A slightly odd workshop took place this week. It was organised by The Centre for Research in the Arts, Social Sciences and Humanities which a first glance doesn't sound very relevant to our work, but in fact the main theme was the use of modelling to predict and inform policy choices, with some specific reference to (though not sole focus on) climate change. There were several eminent names associated with it too, so it seemed like a good idea to attend.
The event kicked off with Mike Hulme, who used to be a normal climate scientist but these days could perhaps be better described as a post-normal climate scientist, mostly interested in the socio-political sphere in which climate science operates. He asked "How do Climate Models Gain and Exercise Authority?", and listed what he thought of as 4 key properties - code precision, output accuracy, methodological quality and social acceptance.
Unfortunately he didn't really seem to me to make much progress towards investigating these questions. In fact he hardly scratched the surface of how even the first three questions might be investigated, let alone the last one. His talk featured a now-famous (well Hans von Storch has also presented it) plot from a certain Hargreaves 2010 publication, and also a quote from Reto Knutti warning of the risks of models being tuned to agree with each other too much. I think this latter worry, while certainly a possible risk to be considered, is often overstated, and much of the analytical support for it seems to be based on the failed concept of a truth-centred ensemble. Back on topic, I think Mike's description of the models as having authority seemed rather exaggerated, since their authority is surely primarily derived from (and entirely dependent on) that of the scientists who develop, use, and choose to trust the models to some extent. Steve Rayner had something to say on this later (see below). Mike also seemed to me to be rather unfairly dismissive of the Climate Code Foundation, basically describing it as a couple of non-scientists with a youtube video and an unrealistic ambition. I tackled him about this the next day, and believe he was at least partially convinced by the end of our conversation. I see the push for more openness as being primarily for the benefit of science, with the PR aspects being rather secondary, but whatever the motivation the current climate is such that this sort of initiative seems to be very much pushing at an open door (eg the recent surfacetemperatures.org workshop). Not that it will necessarily all be plain sailing, but it's a movement whose time is surely coming.
Anyway, the rest of the workshop was a mix of interesting presentations about various aspects of the modelling and policy interaction, interspersed with surprisingly vacuous talks where people basically presented an idea or topic, but seemed to have no analysis or data of any sort to either support or challenge it with. Things like: how models are like/unlike magic (seriously!), or...well actually despite writing this only a few hours later I can't even remember what most of these talks were about. It seems there is quite a cultural difference between the level of support that would usually be considered necessary for a scientific presentation - we can't just talk vaguely about an idea (jules says "the sort of stupid idea you might have in the bath, or drunk"), devoid of any grounding in reality. However there were also plenty of highlights, such as Gerd Gigerenzer berating the medical industry for presenting statistics in ways that are guaranteed to mislead doctors let alone patients (relevance to climate science is left as an exercise for the reader), and Dan Kahan entertainingly demonstrating how people generally filter information through cultural biases, before they start to start to take account of it. Steve Rayner described models as both enabling devices that give people something to mobilise around and also oracles for scientists to hide behind and defer to when presenting advice. This latter point seems to relate to Mike Hulme's talk too.
"Tony" Krebs did a very good Blair impression. I can see why he went so far. Hey guys, you just need to be open and honest, and it will all be just fine. Johan Rockstrom gave an apocalyptic presentation of the precipice we were standing next to (liberally sprinked with hockeystick graphs and real pictures of cliffs). Apparently we have to act in the next 5 years or else! There was also an entertaining debate on whether higher resolution climate modelling is a good investment - the contrary claim being that we need to run ensembles to evaluate uncertainty (Steve Rayner also argued that as far as policy goes, we didn't need any more modelling post-1992, which is probably not far off the mark).
It was striking to us how interesting and switched-on the commenters were. Usually questions after scientific presentations are things like "what was the x-axis on your graph" but these were much more perceptive and interesting. The meeting was also really well-organised, with a good mix of excellent keynote speakers, shorter presentations and debates. Not to mention the poster session...
The event kicked off with Mike Hulme, who used to be a normal climate scientist but these days could perhaps be better described as a post-normal climate scientist, mostly interested in the socio-political sphere in which climate science operates. He asked "How do Climate Models Gain and Exercise Authority?", and listed what he thought of as 4 key properties - code precision, output accuracy, methodological quality and social acceptance.
