Transparency and uncertainty are major challenges for model builders. We can look at modeling as a tool for communication and collaboration among technical experts, modelers, decision makers, and other stakeholders to help build their collective understanding. From that view, it is essential that a model be transparent, to help participants see key model assumptions and implications. Ideally, a computer model and its documentation should be an integrated whole, so that changes to one are automatically reflected in the other. Explicit probabilistic representation of uncertainty also improves transparency, to assess how much -- or little -- confidence is appropriate for the model results. Uncertainty analysis -- assigning credit or blame for uncertainty among the various uncertain inputs and assumptions -- can be a major source of insight, and guide to a more cost-effective process for model development. My colleagues and I designed Analytica as modeling software to explore and illustrate practical and convenient ways to assist modelers in improving transparency and treatment of uncertainty. I will illustrate the presentation with sample Analytica models. These may include -- according to audience interest -- energy savings from daylighting in buildings, TAF (Tracking and Analysis Framework for evaluating the Clean Air Act) and the SEDS (Stochastic Energy Deployment System) under development by NREL, LBL, and other national labs.