the same tools, the same media, the technology of the web, which produced this sea of information, which is at first so disorienting, is also the source of a technology which allows us to do the mapping of it. It is exactly the same technology that provided the problem that should also provide the solution to the problem.

(Latour, 2010, video)

What Latour points to here is the way in which, for his purposes, the web simultaneously generates a problem and a possible solution. Latour argues that, instead of adding something external to the web, the key is to be found through the web itself. We identify a similar dynamic of a simultaneous creation of problem and solution with respect to web algorithms, but the other way around: web algorithms position themselves as first and foremost offering solutions, while their corresponding construction of the problem, to which they are the solution, is less explicated. Still, what happens in practice is that Google crawls the web in order to discover and index new websites, with the result that Google searches return thousands of hits, making its hierarchy-generating PageRank algorithm seem indispensable. With devices like Facebook and Twitter, this dynamic is made somewhat more explicit. These so-called ‘social media’ constantly invite their users to ‘be social’ by submitting new content, which results in a stream of posts, tweets and ‘likes’ that no human user can follow in its ‘raw’ version. The work of algorithms seems absolutely necessary to order this ‘mess’ and deliver a useful Facebook newsfeed and point to top Twitter trends.

The focus of this chapter is the algorithmic methods that web services deploy to order their own disorder. Like all ordering methods, web algorithms enact the social in specific ways. In this chapter, we read the calculative devices of Google, Facebook and Twitter as sociologies and as political philosophies. We raise the question of what it means when Google (2014b) claims that “democracy on the web works” and Twitter (2014) says that it “connects the planet to a global conversation”. What kinds of publics are enacted with these omnipresent calculative devices? And how might we distance ourselves from their positioning as self-evident and indispensable?

We examine these questions in three moves. First, we argue that the important task is to clarify, rather than critique, the political philosophies of contemporary calculative devices. We base this argument on a pragmatist understanding of publics as always in need of orientation in uncertain situations, as developed by Lippmann (1925) and Dewey (1927) and discussed further immediately below. Second, we pursue the ambition of problematising the calculative devices of Google, Facebook and Twitter. We do so by simplifying and contrasting the ‘political philosophies’ that can be derived from their algorithmic assumptions, and use these caricatures on a dataset that is not native to any of them. Third, we take advantage of the pluralistic space of calculated publics that has now been deployed in order to think about how it could have been otherwise. More specifically, we propose two alternative calculative approaches as an intervention to supplement existing calculative publics.

The public and its algorithms

Following a pragmatist understanding, publics always need means for orienting themselves. This is what Walter Lippmann (1925) called the “coarse signs”, with which an always busy and ignorant public can find ways to approach an issue. As Dewey (1927) reminds us, one way in which a public might help itself is by appointing public officials to produce such signs that reduce the amount of uncertainty that has to be dealt with. Recognising this pragmatic need is helpful for avoiding a premature critical stance on web algorithms. Examining such algorithms as a new kind of ‘public officials’ means appreciating their value as the producers of coarse signs with which publics can orient themselves. However, such signs are only useful in so far as they are meaningful, which raises the question of what kinds of publics web algorithms assume and produce. Put differently, if web algorithms constitute a new sort of public officials, by what kind of public would these officials be employed? What public would find the signs produced by Google, Facebook or Twitter meaningful, useful and legitimate?

Answering this question amounts to a clarification of the world in which each web algorithm seems self-evident and indispensable. Here, the work of Boltanski and Thévenot (2006) offers valuable guidance by pointing out that public orderings always come with justifications. There is not only a practical need for orderings, there is also a need for being able to make apparent a world in which these orderings can be justified. Discussing contemporary web algorithm in terms of their production of ‘calculated publics’ thus has the advantage of explicating how such algorithms are simultaneously descriptive and prescriptive (see also Rieder, 2012). Drawing on Boltanski and Thévenot (2006), the algorithmic devices of Google, Facebook and Twitter should be examined as offering situated visions of not simply ‘the public’, but ‘the just public’. The reward of such a move, we hold, is a heightened sensitivity to how calculative devices not only explicitly generate the worlds they claim to describe, but also the moral trope from which we are to judge and act on this world, which is a crucial part of understanding the politics of algorithms. By ‘the public’ we simply mean a public that has been filtered and ordered as to produce a legitimate vision of the public. This has importance in so far as such a vision – always situated in a specific ‘world’ – offers the means of navigation in a critical situation, making it possible to identify, consult, oppose or support those that matter in relation to a given issue.

