Ian Ardouin-Fumat

Overview / Process

I joined the The New School’s Center for Data Arts to visualize the depth and breadth of conversations that took place at one of the world’s most eclectic business conferences.

Together, we put together a physical installation and a 3-day live-coding performance to transcribe the substance of C2 Montreal.

Data Visualization Lead with Center for Data Arts, for C2.


Goodbye cynicism, hello new ideas

C2 Montreal is a three-day event that brings together marketers and creatives to explore trends, opportunities and major shifts on the horizon. Every year, the event drags 5000 attendees to experience and challenge the future of business.

For the conference's 2017 issue, C2 commissioned The New School's Center for Data Arts to create the C2 Brain: a data-based installation that accounts for the many conversations taking place during the event. Conference transcripts, social media posts, and real life conversations are as many data points to be visualized in this 2500 square feet LED installation.

C2 Montreal 2016, main stage

I join CDA's director Ben Rubin and designer Sam Gallison to lead the design and development of real-time graphics for the installation. This work comprises of a week-long preparation phase as well as two 3-day long live-coding performances dedicated to designing visuals based on the conference's on-going content.


Arranging data in space

The installation, an ensemble of 96 LED panels and 20 LCD screens, constitutes one of the center pieces of the conference's venue. It surrounds the Agora, a space meant for relaxing and connecting with other attendees. We meant this installation as an immersive prolongation of the conference's experience, where one can read its latest content as well as getting a broad sense of all the conversations happening around them.

To accommodate this idea, I designed a virtual world that revolves around the installation, and turns the LED screens into a window onto the conference's data. In this visualization, conference quotes, social media content, and individual avatars collide together to form an ephemeral, collective intelligence. By standing at the center of it, one can glance at C2's consciousness.

The graphics were developed as a WebGL application using THREE.js. This allows us to easily iterate and dynamically push new visual content as the conference goes on.

Adjacent is an early prototype demonstrating the multiple development views and their slicing into a 1080p production display that will later be dispatched to the many screens of the Brain.


Live-coding performance

The application runs in a web browser in both development and production environments. This enables us to monitor new data coming in, while easily pushing new visual content to the installation.

After a couple of days spent installing the Brain in Montreal, C2 has finally started. From day one, a team of volunteers under Ben Rubin's supervision (joined later by Patrick Tanguay) has been collecting notes and nuggets of knowledge from the multiple talks and workshops happening simultaneously. This fast-flowing content is transcribed in a collaborative spreadsheet where the data is verified, timstamped, annotated, and processed. In real time, data flows into the Brain's visualization, where all quotes come together.

As the event goes on, Sam and myself continually develop new visualization modes allowing for new insights into the data. Some of them appear in the video below.

Here, one sequence visualizes social media content; another the curated conversations between attendees; the last one shows the stream of thoughts coming from each talk as they're happening.


3 days, 5000 people, thousands of data points

The conference was attended by 5000 people from all over the world. The C2 Brain appeared in hundreds of pictures posted on social media, making it one of the center pieces of the conference. Some of these images appear below.

Very flattering reflections going on with the central booth

Portrait of 2 exhausted designers, after 3 days of intense live-coding