Dr. Solon Barocas
How Data Mining Discriminates
This talk will survey the many ways that machines, drawing lessons from large datasets, can learn to discriminate, even when they have not been instructed to do so, and the pressure this puts on discrimination law and other ways of reasoning about fairness. I will start by tracing discrimination back to five main difficulties in having machines learn from data. I will then explain why attempts to address discrimination stemming from machine learning will be difficult, costly, or controversial.In particular, I will review recent computer science scholarship on discrimination-aware data mining and explore the unsettling finding that attempts to ensure procedural fairness may be in conflict with the imperative to ensure accurate determinations. To conclude, I will argue that machine learning will render increasingly untenable policies that depend on a tidy distinction between procedural fairness and distributive justice.
Dr. Maude Bonenfant
We don’t have to demonstrate anymore that we have entered the digital trace paradigm. The migration of our many daily activities towards the digital landscape causes the massive production of computerized traces, a translation of the world into computer data. This semiotic construction of the trace as data can be regarded as the starting point for the comprehension of the effects of digital traces on individuals, social relationships, collectivity, and society. In this sense, computerized circuits for the automated production of meaning represent the first part of our presentation and will allow us to think about the massive production of traces, also known as the phenomenon of Big Data. Secondly,we will talk about several ethical and epistemological issues that are raised by the question of massive digital traces production. We will put forward the necessity of reevaluating the definition of the trace when we talk about the digital trace in the context of massive digital data production.
Dr. Baki Cakici
Spurred by the possibilities of Big Data, many European National Statistics Institutes are moving away from surveys to new data sources for counting populations. This move requires the implementation of new methods for counting, and these new methods have implications for the counted populations. Specifically, the change in methods involves describing the populations as collections of profiles rather than collections of categories.Based on collaborative ethnographic research by a team of researchers at several European National Statistical Institutes, I analyse the different subjectivities enacted by different methods for counting the population. I describe subjectification through intersecting and interacting modes, which are in turn constituted by situated sociotechnical acts and practices. Focusing on agential acts, I argue that different methods enact different subjectivities, and changes in methods also bring about different subjectivities.
Dr. Sami Coll
Big Data and Privacy: Friends or Enemies?
Big data seems to be the new gold mine of the information era. But what is big data precisely? Is big data a future, an ideal-type or a myth? Although definitions of big data vary, three different perspectives can be distinguished. First, according to the “makers of big data”, that is the engineers and the promoters who develop and sell techniques able to gather, store and analyse data, big data is technically defined as an unstructured data base. Second, from a policy and regulatory perspective, big data is commonly defined with reference to a shifting paradigm involving new knowledge management techniques that allow public services or private companies to make better decisions.Third, there is a growing critical literature expressing concerns about the promises but also the dangers of big data. In this paper, I explore tensions across these competing definitions, focusing especially on brewing conflicts between big data makers and big data regulators. I draw on empirical case studies to show how conflicts—often boiled down to one between big data and privacy—are often resolved with reference to the hermeneutics of risk.
Dr.. Gilbert Émond
Risk Development for Young People in Contact with the HIV World: the Quebec HIV Youth Epidemic or Catching “Small” Events in Big Data Fireworks
Big data are informative but limited about risks. Are they clean, accurate and sense making? For my recent research on the HIV youth epidemic in Quebec I requested the assistance of biostatistician.Good collaboration went on. Making last minute corrections to the abstract, a new number came in, rush rush, no time to ask, we sent it. Preparing final figures, we saw that the usual numbers are off and the youth HIV epidemic was unexpectedly bigger in % but less in number than reported. Big data?… don’t tell why this epidemic is hitting youth particularly hard, or why 18- 35 year olds account for 50% of new infections.
Dr. Martin French
Big Data and the Body-Sensor: Enrolling Bodies into Risk-Sensing Systems
In their article entitled ‘Defining the Sensor Society,’ Mark Andrejevic and Mark Burdon suggest that, in the first instance, the concept of the sensor society refers to ‘a world in which the interactive devices and applications that populate the digital information environment come to double as sensors’ (2015: 20).Riffing on this definition, I emphasize that it is not only interactive devices that have been transformed into sensors by the sensor society. Indeed, the ethos of the sensor society is integral to the development of an emergent ontology of the body, which articulates the body itself as a sensing device. Think, for example, of the ways that human and animal bodies are configured as key material components in integrated, socio-technical circuits of population health surveillance and risk management. Drawing on such examples, I explore some of the implications of this emergent ontology.
