Data scientists have confronted a gap between their desire to improve society and the harmful impacts of their creations. While algorithms once appeared to be valuable tools that advance social justice, today they appear to be dangerous tools that exacerbate inequality and oppression. In this talk, I argue that improving society with algorithms requires transforming data science from a formal methodology focused on mathematical models into a practical methodology for addressing real-world social problems. The current approach to data science—which I call “algorithmic formalism”—focuses on the formal mathematical attributes of algorithms. As a result, even when data scientists follow disciplinary standards of rigor, their attempts to improve society often entrench injustice. In response to these limitations, I introduce “algorithmic realism,” a data science methodology that designs and evaluates algorithms with a focus on real-world impacts. Algorithmic realism expands the data science pipeline, providing concrete strategies for how data scientists can promote social justice. This new methodology also suggests more socially beneficial directions for the burgeoning fields of data science ethics, algorithmic fairness, and human-centered data science.
Censorship studies is a vibrant academic and activist field, producing an important amount of publications that document cases of Internet censorship or shutdowns. It uses mainly network measurements methodologies and tools which rely on the so-called probes (basically, servers connected to a network which scan it and send multiple requests to specific URLs in order to confirm if they are accessible or blocked). Installed on a virtual private server in the country of interest (say, Iran or Russia), these probes pretend to imitate online human behavior. Other approaches consist in packaging the probe on end-user devices (e.g. smartphones) and run tests from there. However, in both cases something important is missing: the real user experience of censorship in its complexity. This presentation focuses on the case of Russia and its "Sovereign Runet" project, that aims at cutting Russian networks from the global net. Based on an approach to censorship as an experience, and in a multidisciplinary approach, this presentation suggests to chase the user and demonstrate how real experience differs from what is seen on the network measurement graphics.
The conference leaves ample time for questions and discussions after each presentation and between sessions.
2:00 pm: Invited talk: Algorithmic Realism: Data Science Practices to Promote Social Justice • Ben Green
2:00 pm: Invited talk: A censored user: how censorship studies ignore user experience • Ksenia Ermoshina
5:30 pm: Panel Session. What sense can we make of computer science research in the Anthropocene? • Panel chair: Sophie Quinton. Offline.
3:20 pm: Panel Session. Critical research and the place of the scientist in the polis • Panel chair: Chantal Enguehard. Offline.
We organise two panel sessions at the end of the day on Tuesday and Wednesday. They are offline exclusively.
As reflected by the numerous submissions on the topic to the conference, many think that today's computer science research is inadequate given the gigantic challenges posed by the Anthropocene. In this panel, we will first discuss why computer science research for the Anthropocene can be considered as undone science. We will then address the many difficulties that appear when looking for what such a research could look like. We will focus in particular on how the Anthropocene challenges some commonly shared assumptions amongst researchers such as the neutrality of science. The objective of the panel is not to reach a consensus, but rather to share experiences, knowledge and questions as well as confront ideas in a respectful and constructive setting.
Bibliography:
Critical research, familiar to the fields of the human sciences, is not so common in computer science. However, some computer scientists have become aware of the undesirable effects of digital technology. During this panel session, we will introduce the notion of critical research and then address the dangers and difficulties of interactions with media, activists and political powers.
Bibliography:
Online user: 2 | Privacy |