One of the induction week activities for our incoming masters and doctoral students is a faculty-lead discussion on the “Grand Challenges” of Internet research — many of which will be considered and tackled by our students during their time with us. This year we heard from Phil Howard (political science), Vicki Nash (public policy), and Joss Wright (computer science). Below are some of the current research challenges that were discussed, starting with Phil Howard, who opened with:
- The challenge of auditing algorithms, understanding how algorithms and code shape our institutions (for example, in banking, insurance, healthcare, and politics), and how they mitigate and exacerbate social problems. The flip-side of this challenge is examining the social institutions that actually produce the code. There is always some social follow through, and we need qualitative critical inquiry to investigate it — for example, there are political implications for the code used in everyday life. The 2016 US Presidential election represents a moment in global politics in which a significant money is being spent on propaganda and manipulation (e.g. in applying complex datasets to manipulate voters) — and what gets practiced in the US tends to get exported a few years later to other democracies, and then to authoritarian regimes. This is an important research challenge in political science.
- There is also the challenge of bringing this all together — combining the quantitative and the qualitative aspects. Humanists tend to ask the questions that others (in the social and computer sciences) will run with for years to come. We need to learn from humanists, we need feedback from computers scientists, and we need to make our findings generalisable.
- We also need to do all of this ethically. It is easy to write code and to run experiments that manipulate users. In order to understand the effect of automated bots on user knowledge and behaviour (e.g. the Russian bots used to manipulate Brexit voters), it would be possible to write a bot that alerted the public to regime interference. However, this may throw up too many ethical questions about its impact on the public to negotiate comfortably. There would be fewer ethical concerns about a behavioural study that used twitter bots to try to convince people that the world is flat.
The discussion then moved on to how to make academic research applicable to public policy, led by Vicki Nash. She opened by commenting that:
- We need to keep an eye on questions that are policy relevant, and think about how our research might be made usable. While also being aware that there may be a mismatch between the questions policy makers are asking and the ones academics are able to answer. In the case (for example) of how internet use is associated with particular health outcomes: we can certainly give specific examples from individual studies, but not general ones. Has the incidence of child abuse gone up — and could this be related to internet use? It is too difficult to get figures on this, and therefore to comment. What has been the impact of social media on political participation? Well, there are lots of different studies, but a failure to provide clear answers might be related to a lack of clear questions — and a lack of theorisation. For example: what exactly do we mean by “participation”?
- Our work needs to be theoretically informed, and we need to use theories more intelligently. We should also check for gaps in methods and methodological innovation — we might need to use different methods to shed light on old questions.
Finally, the discussion considered the interaction of computer science and the social sciences, led by Joss Wright. He asked:
- How are human rights affected by technology? You can define anonymity very formally and mathematically, but can’t define privacy in the same way — it’s too fuzzy. And this more human element of privacy / anonymity makes an attractive research area. The grand challenge is making explicit the rights and freedoms we have on the Internet, and designing a future society. The technologies that effect us are changing far faster than at any previous time in history; they are also being adopted at breakneck pace, i.e. faster than we can catch up, and this has unseen implications. The technology sector is used to asking “can we build it and make money?” — and dealing with the societal implications later.
- Why do governments choose to filter? By understanding the motivations behind these decisions, we can try to affect these systems. There are clear social outcomes to the protocols we design, but computer scientists by and large don’t tend to (and may even actively refuse to) think about these affects. Policy makers and lawyers will know that they don’t understand the technology so they ask about it — but technologists tend not to know what they don’t know, for example, how the law is made.
- We build things on the Internet, but don’t always consider the implications. Silicon valley has a specific view of right and wrong — but this is not necessarily a model that works elsewhere. We are not always going to find a solution to a problem, but we live in a democracy, and need to be part of that ongoing process.
Additional advice and comments that came up in the Q&A session that followed came to focus more broadly on the research process, including advice on research questions and completing theses:
- Ask the question first — then find the data. You need to analyse as well as measure.
- Find a new way of finding data. There’s lots of (e.g.) Twitter data, but finding the data that are hard to find (and ripping out data that weren’t intended to be ripped out..) is more interesting and less lazy.
- Ask good questions. Have hypotheses. Don’t just use data you can see. Analyse it in all its contextual glory. Make your thesis policy-relevant. Help us understand where we’re going with these technologies.
- The OII was founded to examine the interaction of the Internet and society partly because technologies are developing faster than policy can keep up. So do your research quickly!
- Who do academics think of as their main audience? The formal answer is that the audience is other academics (such as when passing the quality barrier of peer review) — i.e. we wouldn’t be here if academia were not our main focus. But also ask “at what stage is this work ready for other audiences” — present it first to academics, but consider that at some point it may become relevant for policy makers or journalists. Start out by thinking about what is interesting academically — and then push it a bit further. And be aware of the interests of different audiences e.g. research on freedom of expression will not be of much interest to business — but there is a high level of interest elsewhere.
- How do you ensure that something will be of lasting interest — even if a platform you’re studying disappears overnight (and possibly partway through your data collection or analysis..). It is important to generalise — e.g. don’t consider “Twitter”, but “social networks”.
- It’s important to think about how your research fails (rather than how it succeeds). If the effect you are hoping for doesn’t happen — will you still get something useful and productive out of the work?
- Choose a topic and question because it relates to a passion of yours, not just for logical methodology reasons. To get through the sleepless nights and the tedium of analysing data, or understanding why a piece of code doesn’t work— you need to be passionate about the question.
- Don’t lose track of the original question, which is the most important thing. Don’t get lost in procedure.
- Don’t think that because you are taught lots of statistics that you have to use them — qualitative methods are extremely useful too.
- Multidisciplinarity is great — especially if you’re fascinated by everything. The OII is a place where everyone has a high level of interest in what everyone else is doing — which will not always be the same in other places. But you will need to negotiate the terms and vocabularies used in very different ways in different disciplines.
- And finally: the dinner party test! Can you explain the importance of your research to other people?
We hope this “Grand Challenges” session was of interest to our incoming students, and look forward to lots of discussion, great research, and interesting theses to come!