Dr. Henning Füller
Managing the Risk of Emergence. The example of ESSENCE syndromic surveillance
Drawing on the implementation of the ESSENCE syndromic surveillance system in the U.S. National Capitol Region, the paper aims to point out truth-effects and epistemological shifts in public health practice related to Big Data.
Considering the discourse of digital health technologies in the National Capital Region as well as its use ‘on the ground‘ in several County health departments, the paper shows how the promise of data-driven detection and early warning is active in reworking public health towards a preemptive rationality. surveillance seems to be the right tool confronting the threat of ‘emerging diseases‘ but it is also establishing this very problem perception. Furthermore, working with this system may lead to a de-qualification of health related truth production and real-time surveillance is re-centering attention and resources towards the proof of the non-event. The paper thus complements critical work on the ongoing securitization of health.
Dr. Marilou Gagnon
Community Viral Load as ‘Big Data’: A Critical Analysis of New Technologies for Tracking, Mapping, and Conceptualizing HIV Risk
Authors: Marilou Gagnon & Adrian Guta (CIHR Post-Doctoral Fellow)
Community Viral Load (CVL) offers a new way of tracking, mapping, and conceptualizing HIV risk by aggregating individual HIV viral loads and changing the way we collect, report, and use laboratory data. CVL maps have been used along with geographic information system technology to demarcate ‘risky’ spaces in the United States and Canada. These CVL maps overlap with other kinds of maps identifying deprivation and spaces where poverty, race, and sexuality have been historically marginalized and policed. Our analysis considers the implications of CVL maps as well as the emergence of CVL as ‘big data’ as its own phenomenon, one that significantly
Much research surrounding the risks of big data and the Internet focuses on the impossibility of maintaining any version of anonymity, privacy, or socio-political freedom while engaging with the web. This fatalistic ideology is risky in itself. It ignores the existence and potential of the Darknet, or Dark Web. Free to use and legal to implement, the Darknet provides users with end-to-end encryption that effectively masks both the content and the user from outside surveillance.Developed and used by the US Government as well as political dissidents, common users with an interest in privacy, public library projects, and social activists, this ‘Internet behind the Internet’ is often misrepresented as a breeding ground or illegal and immoral activity. In fact, the more widely used it is, the more secure it is. It has and will continue to be misrepresented and relatively unknown. Are we content to assert the hardware and software we use will always allow our identity and content to be accessible to governing bodies and commodifying agents? We should not be. The Darknet can not be framed in these terms, and the dissemination of its existence to social analysts and the public should be seen as a crucial project.
Dr. Kelly Hannah-Moffat
Precriminalization and the increased demand for Police Background Checks
There is a vast literature on the impact of the criminal record, but less on the effects of disclosing non-conviction information and/or police contact data (i.e., mental health or domestic violence contacts, suspicious persons alerts, arrests, discharges, acquittals and peace bonds). Although non-conviction records are not at all indications of ‘guilt’ like a conviction, our research shows that the release of non-conviction information has analogous effects as the disclosure of a criminal record. This paper uses Canadian data to examine the consequences of discretionary police disclosures of non-conviction records for individuals and the risks of accommodating such requests (from varied public and private agencies) for police organizations. There is growing capacity of police and private companies to link data sources about individuals which are being mobilized for a variety of non criminal justice decisions. We argue non-conviction and other disclosures raise conceptual and juridical issues about the presumption of innocence, privacy and due process that contradict wider administration of justice objectives.More specifically, we claim that disclosures can undermine the ‘intent’ of peace bonds, diversion sanctions and other efforts to limit the stigma and impact of a criminal sanction on the accused. Second, we argue that non-conviction have social class specific marginalizing costs for individuals at the ‘softer end’ of the criminal justice system and produce precarious public / private assemblages of knowledge.
Dr. Sylvia Kairouz
Dr. Lisa Lynch
Whose data is it now? Ethical norms around the use of data leaks and breaches
As demonstrated this past summer by the heated debates various uses of the leaked data from the Ashley Madison site, it is becoming increasingly important to establish broad ethical guidelines for journalists and researchers who are presented with the opportunity to mine information from data leaks and data breaches. However, this task is challenging for a number of reasons. First, journalists and academics have quite different (and non-monolithic) approaches toward such information use based on their respective institutional norms and histories.Second, there are a series of domestic and international legal norms that sit alongside these institutional norms, at times coming into direct conflict with them (for example, in the case of the prosecution of investigative journalists). This presentation attempts to map out some of the main points of conflict and consensus in this evolving conversation, with reference to several key incidents that have shaped approaches to the use of ‘stolen’ data.
