social network analysis

[Séminaire #ecnEHESS] L’idéologie du machine learning avec Bernie Hogan (Oxford), 4 déc. 2017, 17h

Enseignement ouvert aux auditeurs libres. Pour s’inscrire, merci de renseigner le formulaire.

Après une première séance riche et passionnante, notre séminaire EHESS Etudier les cultures du numérique revient avec un invité de marque. Le 4 décembre, nous aurons l’honneur d’accueillir Bernie Hogan, Senior Research Fellow du Oxford Internet Institute. Les travaux de Hogan portent sur comment les plateformes numériques gèrent l’identité et les données personnelles de leurs usagers, ainsi que sur les modalités de représentation des mondes sociaux à travers les réseaux. Par-delà ses articles publiés dans des revues savantes (Field Methods, Social Networks, New Media and Society) et ses interventions régulières dans les grands médias internationaux (BBC, Reuters, Times), il a participé au développement de plusieurs logiciels de visualisation de réseaux sociaux, dont le dernier en date est Network Canvas.

ATTENTION : Le séminaire aura lieu le lundi 4 décembre 2017, de 17 h à 20 h. Au vu du nombre important (et toujours croissant !) de participants, l’EHESS nous a accordé l’amphithéâtre François-Furet, au 105 bd Raspail 75006 Paris.


Title: Control, Privacy, Generativity: Big Data after the Ideology of Machine Learning

Speaker: Bernie Hogan (Oxford Internet Institute)

Abstract: This talk will unpack the three competing tensions for platforms in the age of big data. Users expect autonomy and privacy, platforms expect control and third parties expect privacy. It is not plausible to optimise for all three. Yet, in the absence of regulations for third parties and platforms, as well as public ignorance of important technical details, platforms will seek ever more control. The purpose of control is foster increasing strategies for predictability and monetization. What is lost when control is gained has been called “generativity” or the capacity to foster innovation and imagination through technology. Social network visualization will be presented as a case study in how both the user and third parties lose when we are left to view data solely through the eyes of the platform’s preferred interface.


La présentation et les débats se dérouleront en anglais.

Les nouveaux défis de l’analyse des réseaux (The Conversation, 2 novembre 2017)

Le 5 et 6 décembre 2017, le colloque Recent Ethical Challenges in Social Network Analysis (RECSNA17) aura lieu à Paris : une rencontre internationale et interdisciplinaire que je coorganise avec une équipe de chercheurs français et anglais. Les inscriptions sont gratuites et ouvertes jusqu’au 30 novembre.

Pour lancer le débat et pour énoncer les thématiques qui seront traitées lors du colloque, le site web The Conversation a publié une tribune que j’ai coécrit avec Paola Tubaro.

Il faut repenser l’éthique de la recherche des réseaux sociaux

(…)

“Depuis longtemps, les universitaires s’interrogent sur ces difficultés éthiques : déjà en 2005 la revue Social Networks dédiait un numéro à ces questions. Les dilemmes des chercheur.e.s sont exacerbés aujourd’hui par la disponibilité accrue de données relationnelles collectées et exploitées par les géants du numérique comme Facebook ou Google. Des problèmes nouveaux surgissent dès lors que les frontières entre sphères « publique » et « privée » se brouillent. Dans quelle mesure a-t-on besoin d’un consentement pour accéder aux messages qu’une personne envoie à ses contacts, à ses « retweets », ou à ses « j’aime » sur les murs des amis ?

Les sources d’information sont souvent la propriété d’entreprises commerciales, et les algorithmes que celles-ci utilisent biaisent les observations. Par exemple, un contact créé par un usager de sa propre initiative, et un contact créé sous le conseil d’un système de recommandation automatisé peuvent-ils être interprétés de la même manière ? Bref, les données ne parlent pas par elles-mêmes, et il faut s’interroger sur les conditions de leur usage et sur les modalités de leur production, avant de penser à leur traitement. Ces dimensions sont profondément influencées par les choix économiques et techniques ainsi que par les architectures logicielles imposées par les plateformes.

Mais une réelle négociation entre les chercheur.e.s (surtout dans le secteur public) et les plateformes (parfois émanant de grandes entreprises multinationales) est-elle possible ? L’accès aux données propriétaires ne risque-t-il pas d’être freiné, ou inégalement distribué (de manière potentiellement pénalisante pour la recherche publique, surtout quand elle est mal alignée avec les objectifs et les priorités des investisseurs) ?

