Back in Seattle, Bovy, Hennawi, Myers, and I discussed the most general extreme deconvolution approach to photometric redshifts, with quasars as the sources of interest. Since it has been a whole week, Bovy has, of course, implemented fully what we had in mind! It performs incredibly well, naturally incorporates GALEX and UKIDSS data where we have them, can be marginalized along directions other than the photometric directions to give fully empirical (hierarchical) prior PDFs over quasar spectral energy distributions, and improves quasar target selection. It is pretty sweet! And XD is so fast now, he was able to optimize the (more than) ten-thousand parameters in an hour or two. Have I mentioned that I love my job?