‘Lock this whore up’: public health legislation & other ‘risks’ to public safety
This paper examines the convergence of a range of forms of information about people living with HIV who have not disclosed their HIV status to sex partners. My analysis is grounded in a detailed examination of how public health orders are taken up in media reports, as evidence to inform court judgements, and in the context of psychiatric testimony by experts to classify ‘offenders’ as future risks to ‘public safety’. With a critical inquiry attuned to the social and historical constitution of the legal, I outline how these orders are underpinned by logic of risk mitigation, a logic aimed to protect the ‘public’ through governing the biologically and juridically marked viral underclass: the person with HIV who has come to be known as ‘unwilling’ or ‘unable’ to take the precautions to protect others from HIV transmission. Using examples from two HIV non-disclosure cases in Ontario, this paper argues that public health legislation is one component of a diverse assemblage of technological informational formations of legal and social governance – comprised of public health law and criminal law, as well as civil law and other extra-legal practices – which have come to order the lives of certain classified people with HIV.Examining the use of public health orders under Section 22 of the Ontario Health Promotion and Protection Act (HPPA), I argue that these orders act as the first point of entry into a broader heterogeneous assemblage of legal actors, institutions, mechanisms and practices that act in concert to enable forms of surveillance and governance, constituting something altogether different than the stated benevolent intentions of public health.I thus argue that public health legislation cannot be understood as easily divorced from this assemblage, or understood as a form of jurisprudence that can be applied in a silo.
Dr. Fenwick Mckelvey
Psycho Paths: Problematic Items in the PCL-R.
Dr. Robert Hare’s Psychopathic Checklist Revised (PCL-R) is widely regarded as the “gold standard” instrument in the assessment and measurement of psychopathy. Having demonstrated its usefulness in determining offender recidivism, the PCL-R is increasingly administered to incarcerated offenders, and has even become an intrinsic component of the VRAG (Violent Risk Appraisal Guide), SORAG (Sex Offender Risk Appraisal Guide), and HCR-20 (Historical, Clinical, Risk-20) risk assessment instruments. The PCL-R consists of 20 items, distributed between four facets and two factors. This essay examines these items on an individual basis, drawing from Hare’s descriptions of how each item should be scored.It looks critically at their subjective and objective criteria to highlight the potential for the assessment of these items to result in a score which indicates lower or higher levels of psychopathy than what actually exists. It concludes by i) proposing parameters for future research experiments to determine whether or not confirmation bias related to the subjectivity of many of these items is occurring, and (ii) amendments to the objective criteria to improve the PCL-R’s accuracy.
Dr. André Mondoux
Our presentation will focus on an political economy analysis of the commodification of personal information (Big Data dynamics) in regard to the foucaldian notion of governementality, more specifically the power relations, the truth regiment (knowledge) and the formation of subjectivity.
Encryption As A Risk Management Solution to Big Data Mining
In recent years, it has become apparent that digital communications come with increasingly heightened risks surrounding surveillance and data mining. As Edward Snowden revealed in 2013, the NSA has been collecting personal information from people in 193 countries: resulting in big data mining of 42 billion internet and cell phone records per month. As a means to circumvent this violation of privacy, encryption software renders messages into a series of codes which are nearly mathematically impossible to decipher without the key.In protest to these online surveillance tactics which, according to Amnesty International, 71% of the global population oppose; this research suggests the widespread use of e-mail encryption as a risk management tool that responds to the unwarranted monitoring and surveillance of private personal data by government agencies and corporations.
“Just one more”: Gambling within modern virtual economies
In recent years, the modern video game industry has seen a growing trend in the format of monetization strategies for many online games. Specifically, this monetization comes in the form of gambling within massively multiplayer online (MMO), first-person shooter (FPS), and multiplayer online battle arena (MOBA) games. This is typically presented as the exchange of real-world currency for virtual currency to be used within the game’s ‘cash shop’. Virtual currency is used to purchase ‘in-game’ items – anything from outfits, to vehicles, to weapons, and more. While potentially problematic in itself, the added component of prize boxes introduces an element of gambling into this format. These prize boxes often contain a mix of items which vary in rarity, the most desired items being incredibly difficult to obtain percentage-wise.What we witness through cash shops is the introduction of gambling in virtual gaming – an almost completely unregulated area of our society. In addition, this model has seen incredible success as far as profit for these game developers based on data collected on in-game purchases. This presentation seeks to discuss this form of gambling and the inherent risks associated for the players.
