Archaeologists are accustomed to thinking about uncertainty in their work, and this volume is an effort to come to terms with uncertainty in formal computational modeling—that is, the simulation of the past and of systems of interest to archaeologists. It is somewhat surprising that there was not any formal treatment of uncertainty in computational modeling in archaeology before the contributors to this book took it on, despite the fact that tools and processes to do so are commonly deployed in other fields. As discussed in the preface, the impetus for this volume came about when Marieka Brouwer Burg was preparing to defend her dissertation, and a geologist on the committee asked whether she had done a sensitivity analysis. This prompted the archaeologists involved to wonder what that would even look like, and products of this effort were an SAA symposium and this volume. Although all the chapters in this book engage somewhat with the general uncertainties involved in computational modeling (data, model choice, model precision, coding errors, etc.), the primary focus throughout is sensitivity analysis (SA), which examines the effect of input parameters on model output and performance.

The first two chapters are introductions to the volume. The first chapter—by Brouwer Burg, Hans Peeters, and William Lovis—does an excellent job of defining and situating uncertainty studies in archaeology. It also nicely surveys how other fields have done this and how these might be applied by archaeologists. Chapter 2, by Lovis, summarizes attitudes by archaeologists toward computational modeling, and Lovis champions a pragmatic role for SA as a way to better understand the models archaeologists do create and to maximize what they learn from them by engaging in controlled experiments.

The middle chapters are all case studies, and each has a very different approach to SA, demonstrating the diversity of forms this approach can take. Chapter 3, by Peeters and Jan-Willem Romeijn, analyzes a simulation of hunter-gather behavior and land use in the postglacial Netherlands. They analyzed the robustness of results to variations in parameters to identify thresholds of importance for further consideration. Chapter 4, by Brouwer Burg, deploys formally sophisticated forms of SA on a model of hunter-gatherer land use. In a corner-test SA, she changed the parameter expressing desirability of different game species to extreme values, and in a Design of Experiments analysis, she varied these independently to determine which values were most significant to the outcome. In Chapter 5, John Carroll analyzes a very abstract model of cultural transmission. The chapter demonstrates interesting nonlinear results when the parameter showing the percentage change that someone can be influenced in a cultural transmission event is modified. In Chapter 6, Joshua Watts revisits work that was done to analyze both how long (in simulation time) a simulation of Hohokam pottery exchange should be run and in how many different iterations of each model configuration should be generated. He argues that analysts should be thoughtful about such values, rather than just picking arbitrary ones, and he demonstrates the steps that were used in this example. Chapter 7, by Andrew White, analyzes a demographic model for hunters and gatherers. The model has a feedback value that increases or decreases overall mortality as total population deviates from a threshold. This chapter explores the implications of sweeping that variable across various levels on the results.

The final chapters reflect on archaeological simulation more broadly. Chapter 8, by Thomas Whitley, considers the fundamental purpose of digital archaeological models. He encourages analysts to ask why a model is being built in the first place in order to determine how to verify its performance. This chapter astutely notes that archaeologists in general seem more open to accepting “vague nonspecific generalized and qualitative models of human behavior than simulations based on concrete numerical assumptions and scale-dependent digital datasets . . . because there are very often too many arguments for alternative assumptions about the inputs for each model in the simulation” (p. 150), and there are no easy alternatives. Finally, the concluding chapter by Sander van der Leeuw effectively summarizes all of the chapters. He encourages archaeological modelers to treat uncertainty generated by fragmentary knowledge about the past differently from uncertainty related to ontological concerns or model choice. This is because the first can never be reduced by modelers, and their contributions should work to actively reduce uncertainty in the other domains.

Collectively, chapters in this volume tend to share similar philosophies about modeling. Most advocate for simple models with fewer parameters that work (as van der Leeuw argues in Chapter 9) to develop “tools for thought” (p. 160) rather than detailed models of reality. They also tend to be pessimistic (as is Lovis in Chapter 2) that archaeological simulations will usually be able to be as precise as some models in other fields because of inherent archaeological uncertainties and biased and missing data. And they all struggle with the fact (observed by Whitley in Chapter 8) that fully documenting and exploring a model can take a lot of time, and it is nearly impossible to ever present all of these explorations to readers.

This is a well-crafted volume that is essential reading for those who construct archaeological simulations, and it is an educational read for any other archaeologist to reflect on how to grapple explicitly with uncertainty in our field.