Tuesday, October 6, 2009

Beware of black swans!

Some thoughts on modeling, from IRIN: Predicting the impact of climate change requires studying changes in the long-term averages of daily weather patterns and many other factors, and can be a tricky business. Scientists use climate models that simulate the possible impact of variables like radiation, moisture content, and the movement of air and temperature over a given period of time to help project what could happen.

To make forecasting the possible impact of climate change as comprehensive as possible, and make the connection between current events and future consequences clearer, scientists and academics have been expanding the list of variables such as sea level rise and even food price increases and malnutrition statistics.

"The lesson is to understand what models do," said Peter Walker, director of the Feinstein International Centre at Tufts University in the US. "They show you a very simplified picture of what may happen under very tightly proscribed conditions. They do not predict the future!"

Models show "what will happen if the variables looked at behave as expected - i.e., all other things being equal, and our predictions over the changes in variables being right - this is what will happen," he explained.

"Where models fall down is if the accrual variables affecting the future turn out to be ones they [scientists] have not mapped. So, what if conflict triggered by climate changes turns out to be a big issue [in future]? What if the relationships between the variables change as a different future evolves?" Walker commented.

…."Finally, it is an irritating reality that much change in history has not been driven by the predictable events, but by the unpredicted - the big, unexpected events. This is referred to as the 'Black Swan Theory'."...

A black swan in the Singapore Botanical Gardens, shot by Calvin Teo, Wikimedia Commons, under the Creative Commons Attribution ShareAlike 3.0 License

No comments: