Blog - BS Data, hackery, stories

Notes from OpenVis 2018

This year, the OpenVis conference took place in Paris. And it was awesome.

Formerly Boston-based, OpenVis made the trip to France and set up shop next to the Tour Eiffel thanks to the support of EM Lyon business school (and of the sponsors), which supported Lynn's desire to keep the conference alive.

Practitioners from around the world are surely grateful for what Lynn and the committee put together this year. I certainly am!

The media folks

Good to see media organisations were well represented, and that some great work keeps coming out and pushing the field. Interestingly though, it seems that the admiration of the NYT is not confined to interactive journalists!

There were a fair few journalist types at OpenVis, in addition to those who occasionally work for publications. But there were many, many colleagues in attendance, which made for fun gossiping in the evening.

And as far as the London media dataviz scene in concerned, more soon…

Machine learning

t-SNE and image classification were all the rage. Great data work is alive and well, but it seemed to be always somehow linked to Big Tech.

Performance in the browser

No fewer than two talks were solely devoted to 3D and to animating heavy models with either thousands of points or complex rasters and textures. WebGL and Canvas are still where the money is, but the challenge remains for lower-performance smartphones.

AR/VR

Coincidentally, little was said about AR and VR. For which I'm not tremendously sad.

A shared set of practices?

As one of the speakers said, "we should get ourselves some cool acronyms to describe our methods too." Are data scientists still suffering from a lack of recognition?

Tools

Open source is pretty much ruling here. I say almost because Tableau's market cap still is $7.6b as of May 2018.

I'm part happy, part disappointed we weren't put through hours of live coding or technical demos. Only Jo Wood, from City University, presented a tool. The two-day conference was however followed by a day of workshops - which I sadly could not attend. Shame I missed the opportunity to have my typos spotted by Mike Bostock.

Various levels of technicality

In turn, the speakers took us from advanced machine learning algorithms applied to solving their viz issues to artsy design-led thinking and Illustrator; from heavy statistics to gif-making; and from theoretical discussions about how to think about "the AR canvas" to common numerical and statistical fallacies.

Regardless of our background and experience, there was enough of everything for everyone in the room.

Great scheduling

Hat-tip to the committee once again for crafting such a splendid schedule. Presentations echoed each other, and at times replied to one another. The best example was Federica Fragapane taking the stage following Steven Franconeri: she is such an incredible artist and embodies "The Italians" category, as per Elijah Meeks and Susie Lu's definition. Her talk about visual narratives for empathy was visually rich, and her work felt like the polar opposite of Steven's demonstration, an hour earlier, of how our brain takes such a long time to read visualisations, colours, and patterns.

A fantastic discussion that subconsciously happened thanks to scheduling mastery.


All this to say that I can't wait for the 2019 edition of OpenVis! Thank you again to the committee and volunteers for planning and running such a great event.