1/3 = 1 - 2/3.
Observers often complain about the problems posed by foregrounds in their data. As a theorist the solution is obvious: just don't observe the foregrounds.
It is important when answering such questions to be objective, to look at each experiment on its own merits, and not to be swayed by any personal biases. So my answer is the Balloon-borne Anisotropy Measurement at UBC, and the Low Frequency Instrument part of the Planck mission (both of which I am coincidentally involved with).