At its simplest, all information processing that people and organizations perform is a series of Input-Processing-Output steps, chained together in complex and iterative ways. We often speak of "value-added" processing as the steps that really transform the data into something significant to an organization's strategic objectives or needs. For an oil company, that might be the step that takes data from seismic tests and field surveys and transforms it into the "drill a test well here" decision; a refinery operator in the same company might see the real value-added processing as the real-time process control actions that run the refinery, moment by moment. In one case, safety and responsiveness require the value-added processing to be very, very local to the refinery; in other cases, where the processing is done vs. where the input and/or output occur can be many, many "network hops" apart. How does the rapidly-changing set of network capabilities, and their costs, affect how a systems designer needs to think about this "split" of what processing occurs where in their system, with what kind of network links in between?