Problem with denoising BICEP2 data? This question pertains to this article which talks about why the BICEP2 measurements of B-mode polarization in Cosmic microwave background radiation turned out to be noise from galactic stardust. They go on to add that the Plank data of the noise does not lend to getting a good B-mode estimate from the BICEP2 data.

However, the results of the joint assessment would suggest that
  whatever signal BICEP2 detected, it cannot be separated at any
  significant level from the spoiling effects. In other words, the
  original observations are equally compatible with there being no
  primordial gravitational waves. "This joint work has shown that the
  detection of primordial B-modes is no longer robust once the emission
  from galactic dust is removed," Jean-Loup Puget, principal
  investigator of Panck's HFI instrument, said in the Esa statement.
  "So, unfortunately, we have not been able to confirm that the signal
  is an imprint of cosmic inflation."

Can someone explain why exactly the detection of primordial B-modes is not robust even after the emission from galactic dust is removed? And what are the proposed workarounds to tackle this problem?
 A: The short answer is that galactic dust can fully account for the findings, with or without the existence of Primordial Gravitational Waves, which amounts to "no proof." 
The longer answer is that they were looking for a particular pattern in the polarization of the CMB called B mode polarization, as this was the pattern they expected to see under gravitational wave conditions. B mode results in fields that tend to look like magnetic fields (the curls tend to point in circles).
But the CMB can also be polarized in E mode, which produces fields that tend to point in particular directions (similar to electric field lines). Here's a basic pic-

The problem is that not only can galactic dust reflect or refract the photons it interacts with, but galactic dust is also often susceptible to the effects of electromagnetic fields in space. As a result, dust particles can align within these magnetic fields, creating an electromagnetically-induced illusion of a gravitationally-distorted spacetime. So if enough dust particles reside within a given region, we must conclude that to an observer, a B-mode-like polarization pattern will form, with or without the presence of Primordial Gravitational Waves.
I believe there is (or was) a third collaboration in process, however (drum roll please) . . . LIGO announced just last week the confirmation of gravitational waves resulting from the merger of two black holes they were able to observe in September 2015, which produced gravitational waves strong enough to be measured. 
I see what you mean about Puget's statement, though. It seems almost misleading at first, but could simply be a translation issue (English doesn't appear to be his language of origin). In any case, it sounds like he was trying to say that the CMB signal isn't robust enough to be able to discern from the interference from galactic dust.
A: I would like to add to the other correct answer, that BICEP2 is an amazing detector, and the original announcement came out having subtracted the dust signal in the area they were looking using published Planck data at the time. Unfortunately Planck came up with new dust maps of the sphere which showed a lot of dust in the angular window BICEP was sensitive to. The analysis with Planck discussed in your question summarizes the joint publication.
Nevertheless, combining data from Planck, which relies on fitting mathematical curves to the whole sky and has large errors, to get at polarizations, with data from BICEP2 which has extreme local accuracy in a narrow angular window, is not the best way to treat the problem.  Ongoing is work on BICEP3 and we will have to see what happens.