Unfortunately he didn't really seem to me to make much progress towards investigating these questions. In fact he hardly scratched the surface of how even the first three questions might be investigated, let alone the last one. His talk featured a now-famous (well Hans von Storch has also presented it) plot from a certain Hargreaves 2010 publication, and also a quote from Reto Knutti warning of the risks of models being tuned to agree with each other too much. I think this latter worry, while certainly a possible risk to be considered, is often overstated, and much of the analytical support for it seems to be based on the failed concept of a truth-centred ensemble. Back on topic, I think Mike's description of the models as having authority seemed rather exaggerated, since their authority is surely primarily derived from (and entirely dependent on) that of the scientists who develop, use, and choose to trust the models to some extent. Steve Rayner had something to say on this later (see below). Mike also seemed to me to be rather unfairly dismissive of the Climate Code Foundation, basically describing it as a couple of non-scientists with a youtube video and an unrealistic ambition. I tackled him about this the next day, and believe he was at least partially convinced by the end of our conversation. I see the push for more openness as being primarily for the benefit of science, with the PR aspects being rather secondary, but whatever the motivation the current climate is such that this sort of initiative seems to be very much pushing at an open door (eg the recent surfacetemperatures.org workshop). Not that it will necessarily all be plain sailing, but it's a movement whose time is surely coming.
Anyway, the rest of the workshop was a mix of interesting presentations about various aspects of the modelling and policy interaction, interspersed with surprisingly vacuous talks where people basically presented an idea or topic, but seemed to have no analysis or data of any sort to either support or challenge it with. Things like: how models are like/unlike magic (seriously!), or...well actually despite writing this only a few hours later I can't even remember what most of these talks were about. It seems there is quite a cultural difference between the level of support that would usually be considered necessary for a scientific presentation - we can't just talk vaguely about an idea (jules says "the sort of stupid idea you might have in the bath, or drunk"), devoid of any grounding in reality. However there were also plenty of highlights, such as Gerd Gigerenzer berating the medical industry for presenting statistics in ways that are guaranteed to mislead doctors let alone patients (relevance to climate science is left as an exercise for the reader), and Dan Kahan entertainingly demonstrating how people generally filter information through cultural biases, before they start to start to take account of it. Steve Rayner described models as both enabling devices that give people something to mobilise around and also oracles for scientists to hide behind and defer to when presenting advice. This latter point seems to relate to Mike Hulme's talk too.
"Tony" Krebs did a very good Blair impression. I can see why he went so far. Hey guys, you just need to be open and honest, and it will all be just fine. Johan Rockstrom gave an apocalyptic presentation of the precipice we were standing next to (liberally sprinked with hockeystick graphs and real pictures of cliffs). Apparently we have to act in the next 5 years or else! There was also an entertaining debate on whether higher resolution climate modelling is a good investment - the contrary claim being that we need to run ensembles to evaluate uncertainty (Steve Rayner also argued that as far as policy goes, we didn't need any more modelling post-1992, which is probably not far off the mark).
It was striking to us how interesting and switched-on the commenters were. Usually questions after scientific presentations are things like "what was the x-axis on your graph" but these were much more perceptive and interesting. The meeting was also really well-organised, with a good mix of excellent keynote speakers, shorter presentations and debates. Not to mention the poster session...
4 comments:
The Mayans figured that 2012 CE was the last year; maybe we don't even have 5 years left...
Regarding the Foundation, (your report of) Mike's impression identifies a branding problem, which I'm already aware of and which I'm hoping to fix over the next couple of weeks.
Originally in 2008 we were the Clear Climate Code project, which now has two years' worth of spare-time stuff including real and IMO interesting science (some of which we are now writing up for publication).
Then this summer we set up the Open Climate Code project, which does indeed have a youtube video and not much else: people arriving through that door are not going to be impressed and it's on my list to fix this.
Finally I put the day job on one side to make time to set up the Climate Code Foundation, bringing the projects together into a framework to do more and better, with full-time effort. I have about a dozen pages still in draft form for the new website ('Board', 'Advisory Committee', 'Testimonials', 'Press', etc). I'm planning to put the shine on that over the next couple of weeks, before certain events which should generate some press interest.
In short, if we look like spare-timers, that's because we have been exactly that until very recently. If we look unfunded, that's because we are, and you can help to fix that. But if you think we're not serious, or that we don't know what we're doing, or that we've underestimated the size of the challenge, think again.
"(A) few hours later I can't even remember what most of these talks were about."
"It was striking to us how interesting and switched-on the commenters were. Usually questions after scientific presentations are things like "what was the x-axis on your graph" but these were much more perceptive and interesting. The meeting was also really well-organised, with a good mix of excellent keynote speakers, shorter presentations and debates."
I experience cognitive dissonance.
Steve, *I* experienced cognitive dissonance too. Much of the workshop was very good, and it was very well organised. The content-free talks left me genuinely baffled as to how anyone could have thought it was worth standing up and talking for 15 minutes about so little.
Post a Comment