Perhaps one of the most interesting parts of trying to clean up the environment is the need to balance individual action with global behavior. Actions that are optimal at the individual level are often what lead to depletion of resources and environmental damage at the global or regional scale.
Take climate change for example – even though the effects of a changing climate promise to be devastating to many countries and places around the world, it has often worked better for countries to focus on their short-term economic goals rather than look at what would work best for the economy and environment in the long term.
A recent approach from scientists at Georgia Tech looks at how game theory can be used to help solve this problem. An assumption inherent in how many environmental policies and markets are designed is that actors will act rationally in their own interest and that the system doesn’t change drastically. Now, this is an assumption that doesn’t necessarily hold true in many cases – especially in cases where overuse of a resource like forests or water results in significantly changing the system.
What the scientists have done is build a new algorithm that couples the state of the system (the environmental resource in this case – water or carbon or forests) with the incentives for each actor. The results of this algorithm showed that a key factor in ensuring that the environmental resource was not completely depleted was to incentivize cooperation between most of the actors in the system, even if there are a few bad actors (depletors in the study’s language).
A model like this provides insights into how people and organizations can improve environmental outcomes and ensure that natural resources are not completely depleted.
This is not necessarily a classical data science algorithm, but what studies like this show is how new algorithms can be built to make better predictions and design more efficient systems. Deep learning after all was an esoteric algorithm developed in universities and research labs around the world and has come out to of the lab in the last five years to become one of the most powerful algorithms used by data scientists in a wide variety of fields. Who knows, a predictive algorithm like the one in this study may well become a standard part of the clean tech data scientist’s toolkit!