Building Bridges Between Insular InfoVis Papers
Can you describe what information visualization research does in a few words? What are the driving questions and problems right now? It’s harder than it might seem. I believe that the lack of cohesion in the field is due, at least in part, to how we publish research.
The State of the Field
And it’s not just where we are now. How do the current themes compare to the ones from ten years ago? What might we be thinking about ten years hence?
Our difficulty in answering these questions strikes me as a symptom of a general lack of vision and identity. That’s not to say that there isn’t good work being done, but it’s mostly disconnected. Many people are doing interesting things and sometimes those coincide and seem to form a pattern. But they are largely random, not driven by bigger goals (bigger than those of researchers or groups).
Individual projects and creativity are important, of course. But a field also needs overarching themes and challenges. It’s hard to build on pieces of work that don’t connect to other work – which ends up perpetuating the problem.
Journal Papers and Their Discontents
Papers published at EuroVis, InfoVis, VAST (with exceptions, since VAST also has a conference-only track), and SciVis are published in special editions of journals. CHI proceedings are not technically a journal, but the bar is very similar.
Journal papers are great because they count much more than conference papers (especially for people unfamiliar with publishing in computer science) and thus make for a better CV. Journal papers are problematic because the bar for novelty is high and anything considered incremental is sure to get rejected.
For my talk at EuroRVVV last week, I tried to show my view of the field with this little sketch. The big red blobs are journal papers, the small circles are the minor workshop and short papers that drift around the periphery.
This is what I think we should be doing, however: use the smaller pieces to fill in the gaps between the large ones. Make connections. Dig deeper. Iterate and increment. There is a lot of value in doing this if done right, but it’s not done much right now. But the picture would look much more like this.
This is the cohesion I’m thinking about. Small papers that build bridges and that end up clustering the larger pieces together into themes. These themes would then inspire more work, on any level, on related topics. We would know what we’re doing.
The Value of Incremental Work
Attacking a problem from many different angles, and really deeply understanding it, means being able to publish closely related papers (whether by the same or different people). Such work is easy to dismiss as derivative or incremental, but it’s actually quite important. If every paper has to be entirely novel, it can’t at the same time be closely tied into the existing work in the field.
The criteria for this sort of work need to be established, and they’re more subtle than the usual novelty question. Does this paper ask an interesting question about something we’ve already seen some work about? Does it illuminate a point further than the existing work? Does it provide additional evidence to flesh out a theory? Does it draw connections that give us a new insight, potentially paving the way for interesting new work?
What this sort of work does is fill in the gaps and provide more of the connective tissue between the larger pieces of more traditional research (when I talked to Fernanda Viégas at EuroVis, she called it the glue – another good term if you don’t like my anatomical metaphor). It then allows us to see more connections and build on that work more easily than we have been able to so far.
Collections and Clusters, Not Grand Challenges
In the past, there has been talk of Grand Challenges. The problem with those, in my opinion, is that these captial-letter Grand Challenges are hard to define and, when they’re Grand enough, they end up being quite abstract.
Instead, perhaps we need to start smaller, with something lower-case and tractable. We don’t have to boil the field down to a dozen or so topics, simply have an idea of interesting, challenging problems that go beyond individual people and projects.
A good name for them might be open questions, research areas, or clusters. Somebody come up with a clever name for them that doesn’t try to imply too much. The idea is for them to be crystallization points for more work that connects to the other parts of that collection. Over time these will grow, as will our understanding of the topic.
These clusters are not meant to live forever as topics, quite the opposite. The size and difficulty should probably be such that they can be solved, or reasonably well understood, by a handful of researchers within five years or so – about the length of a Ph.D.
Publishing Smaller
There is currently a big gap between the valuable, fully peer-reviewed, low acceptance-rate, archival journal papers and the minor, barely-even-published, much less thoroughly reviewed workshop papers where you don’t always even know the acceptance rate (and it can be very high). What we need, I believe, is an outlet for things like workshops, but also a general way of publishing smaller ideas and results, and do that faster.
One venue for this are short papers. EuroVis already has them, but why not make them a publishing model rather than just a track? Let the workshops publish their accepted papers (or perhaps a selection of the best work) as short papers. The same is true for VIS: add short papers, let the workshops take advantage of that. This would raise the workshops’ appeal and lead to better papers there, and it would mean more material for such a track.
Depending on how deadlines are handled, this could also bring back the late-breaking hot topics that were a thing at Vis back in the day. Today, that role is largely, if poorly, served by posters. I believe that a lot more work could be published, publicized, and iterated on much earlier if there was a better step in between posters and full papers.
Cohesion and Communication, Not Just Novelty
I’m not against novelty or big journal papers. The move to publishing conference proceedings in journals was a good one. But to move faster and create a richer, more connected field, we need to have another way of getting work out. Short papers are not the only way, but they already exist and they could be made much more useful and valuable.
Smaller pieces of work are the lifeblood of many sciences. Visualization has moved up from workshops to conferences to journals. We now need to move down again, at least for some of our work, to allow that sort of research to happen and be published.
Posted by Robert Kosara on June 26, 2017.