D’autres problèmes surgissent dans la mesure où un.e chercheur.e peut même avoir recours à du crowdsourcing payé pour produire des données, en utilisant des plateformes comme Amazon Mechanical Turk pour demander à des foules de répondre à un questionnaire, ou même de télécharger leurs listes de contacts en ligne. Mais ces services mettent en cause d’anciens acquis en termes de conditions de travail et appropriation de son produit. L’incertitude qui en résulte entrave des recherches pourtant susceptibles d’avoir des retombées positives sur la connaissance et la société au sens large.

Les possibilités de détournement des résultats de la recherche pour des finalités politiques ou économiques sont multipliées par la disponibilité d’outils de communication et publication en ligne, que les chercheur.e.s sont désormais nombreux à saisir. Si l’intérêt des milieux militaires et policiers pour l’analyse des réseaux sociaux est bien connu (Osama Ben Laden aurait été localisé et neutralisé suite à l’application de principes d’analyse des réseaux sociaux), ces appropriations sont plus fréquentes aujourd’hui, et moins facilement contrôlables par les chercheur.e.s. Un risque non négligeable est l’usage de ces principes pour réprimer des mouvements civiques et démocratiques.”

[Read more]

Nouvelles d’ANAMIA : conférences, vidéos et outils de visualisation de données

Je suis actuellement à Québec (plus précisément sur le campus de l’Université Laval) où j’ai été invité à présenter ANAMIA, notre projet ANR sur les communautés anorexiques et boulimiques du Web, dans le cadre du méga-congrès francophone canadien ACFAS. Que Georges Canguilhem ne m’en veuille pas trop, l’intitulé de ma présentation est Le normal et le parfait. Rapport au médical et émergence de normes corporelles au sein des communautés anorexiques du Web (lundi 6, 15h30, salle 3850 du Pavillon Alexandre-Vachon, colloque du CELAT, session “Corps et médias : énonciation, négociation, contestation et réaffirmation, présidée par Madeleine Pastinelli).

Par ailleurs, avec les autres membres du projet, nous venons de lancer en ligne la série de Conférences ANAMIA : des vidéos et des slides de présentations de membres de notre équipe de recherche et de spécialistes français apportant un éclairage sociologique, historique, psychologique sur les Web des troubles alimentaires et sur le phénomène pro-ana. La première vidéo est celle de La minceur, obsession ou danger, conférence de l’historien Georges Vigarello (directeur d’études à l’EHESS et auteur, entre autres, de La silhouette, 2012 ; Les métamorphoses du gras, 2010 ; Histoire de la beauté, 2004). Le montage a été réalisé par Argyro Paouri, de la cellule audiovisuelle du CEM IIAC CNRS/EHESS.

VigarelloANAMIA[Conférence ANAMIA] Georges Vigarello « La minceur n’est pas une obsession exclusivement moderne »

Pour terminer, un teaser de quelque chose sur laquelle nous avons travaillé ces derniers mois avec le très talentueux designer Quentin Bréant : un tools de visualisation des données collectées dans le cadre de nos enquêtes sur les utilisateurs de sites Web liés aux troubles alimentaires en France et au Royaume. Nous allons faire une présentation live sur le site Web du projet ANAMIA prochainement. Entre temps (et sans autre explication) voilà une petite galerie… question de vous donner un avant-goût.

Visualisations des données de l’enquête ANAMIA en France et au Royaume-Uni

Please Your Holiness, feed the #troll (an open letter to the Pope)

[Update Feb. 11 2013: This post has been also published in the French edition of the Huffingtonpost and featured in Andrew Sullivan’s The Dish as well as in the Catholic Herald.]

> To : Pope Benedict XVI
> From : Antonio Casilli
> Sent:  Sun Feb. 3 2013 03:52:22 PM
> Subject: What to do about Twitter trolling
>

Your Holiness,

let me start by saying that I am not a christian, plus I am not particularly appreciative of your work. I am but a modest scholar of digital cultures who has been following, with a professional eye, your recent effort to rebrand your online image. By now, the general public is aware that you and your staff operate the Twitter account @pontifex – and its multilingual permutations @pontifex_fr, pontifex_it, pontifex_es… My sources indicate that this is the brainchild of Jesuit cybertheologian Antonio Spadaro, counselor of the Pontifical Council for Social Communications 1. So it seems only obvious that the Jesuit magazine Popoli commissioned a study to assess how well your online presence has been perceived after a month on Twitter. On the face of it, you did fine. You have been sending out approximately 100 messages in 9 languages, and earned more than 2 million followers altogether. Moreover, you have generated 270,456 responses from your fellow users.