Dr. Joanna Redden
Governments are using big data analysis to respond in real time to social and environmental problems, improve public services, and develop a greater understanding of citizen needs. However to date, there is little research which provides an actual account of how government departments are making use of big data, and how big data is changing the way governments research, prioritize, and act in relation to social issues. An important emerging body of work warns that big data processes may contribute to poverty, inequality, and discrimination.Given the stakes, an investigation of actually existing practices and a consideration of their implications is necessary. But this kind of research comes with its own set of unique challenges. In this paper I discuss findings, implications, and challenges identified through my study of the Government of Canada’s uses of big data.
Professor Gerda Reith
Techno Economic Systems at the ‘New Frontier’ of Mobile and Social Gambling
This paper draws on Castells’ (1996) notion of techno-economic systems to explore the ways that intersections between technology, capital and states have created environments of increasingly intensified consumption. It argues that gambling — and particularly new forms of mobile and social gambling (MSG) — can be seen as a paradigmatic example of this. The technological and commercial drivers of MSG, such as the deployment of geolocational and data tracking technologies, the increasingpersonalisation and targeting of advertising, as well as strategies that work to harness the power of online social networks, bring what I describe as ‘turbo charged’ features to experiences of gambling. They also enact a form of ‘algorithmic control’ through the continuous monitoring and surveillance of online behaviour, in a feedback loop which both produces, and is produced by, the intensification of gambling as within techno economic systems.
Big Data, Ethics and Social Network Gambling: An Examination of Environmental Design and Intention
A surge of unregulated gambling opportunities, referred to as social network gambling have recently emerged on social media platforms such as Facebook. With a minimum age requirement of thirteen to create a profile on Facebook, this inevitably adds a new layer of risk and ethical considerations for public health, particularly with respect to developer’s use of game mechanics and advanced data analytics.This presentation will call attention to potential ethical and risk concerns about game mechanics, and the use of big data to personalize players’ gameplay to help optimize player engagement and monetization. Understanding the culture and environmental context that surrounds gambling opportunities on Facebook is crucial, particularly with respect to the protection and prevention of gambling-related harms to vulnerable populations, such as youth.
Dr. Chantal Robillard
Fairness of the designer God? Players’ perception of disruptive and exemplary behaviours regulation in League of Legends
Rooted in the ethnography and phenomenology of online gameplay, our SSHRC-funded Corporeal Intersubjective Battlefield for the Resistance of Gamers (CIBRG) project examines the subjective experience of players in the use of champions in the context of a massive multiplayer online battle arena game, League of Legends. This paper reports on preliminary findings from that project, framed with reference to the modes of surveillance and regulation of sanctioned (e.g., disruptive) and rewarded (e.g., exemplary) game behaviours aims to identify the surveillance assemblage which relies on multiple forms of data (from big data to subjective evaluations), different configurations of actors (Game company to players), and varying degrees of players’ agency.By distinguishing between these modes of surveillance, this paper allows for a consideration of the ways subjects (or objects) of automated monitoring of their behaviours (big data) perceive, accept or resist these forms of risky gameplay surveillance.
Dr. Lindsay Thomas
Pandemics of the Future: Disease Surveillance in Real Time
This paper considers the rise of global health security as a form of risk management by focusing on the media through which this risk is mediated. Specifically I detail two disease surveillance systems: the World Health Organization’s Global Influenza Surveillance and Response System, which relies on human data input to track and map the speed of diseases; and Google’s Flu Trends, which automatically harvests Google search data to anticipate where disease outbreaks will occur next. Despite their differences, both of these systems are designed to help officials define protocols for real-time, instantaneous response in the event of a pandemic. Yet the idea that we can track diseases instantaneously, in “real time” – that “real time” can obliterate mediation – is an effect of these disease surveillance systems themselves. Rather, real time is always the experience of real time; it is always mediated.Instead of catching us up to the simultaneous present, these systems proliferate and thereby forestall the present, producing many different present versions of the catastrophes for which they seek to prepare. I argue this proliferation of the continually catastrophic present positions future disaster as part of everyday present life.
Dr. Jennifer R. Whitson
Games of Risk: Making sense of data analytics in culture industries
Drawing from ethnographic research, this talk evidences how game developers use big data in a way that is much more complex and counter to accepted definitions of the “value” of big data. Big data in the game industry is marketed as the ability to use player data to create “actionable insights” that help developers re-design games on the fly, increasing player engagement and profitability. However, these analytics services rarely deliver on this promise. In the face of repeated failure, why do independent developers use big data services? In this paper, I argue that big data analytics are important to indie developers not for what they say about games and players, but because they legitimate developers, providing access to exclusive financial networks and social worlds. Big data, in this sense, is less about actionable insight for game developers, and more about risk management for game investors.