Interview with LML Laboratory Director Colm Connaughton

As a new feature of the LML blog, I will be running a series of occasional interviews with some of the LML Fellows and other individuals linked to the laboratory. This is an interview with Colm Connaughton, current Laboratory Director, and also Director of the Centre for Complexity Science at Warwick University. Colm’s research interests include non-equilibrium statistical mechanics, fluid dynamics and turbulence, nonlinear waves and a host of related topics. A list of his publications is available here.
Mark Buchanan
 
Mark Buchanan:
In an email, you told me that your research is, in terms of topics, “all over the place.” I had a quick look at your most recent papers and that seems about right. So I wonder — how does that happen? Do you organize your research by problem area, or is it by methods, or maybe by the  people you work with — or is it something else?
Colm Connaughton:
Well, I think there are two strands of my work. There’s a science driven strand, which really comes from my interests in non-equilibrium statistical physics, which is my core domain of expertise. I’ve always been interested in things like turbulence and kinetic theory and non-equilibrium systems of various kinds. And then there’s a more problem driven strand, which is often the more interesting one, and this really makes things all over the place. This comes about because I’ve been involved in MathSys — a very applied doctoral training program at the University of Warwick and these problems just come to us, as people need help on different problems. One common thread linking lots of these problems is a reliance on applied data science.
Mark Buchanan:
Indeed, one of the specific things I wanted to ask you about was your work on smart motorways. This seems to me driven by the importance of the problem, but also by the availability of data to do analyses that weren’t possible before.
Colm Connaughton:
Yes. The volume and diversity of the data available now, even compared to 10 years ago, are really enormous. This work on smart motorways really came about when a big engineering company in the U.K. came to us and said “we have all of this data and we use it for a few things, but we’d like to know what else it can be used for.” So we were given a free ride to just do curiosity-driven research, the only criterion being that we use this resource.
One thing we considered was evidence for or against the phenomenon of so-called secondary accidents on smart motorways. The idea is that if there’s an accident, the resulting disruption makes it more likely that another accident will happen nearby soon afterwards. This was quite an interesting project because it required some fine-grained statistical tools to convince ourselves that we were seeing something real, and not just a phantom  generated by the methods we were using. In the end, we found that from the data alone not very many accidents – about 5% — could justifiably be attributed to this secondary mechanism. Also coming out of this work was an improved spatio-temporal map of accident hazards on the M25, which should be useful for taking steps to better road safety. You don’t know this until you get some hard numbers for it, but the 5% finding alone suggests that efforts to reduce accident risk would be better targeted at reducing primary accidents rather than secondary accidents specifically.
Mark Buchanan:
There’s another paper I want you to tell me about, your curious paper on The Song of Ice and Fire. How did that come about?
Colm Connaughton:
This is an interesting story, actually. The paper is studying the social network of the cast of characters in The Song of Ice and Fire, a series of books on which the TV series Game of Thrones is based. This came about due to an old friend, Ralph Kenna at Coventry University, who was a post-doc with me when I was a PhD student in Dublin. He’s a physicist, but also leads an interdisciplinary research programme called “Maths Meets Myths” that tries to quantify the structure of mythological narratives. I ended up having a chat with him and some colleagues over a few beers and we thought it would be great fun to do this for the Game of Thrones. Then a few months later he called me up and said he’d found someone – a kind of super-fan of the books – who was willing to read through the entire series and map out the social networks of all the characters. That was the Genesis of this project. It then took quite a long time to analyse these networks and what they reflect about the nature of the story.
We were able to measure the time evolution of the network structure and found an interesting distinction between story time and discourse time, the first being the time measured in actual dates as depicted in the story, and the second the time as experienced by the reader, measured in chapters and pages. It turns out that the structure is very different when viewed in these two ways of conceptualizing time. If you look at the distribution of times between significant deaths in the story timeline – the time experienced by the characters – it has a kind of power law distribution that is characteristic of lots of other human activities. It’s a bursty pattern, and makes for a credible world in the way events happen, reflective of actual human interactions.
But when that timeline is translated into the discourse time, the distribution of times between events instead basically an exponential distribution, which is a special memory-less distribution. This lends a feeling of randomness to the story, as anyone can die at any time, which is the feeling that you have when you read these books.
Mark Buchanan:
It’s quite incredible that the author manages this delicate balancing act. It means they must have really had a good intuitive feel for the story, I’m sure it’s not an intellectually designed thing.
Colm Connaughton:
Yes, I think this is just the intuition of a good writer.
Mark Buchanan:
To change pace a little, you’ve changed roles at LML and you’re now the laboratory director. As you’re feeling your way into this new position, I wonder what you think are the main challenges for LML right now? What kinds of things would you like to do differently — or keep the same?
Colm Connaughton:
Well, LML has now existed for 10 years, trying to show how research can be done a little bit differently. I think this is even more important now than before, particularly in the mathematical sciences. And so I really would like to see LML continue to support the freedom to pursue questions of interest that don’t quite fit the mainstream view of how mathematical research should be done. The way university research is going – it’s become so competitive and the measures of success have become so narrowly defined – that it’s actually hard for many people to freely pursue many questions that are interesting and/or useful. LML was initially founded with a view towards providing an environment to avoid this. We’re small, and we can’t fix this problem. But I strongly believe we can have an impact just by showing that research can be done differently if there’s a will to do it.
So no, there’s not really so much that I would say I will do differently, but I would like us to be better known. And in particular for people to better understand what LML is aiming for, and see us not just an interesting stand-alone experiment but as part of a broader evolution of the research ecosystem that ultimately aims to make science and society work better.

Leave a Reply

Your email address will not be published. Required fields are marked *