"Galaxy formation is a solved problem." That was the most memorable quote from the last meeting I attended at Cambridge, the 1996 conference on the Hubble Deep Field. The prominent astronomer who offered this sentiment was perhaps a little premature, but his excess of enthusiasm was forgivable considering the stream of beautiful data on the high redshift universe that had just begun to emerge. That stream has exponentiated over the last 8 years, and it may well be that we can solve the problem of galaxy formation within the next couple of decades.
Before that is possible, however, we need to understand star formation. We have already solved the basic problem of stellar structure and evolution. That can fairly be said to be the primary accomplishment of astrophysics in the 20th century (especially because it had been several million years since humans had first wondered about the stars!). There are only a few remaining dark corners of the evolutionary process. But one is crucial: star formation. This is central to galaxy astrophysics, but the deficiencies in our understanding are obvious. For instance, faulty prescriptions for star formation are thought to be the culprits in discrepancies between predictions of CDM models for galaxy formation and the observations.
A great deal of observational firepower will be directed at the problem of star formation in the coming years, but we already know one essential fact: star formation is a collective process. Most stars (perhaps nearly 100% in our Galaxy) form in clusters. There are strong interactions with the surrounding environment and among protostars. Quantifying feedback processes, both positive and negative, is a key to understanding star formation. All this means that star formation is a more difficult problem than was the astrophysics of isolated stars. Progress will be importantly informed by observations of other galaxies and a wider range of environments than are found in our Galaxy.
Starbursts are important because they are clearly a collective phenomenon and represent one extreme of the star formation process. Furthermore, they are bright enough to be detected throughout the observable universe and can serve as tracers of the cosmic history of star formation. This conference provided vivid testimony to the importance of starbursts as keys to galaxy evolution and to the speed with which our understanding of them is being transformed by the flood of new data, especially at high redshifts.