To develop a truly holistic view of human ecosystems—to identify feasible pathways towards their sustainability—it is necessary to make use of tools and techniques that are able to grapple with the complex interactions between human activities and the stability or resilience of the non-human ecosystem habitat. Various chapters in this book show that human behaviour can be complex in itself, necessitating its representation in a more sophisticated manner in computational approaches that aim to improve our understanding of the collective outcomes under alternative management strategies. Agent-based modelling is one promising way of exploring the possible collective implications of more complicated behaviour of human agents in an ecosystem context. Role-playing games are another valuable method of engaging with stakeholders and eliciting their views in an ecosystem context.

Opportunities for further, impressive progress exist. Recent developments in object-oriented programming languages have allowed users to create autonomous modules which can interact with each other even when they have been designed by different people, different teams or different companies (Ferber 1999). Agent-based models and Multi-Agent Systems (MAS) have an important role to play in this endeavour by serving as possible successors to object-oriented systems and combining local behaviours with autonomy, best-practice agent modules and distributed decision making. Thus it seems very likely that the software engineering of tomorrow for addressing more complex societal problems will be agent-oriented, just as that of today is object-oriented.

To make our new kind of science truly polymorphic, the mapping of relationships between agents needs to be continuously updated depending on circumstances, proficiencies, perceptions, tasks to accomplish, or relational rules based on the social contracts established. This means that the simulation models need to recognise different views, biases or expertise. Such simulation tools are more participatory in nature, and should lead to further exciting developments in the future.

In the field of companion modelling, for example, convergent views may emerge after allowing several recognised experts to take the lead one after the other. This is similar to a cooperative team sport like football; for instance, the right-winger becomes the leader when the ball comes into his space, but it can happen that he becomes the goalkeeper when the situation requires it. This flexible kind of companion modelling forms the basis of AtollGame (chapter 12), in which the agent-based model and the corresponding role-playing game were designed according to different viewpoints—converging or conflicting—that were recorded during a series of interviews.

The sixth chapter of this book presents an argument by Gabriele Bammer to think differently about the future of current science and the way it engages with decision-makers. According to the author, developing a new specialisation—Integration and Implementation Sciences—may be an effective way to draw together, and significantly strengthen, the theory and methods necessary to tackle complex societal issues and problems. It would place research on human ecosystems in broader context and link it with a range of complementary concepts and skills.

There is indeed a pressing need to integrate not only the participation, but, more importantly, the engagement of local stakeholders in projects that concern their future. Aslin and Brown (2004) argue that local communities need to be involved in the analysis of the results (consultation) and the choice of the possible scenarios (participation), but also in the knowledge creation itself (engagement). This is a post-normal posture adopted, for example, within companion modelling approaches (Bousquet et al. 2002), for which Anne Dray and colleagues (chapter 12) provide a good illustration.

Research on sustainable development too often relies on deductive scientific approaches to reach outcomes that require more inductive and flexible solutions. But flexibility means that one must assume some uncertainty during implementation and must get away from traditional reliance on deterministic and predictable solutions. What is true at the technical level becomes paramount at the political level. As stated by Bradshaw and Borchers (2000: 1):

One of the most difficult aspects of translating science into policy is scientific uncertainty. Whereas scientists are familiar with uncertainty and complexity, the public and policy makers often seek certainty and deterministic solutions.

This call for a meta-integration of science takes us back to the beginning of this book. Roger Bradbury’s vision of a complex and adaptive science is nothing less than a fantastic opportunity for scientists to think differently about the goals of their own research and their relations to others.