In this synthesis, we hope to accomplish two things: 1) reflect on how the analysis of the new archaeological cases presented in this special feature adds to previous case studies by revisiting a set of propositions reported in a 2006 special feature, and 2) reflect on four main ideas that are more specific to the archaeological cases: i) societal choices are influenced by robustness–vulnerability trade-offs, ii) there is interplay between robustness–vulnerability trade-offs and robustness–performance trade-offs, iii) societies often get locked in to particular strategies, and iv) multiple positive feedbacks escalate the perceived cost of societal change. We then discuss whether these lock-in traps can be prevented or whether the risks associated with them can be mitigated. We conclude by highlighting how these long-term historical studies can help us to understand current society, societal practices, and the nexus between ecology and society.

For at least the past 8000 years, small-scale farmers in semi-arid environments have had to mitigate shortfalls in crop production due to variation in precipitation and stream flow. To reduce their vulnerability to a shortfall in their food supply, small-scale farmers developed short-term strategies, including storage and community-scale sharing, to mitigate inter-annual variation in crop production, and long-term strategies, such as migration, to mitigate the effects of sustained droughts. We use the archaeological and paleoclimatic records from A.D. 900-1600 in two regions of the American Southwest to explore the nature of variation in the availability of water for crops, and the strategies that enhanced the resilience of prehistoric agricultural production to climatic variation. Drawing on information concerning contemporary small-scale farming in semi-arid environments, we then suggest that the risk coping and mitigation strategies that have endured for millennia are relevant to enhancing the resilience of contemporary farmers’ livelihoods to environmental and economic perturbations.

The value of “diversity” in social and ecological systems is frequently asserted in academic and policy literature. Diversity is thought to enhance the resilience of social-ecological systems to varied and potentially uncertain future conditions. Yet there are trade-offs; diversity in ecological and social domains has costs as well as benefits. In this paper, we examine social diversity, specifically its costs and benefits in terms of decision making in middle range or tribal societies, using archaeological evidence spanning seven centuries from four regions of the U.S. Southwest. In these nonstate societies, social diversity may detract from the capacity for collective action. We ask whether as population density increases, making collective action increasingly difficult, social diversity declines. Further, we trace the cases of low diversity and high population density across our long-temporal sequences to see how they associate with the most dramatic transformations. This latter analysis is inspired by the claim in resilience literature that reduction of diversity may contribute to reduction in resilience to varied conditions. Using archaeological data, we examine social diversity and conformity through the material culture (pottery styles) of past societies. Our research contributes to an enhanced understanding of how population density may limit social diversity and suggests the role that this association may play in some contexts of dramatic social transformation.

What relationships can be understood between resilience and vulnerability in social-ecological systems? In particular, what vulnerabilities are exacerbated or ameliorated by different sets of social practices associated with water management? These questions have been examined primarily through the study of contemporary or recent historic cases. Archaeology extends scientific observation beyond all social memory and can thus illuminate interactions occurring over centuries or millennia. We examined trade-offs of resilience and vulnerability in the changing social, technological, and environmental contexts of three long-term, pre-Hispanic sequences in the U.S. Southwest: the Mimbres area in southwestern New Mexico (AD 650–1450), the Zuni area in northern New Mexico (AD 850–1540), and the Hohokam area in central Arizona (AD 700–1450). In all three arid landscapes, people relied on agricultural systems that depended on physical and social infrastructure that diverted adequate water to agricultural soils. However, investments in infrastructure varied across the cases, as did local environmental conditions. Zuni farming employed a variety of small-scale water control strategies, including centuries of reliance on small runoff agricultural systems; Mimbres fields were primarily watered by small-scale canals feeding floodplain fields; and the Hohokam area had the largest canal system in pre-Hispanic North America. The cases also vary in their historical trajectories: at Zuni, population and resource use remained comparatively stable over centuries, extending into the historic period; in the Mimbres and Hohokam areas, there were major demographic and environmental transformations. Comparisons across these cases thus allow an understanding of factors that promote vulnerability and influence resilience in specific contexts.

Human activity recognition is the task of identifying a person’s movement from sensors in a wearable device, such as a smartphone, smartwatch, or a medical-grade device. A great method for this task is machine learning, which is the study of algorithms that learn and improve on their own with the help of massive amounts of useful data. These classification models can accurately classify activities with the time-series data from accelerometers and gyroscopes. A significant way to improve the accuracy of these machine learning models is preprocessing the data, essentially augmenting data to make the identification of each activity, or class, easier for the model. <br/>On this topic, this paper explains the design of SigNorm, a new web application which lets users conveniently transform time-series data and view the effects of those transformations in a code-free, browser-based user interface. The second and final section explains my take on a human activity recognition problem, which involves comparing a preprocessed dataset to an un-augmented one, and comparing the differences in accuracy using a one-dimensional convolutional neural network to make classifications.