This impressive amount of comments was also used to perform a “sentiment analysis”, to determine the general attitude of the Twittersphere. About 82% of the feedback received was “neutral”, a meager 10% was positive, and 8% negative. Let me break it down for you, Your Holiness: sentiment-wise, your entrance on Twitter has been saluted by a roaring “meh”… The not so brilliant results are summarized in the following infographic:

The Pope on Twitter. Source: Oogo.com (more…)

  1. Turns out my sources were not completely accurate after all. On Twitter, jesuit theologian Antonio Spadaro clarifies that:

Slides du séminaire EHESS d'Antonio Casilli “Contre l’hypothèse de la fin de la vie privée” (20 nov. 2012)

La première séance de mon séminaire EHESS Étudier les cultures du numérique : approches théoriques et empiriques pour cette année universitaire a eu lieu le mardi 20 novembre 2012 à l’EHESS. Merci à tou(te)s les participant(e)s pour leur présence, leurs commentaires et leur enthousiasme. Voilà les slides de ma présentation.

TITRE : “Contre l’hypothèse de la ‘fin de la vie privée’ sur les médias sociaux : négociabilité et cyclicité de la privacy”

RESUME : “Au sein de la communauté internationale plusieurs voix se lèvent pour dénoncer l’érosion inexorable de la vie privée dans le  contexte des usages actuels du Web social. En s’adonnant à une surveillance mutuelle et participative, les internautes renoncent-ils volontairement à la protection de leurs données personnelles ? Cette intervention adopte une approche ethno-computationnelle des controverses relatives aux politiques de négociation des paramètres de confidentialité en ligne pour montrer que la vie privée a encore de beaux jours devant elle. Sous certaines conditions, des « cycles de privacy » se mettent en place. Au travers du travail des associations d’usagers et des organismes préposés à la défense de leurs droits, ces conditions peuvent être remplies.”

Lectures :

Susan B. Barnes (2006) A privacy paradox: Social networking in the United States, First Monday, 11 (9)

danah boyd (2008) Facebook’s Privacy Trainwreck: Exposure, Invasion, and Social Convergence, Convergence, 14 (1): 13-20

danah boyd & Eszter Hargittai (2010) Facebook privacy settings: Who cares?, First Monday, 15 (8)

Anders Albrechtslund (2008) Online Social Networking as Participatory Surveillance, First Monday, 13 (3)

(more…)

The academic, the wikipedian, the vandal [Full version, updated 05.11.2012]

The French translation of this essay is available on OWNI (part 1 and part 2), as installments of my column Addicted To Bad Ideas.

With the new academic year kicking in, my colleagues and I have decided to add a little wiki twist to a couple of courses we teach at Telecom ParisTech. I started a Wikispace for my digital culture class, and with Isabelle Garron and Valérie Beaudouin we’ve made compulsory for first year students to try and edit and discuss at least one Wikipedia page, as part of their initiation to online writing.

Sure, Wikipedia has been used as teaching tool in academia for some years now, to say nothing about its increasing popularity as a research topic. But the main rationale for using it in the classroom is that it has become the one-stop-shop for bibliographical research and fact-checking.

Challenging the Academic Mindframe

Think about your own online information habits. What do you do when you don’t know the first thing about a given topic? You probably google it, and the first occurrence is most likely a page from Jimbo Wales’s brainchild. You do it, we do it, our students do it. So we have to incorporate Wikipedia in our academic activity, not because it’s a cool gadget, but because otherwise it will create a dangerous blind spot.

[Don’t panic… Ok, panic]

And yet, admitting to this without panicking is not simple. At least here in continental Europe, ill informed judgments about the allegedly poor quality of Wikipedia articles are still commonplace in higher education. Some – like the French high-school teacher Loys Bonod, who had his 15 minutes of fame earlier this year – go as far as to add false and misleading information to Wikipedia, just to demonstrate to their students that it… contains false and misleading information.

Such paradoxical reactions are a case in point. Wikipedia is just as accurate and insightful as its contributions. Hence, the need to encourage its users to relinquish their passive stance and participate, by writing about and discussing relevant topics. Of course, one might say, when it comes to Wikipedia the Internet iron law of 90–9–1 participation applies: for 90 simple readers of any article, there will be only 9 who will make the effort to click on the “modify” tab to actually write something in it, and maybe just 1 motivated enough to click on the “discussion” tab and start a dialogue with other wikipedians.

