Author: Stefania

[BigDataSur] Beyond Touchscreens: The perils of biometric social welfare in lockdown

In the context of COVID-19, what are the perils involved by the perpetuated subordination of social welfare access to biometric identification?

Silvia Masiero

Gradually over the last years, India has introduced biometric identification of users in most of its social welfare schemes. One of the main such schemes is the Public Distribution System (PDS), the nation’s largest food security programme, which provides rationed subsidised commodities to the nation’s poor through a network of ration shops. Biometric access to the PDS is largely operated through Aadhaar, the world’s largest digital identification scheme which provides enrolees with a 12-digit number and capture of biometric credentials (ten fingerprint and iris scans) for recognition. While modalities of identity identification and authentication differ through the country, states that adopted an Aadhaar-enabled PDS require recipients to authenticate through a biometric point-of-sale machine to receive rations.

As a consequence of the COVID-19 crisis, biometric authentication in ration shops has been suspended in several Indian states. A commonly given reason for this is the risk of disease transmission associated to users’ fingerprint contact with the machine, which falls under the broader remit of social distancing measures taken within the ongoing pandemic. The Indian case epitomises, however, a global trend of transition to biometric identification in anti-poverty programmes, programmes that – in the light of very serious effects of the COVID-19 crisis on vulnerable groups – are now more crucial to their recipients than ever. In the context of COVID-19, what are the perils involved by the perpetuated subordination of social welfare access to biometric identification?

The Trade-Offs of Digital Identity

With reference to the use of biometrics in India’s social welfare system, many researchers have highlighted the dichotomy between an anti-leakage rationale and the exclusionary effects yielded by such technologies. Latest in chronological order, Muralidharan et al. (2020) report on a large-scale experiment conducted in Jharkhand, a state where deaths by starvationdue to failed Aadhaar-enabled authentication of PDS beneficiaries were previously reported. The results of the study reveal a 10 percent reduction in benefits for recipients (23 percent of the total) who had not linked Aadhaar credentials to benefit rolls, with 2.8 percent receiving no benefits at all. Such exclusionary effects mirror previous studies of the Aadhaar-based PDS in the same state, with Drèze et al. (2017) reporting, among others, the anxiety brought in poor people’s lives by the uncertainties of biometrically-enabled foodgrain distribution.

The policy vision behind biometric anti-poverty schemes can be summarised in terms of two different types of error being tackled. In targeted welfare schemes, an exclusion error means the exclusion of genuinely entitled subjects, while an inclusion error indicates the erroneous inclusion of non-entitled subjects into provision. By matching biometric records (collected through databases such as Aadhaar) with records of recipients’ entitlements, biometric anti-poverty schemes promise to maximise the affordance of proper targeting, offering credentials to the excluded and preventing access to the erroneously included. This rationale lies at the basis not only of the global proliferation of digital identity schemes, but of their ever-increasing incorporation in anti-poverty programmes, of which the Aadhaar-enabled Indian system constitutes a notable example.

A ration shop in Tumkur, Karnataka, April 2018

But the reality revealed by extant research, including our previous work on the Karnataka PDS, differs from the orthodoxy of good targeting. First, as illustrated most recently by Hundal et al. (2020), requiring biometric identification at the ration shop does not prevent diversion, because the system affords recording successful disbursement even if rations are not provided as per eligibility. Second, there is a trade-off between anti-leakage affordances–in the form of accurate recognition at the point of sale–and the repeated exclusions of entitled beneficiaries, for reasons that range from machines’ malfunctioning to issues of fingerprint readability reported to affect, in particular, the elderly and those in manual labour. In the Aadhaar-enabled PDS, the need for multiple fragile technologies to work at the same time, as highlighted by Jean Drèze, poses a problem of practical feasibility of the system, which is crucial to those parts of the country that are most subjected to infrastructural issues. While inclusion errors are at least in principle targeted by the rationale of biometrics, exclusions keep happening, and put into serious predicament a social welfare system that should cover exactly the most vulnerable groups.

