First, a (too) spectacular claim, then a spectacular {statistically} insignificant result!

First detection of inflationary gravitational waves probably did not occur in 2014

At the recombination epoch, the inflationary gravitational waves (IGW) contribute to the anisotropy of the CMB in both total intensity and linear polarization. The amplitude of tensors is conventionally parameterized by r, the tensor-to-scalar ratio at a fiducial scale. Theoretical predictions of the value of r cover a very wide range. Conversely, a measurement of r can discriminate between models of inflation. Tensor modes produce a small increment in the temperature anisotropy power spectrum over the standard [cosmological model] ΛCDM scalar perturbations at multipoles l<∼60; measuring this increment requires the large sky coverage traditionally achieved by space-based experiments, and an understanding of the other cosmological parameters. The effects of tensor perturbations on B-mode polarization is less ambiguous than on temperature or E-mode polarization over the range l<∼150...

Interstellar dust grains produce thermal emission, the brightness of which increases rapidly from the 100– 150 GHz frequencies favored for CMB observations, becoming dominant at ≥ 350 GHz even at high galactic latitude. The dust grains align with the Galactic magnetic field to produce emission with a degree of linear polarization [16]. The observed degree of polarization depends on the structure of the Galactic magnetic field along the line of sight, as well as the properties of the dust grains (see for example Refs. [17, 18]). This polarized dust emission results in both E-mode and B-mode, and acts as a potential contaminant to a measurement of r. Galactic dust polarization was detected by Archeops [19] at 353 GHz and by WMAP [2, 20] at 90 GHz. 

BICEP2 was a specialized, low angular resolution experiment, which operated from the South Pole from 2010 to 2012, concentrating 150 GHz sensitivity comparable to Planck on a roughly 1 % patch of sky at high Galactic latitude [21]. The BICEP2 Collaboration published a highly significant detection of B-mode polarization in excess of the r=0 lensed-ΛCDM expectation over the range 30 < l<150 in Ref. [22...]. Modest evidence against a thermal Galactic dust component dominating the observed signal was presented based on the cross-spectrum against 100 GHz maps from the previous BICEP1 experiment. The detected B-mode level was higher than that projected by several existing dust models [23, 24] although these did not claim any high degree of reliability.  

The Planck survey released information on the structure of the dust polarization sky at intermediate latitudes [25], and the frequency dependence of the polarized dust emission at frequencies relevant to CMB studies [26]. Other papers argued that the BICEP2 region is significantly contaminated by dust [27, 28]. Finally Planck released information on dust polarization at high latitude [29, hereafter PIP-XXX], and in particular examined a field centered on the BICEP2 region (but somewhat larger than it) finding a level of polarized dust emission at 353 GHz sufficient to explain the 150 GHz excess observed by BICEP2, although with relatively low signal-to-noise. [...] 

In this paper, we take cross-spectra between the joint BICEP2/Keck maps and all the polarized bands of Planck. [...]

Upper: BB spectrum of the BICEP2/Keck maps before and after subtraction of the dust contribution, estimated from the cross-spectrum with Planck 353 GHz. The error bars are the standard deviations of simulations, which, in the latter case, have been scaled and combined in the same way. The inner error bars are from lensed-ΛCDM+noise simulations as in the previous plots, while the outer error bars are from the lensed-ΛCDM+noise+dust simulations. Lower: constraint on r derived from the cleaned spectrum compared to the fiducial analysis shown in Figure 6.

[...] The r constraint curve peaks at r = 0.05 but disfavors zero only by a factor of 2.5. This is expected by chance 8% of the time, as confirmed in simulations of a dust-only model. We emphasize that this significance is too low to be interpreted as a detection of primordial B-modes. [...] 

In order to further constrain or detect IGW, additional data are required. The Planck Collaboration may be able to make progress alone using the large angular scale “reionization bump,” if systematics can be appropriately controlled [50]. To take small patch “recombination bump” studies of the type pursued here to the next level, data with signal-to-noise comparable to that achieved by BICEP2/Keck at 150 GHz are required at more than one frequency... During the 2014 season, two of the Keck Array receivers observed in the 95 GHz band and these data are under active analysis. BICEP3 will add substantial additional sensitivity at 95 GHz in the 2015, and especially 2016, seasons. Meanwhile many other ground-based and sub-orbital experiments are making measurements at a variety of frequencies and sky coverage fractions.

A Joint Analysis of BICEP2/Keck Array and Planck

DataBICEP2/Keck and Planck Collaborations

30 January 2015