Social scientists can come up with many explanations for this situation. The claims about the dawn of online participatory culture might have been largely exaggerated. Or maybe the encyclopaedic form tends to recreate cultural dynamics that are more coherent with an “author vs. reader” dichotomy than with many-to-many communication. Or maybe Wikipedia editors tend to intimidate other users in an effort to increase their own social status by implementing specific barriers to entry.

Try starting a new article. In all probability, its relevance will be challenged by some editor. Try starting the biography of a living public figure. Chances are that a discussion will ensue, focussing not on the public figure in question, but on the private qualities of the biographer. Is the author just an IP-based anonymous, or a legit logged-in user with a recognized contribution track record?

(more…)

Censorship and social media: some background information

[Update July 27, 2012: so far, our study has been featured in a number of media outlets in UK, India, Algeria, US, Oman, Indonesia… These are just the ones we know of: The Daily Mail, Yahoo Lifestyle, CNN, Technorati, The Times of India, GigaOM, Buzzfeed, National Affairs, Sify News, Phys.org, Science Daily, Zee News TV India, Oman Tribune, The Free Library, L’atelier, Sciencenewsline, Le Soir d’Algérie, Tempo Indonesia. We’re particularly impressed by this response, and would like to thank the researchers, journalists and activists who’ve been spreading the news.]

Hello everyone,

You have probably reached this page after reading in the international press about our study “Social Media Censorship in Times of Political Unrest – A Social Simulation Experiment with the UK Riots” (published in the journal Bulletin of Sociological Methodology, vol. 115, n. 1). This post will provide some background information.

Read the study

First of all, if you are interested in reading the paper, you can purchase the article from SAGE website. Anyhow, here’s a preprint version you can download for free. Just saying.

About the authors

If you are looking for the authors’ bios:

 Antonio A. Casilli, is an associate professor of Digital Humanities at Telecom ParisTech and a researcher in sociology at the Edgar Morin Centre (EHESS), Paris, France. He is the author of the social media theory book Les liaisons numériques [Digital Relationships], published by the Editions du Seuil. He blogs at Bodyspacesociety.eu, tweets as @bodyspacesoc, and is a regular commentator for Radio France Culture. You can contact him here.

 Paola Tubaro, is a senior lecturer in Economic Sociology at the Business School of the University of Greenwich, London, UK, and associate researcher at the Centre Maurice Halbwachs (CNRS) Paris, France. Economic sociologist with interest in social networks and their impact on markets, organisations, consumer choice and health, her research also includes work in the philosophy and methodology of economics and social science. Her blog is here, plus you can contact her here.

The story, so far

In the wake of the August 2011 UK uprisings, Casilli and Tubaro built a rapid response study. Using computer simulation, the investigators showed that any move by the government to censor social media was likely to result in more civil unrest, higher levels of violence, and shorter periods of social peace. Released as a joint post on their websites and subsequently available as a working paper on SSRN (Social Science Research Network), the study was widely shared online and in the press.

Such an enthusiastic response prompted them to continue their research. Presently, they are launching follow-ups and new developments, both empirical and theoretical, in other European and MENA countries. They are members of the scientific committee of Just-In-Time Sociology (JITSO), an EPFL Geneva-based program gathering international researchers that try “to understand social phenomena as they unfold”.

TEDx talk, simulations and other stuff

If you want to watch a video presentation of the study, here’s Antonio Casilli’s TEDx talk (in French, with English subtitles), “Studying censorship via social simulation”, TEDx Paris Universités, May 19, 2012.

If you want to know more about our ongoing research, Internet Censorship and Civil Unrest (ICCU), here’s the project’s wiki.

If you want to download the computer simulation, here you’ll find a detailed technical description of the model. The model file (Netlogo and Java applet versions) is available here . You should: 1) unzip and save all three files in the same directory; 2) either open the .nlogo file from your computer in Netlogo, or open the .html file in your browser).

Enjoy!

Trollarchy in the UK: the British Defamation Bill and the delusion of the public sphere

[UPDATE 26.06.2102: A French version of this post is now available on the news website OWNI. As usual, thanks to Guillaume Ledit for translating it.]

These days, the House of Commons has been debating an amendment to the British Defamation Bill specificially designed to tackle Internet trolls. Now website owners and internet access providers will be forced to reveal the IP and personal information of users identified as authors of ‘vile messages’. It is business as usual: whenever some ICT-related news story catches the public eye, British policy makers come up with an ad hoc law. Preferably, one mindlessly disregarding privacy and free speech.