COVID-19: A Reshuffling of Priorities

In the midst of the ongoing crisis, many studies are being conducted on the effects of COVID-19 on health infrastructures and, crucially here, economic vulnerabilities in the Global South. Studies of factory workers, gig-workers and low-income households all point in the same direction: the economic impact brought by national lockdowns is disproportionally affecting the poor and vulnerable, large proportions of whom are recipients of social welfare systems. Where such systems have limited reach or are not available, measures of immediate assistance are invoked, such as the provision of universal basic income or emergency social safety nets. In the Indian PDS, the promise of doubling foodgrain rationsalong with providing extra commodities increases the scheme’s cruciality, in a situation in which new vulnerabilities–such as that of migrant workers exposed to distress and food insecurity–have emerged in the wake of lockdown.

In these times of heightened crisis, severely affecting the users of anti-poverty schemes, the exclusion errors induced by mandatory biometric access are a risk that social protection schemes cannot afford to take. While the incorporation of biometrics is purposefully designed to improve targeting, the crisis poses us in front of the priority of reaching out to the most affected, adapting systems in such a way that biometric recognition–or its digital equivalents–are suspended at the very least. While the problem of touchscreen-induced disease transmission is in itself a valid reason for doing so, the inclusion-exclusion trade-off illustrated here equally poses a problem that needs consideration. As systems are adapted to coping with the COVID-19 crisis, the need for assisting the affected needs to prevail on the stringent adoption of biometric credentials.

Photo: A ration shop in Tumkur, Karnataka, April 2018

 

[BigDataSur] A widening data divide: COVID-19 and the Global South

COVID-19 shows the need for a global alliance of experts who can fast-track the capacity building of developing countries in the business of counting.

Stefania Milan & Emiliano Treré

The COVID-19 pandemic is sweeping the world. First identified in mainland China in December 2019, it has rapidly reached the four corners of the globe, to the point that the only “corona-free” land is reportedly Antarctica. News reports globally are filled with numbers and figures of various kinds. We count the number of tests, we follow the rise of the total individuals who tested positive to the virus, we mourn the dead looking at the daily death toll. These numbers are deeply ingrained in their socio-economic and political geography, as the virus follows distinct diffusion curves, but also because distinct countries and institutions count differently (and often these distinct ways of counting are not even made apparent). What is clear is that what gets counted exists, in both state policies and people’s imaginaries. Numbers affect our ability to care, share empathy, and donate to relief efforts and emergency services. Numbers are the condition of existence of the problem, and of a country or given social reality on the global map of concerns. Yet most countries from the so-called Global South are virtually absent from this number-based narration of the pandemic. Why, and with what consequences?

Data availability and statistical capacity in developing countries

If numbers are the conditions of existence of the COVID-19 problem, we ought to pay attention to the actual (in)ability of many countries in the South to test their population for the virus, and to produce reliable population statistics more in general–let alone to adequately care for them. It is a matter of a “data gap” as well as of data quality, which even in “normal” times hinders the need for “evidence-based policy making, tracking progress and development, and increasing government accountability” (Chen et al., 2013). And while the World Health Organization issues warning about the “dramatic situation” concerning the spread of COVID-19 in the African continent, to name just one of the blind spots of our datasets of the global pandemic, the World Economic Forum calls for “flattening the curve” in developing countries. Progress has been made following the revision of the United Nations’ Millennium Development Goals in 2005, with countries in the Global South have been invited (and supported) to devise National Strategies for the Development of Statistics. Yet, a cursory look at the NYU GovLab’s valuable repository of data collaboratives” addressing the COVID-19 pandemic reveals the virtual absence of data collection and monitoring projects in the South of the emisphere. The next obvious step is the dangerous equation “no data=no problem”. 

