Darwinian evolution is easier to understand at lower levels, where we see populations of individuals with heritable variation. Understanding higher levels, where only a few ‘individual’ systems are present, is always harder: constraints, principles of self-organisation, transformation or emergent properties all might be needed for discussing the actual problems. To better understand both the parts and the whole, we need to focus on the (multiple) interactions connecting the components. These interactions serve as the mechanistic basis of system dynamics, and they can be flexible and adaptable, contingent or relatively hardwired. These interactions are the grist for the mills of system scientists.