Most scientific data will never be directly examined by scientists; rather it will be put into online databases where it will be analyzed and summarized by computer programs. Scientists increasingly see their instruments through online scientific archives and analysis tools, rather than examining the raw data. Today this analysis is primarily driven by scientists asking queries, but scientific archives are becoming active databases that self-organize and recognize interesting and anomalous facts as data arrives. In some fields, data from many different archives can be cross-correlated to produce new insights. Astronomy presents an excellent example of these trends; and, federating Astronomy archives presents interesting challenges for computer scientists.