allowed the two to influence one another, they couldn’t capture this critical
phenomenon.
Nevertheless, climate modelers have only just begun to include suchfeedbacks into their models. The delay hasn’t been because they haven’t thought
such feedbacks were important; it’s been because modeling them is extremely
tricky. Feedback effects can make tiny inaccuracies blow up into massive errors
over time. The models need to be designed with enormous care to control for
this.
And no one really knows how to do it. Climate modelers are trying to figureit out in the context of these enormously complex models, ones which no single
person could possibly understand in their entirety. It’s an overwhelmingly
confusing and difficult task (that can come down to the impossibility of solving a
large system of partial differential equations exactly through “discrete
approximation”). What they need is something analogous to what biologists have
in the fruit fly: a simpler case to study to develop a basic understanding of how
things work. Long before tackling the horrifically complex human genome, for
example, biologists cut their teeth by sequencing the fruit fly. Armed with a
simple model like the fruit fly, climate scientists would have a vastly easier time
unraveling how the various components of climate interact with one another.
Mathematical scientists can play a leading role in providing this simplermodel. They specialize in abstraction, reaching beneath the messy details of real
life to expose the skeleton beneath. They can learn from climate scientists what
the most important elements are and then explore how those elements interact
by analyzing relatively simpler mathematical models that can be thoroughly
understood. However, there is a danger here: We must be careful not to think
that solving a simpler, though still relatively complex, mathematical problem is
the end of the story. It is only the beginning.
Mathematical scientists can also work with climate scientists to deal withthe uncertainty in their models that is inevitable because these models are of
complex adaptive systems. Scientists will never, for example, be able to produce
a model that can tell us the precise lowest wintertime temperature in Manhattan
in fifty years. Accepting that limitation allows scientists to focus on the questions
that can be answered, like, what’s the best estimate they can make, and what is