Myths


The map must be scanned at many angles - "cross linked"  
Not true if you just want the power spectrum from a non full sky experiment

A given map pixel must be measured at many polarization angles
Not true - 2 (well separated) angles is in principle sufficient, and a small range of specific angles will supress systematics just as well as a continuous distribution.

The relative gains of the A/B detectors must be known with exquisite precision when doing direct pair difference to avoid T->Pol leakage
Not true so long as the gain errors are random

A/B beam differences are a big problem
But only at angular scales comparable to those differences