Retreating from conferences

Picture it. You’ve been working on your presentation for a big conference for months. Last night, insomnia took hold and you were running through your slides in a sweat. You’re there, ready to present, and you look around the room and you see a sparse sea of faces, most of whom look tired or disinterested. It’s the last session of the day, so it’s understandable. A professor in the corner is tapping away distractedly on her laptop, trying to clear her emails. You sprint through your talk in the allotted twelve and half minutes, desperately trying to get across the essence of your project, your baby. A cursory round of applause follows. A couple of half-hearted questions are followed by “not a question, more of a comment”, lengthily detailing some irrelevant work in a knowing tone.

You feel deflated. What’s the point of these things anyway? What did you gain from this? Perhaps all this energy could have been better spent writing, or starting a new project, or learning new techniques. The next day, however, in the coffee break, the tapping professor approaches you and thanks you for your interesting talk, and asks if you would consider extending the work in another direction which you hadn’t previously considered. A helpful bystander suggests a data-source, and proposes collaboration since it ties in with his work. Perhaps conferences are not so bad after all!

This is a caricature, of course, and not all conferences are like this, but perhaps there is enough truth to it to lead us to think more carefully about academic gatherings and how they should be structured. The Rostock Retreat on visualisation, which I attended back in June 2017, experimented with a different format, aiming to cut out the formulaic stiffness of many big academic conferences, and make the whole event more like an extended, but super-productive version of the coffee break.

Visualisation genres

The retreat did involve a few set-piece presentations, however, and for good reason. Visualisation is a tool for getting across our message. The way in which the message is delivered, however, depends crucially on the audience, and modern academics have to be able to make their pitch to everyone from practitioners to the general public. The retreat recognised this problem, and featured a selection of keynote speakers with visualisation stories from very different genres.

We forget how easy we have it in the modern era. Data analysis software liberates the analyst from the hard technical labour of producing high quality plots. But some of the most beautiful and influential visualisations were produced by hand. Michael Friendly‘s keynote talk gave a dive through history, starting in the ‘Golden Age’ of visualisation in the second half of the nineteenth century, and surfacing to detail his own working visualising the statistical properties of data in a geometric fashion.

Dino Citraro from Periscopic, a data visualisation company based in Portland Oregon, provided a useful counterpoint to the academic experience of data viz, which usually entails working out how to present and prettify your latest finding, in a field you have been working in for years. In contrast, for Periscopic, the visualisation workflow involves receiving a new brief from a client, learning about the subject area, brainstorming ideas and finally producing a striking and elegant graphic that efficiently conveys the desired message. This gun crime visualisation is a moving example of how to make people care about your message through data.

When done badly, however, data visualisation can have the opposite effect from what was intended. James Cheshire, author of two books aimed at communicating data and scientific results to the general public, delivered us from evil by alerting us to the Seven Sins of Visualisation. Often, academics treat presentation of results as a sort of afterthought and are happy with voluminous tables and ugly, unclear, and poorly labelled graphs. This does your research a disservice, and is a discourtesy to your readers.

Finally, Anika Rasner from the German Federal Chancellery showed us how data viz can play an important role in communication between government and its citizens. In the age of ‘fake news’ and declining trust in government, transparency and simplicity are important in building relationships with those that the government represents. The Federal Government report on Wellbeing displays both of these qualities, and is visually very engaging and appealing.

Retreating together

Apart from the keynotes, the retreat featured poster sessions, and a ‘slam’ event in which participants had just a few minutes to present speculative and imaginative visualisation that might not generally be at home in academic context. But perhaps the most productive and thought-provoking element of the retreat was the group work. The participants were divided into four groups and assigned a data-set, with the task of reporting back after a couple of sessions high-intensity data-munging and graphing. The results were impressive, as can be seen from some of the contributions on the @RostockRetreat twitter feed. Most importantly, however, it provided a real opportunity for attendees to learn from each other in an actual live analysis environment. Talking through a task and actually working together on it with colleagues reveals more about their thought process and approach to tackling a problem than can be revealed in a 15 minute presentation.

This geofaceted graphic, created during the visualisation challenge by a team including Nikola Sander, Ilya Kashnitsky, Jorge Cimentada, Corina Huisman, Michael Boissenneault, and Juan Galeano, examines years of lost life by cause in Mexico.

Hacking and manipulating

I took two things from the Retreat. The first is that alternatives to the traditional conference format for academic meetings in demography have a lot to recommend them. Smaller workshops of course have long been popular, but perhaps experiments with even more focussed formats like the hackathon, where the expectation is that the meeting will result in some final product, may be worth investigating given the time pressure many academics are under.

Secondly, there are also many opportunities to improve the impact of our research through better use of interactivity in visualisation. The human mind has evolved to understand its environment through manipulating it; we are better able to understand relationships which we can directly affect and experiment with. Many participants had made good use of frameworks like Shiny, which provides a toolkit with which to create interactive visualisations through R, and javascript’s D3 library, and tools like these may better allow us to leverage these peculiarities of the human condition to get our messages across.

The next Rostock Retreat focuses on Causality and Causal Inference. A list of confirmed speakers is available on the retreat website. The application deadline is 28 February, which is next week!

The Rostock Retreat on Data Visualisation was organised by Tim Riffe, Sebastian Klüsener and Mikko Myrskylä, and hosted by the Max Planck Institute for Demographic Research.