Historically, landslides have had serious socio-economic and ecological effects. In response, many researchers have developed landslide models to help decision making and policy to reduce landslide-related losses. These models are extracted from geotechnical engineering or geomorphology rather than designed explicitly to aid decision making. Engineering and geomorphology have disparate practices regarding what models are useful and how models are used in the pursuit of successful work. Geotechnical engineers employ a few familiar techniques of slope-stability analysis to design structures that minimize or eliminate the chance of slope failure. Geomorphologists are concerned with understanding the origins and evolutions of landforms, focusing on the processes of erosion, transport, and deposition. Depending on the particular research questions, a geomorphologist may choose from many models and spatio-temporal scales to develop insight, motivate experiments, and inspire new questions. Many landslide modelers seem to lose sight of the discipline in which particular models originated and how this influences the applicability of a model to dissimilar contexts. Careless extraction of models from geotechnical engineering or geomorphology to public decision situations can lead to confusion, disagreement, and disappointment among public participants, engineers, and scientists. Landslide models should be developed or appropriated with the foremost consideration being the intended application, whether it is engineering design, geomorphic research, or decision support. The consensus building approach to group decision-making is presented as a normative application with which to rethink landslide models for decision support. Five considerations or criteria for evaluating models are proposed: adaptability, transparency, discursiveness, plurality, and parsimony. Applying these criteria to existing modeling methodologies -- including heuristic, historical, process-based, and probabilistic -- it is suggested that, presently, the characteristics of heuristic and process-based models are best suited for decision-support applications.