The main ideas is that we need different types of models for different scopes, and that this models can be implemented with sound informatics, and without having to redo it again from the scratch. The talk, in a sense, complements the post I made on my future research activities, by specifying the "methods" with which I will envision to do them. Many live for the motto: getting the right answer for the right reason. I support the idea that for getting the right answer you need sound models. The link under the picture will bring you to the presentation.