Disease and “whiteness”

Epidemiology and pharmacogenetics (i.e. the study of the genetic basis of how people respond to pharmaceuticals), to name but a few amongst the number of concerned life sciences, are largely based on the “inclusion of white/Caucasians in studies and the exclusion of other ethnic groups” (Tutton, 2007). In other words, modeling of disease evolution and the related solutions are based on datasets that take into account primarily–and in fact almost exclusively–the caucasian population. This is a known problem in the field, which derives from the “assumption that a Black person could be thought of as being White”, dismissing specificities and differences. This problem has been linked to the “lack of social theory development, due mainly to the reluctance of epidemiologists to think about social mechanisms (e.g., racial exploitation)” (Muntaner, 1999, p. 121). While COVID-19 represents a slight variation on this trend, having been first identified in China, the problem on the large scale remains. And in times of a health emergency as global as this one, risks to be reinforced and perpetuated.

A succulent market for the industry

In the lack of national testing capacity, the developing world might fall prey to the blooming industry of genetic and disease testing, on the one hand, and of telecom-enabled population monitoring on the other. Private companies might be able to fill the gap left by the state, mapping populations at risk–while however monetizing their data. The case of 23andme is symptomatic of this rise of industry-led testing, which constitutes a double-edge sword. On the one hand, private actors might supply key services that resource-poor or failing states are unable to provide. On the other hand, however, the distorted and often hidden agendas of profit-led players reveals its shortcomings and dangers. If we look at the telecom industry, we note how it has contributed to track disease propagation in a number of health emergencies such as Ebola. And if the global open data community has called for smoother data exchange between the private and the public sector to collectively address the spread of the virus,in the absence of adequate regulatory frameworks in the Global South, for example in the field of privacy and data retention, local authorities might fall prey to outside interventions of dubious nature. 

The populism and racism factors

Lack of reliable numbers to accurately portray the COVID-19 pandemic as it spreads to the Southern hemisphere also offers fertile ground to distorted and malicious narratives mobilized for political reasons. To name just one, it allows populist leaders like Brazil’s Jair Bolsonaro to announce the “return to normality” in the country, dismissing the harsh reality as a collective “hysteria”. In Italy, the ‘fake news’ that migrant populations of African origin would be “immune” to the disease sweeped social media, unleashing racist comments and anti-migrant calls for action. While the same rumor that has reportedly been circulating in the African continent as well and populism has been hitting hard in Western democracies as well, it might be have more dramatic consequences in the more populous countries of the South. In Mexico, left-wing populist president Andrés Manuel López Obrador responded to the coronavirus emergency insisting that Mexicans should “keep living life as usual”. He did not stop his tour in the south of the country and frequently contradicted the advice of public health officials, systematically ignoring social distancing by touching, hugging and kissing his supporters and going as far as considering the pandemic as a plot to derail his presidency. These dangerous comments, assumptions and attitudes are a byproduct of the lack of reliable data and testing that we signal in this article. 

The risk of universalising the problem

Luckily, the long experience and harsh familiarity in coping with disasters, catastrophes and emergencies has also prompted various countries from the Global South to deploy effective measures of containment more quickly than many countries in the Global North. 

In the lack of reliable data from the South, however, modeling the diffusion of the disease might be difficult. The temptation will likely be to ”import” models and “appropriate” predictions from other countries and socio-economic realities, and then base domestic measures and policies on them. “Universalizing” the problem as well as the solutions, as we warned in a 2019 article, is tempting, especially in these times of global uncertainty. Universalizing entails erroneously thinking that the problem manifests itself in exactly the same manner everywhere, disregarding local features to “other” approaches. Coupled with the “whiteness” observed earlier, this gives rise to an explosive cocktail that is likely to create more problems than it solves. 

Beyond the blind spot? 

