© Copyright JASSS

JASSS logo ------

Uncertainty and Sensitivity Analysis in Archaeological Computational Modeling (Interdisciplinary Contributions to Archaeology)

Burg, Marieka Brouwer, Peeters, Hans and Lovis, William A. (eds.)
Springer-Verlag: Berlin, 2016
ISBN 978-3319278315 (hb)

Order this book

Reviewed by Tom Brughmans
Department of Computer and Information Science, University of Konstanz

and Iza Romanowska
Faculty of Electronics and Computer Science, University of Southampton

Cover of book This edited volume aims to address the challenges related to assessing uncertainty in the design, creation and publication of archaeological computational models, and to impress upon archaeologists the importance of doing this explicitly as part of their computational modelling work. It signals an end to the uncritical enthusiasm that characterised the early adoption phase of archaeological computational modelling (in particular complex systems simulation) and marks the beginning of a more critical and self-reflective phase, with researchers paying more attention to the epistemological principles behind simulation and ways of fine-tuning computational models to make solid contributions to the study of the human past.

The introduction by the editors sketches out the need for more discussion on the later stages of model development: experiment design, recognising patterns in results and analysing them. The subsequent three chapters address, respectively, challenges to research design and theory building in archaeological modelling (by Lovis), dealing with uncertainties and model selection in exploratory computational modelling (by Peeters and Romeijn), and Geographical Information Systems (GIS) methods for exploring human socionatural dynamics (by Brouwer Burg). The remainder of the chapters deal explicitly with simulation and with agent based modelling (ABM) more specifically, the most popular simulation technique among archaeological modellers. Carroll performs sensitivity analysis to detect nonlinear patterns in ABM results, and White tests the impact of mortality on other demographic variables in his ABM. The papers by Watts and Whitley are of particular interest. Watts provides a summary of techniques to determine the parameters of an experiment design – number of time steps, agents and runs. Particularly useful is his practical demo of how to implement these statistical techniques in an agent-based model. In contrast, Whitley focuses on the theoretical, epistemological aspects of archaeological simulation. He sets out to define and categorise archaeological simulations and their goals. In the process, he points out the long history of an archaeology-specific ‘analogue simulation’, that is, experimental studies such as tool making experiments or battle re-enactments.

This is a volume written and edited by archaeologists for archaeologists, rather than computer scientists. This is evident from a number of issues problematized throughout the volume that have already been extensively and repeatedly addressed by the modelling community and through best practice guidelines. For example, code verification is repeatedly invoked as a key step in model development yet none of the authors mention any of the common techniques used by research software engineers, such as unit testing, defensive coding or test-driven development. Archaeology also has a strong tradition of spatial analysis research, which is reflected in the prominence of GIS approaches in the volume. Almost half of the papers are dedicated to or draw examples from spatial predictive modelling and palaeoenvironmental modelling, and the prominent role of spatially explicit environments in archaeological computational models is often referred to. Van der Leeuw’s summary of the volume provides a digression. The majority of authors present data-heavy and (often) spatial models comprising many variables striving to capture as much real-world complexity as possible. In contrast, Van der Leeuw grounds archaeological simulation in complexity science and its simple abstract ‘toy-models’ which assemble into, in van der Leeuw’s words, a composite “bee’s eye view” of different but overlapping perspectives.

This volume will no doubt stay on archaeological modellers’ bookshelves for years to come. In addition, the papers by Whitley, Van der Leeuw and the introduction to the volume serve as a useful reference to archaeological computational modelling for graduate students, non-modellers, and non-archaeologists.

-------

ButtonReturn to Contents of this issue