More bark than bytes? Reflections on 21+ years of geocomputation
Richard Harris; David O’Sullivan; Mark Gahegan; Martin Charlton; Lex Comber; Paul Longley; Chris Brunsdon; Nick Malleson; Alison Heppenstall; Alex Singleton; Daniel Arribas-Bel; Andy Evans (2017). Environment and Planning B: Urban Analytics and City Science, 44(4), 598-617. DOI: 10.1177/2399808317710132
Abstract
This year marks the 21st anniversary of the International GeoComputation Conference Series. To celebrate the occasion, Environment and Planning B invited some members of the geocomputational community to reflect on its achievements, some of the unrealised potential, and to identify some of the on-going challenges.
Extended Summary
This research examines the evolution and current state of geocomputation as a field, marking 21 years since the inaugural International GeoComputation Conference in Leeds in 1996. The paper presents reflections from eleven prominent geocomputation researchers who evaluate the field’s progress, challenges, and future directions in the context of Big Data and urban analytics. The research methodology involves expert commentary and retrospective analysis of the field’s development against the original eight challenges outlined at the first conference. These challenges focused on leveraging computational power to achieve better analytical solutions to geographical problems, including improved resolution of computational models, intensive statistical methods, optimisation techniques, machine learning applications, and spatial data mining. The study finds that geocomputation has achieved mixed success against its original objectives. Significant progress has been made in machine learning applications, with neural networks, decision trees, and genetic algorithms now regularly appearing in geographical literature. The development of geographically weighted statistics represents a particular success, enabling researchers to explore spatial variation rather than averaging it away. However, progress in high-performance computing has been limited, with few spatial algorithms successfully parallelised for modern computing architectures. The research identifies several ongoing challenges facing geocomputation. The field struggles with defining its distinct identity beyond ‘doing geography with computers’, though this broad scope remains both a strength and weakness. Big Data presents opportunities but also risks, particularly around data quality, theoretical grounding, and the tendency toward data mining without clear research questions. Agent-based modelling shows promise for urban systems but faces significant calibration and validation challenges. Key concerns include the need for reproducible research practices, better integration between academia and industry, and addressing gender balance and geographical diversity within the geocomputation community. The paper argues that improved engagement with behavioural frameworks and ‘softer’ individual-level data is essential for realistic modelling of geographical systems. The broader significance lies in geocomputation’s potential contribution to urban analytics and smart cities initiatives. The research suggests that combining Big Data, agent-based modelling, and dynamic calibration could enable reliable forecasting of urban dynamics. However, this requires addressing methodological challenges around data assimilation, population bias in digital engagement, and ethical considerations. The study concludes that geocomputation’s future success depends on maintaining core principles of rigour, sympathy, and imagination whilst engaging more effectively with contemporary data science developments and societal challenges including sustainability and social equity.
Key Findings
- Geocomputation has made significant progress in machine learning applications but limited advancement in high-performance computing parallelisation over 21 years.
- Geographically weighted statistics represent a major success, enabling exploration of spatial variation rather than averaging away geographical differences across maps.
- Big Data opportunities risk producing answers to arbitrary questions without theoretical grounding, requiring better integration of exploratory and confirmatory analysis approaches.
- Agent-based modelling shows promise for urban systems but faces critical challenges in calibration, validation, and capturing realistic individual human behaviour patterns.
- The field needs stronger industry-academia collaboration, reproducible research practices, and better integration of emotional and belief-centred relationships between society and space.
Citation
@article{harris2017more,
author = {Richard Harris; David O’Sullivan; Mark Gahegan; Martin Charlton; Lex Comber; Paul Longley; Chris Brunsdon; Nick Malleson; Alison Heppenstall; Alex Singleton; Daniel Arribas-Bel; Andy Evans},
title = {More bark than bytes? Reflections on 21+ years of geocomputation},
journal = {Environment and Planning B: Urban Analytics and City Science},
year = {2017},
volume = {44(4)},
pages = {598-617},
doi = {10.1177/2399808317710132}
}