When thinking about the ethical and political implications of surveillance technologies researchers often suggest that we locate the relevant agency either in the social (users) or in the technical domain (technology). They either claim that the problem is that people tend to use the emerging surveillance technology in a normatively questionable way (thus we simply have to regulate the use of the technology more effectively), or they claim that the design of the technology affords practices that are normatively questionable (thus we need to regulate the design of technology more effectively).   I want to argue that this type of account, although useful at some level, may in fact hide more than it reveals. I would suggest that socio-technical agency is distributed in a much more complex and subtle way.   In opposition to this account of socio-technical agency I will suggest a co-constitutive account of agency based on the work of Martin Heidegger and some of the more recent work of Bruno Latour. I hope to show how such a co-constitutive view of agency will provide us with a better understanding of the social, ethical and political implications of surveillance systems such as facial recognition systems and plagiarism detection systems. I will do this by providing an account of co-constitutive agency through an analysis of plagiarism detection systems. In providing such an analysis I hope to show the particular importance of such an approach particularly for algorithmic based surveillance technologies. I will conclude by proposing disclosive ethics as a possible way forward in dealing with   the ethical and political implications to electronically mediated surveillance practices.

In asking the question “what is it that surveillance does” my concern here is first and foremost with the normative and politics implications of electronically mediated surveillance practices—more specifically with algorithmic based surveillance practices. As more and more organisational processes becomes electronically mediated (or automated) our activities are becoming increasingly recorded as an integral part of the ongoing operation of these systems. This in itself is not remarkable. However, as Zuboff (1996)   notes these traces generate new forms of reflexivity:

“[W]hen information technologies are applied to automate, they simultaneously set into motion an entirely unique set of reflexive processes, translating newly automated activities into data and information for wide-ranging display. Information technologies symbolically render processes, objects, behaviors, and events so that they can be seen, known, and shared in a new way. In other words, these technologies codify and illuminate interior detail, creating transparency where there was opacity, an explicit public text where there was once fragmentation, privacy, and intuition.”    (15)

Moreover, through these new modes of reflexivity new possibilities are imagined and afforded through the plasticity of the technology. Thus, what we are observing, I would claim, is an coincidental—not necessary intended—landscape of emerging opportunities for electronically mediated surveillance practices to be established. The examples of this in the UK are easy to point to. The implementation of a congestion charge system in London that automatically charges vehicles entering the charging zone creates a database of every vehicle (identified through its numberplate) entering in order to charge the owner (identified in the licensing database) for such entry. Inadvertently, as a matter of its standard operation, this transaction also creates a record that locates the movement of the individuals, connected to the vehicle, to a specific time and place. It is easy to imagine that there would be a host of other parties that may declare and interest in this information. We can also add to this the Oyster cards (use of the London Underground) that records every train journey of the card holder. Mobile phone numbers and signals, necessary for the normal operation of the service, now allow parents to locate there children in the form of the ChildLocate service. This process is often referred to as ‘function creep’. In other words new, sometimes unforeseen, modes of action emerge from something designed for a very specific domain of action—such as being able to charge for a service provided. This line of argument, about the implicit—and often necessary—emergence of an multiple surveing opportunities, has been developed by many, most notably David Lyon in his series of books (2001, 2002). I do not want to pursue this further here. For me the key question comes subsequently. If we accept this analysis and we judge it to be normatively problematic then we need to decide what we are to do about it? Some argue that we should regulate better the use of information   technology, for example prevent function creep. Their argument is that the problem is not the affordances   or prohibitions of surveillance technology as such but rather the way these are being seized upon by different actors for the realisation of their intentions or interests—i.e. their use of it. Differently stated: for them the principle agent is the human and human intentionality. Others argue that the problem is the affordances or prohibitions of surveillance technology itself.   For them—as argued by Winner (1980)—the problem is that the affordances or prohibitions of the surveillance technology, intentionally or unwittingly, already embody certain values and interests from the start. They argue that certain affordances or prohibitions should simply not be available, the problem is the technology as such, not merely its use—i.e. for them the principle agent is the intentionality embedded in the technology itself.

