We provide evidence that the obliquities of stars with close-in giant planets were initially nearly random, and that the low obliquities that are often observed are a consequence of star-planet tidal interactions. The evidence is based on 14 new measurements of the Rossiter-McLaughlin effect (for the systems HAT-P-6, HAT-P-7, HAT-P-16, HAT-P-24, HAT-P-32, HAT-P-34, WASP-12, WASP-16, WASP-18, WASP-19, WASP-26, WASP-31, Gl 436, and Kepler-8), as well as a critical review of previous observations. The low-obliquity (well-aligned) systems are those for which the expected tidal timescale is short, and likewise the high-obliquity (misaligned and retrograde) systems are those for which the expected timescale is long. At face value, this finding indicates that the origin of hot Jupiters involves dynamical interactions like planet-planet interactions or the Kozai effect that tilt their orbits rather than inspiraling due to interaction with a protoplanetary disk. We discuss the status of this hypothesis and the observations that are needed for a more definitive conclusion.