Cosmic covariance and the low quadrupole anisotropy in the WMAP data

Low quadrupole power in the cosmic microwave background (CMB) temperature anisotropies has been a puzzle since WMAP data release. In this talk I will demonstrate that the minimum variance optimization (MVO), a methodology used by many authors including the WMAP science team to separate the CM13 from foreground contamination, serves not only to extract the CM13, but to subtract the "cosmic covariance", an intrinsic correlation between the CM13 and the foregrounds. Such subtraction induces low variance in the signal via MVO, which in turn propagates into the multipoles, causing a quadrupole deficit with more than 90% CL. As we do not know the CM13 and the foregrounds a priori, and their correlation is subtracted by the MVO in any case, there is therefore an unknown error in the quadrupole power even before the cosmic variance interpretation. We combine the MVO and Monte Carlo simulations, assuming CM13 is a Gaussian random field, and the estimated' quadrupole power falls in [308.13, 401.97] mu K-2 (at 1 - sigma level).