Why mainstream media are scared of trolls

In a remarkable effort to lull the general public in a false sense of understanding digital cultures, The Guardian has devoted a special session of its June 12, 2012 edition to this peculiar online phenomenon. The pièce de résistance is Zoe Williams’s What is an internet troll?. An article concocted using the usual troll news story recipe: one part pyschology professor delivering highbrow quotes about the ‘disinhibition effect’ of electronic media, one part journalist whining about today’s diminishing education standards and pervasive hate speech, two parts sad anecdotes about some celebrities we’re supposed to sympathize with. The conclusion of this tone-setting essay (“We shouldn’t call them ‘trolls’. We should call them rude people.”) is probably best rendered when pronounced with a high-pitched monty pythonesque voice, like in The Life of Brian‘s “He’s not the Messiah. He’s a very naughty boy!”.

(more…)

Would online censorship be effective? Evidence from two research projects proves the opposite

Hi,
you’ve probably reached this blog after listening to my interview with Jian Ghomeshi on CBC Radio Canada’s programme Q. In case you missed it, here’s the podcast:

In this post, you’ll find some background information about my ongoing research on internet censorship – mainly in collaboration with Paola Tubaro (University of Greenwich, UK) and other colleagues. Our focus is on unintended and negative effects of censorship, based on analyses of social media use conducted in the last few years.

In my latest book Les liaisons numériques. Vers une nouvelle sociabilité? [Digital Relationships. Towards a New Sociability?, Paris, Seuil, 2010] I dealt with the topic of pro-ana (short for “pro-anorexia”) and pro-mia (“pro-bulimia”) websites, blogs and forums of persons with eating disorders. The most controversial among them have gone as far as to claim that eating disorders are a choice or a lifestyle, rather than conditions. A grant from the French National Research Agency (ANR) allowed me and my colleagues to lauch ANAMIA, a large-scale study on eating disorder-oriented online communities.

ANAMIA research project – featured on Boing Boing

Since the early 2000s, fears that these websites may induce unhealthy behaviours (possibly in young and adolescent viewers), have prompted many web services to remove them, while some countries have considered outlawing them. Yet eating-disorder related Web communities continue to proliferate. They have migrated to more hidden platforms, barred entry to outsiders, concealed their true nature, and relocated in foreign countries. In a previous post published on Bodyspacesociety blog, I have dubbed this the “toothpaste tube effect“: squeezed from one service, controversial contents re-group elsewhere. Paradoxically, censorship multiplies these websites – if only because of the urge to duplicate contents for backup purposes, in case they have to shut down and move!


Mapping pro-ED websites (France, 2010-2012) – ANAMIA research project

Today, these websites are less open and less visible, though still numerous and densely connected with one another. Thus, they can still influence their users, just as before; but it has become harder for health and nutrition campaigns to locate them and reach out to their users.

Our results indicate that Internet censorship is ineffective and inefficient: it has failed to stop “negative” influences, and has made it more difficult for “positive” influences to operate.

(more…)

Banning pro-ana websites? Not a good idea, as Web censorship might have a 'toothpaste tube effect'

[Update 05.04.13: A longer version of this post, with revised results, has evolved into a full-fledged article published by the UK Royal Society. To cite the article: A. A. Casilli, F. Pailler, P. Tubaro (2013). Online networks of eating disorder websites: why censoring pro ana might be a bad idea, Perspectives in Public Health , vol. 133, n.2, p. 94 95. As part of our research project ANAMIA (Ana-mia Sociability: an Online/Offline Social Networks Approach to Eating Disorders), the post has been featured in a number of media venues, including The Economist, Libération, Le Monde, Boing Boing, The Huffingtonpost, CBC Radio Canada, DRadio Wissen, Voice of Russia.]

Tumblr, Pinterest and the toothpaste tube

On February 23rd, 2012 Tumblr announced its decision to turn the screw on self-harm blogs: suicide, mutilation and most prominently thinspiration – i.e. the ritualized exchange of images and quotes meant to inspire readers to be thin. This cultural practice is distinctive of the pro-ana (anorexia nervosa), pro-mia (bulimia) and pro-ED (eating disorders) groups online: blogs, forums, and communities created by people suffering from eating-related conditions, who display a proactive stance and critically abide by medical advice.

A righteous limitation of harmful contents or just another way to avoid liability by marginalizing a stigmatized subculture? Whatever your opinion, it might not come as a surprise that the disbanded pro-ana Tumblr bloggers are regrouping elsewhere. Of all places, they are surfacing on Pinterest, the up-and-coming photo-sharing site. Here’s how Sociology in Focus relates the news: (more…)