Critical Climate Machine 4/4 : Skeptical Science

Artistic Residencies: The Blog

For his Critical Climate Machine - and especially for the algorithmic portion of the analysis of the climatoskeptic arguments which makes up the raw material of this "data sculpture" - Gaëtan Robillard has largely relied on the work of John Cook, an Australian researcher in cognitive psychology at Monash University (Melbourne). This is an opportunity for us to learn a little more about the science of "climate communication".

John Cook, how did you come to devote your scientific career to "climate denial"?

It started with a debate about global warming that I had with my father-in-law at a family party. He was a climate skeptic, and I wasn't really involved in the issue yet. Following these conversations, I began to research the different arguments he had put forward, and I realized that none of them were seriously supported by science.

Like any self-respecting son-in-law who hates to let his father-in-law have the last word, I started to make a list of all the arguments he could come up with at future family gatherings, along with their scientific rebuttals. I studied physics, but after graduation I turned to graphic design, so something else entirely. That said, this initial scientific training was invaluable in being able to read and understand the most serious articles on the subject especially since I tended to go to the primary sources, relying exclusively on papers published in peer-reviewed journals. Over time, I thought this list might be useful to others, so I posted it on

This is how I inadvertently got back into science. What I didn't know is that this research work is similar to research in "Climate Communication", which is now a scientific discipline in its own right! Coming from physics, I was naive enough to believe that transmitting scientific results as they were was enough. This, especially when it comes to climate, is anything but true: many aspects, notably the psychological ones, come into play in the communication of established facts. One thing led to another, and I ended up doing a PhD in cognitive psychology, on the subject of climate misinformation.

You have developed algorithms for detecting and classifying climate-skeptic arguments online - algorithms that served as the basis for those used by Gaëtan Robillard for Critical Climate Machine. How did the idea come about? What do these algorithms do?

Several years ago, I was contacted by two scientists: Travis G. Coan, a researcher in Computational Social Science at the University of Exeter (UK) and Constantine Boussalis, a researcher in Political Science at Trinity College Dublin (Ireland). They had just published a research paper on models for analyzing climate-skeptic themes. Simply put, they were scanning the Internet for climate misinformation, to identify the main patterns. Knowing my work, they wanted my opinion on their research.

In addition to my feedback, I suggested that they use the machine learning methods they had developed as part of their work to test climate skeptic claims (such as "the climate is not warming" or "humanity is not to blame," etc.). This is how we launched ourselves into a long-term research project. The first step was to draw up a taxonomy of all the climatoskeptic arguments and claims, which we used to methodically train the computer models. In principle, it sounds simple: you find a text and, if you detect a false or misleading claim, you classify it according to the taxonomy and tell the machine. The problem is, of course, that this involves repeating the process tens of thousands of times until the machine is sufficiently trained.

What was the aim of this work?

The ultimate goal is what I would call the "Grail of fact-checking": that is to say, to detect disinformation in real-time and "debunk" it immediately, i.e. to counter it with the correct argument before it has time to spread. Since misinformation has the ability to spread almost instantaneously, the idea is to catch it before it spreads.

The "machine learning" part I just described is therefore only the first step in a larger process, which we have called the "4Ds": Detect-Deconstruct-Debunk-Diffuse.

How far along are you in this process?

We are at the first "D": detection. You have to find out how to deconstruct a false assertion very quickly (understand what is false in the assertion, which can be quite complex since the same assertion can call upon various arguments that must each be understood and refuted), immediately associate the appropriate refutation with it (automatically generate an adapted fact-checking text), and then circulate it.

This last step is just as complex: how do you implement the tool you have developed so that it has a real impact? For example, for the past year we have been developing an extension for Internet browsers which will flag false arguments and assertions in a visited page with a pop-up, for example, and immediately debunk them.

We hope to release a public version by the end of the year.

© Gaëtan Robillard

What was your reaction when Gaëtan Robillard told you about his installation Critical Climate Machine?

After my degree in physics, in addition to graphic design, I also worked on some comic books. I have a scientific background, but also an artistic one and I am totally convinced that creative arts can be an efficient tool to raise awareness on important issues. I was immediately very excited about the idea of an artwork that, by finding a creative outlet for all this data, would raise awareness about not only global warming, but also research on climate misinformation.

Similarly, after seeing the Cli-Fi film Don't Look Up, I wondered if we could use this work in a fictional setting: for example, why not harness the taxonomy we've established to tell a story? Using creativity to raise awareness can be extremely powerful. And any way you can reach people is good.

Interview by Jérémie Szpirglas

Portrait credits : © John Cook

Critical Climate Machine will be shown in its complete version (including all aspects: sculpture, sound installation, etc.) October 12—22 in Grenoble, as part of EXPERIMENTA, Biennale Arts Sciences.
Critical Climate Machine is part of the MediaFutures project, of which IRCAM is a partner. In this context, Critical Climate Machine has received funding from the European Union's Horizon 2020 framework program for research and innovation, under grant agreement n° 951962. Critical Climate Machine (Patterns of Heat) is also part of The Intelligent Museum, funded by the Digital Culture Program of the German Federal Cultural Foundation (Kulturstiftung des Bundes) and the Federal Government Commissioner for Culture and Media (Beauftragte der Bundesregierung für Kultur und Medien).