Scientific expertise does not carry over to new fields

Chuck Doswell has an extremely good blog post on allegations by climate change deniers that scientists raising alarms about climate change are doing so because that's one way to get more funds. He ends the post saying:

I have no logical claim to seek to overthrow the IPCC consensus until such time as I become a participant in the process and have established my credibility in that domain, via the process of publishing research results in refereed journals. Since that's not likely to happen, I will defer to the experts. I believe most of the critics of global climate change science should do the same.

Because scientific expertise does not carry over between fields, it always surprises me that people who with no expertise in a field can so easily imagine themselves on par with smart people who've engaged with those problems for years.

Unfortunately, this is not limited to politicized areas like global warming. Because I work in a cross-disciplinary field (the intersection of image processing, statistical learning and meteorology), I get to interact with specialists in each of these fields. A meteorologist assumes that statistical learning is just a matter of throwing variables in a hopper and out comes a system that's learnt all the relevant details -- it's hard to convince him that things are trickier than that. A mathematician assumes that comparing two images is as simple as computing a distance metric between objects; it is a long time (weeks!) before he realizes that objects in meteorological images are ill-defined and that metrics like the Hausdorff distance produce counterintuitive results all too often. A statistician can not believe that I built a real-time statistical model based on a handful of cases. He fails to realize the challenges inherent in working with rare events that are in the extremely long tail of possible outcomes. An image processing researcher can not believe that incorporating contiguity constraints into a well-known clustering procedure can have any impact. And so on and so on. The commonality is each of the specialists is extremely knowledgeable about the outstanding issues in his/her field but is all too willing to take a simplistic understanding of other scientific disciplines to the battlefield.

Television meteorologists who deny climate change are making the same mistake -- they are extrapolating the difficulties in short-term weather forecasting (models that get progressively less accurate over time periods of a few days) to a completely different scientific discipline. The problems and issues are different. To take a simplistic example, it's quite easy for a climate model to predict that next February, it'll be colder in Oklahoma than it will be this July even though a numerical weather prediction model couldn't begin to tell you what the temperature will be on February 17, 2011.

As Chuck says, I'm not a climate scientist, so I'll defer to those who are. Just as when cancer researchers said that they had proof that tobacco causes lung cancer, I believed them too.