While many have enough to worry about “at home”, the largest portion of the world population today resides in the so-called Global South, with all the very concrete challenges of the situation. For instance, for a good portion of the 1,3 billion Indian citizens now on lockdown, staying at home might mean starving. How can the global community–open data experts, researchers, life science scholars, digital rights activists, to name but a few–contribute to “fix” the widening data divide that risks severely weakening any local effort to curb the expansion of COVID-19 to populations that are often already at the margins? We argue that the issue at stake here is not simply whether we pump in the much-needed resources or how we collaborate, but it is also a matter of where do we turn the eye–in other words, where we decide to look. COVID-19 will likely make apparents the need of a global alliance of experts of various kinds who, jointly with civil society organizations, can fast-track the capacity building of developing countries in the business of counting. 

This article has been published simultaneously on the the Big Data from the South blog and on Open Movements / Open Democracy.

Cover image credits: Martin Sanchez on Unsplash

Acknowledgements. This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 639379-DATACTIVE; https://data-activism.net).

[BigData Sur] Exploring Facebook’s role in Ethiopia’s rising ethnic tensions

by Syver Petersen

As the world’s most widely used social media platform, Facebook has become a vehicle for extreme political forces and a breeding ground for pernicious stories bent on instigating conflict among groups. In Myanmar for example, an independent United Nations Human Rights Council Fact-Finding Mission found that Facebook played a ‘determining role’ in the recent mass atrocities committed against Rohingya people. Parallel to this, Facebook seeks to expand its userbase in the Global South(s) and, underpinned by modernisation narratives it proclaims its services as supportive of international development agendas.

In Ethiopia, recent ethnic conflict and resulting mass displacement are being linked to social media disinformation and hate speech. This blog reflects on my research on the role of Facebook in this politically polarised and culturally diverse country, with more than 80 languages and ethnic groups and undergoing a historical political transition.

Background

In Africa’s second most populous country, ethnic-based violence has sharply risen in the last couple of years. Since Prime Minister and recent Nobel Peace Prize Laurate Dr. Abiy Ahmed took office in early 2018 Ethiopians have experienced a small ‘political revolution’. The Ethiopian government has gone from using widespread authoritarian practices to releasing political prisoners and journalists, even inviting back previously banned opposition groups.

However, in terms of ethnic relations, political commentators have referred to the opening the political space as taking the lid off a pressure cooker. Mass protest and violence has left more than a thousand dead, and in 2018 close to thee million were displaced due to ethnic conflict and violence, the largest increase in internal displacement globally that year.

Broadly speaking, ethnicity has been and still is one of the most important identities structuring the Ethiopian society. Especially since the fall of the ‘communist’ Derg regime in 1991, ethnic identity has been intentionally emphasised and promoted by the ruling political elites, and the country’s administrational regions were re-organised along ethnic lines. This only further entangled ethnic identity and politics.

The government largely assigns blame to social media for the recent ethnic turmoil. Upon receiving the Nobel Peace Prize, Prime Minster Abiy Ahmed stated that “The evangelists of hate and division are wreaking havoc in our society using social media”. Although social media cannot take full responsibility for the current situation, there is no doubt it plays an important role in shaping political discourse.  

In a context where neighbours kill each other on the basis of ethnic identity, and ethnic tension has the very real potential spin out of control, the potentially conflict-inducing effects of social media is an urgent issue. 

As has been noted in other countries, filter-bubbles, as a result of algorithms personalising online experiences such as the Facebook News Feed, can trap individuals and groups in a state of intellectual isolation. This can reinforce already held viewpoints without challenging them, encouraging partisanship and tribalism. In the highly polarised situation Ethiopia finds itself, I believe this phenomenon merits particular concern. As speech aimed at creating suspicion, spreading fear and encouraging violence, increasingly circulates on Facebook in Ethiopia, Facebook’s personalised algorithmic filtering might further polarise ethnic relations.

Furthermore, in August 2020, Prime Minister Abiy Ahmed plans to hold Ethiopia’s first ‘free and fair’ election since 2005. As the election approaches, the ethnic-based tension and conflict has taken centre stage and the challenge poses a real threat to the country’s political stability. 

