Two Ideas for a Better Visualization Web
eagereyes 2016-09-03
There is a reasonable amount of information about visualization available on the web. There are still huge gaps though, especially when it comes to bridging the gap between academic research and the rest of the world, though. Here are two ideas: one simple, one rather involved.
Ben Shneiderman has recently been talking to a number of people about a resources collection for visualization similar to KDD Nuggets. I'm skeptical about the need for such a thing. Lists of resources need to be kept up to date, and they can lose their utility quite quickly when they go for completeness rather than usefulness (Andy Kirk's list of everything ever is a good example).
This had me thinking about better formats, though. How can we make information more accessible? How can we make research results available and interesting to people? And how can we drag more of research out onto the web?
The Paper Explainer
Papers are what academics publish. But they are often quite terse, and are usually written in a mad rush to beat a deadline. Going back to a paper from a few years ago can be painful: typos, tortured writing, and some really interesting points were buried rather than made the center of the paper.
Carlos Scheidegger thought about this recently too:
Lately I've become tempted to rewrite some of my recent papers into webpages that actually -- and slowly -- explain everything in it. (1/n)
— Carlos Scheidegger (@scheidegger) August 11, 2016
The posting I recently did to walk through the pie chart study results was along similar lines: find a better way to explain the results in a way that I now feel is much clearer and helpful than the paper (it's also glossing over many details, admittedly, but otherwise it would have gotten out of hand).
What Carlos is suggesting would be tremendously useful, both for people interested and wanting to read about published ideas in more depth and for the author of the paper. I learned quite a bit when looking through the results again and thought about how I would explain them to an audience that doesn't know what a confidence interval is.
It's a lot of work, for sure. I probably wouldn't have done this if I hadn't created the charts for my Information+ talk. Even so, just writing it all up took a while. But now a lot of people are reading about the work Drew Skau and I did. It's actually reaching people. Far, far fewer people are ever going to bother reading the papers.
If you're an academic who has considered blogging, the paper explainer is the ideal starting point. Grab a few of your papers and start writing! Maybe you have some early work that you now think is funny because you were so naive. Or there's an idea you've thought about in multiple places but never quite put into words. Or just dive deeper into one paper. Even if nobody else reads it, you'll learn a lot yourself.
The VisRxiv or DataRxiv
On the other end of the spectrum is the idea of a pre-print server for data visualization. The physics community has been doing this with great success for a number of years now with arXiv. The site lets people upload papers and get feedback before submitting to a journal. It helps to get ideas out earlier and to hone a paper before it even goes into review.
Other fields have taken note. Biology just started BioRxiv, and Chemistry is now pondering their own server, ChemRxiv. Clearly, an unpronounceable name is the first requirement, but there are a few more hurdles.
Visualization by itself is probably too small to really make this work. But if this could include a number of fields like statistics and parts of psychology, it could work.
arXiv has a computer science section with an HCI subsection. That's where you'll find a handful of visualization papers. This follows the age-old tradition of visualization being ignored in classification schemes, and it clearly doesn't help get noticed or get feedback for work uploaded there.
There are lots of things a more visualization-centric site could do that arXiv doesn't: a stronger focus on images, embedded demos of techniques; interactive widgets to explore data from studies, etc. There are so many things that could be done beyond just hosting PDFs that would create more interest and engagement.
To make sense, a VisRxiv or DataRxiv would need some serious long-term support, and probably some sort of institutional backing. If the site is funded from a three-year grant, or run as a hobby by somebody, it's going to disappear or fall apart after a few years.
Still, I think it would be incredible helpful to have such a site. It would create a different kind of conversation than exists right now, and help push publishing in visualization towards a model that's a bit more forward-looking than what we have right now.