Inadequate decision-making strategies
A wide variety of strategies exists for solving the decision-making dilemma in climate and environmental protection measures:
Some of these strategies are not at all, others are at best moderately good at generating solutions to complex problems like climate change.
Numerous strategies are based on models to derive possible future developments. There are several types of such models, for example these:
So far, all of the models mentioned have in common that human behaviour is represented in them by formulas, rules and algorithms. In many economic models, all people are even assumed to be a single “homo economicus” who seeks to increase his economic benefit in a completely rational way.
Agent-based models – which can include an integrated-assessment model – are the most advanced approach so far to introduce human behaviour in models. In them, potentially very many people are modelled in different roles, all acting individually. But here, too, the rules for the human behaviour to be simulated are derived from economic and sociological behavioural experiments and are sometimes even determined on the basis of a simple plausibility assessment by the modeller.
Another limitation of the models is that the people modelled in them cannot discuss their behaviour and perception of the world with each other – that would be far too complex!
No wonder that predictions based on such models regularly miss the mark (see also Towards representing human behaviour and decision making in Earth system models – an overview of techniques and approaches, Müller-Hansen 2017, Journal of Earth System Dynamics).
The best statements that can be made with this are “if people behaved as assumed in our model, this would be the outcome”.
But that’s not good enough, because people don’t behave like wind-up robots. Decision support for politics and economics on climate change, but also on many other issues, needs statements about the actual likely and unlikely behaviour of the people affected in order to identify the effective and socially acceptable measures.
The SCIARA platform is a tool to be able to determine such statements for a variety of issues.
The SCIARA idea: In the agent-based models, let’s control the agents through the neural networks best trained for human behaviour that we know – the ones in the heads of real people!