A key question in cosmology is how to extend the Big Bang standard cosmology to create structures in the early universe, in other words, to create a theory explaining galaxy formation. An important observational goal is to find the "characteristic scale" of the early universe, a quantity which represents the clumping of matter at that stage in the evolution of the universe. (For example, the characteristic scale of a city would be about the size of a building, because that is the scale on which details in the city change significantly to a distant observer). The characteristic scale of the universe is determined by factors such as the Hubble parameter and the curvature of the universe (omega) and is thus of enormous interest to cosmologists.
How would the first structures in the universe form? Primordial fluctuations would be amplified exponentially by cosmic inflation. Or cosmic strings could pull matter together at a later time. The addition of dark matter complicates the scene. In a Hot Dark Matter model galaxies don't cluster very well, and galaxies would have to form by fragmentation of a huge supercloud. In a Cold Dark Matter, the reverse is true, galaxies would have to form by agglomeration of relatively small clouds. However, simulations indicate that the universe is best modeled by a mixed dark matter model.
COBE has been taking data for four years and has yielded important results regarding this characteristic scale, probing anisotropies (shown by temperature fluctuations) down to a 6o scale. Fluctuations on such a large scale have been found and have been confirmed by other experiments.
Going beyond to smaller-scale fluctuations may be difficult. They may be masked by a period of intense star formation in the early universe (at z=100) that re-ionized much of the gas, wiping out small-scale fluctuations. In short, COBE data is key in the fine-tuning of the standard model.