It is tempting, and perhaps even expected, to end a book as this by looking ahead, to speculating what should or could be done next to advance resilient performance. A premise for that is, of course, that something like resilient performance will still be needed. That this is so hardly needs any argument. The time when systems and societies could perform efficiently if everyone conscientiously followed simple rules and procedures ended at least a century ago, if not much earlier. Machines and systems are today non-trivial rather than trivial (von Foerster & Poerksen, 2002), in the sense that the transformations that link causes and effects are unknown rather than known. Since systems – and individuals – therefore must perform under conditions that are incompletely known, hence partly unpredictable, they must be able to adjust what they do in order to succeed. Any proposal for how something should or could be done – for Work-as-Imagined – therefore implies a set of assumptions about what the conditions will be about the World-as-Imagined.

On the positive side, people and systems still by and large perform well enough and are stable enough for societies to function, even during major disruptions. There is, therefore, ample evidence of what we may call resilient performance. The question is how we can understand what goes on, given that it by definition cannot be trivial, and how we can sustain and improve that. There has certainly been some progress since the heydays of human factors and behaviour-based safety, but much remains to be done. Performance can be resilient when there are recognisable patterns in how systems perform, but we need to understand both how these patterns emerge and how people recognise and respond to them. We need to understand better how we come to accept certain assumptions about what Work-as-Imagined is and understand better how the regularity of Work-as-Done is established. We need to know what is required for a system to perform acceptably well now and in the future. To do so, it will be more important to ask the right questions than to hunt for improved answers to the questions that we uncritically – and often unsuccessfully – have tried to answer in the past.