The recent discovery of a high optical depth tau to Thomson scattering from the Wilkinson Microwave Anisotropy Probe (WMAP) data implies that significant reionization took place at redshifts z > 6. This discovery has important implications for the sources of reionization and allows, for the first time, constraints to be placed on physical reionization scenarios out to redshift z similar to 20. Using a new suite of semianalytic reionization models, we show that the high value of tau requires a surprisingly high efficiency epsilon of the first generation of UV sources for injecting ionizing photons into the intergalactic medium. We find that no simple reionization model can be consistent with the combination of the WMAP result with data from the z less than or similar to 6.5 universe. Satisfying both constraints requires either of the following: (1) H-2 molecules form efficiently at z similar to 20, survive feedback processes, and allow UV sources in halos with virial temperatures T-vir < 10(4) K to contribute substantially to reionization, or (2) the efficiency epsilon in halos with T-vir > 10(4) K decreased by a factor of greater than or similar to 30 between z similar to 20 and z similar to 6. We discuss the relevant physical issues to produce either scenario and argue that both options are viable and allowed by current data. In detailed models of the reionization history, we find that the evolution of the ionized fractions in the two scenarios has distinctive features that Planck can distinguish at greater than or similar to 3 sigma significance. At the high WMAP value for tau, Planck will also be able to provide tight statistical constraints on reionization model parameters and elucidate much of the physics at the end of the dark ages. The sources responsible for the high optical depth discovered by WMAP should be directly detectable out to z similar to 15 by the James Webb Space Telescope.