Ecosystem structures and processes are always changing. The past never fully repeats itself. Some changes are routine or somewhat predictable, and can therefore be anticipated or even managed. Other changes are radical, unpredictable and uncontrollable. Such big changes derive from external disturbances, structural instabilities of ecosystems, or both. Recovery from big changes may be smooth, delayed or impossible. Resilience is the magnitude of disturbance that an ecosystem can undergo without exhibiting radical change whereas renewal depends on capability for reorganization following radical change. Certain aspects of ecosystem change are especially challenging for society: (1) radical change is hard to anticipate or manipulate and may lead to long-term loss of ecosystem services; (2) capacity for ecosystem renewal is poorly understood and may be undermined by human action. Economists define sustainability as "non-decreasing wealth" and their formulation is compatible with ecosystem change. However, practical implementation requires projecting future ecosystem processes (or probability distributions of ecosystem processes) over long periods of time. Such projections are not possible at present. Given this high uncertainty, policymakers should build flexibility to adapt to information as it comes in while science focuses on key information gaps for social-ecological systems (SES). How do external drivers and system features cause routine and radical change in SES? Following radical change, what determines the characteristics of subsequent routine changes? How is capacity for renewal maintained or enhanced in SES? These and other fundamental questions demand place-based interdisciplinary study. During the course of this research we are likely to discover new ideas for improving ecosystem services while promoting renewal of SES. Sustainability depends on seeing, and then adopting, better ways of living. Science should help induce new ways of seeing.