One of the more provocative conversations I’ve heard at CSCW was a panel called Computational Social Science: CSCW in the Social Media Era, organized by Munmun De Choudhury with Eric Gilbert, Brian Keegan, Mor Naaman* and Hannah Wallach as panelists. ‘Computational social science” is a term that I hear a lot, sometimes bewilderingly in the context of my own work. Rather than a play-by-play live blogging (live tumbling?), I’m doing nearly-live tumbling, fleshing out some of the not-at-all comprehensive notes I took and occasionally adding my own responses. The panel was structured around some key questions, as well as some questions that surfaced from the audience.
What is computational social science?
Brian Keegan says it’s a set of methods for interrogating social science questions, with a core belief that social behaviors are encoded in databases and other digital artifacts and systems. It’s important to note that often times graduate programs don’t necessarily support this kind of work.
How does computational social science relate to CSCW?
Asked specifically in the context of being a first year attendee to CSCW, Hannah Wallach talked about the spectrum of computational social science between computing and social interactions. CSCW brings together people where there’s an overlap in both techniques and data sources with fields like sociology and political science, but has there been sufficient discussion from disciplinary communities? So it’s less in the overlap in techniques, but the similarities of data.
Is computational social science in the service of social science?
Mor says, we’re not here to study society, we’re here (citing Mark Ackerman) to bridge the gap between the socially desirable and the technologically feasible. We’re not here to answer questions about is the world more polarized, we’re here to design systems, to address societal needs through design interventions. For Mor, the issue becomes what’s the question behind the project? Is it to explain human behavior or look at a technological system out in the world?
Hannah pointed out that Mor’s definition privileges computer science over social science, so computation in the service of whom? Social science or computer science?
Mor says that in this community we’re doing computational social systems, we’re looking at behavioral data, and the questions we’re investigating can now use a new set of tools, a new set of systems. Throughout the panel, Mor returns to the notion that particularly at CSCW but also more broadly, we should do “computation social systems” that should really be about the study and design of systems. It’s unclear to me, though, how this body of research would relate to other conference based collectives, like the iconference’s annual Consortium for the Science of Sociotechnical Systems or ASIST’s social informatics SIG.
What are the big open questions?
Hannah noted that given that we’re still trying to figure out what computational social science is, it’s hard to figure out what the big questions are. Doesn’t want to stress methods as a key question.
Munmun notes that part of figuring out what computational social science means is to consider what the contribution of a project will be,
Brian notes that social science was meant as a provocation, a way for quantitative folks to colonize social science, and much of this work is in fact quantitative. How can we develop quantitative social science that can speak to more qualitative questions?
How do computer scientists and social scientists collaborate?
I used this question to raise the issue of the tendency to totalize social science, where a particular pet peeve of mine is the way that computer scientists (and others!) will refer to social scientists as a coherent referent, nevermind that that lumps together disciplines as disparate as sociology, anthropology, communication, economics. This does a disservice to the complexity of the disciplinary boundaries at stake and also betrays a borderline insulting lack of awareness of what it is that people in these fields do.
Libby raised the awesome point that when we call data easily available, that is a big assumption that assumes that APIs are available or usable. Social scientists might as easily say to computer scientists looking to explain human behavior, theory is easily available, just go out and read** it.
Mor says it is the responsibility of information scientists, if not specifically CSCW, to develop tools for computational social systems work, even as social scientists should accept the work of learning what is possible computationally.
Given concerns about mass surveillance of datasets, what are the implications for computational social science?
Raising the question of ethics of comp ss in the context of NSA disclosure, Brian noted that other disciplines have confronted these issues, like physicists after nuclear weapons, biologists in the 70s after recombinant DNA. Munmun noted that doing research on datasets that involve historically marginalized groups can generate deeply troubling ethical concerns.
I voiced something that’s been on my mind for a while, which is that one of the weirdly welcome consequences of the NSA disclosures is the validation of paranoia and an increased literacy among the general public about issues and technologies of surveillance. To Munmun’s point, I noted that many of these issues have already been raised in ethnographic literature, so it’s important to reach out to other disciplines that have experience and advice with these issues of methodological ethics. Beyond turning to our academic colleagues, I also pushed for reaching out to activists and community members themselves to adjudicate issues of membership.
*Statement of disclosure: Mor was my PhD advisor, and he rocks.