Social media use in Ethiopia

Ethiopians using social media are a subset of a subset of another subset–those who have access to electricity, those who have access to the internet and finally those who have accounts on social media. This group is estimated to be only 6 percent of the population. Much of the ethnic-based violence occurs in areas where even electricity is scarce, let alone internet access, like in the case of Guji-Gedeo in Southern Ethiopia, where recent conflicts have displaced close to one million people. Despite this, Ethiopia has one of the world’s fastest growing social media user rates, and social media are becoming increasingly important especially among the massive youth population. Furthermore, through word of mouth, which is still the main source of news for many Ethiopians, social media content appears to reach far beyond the fraction of the population with direct access to social media services. 

My research

As a case for exploring how Facebook influences Ethiopian ethnic-based conflict, I will study its role in recent conflicts between students at the Debre Berhan University, situated in central Ethiopia, about 120 kilometres North-East of the capital Addis Ababa.

Ethiopian universities have long been a hotspot for ethnic riots and violence. In Debre Berhan University, two dormitories were set on fire and a student was killed just off campus last year. Both incidents are suspected to be connected to the wider ethnic conflicts around the country. 

The potentially divisive effects of personalisation algorithms have sparked debates across various scientific disciplines. Some claim that effects are negligible while others highlight them as determining. My study seeks to contribute to this debate by exploring filter-bubble effects in the Facebook News Feed, but also how the students in this specific socio-cultural context relate to and are influenced by their social media information diets.

In order to get access to the student’s personalised Facebook News Feeds, I have asked participants, in their presence, to share access to their social media accounts. I have then used the facebook.tracking.exposed web browser extension– developed as part of the Algorithms Exposed project, a DATACTIVE spin-off– to collect, sort and analyse content from the student’s News Feeds. The idea is to compare the information diets of students from the two main conflicting groups. This will hopefully reveal the extent of filter bubbles, as well as what content participants are actually exposed by the News Feed algorithm. 

As Facebook aggressively expands into new market territory, critical engagement with its context-specific societal effects is pivotal. This is particularly urgent in the context of the fragile Ethiopian political situation, where there is a pressing need for more knowledge about the role of social media in mediating ethnic conflict. 

Syver Petersen

Syver is studying a MSc in International Development Studies at the University of Amsterdam. His academic interests are oriented around how digital technology and big data impact power relations, political engagement and conflicts in the Global South(s). 

Davide in Lugano with paper on algorithms as online discourse (January 30)

Davide Beraldo will be in Lugano, Switzerland, to present a paper on ‘Algorithms as Online Discourse. Exploring topic modeling and network analysis to study algorithmic imaginaries’, co-authored with Massimo Airoldi (Lifestyle Research Center, (EMLYON Business School). The paper is a contribution to the ‘Rethinking Digital Myths. Mediation, narratives and mythopoiesis in the digital age’ workshop hosted at the Università della Svizzera Italiana.

new article out: “Enter the WhatsApper: Reinventing digital activism at the time of chat apps” (First Monday)

Our first article of 2020 is out! Entitled “Enter the WhatsApper: Reinventing digital activism at the time of chat apps”, it reflects on the evolution of political participation and digital activism at the time of chat applications. It is part of a special issue of the open access journal First Monday dedicated to the (first) ten years of WhatsApp. The abstract is below. The article can be read at this link.

