We are recording and analyzing a set of news feeds and constructing a real-time database of the content as represented by what is said, whose image is present, and the emotional state of the images. This data is then being used to create elements that can be assembled both automatically and interactively into a re- al-time newscast that is no more than 5 seconds behind the actualities in the real world. The system uses a Scratch-like graphical programming environment to control assembly, and allows people to create their own content-based macros and programs so that media distribution can be data rather than video streams.