In an important sense, auction theory represents one of the most significant developments in economics and game theory in the second half of the twentieth century. Much of game theory, as with general equilibrium theory, is stillborn, unable to guide meaningful empirical investigation because of its failure to come to grips with exchange institutions and thus with process. But the modeling of auctions is predicated directly upon the allocation and message rules of alternative auction institutions. The distinction between agent messages and agent allocations, the effect of the environment (the number of bidders, parameters defining the probability distributions of values or cost, etc.), and assumptions about agent expectations and agent maximizing behavior are all explicit and clearly specified. In environments in which alternative auction institutions are equivalent, this institution-free property is derived as a theorem instead of being an implicit assertion. Auction theory does more than begin with the extensive form of a game, it begins with the various extensive forms we observe in the economy (with the exception, perhaps, of the Dutch auction, as discussed below). Consequently, it is able to guide empirical testing programs with a minimum overlay of empirical interpretation, or what Lakatos calls “initial conditions” and ceterus paribus clauses.