http://arxiv.org/abs/hep-th/0610211 Authors: Raphael Bousso Comments: 38 pages, 6 figures. Based on a talk given at the conference in honor of the 90th birthday of Charles Townes (Berkeley, October 2005), and on a colloquium at Stanford University After reviewing the cosmological constant problem - why is Lambda not huge? - I outline the two basic approaches that had emerged by the late 1980s, and note that each made a clear prediction. Precision cosmological experiments now indicate that the cosmological constant is nonzero. This result strongly favors the environmental approach, in which vacuum energy can vary discretely among widely separated regions in the universe. The need to explain this variation from first principles constitutes an observational constraint on fundamental theory. I review arguments that string theory satisfies this constraint, as it contains a dense discretuum of metastable vacua. The enormous landscape of vacua calls for novel, statistical methods of deriving predictions, and it prompts us to reexamine our description of spacetime on the largest scales. I discuss the effects of cosmological dynamics, and I speculate that weighting vacua by their entropy production may allow for prior-free predictions that do not resort to explicitly anthropic arguments.