Barriers and model curation issues associated with rapid adaptation of empirical legacy ABM in response to urgent policy maker queries

In this paper, we present the methodology and results of the first test of the front-end of a larger protocol guiding a participatory modelling process to rapidly adapt empirical legacy ABMs to effectively respond to policy queries. The ultimate goal of the research is to create a process that could be completed in timespans of weeks rather than several months. The five stages of the front-end protocol tested here cover the formal elicitation of the policy query, the co-construction with the policy maker of requirements for an adapted ABM to answer it, and the process leading to the re-coding of that model. The protocol is designed for the involvement of four actors: a policy maker; a knowledge engineer; a modeller; and a programmer. The paper concludes with a description of the key barriers identified by the test. These barriers include problems of re-running the legacy ABM from existing code due to changes in software versions since its original development; differences in coding styles between the original developers and the modelling team charged with its adaptation; limits to the descriptive power of ODD; dataset confidentiality and differences in data processing methods; time pressures resulting in design/coding decisions that might lead to errors or model artefacts, etc. We suggest investment in professional model curation as a way to overcome these barriers. The goal of which would be the identification, preparation, and maintenance of legacy ABMs in readiness for their rapid adaptation by modelling teams to meet future policy maker requests.