The architecture of robustness

It is becoming increasingly evident that real-world complex-adaptive systems, from ecosystems to economies and political systems, are prone to systemic risks and collapse (Frank et al., 2014; May, Levin, & Sugihara, 2008). Despite this, the underlying dynamics that create such risks are not only poorly understood, but also seldom acknowledged by policy-makers. This chapter is inspired by and complementary to Mauro Guillén’s recent book The Architecture of Collapse (2016) and, in turn, to Herbert Simon’s monumental ‘The architecture of complexity’ (1962), in attempting to provide a framework for examining the structural features that make systems robust and resilient in the face of challenges, internal and external. In so doing, I will draw heavily on ecological theory, which has wrestled with such questions throughout the history of the development of the subject. (Levin, 1999; MacArthur, 1955; May, 1973). Indeed, in ecological theory, one of the most pervasive themes has been in understanding the relationships among diversity, complexity and stability, and resilience or robustness. Each of these of course has many meanings, and therein lies the potential for substantial confusion and different interpretations of the expected connections among them (Levin & Lubchenco, 2008). In particular, one line of reasoning says that more complex systems and more diverse systems should be more stable (MacArthur, 1955), while another line of reasoning argues just the reverse (May, 1972). Particularly relevant for this chapter is that complexity as measured by interconnectedness can be destabilizing, because it diminishes the capacity of the system to adapt by adjusting internally. That has long been hypothesized for ecological communities (May, 1972), and it undoubtedly played a role in the financial collapse of a decade ago (May et al., 2008); it is what led to adoption of the concept of TCTF (too connected to fail), as opposed to TBTF (too big to fail), as a tool to identify financial institutions that could not be allowed to fail because of systemic risk to the broader financial ecosystem (Haldane & May, 2011; Kambhu, Weidman, & Krishnan,