Following Boltanski and Thévenot, soliciting these worlds requires a positive, even deliberately naive, analysis that does not rush in with critiques external to these worlds. Instead, one must attend to internal references to what constitutes ‘truth’ and what it means to contribute to ‘the common good’ in each world. The advantage of such an approach, for us, is that it allows us to ‘politicise’ web algorithms ‘from the inside out’. By taking these devices seriously as ordering practices that care about justifying themselves, we reach a position from which the political philosophies that come with web algorithms can be deployed on their own terms. Against the backdrop of a pluralism of calculative publics, it becomes easier, we hope, to formulate positive alternatives. In other words, we play with the idea that to politicise contemporary web algorithms, one does not have to reveal ‘true’ economic interests hidden behind rhetoric of democracy and human development (Mager, 2012; Vaidhyanathan, 2011). Instead, there is an option of taking this rhetoric seriously, and using it actively to create political differences, which has the advantage of explicating the several distinct ways in which orderings of web data are being done and justified in practice.

Three calculated publics and their consequences

In what follows, first, we combine a reading of the algorithms of Google, Facebook and Twitter with an examination of how they justify their algorithmic orderings. We do so in order to identify the political philosophy, or the vision of the just public, embedded in each device. Second, we caricature these visions in a way that maximises the contrast between them and makes them easy to operationalise. We use these caricatures to illustrate the consequences of each political philosophy on a small dataset of economics research papers selected for the purpose. Illustrating the consequences of calculative devices in this staged way, serves to open a pluralistic field of workable orderings, whose politics do not need to be unveiled or undercut. The point is to avoid the assumption, implicit in a strong critical stance, that ordering can somehow be done without. Instead, we wish to highlight the work it takes to produce orderings, which also means appreciating what a daunting task it would be to replace the orderings of contemporary web algorithms with something else entirely. The advantage of such an appreciation is a more positive description of existing devices that opens for a specification of what descriptive/prescriptive work these devices do.

Google

Google’s mission is to organize the world’s information and make it universally accessible and useful.

(Google, 2014a)

This dual ambition of being both universal and useful poses the challenge of being inclusive and exclusive at the same time, which is indeed a requirement for any just vision of the public. According to Boltanski and Thévenot (2006), any order of worth must be accompanied by an argument for how its hierarchisations benefits everyone. In Google’s case, the attempt to “bring order to the web” is based on the observation that “democracy on the web works”, as we have already mentioned. This is ‘the truth’ that makes the common good of universal and useful information possible, and it is the principle of equivalence that is built into the PageRank algorithm: democracy ‘works’ in the specific sense that running a continuous vote among web sites results in an algorithmic ordering that is both just and useful. In the same way as a survey enacts an opinionated person (Osborne and Rose, 1999), Google can be said to enact an opinionated website. Crucially, this mechanism is not just the tyranny of the majority, because Google’s algorithm assigns different weights to different ‘voters’, as the name ‘PageRank’ indicates. This arguably introduces a conservative flavour in the political philosophy of the algorithm. Older websites have had more chances to accumulate hyperlinks pointing to them, something that Google interprets as a sign of ‘wisdom’ and values with a higher PageRank, that is, a vote that counts more. As such, Google can be said to show a certain respect for the ‘elders’, with arguably positive and negative implications. Google may be granted to ‘bring order’, but this order has been pointed out to come at the price of ‘winner takes all’ effects (Marres, 2012c). We suggest that the caricature of this vision of the just public, which will allow us to operationalise it for the purpose of ordering a dataset of research papers, is the following rule: articles vote for each other through links (citations), and votes from articles that have received many links themselves (that have been cited more), count more.

Facebook

At Facebook (2014), they seek to: “give people the power to share and make the world more open and connected”. Here we also find an indication of a comm...

Table of contents

• Cover
• Title
• Copyright
• CONTENTS
• List of illustrations
• Notes on contributors
• Acknowledgements
• Introduction
• PART I Algorithmic life
• PART II Calculation in the age of big data
• PART III Signal, visualise, calculate
• PART IV Affective devices
• Index