A diagnostic model design system is provided that is driven by a combination of statistical aggregation methods, ensemble analysis techniques, and model specification strategies. The system takes as input a collection of disparate information (e.g. predictive machinery, numerical models, conceptual models, and/or expert opinions) where the information is attempting to estimate a common process or outcome. Optionally the system can also take real-world observation data. Through the system: the strengths and weaknesses of information and information sources can be identified; information sources can be ranked by reliability; and, a basis for designing and validating new, more reliable and accurate information sources (new models, machinery, etc) is provided.