I would argue that part of the problem of a debate so conceived is our lack of understanding of what and how technology mediated surveillance practices does what it does, when it does it. More specifically, in trying to deal with technologically mediated surveillance practices, it is difficult for us to locate agency in a straightforward manner and therefore to locate responsibility and accountability in a straightforward manner. The difficulty with this inability to locate agency in a straightforward manner is that we do not know how to go about addressing the normative and political issues that these practices quite evidently raise. If the problem was simply that people tend to use the emerging surveillance technology in a normatively questionable way then we simply have to regulate the use of the technology more effectively. If the problem, on the other hand, is the fact that the design of the technology afforded practices that were normatively questionable then we need to regulate the design of technology more effectively—if this can be done at all.   If however, agency is distributed in more complex and subtle way, as I would suggest (following Latour and Heidegger), then the issue of the politics and ethics of technology is itself distributed in more complex and subtle ways, i.e. not open to simple intervention and correction (such as to regulate the use or to regulate the design). A recent discussion on the surveillance list—surveillance@jiscmail.ac.uk—reflects this problem of accounting for agency, and the implied normative judgement, well. The discussion started as a result of a post indicating a news item about the resistance to speed cameras in the UK by various individuals and groups. Here is an excerpt from that thread of discussion.

David Lyon

Surveillance is not sinister as such; resistance is not laudable as such. The more surveillance studies can get away from knee-jerk paranoid responses to all forms of identification, monitoring, tracking etc. the more everyday responses to surveillance can be analysed in properly political and social-ethical frames, the better.

Roger Clark

I take particular exception to the statement that “Surveillance is not sinister as such”. Surveillance of objects perhaps not; but surveillance of humans has an inevitable and substantial impact on human behaviour. A corollary of the Hawthorne experiments was that humans aren’t the same when they’re observed.   Many of us would call that process ‘de-humanisation’.

Kevin D. Haggerty

Sorry, but I'm in the “Surveillance isn't sinister as such” camp. Surveillance of humans is now too widespread, used for too many purposes - many of which are laudatory, to simply say that 'surveillance is evil.' However, that does not mean that we should avoid normative statements and value judgements. What it means is that we have to be conscious that our normative statements are directed at evaluating particular surveillance practices in the context of a program of governance. We pass judgement not on surveillance tout court but on how it is instantiated at particular times, directed at particular people, overseen by specific organizations all done for concrete ends. Consideration of the interplay of all of these diverse factors have to be part of any normative reaction towards surveillance.

Roger Clark

I'm concerned about the ease with which originally two and subsequently several further members of the list were prepared to treat surveillance as ‘just another technology’, and to invoke that daft old fall-back of ‘technology is neutral; it depends what you do with it’.

What we see in this encounter is a problem of how to account for normative agency in surveillance practices, especially in electronically mediated surveillance practices. Roger Clark suggests that surveillance is not ‘just another technology’ as that would imply that ‘technology is neutral’ (i.e. it has no normative agency as such). For him surveillance technology is a normative agent as such.   Kevin Haggerty suggests “that our normative statements [evaluations] are [should be] directed at evaluating particular surveillance practices in the context of a program of governance... on how it is instantiated at particular times, directed at particular people, overseen by specific organizations all done for concrete ends” His claim is that it really matters how surveillance technology is deployed in its particularity (i.e. the normative agency is located in the way it is deployed or used).   This is obviously a rather crude summary of their positions. Nevertheless, it shows at least the difficulty of locating or accounting for the normative agency of socio-technical systems.

I would claim that without a satisfactory account of socio-technical agency we will not be able to address adequately the normative and political implications of electronically mediated surveillance practices—practices that are becoming so pervasive in our society (to the point that we are starting to talk of our society as a ‘surveillance society’).   More simply put: if we want to challenge, critique or change technologically mediated surveillance practices—normatively that is—then we need to know who (in terms of human and non-human actors) is doing what, when and how, i.e. we need to get a grip on the problem of socio-technical agency.   This is especially true when it comes to the increasing use of algorithmic based ‘automatic’ surveillance practices, which is my particular concern. Obviously it is not just the surveillance scholars who are worried about accounting for socio-technical agency and its normative implications. Many scholars in the STS tradition has written about it, most notably Bruno Latour (1991, 1992,   1995, 1999, 2002, 2005).  

In the paper I would like to explore, in a tentative way, the problem of soci-technical agency in algorithmic surveillance practices by drawing on the work of Heidegger and Latour. I will propose a notion of socio-technical agency, which I will call co-constitutive agency, as a way to overcome the dualistic tendencies that often characterised the debate—as seen above. I will then use plagiarism detection systems as an example of how co-constitutive agency can help us to make sense of the normative implication of electronically mediated surveillance practices. Finally, I will propose disclose ethics as an approach to unpack the normative implications of electronically mediated surveillance practices.

    I am here using ‘affordance’ (and its opposite prohibition) in the sense that Norman (1988) uses it: “...the term affordance refers to the perceived and actual properties of the thing, primarily those fundamental properties that determine just how the thing could possibly be used. [...] Affordances provide strong clues to the operations of things. Plates are for pushing. Knobs are for turning. Slots are for inserting things into. Balls are for throwing or bouncing. When affordances are taken advantage of, the user knows what to do just by looking: no picture, label, or instruction needed.” (p. 9)