The Case for Repeatable Analysis with Energy Economy Optimization Models
Energy economy optimization (EEO) models employ formal search techniques to explore the future decision space over several decades in order to deliver policy-relevant insights. EEO models are a critical tool for decision-makers who must make near-term decisions with long-term effects in the face of large future uncertainties. While the number of model-based analyses proliferates, insufficient attention is paid to transparency in model development and application. Given the complex, data-intensive nature of EEO models and the general lack of access to source code and data, many of the assumptions underlying model-based analysis are hidden from external observers. This paper discusses the simplifications and subjective judgments involved in the model building process, which cannot be fully articulated in journal papers, reports, or model documentation. In addition, we argue that for all practical purposes, EEO model-based insights cannot be validated through comparison to real world outcomes. As a result, modelers are left without credible metrics to assess a model's ability to deliver reliable insight. We assert that EEO models should be discoverable through interrogation of publicly available source code and data. In addition, third parties should be able to run a specific model instance in order to independently verify published results. Yet a review of twelve EEO models suggests that in most cases, replication of model results is currently impossible. We provide several recommendations to help develop and sustain a software framework for repeatable model analysis.