Most data sets have more than 3 variables, which makes it hard for you to visualize how observations are distributed in data space. Displaying one plot at a time focusing on a few variables is like peeping at the world through a straw. You can do better by juxtaposing several plots, just as you can juxtapose plot printouts on a desk. Although this approach provides you with many viewpoints to look into the data set at the same time, it is difficult to integrate the information contained in individual plots.
Argos uses multi-window graphics to look into the data from different angles. Argos's graphic windows emulate viewports on the ``wall'' that is between the data space and you. Not only can you view certain facets of the data in a viewport, but you can also manipulate the data by operating a ``manipulator''. The data visible through a viewport can be the complete data set being analyzed or a subset of data. Real viewports have the property that changes done to objects through one viewport will be visible through all viewports which can see those objects; so do Argos's viewports. It is this real-viewport-like behavior that helps you integrate the information contained in separate graphic windows into a coherent image of the data as a whole.