This paper investigates how the appropriation of chat apps by social actors is redesigning digital activism and political participation today. To this end, we look at the case of #Unidos Contra o Golpe (United Against the Coup), a WhatsApp “private group” which emerged in 2016 in Florianópolis, Brazil, to oppose the controversial impeachment of the then-president Dilma Rousseff. We argue that a new type of political activist is emerging within and alongside with contemporary movements: the WhatsApper, an individual who uses the chat app intensely to serve her political agenda, leveraging its affordances for political participation. We explore WhatsApp as a discursive opportunity structure and investigate the emergence of a repertoire specific to chat apps. We show how recurrent interaction in the app results into an all-purpose, identity-like sense of connectedness binding social actors together. Diffuse leadership and experimental pluralism emerge as the bare organizing principles of these groups. The paper is based on a qualitative analysis of group interactions and conversations, complemented by semi-structured interviews with group members. It shows how WhatsApp is more than a messaging app for “hanging out” with like-minded people and has come to constitute a key platform for digital activism, in particular in the Global South. DOI: https://doi.org/10.5210/fm.v25i12.10414

Cite as 

Milan, S., & Barbosa, S. (2020). Enter the WhatsApper: Reinventing digital activism at the time of chat apps. First Monday, 25(1). https://doi.org/10.5210/fm.v25i12.10414

***hot off the press*** Working Paper “Big Data from the South: Towards a Research Agenda”

What would datafication look like seen… ‘upside down’? What questions would we ask? What concepts, theories and methods would we embrace or have to devise? These questions were at the core of the two-day immersive research workshop ‘Big Data from the South: Towards a Research Agenda’ (University of Amsterdam, 4-5 December 4-5 2018). The event was the third gathering of the Big Data from the South Initiative (BigDataSur), a research network and program launched in 2017 by Stefania Milan (University of Amsterdam) and Emiliano Treré (Cardiff University).

The workshop report has finally been released and is ready for download! Special thanks go to the workshop participants, the authors of the thematic areas Anna Berti Suman (Tilburg University), Niels ten Never, Guillén Torres, Kersti R. Wissenbach and Zhen Ye (University of Amsterdam), and to Tomás Doods, Jeroen de Vos and Sander van Haperen for the editorial assistance.

We take the opportunity to once again thank the sponsors that made the event possible, namely the European Research Council (grant agreement No 639379-DATACTIVE; https://data-activism.net), the Amsterdam Center for Global Studies, the Amsterdam School of Cultural Analysis and the Amsterdam Center for European Studies. Our gratitude extends also to SPUI25, the University of Amsterdam and Terre Lente for hosting us.

Stefania in Tel Aviv for the workshop “Algorithmic Knowledge in Culture and in the Media” (October 23-25)

On October 23-25, Stefania will be in Tel Aviv to take part in the international workshop “Algorithmic Knowledge in Culture and in the Media” at the Open University of Israel. The invitation-only workshop is organized by Eran Fisher, Anat Ben-David and Norma Musih. Stefania will present a paper on the ALEX project, DATACTIVE’s spin-off, as an experiment into algorithmic knowledge.

Unpacking the Effects of Personalization Algorithms: Experimental Methodologies and Their Ethical Challenges

Stefania Milan, University of Amsterdam

With social media platforms playing an ever-prominent role in today’s public sphere, concerns have been raised by multiple parties regarding the role of personalization algorithms in shaping people’s perception of the world around them. Personalization algorithms are accused of promoting the so-called ‘filter bubble’ (Pariser 2011) and suspected of intensifying political polarization. What’s more, said algorithms are shielded behind trade secrets, which contributes to their technical undecipherability (Pasquale 2015). Against this backdrop, the ALgorithms EXposed (ALEX) project, has set off trying to unpack the effects of personalization algorithms, experimenting with methodologies, software developments, and collaborations with hackers, nongovernmental organizations, and small enterprises. In this presentation, I will reflect on four aspects of the ALEX project as an experiment into algorithmic knowledge, and namely: i) software development, illustrating the working of the browser extensions facebook.tracking.exposed and youtube.tracking.exposed; ii) experimental collaborations within and beyond academia; iii) methodological challenges, including the use of bots; and iv) ethical challenges, in particular the development of data reuse protocols allowing users to volunteer their data for scientific research while individual safeguarding data sovereignty.

two new articles out

We are happy to announce the publication of two new open access contributions: