BICEP-2 Data Probably Flawed

A flaw in methodology has been identified that potentially undermined the key conclusion of the BICEP-2 experiment that it had detected significant tensor mode gravitational waves at a high level of statistical significance (a conclusion strongly supporting certain kinds of cosmological inflation theories). (One of the first blogs to note this fact, strongly hinted at in a Planck experiment sponsored pre-print, has a post here that took several days to gain traction).

Apparently, a piece of source data from the Planck experiment that was used to adjust the BICEP-2 data was misinterpreted. Once corrected, the BICEP-2 result on tensor modes will definitely be less dramatic, although it may hint at some tensor modes, but without the same statistical significance and without the same intensity.

From the start, the inconsistency between the BICEP-2 result and the preliminary Planck results had cast doubt on the results. As I explained when the results were announced:

Given the previous data, the best combined fit is now r=0.10 to 0.11, assuming that the BICEP-2 result isn't flawed in methodology in some way, which is an entirely plausible possibility that will look more plausible if it is not confirmed by the Planck polarization data later this year, and several other experiments that will be reporting their results within the next year or two. Skepticism of the result, in the absence of independent confirmation by another experiment (Jester puts the odds that this result is right at only 50-50) flows from the fact that the value reported is so different from the consensus value from all previous experiments, with the results in a roughly three standard deviation tension with each other.

Now, it seems that the doubt was justified. At least, these methodological errors were discovered much sooner than the notorious OPERA experiment's superluminal neutrino results, although still, as many as 300 papers based on the early and probably flawed BICEP-2 results from this March have already been written and published in pre-print form so far.

Another important consequence of the BICEP-2 results, had they been true, would have been that they would have caused the cosmological evidence to clearly favor the existence of four, rather than three, kinds of light neutrinos. The evidence that BICEP-2 had provided for this beyond the Standard Model number of neutrino species has also been called into question.

UPDATE (May 12, 2014 5:20 p.m. MDT):  Jester adds the following update in the comments:

One comment: indeed, the issue affects only BICEP's DDM2 foreground emission model. But DDM2 is what they single out in the paper as their best model. Recall that in the original analysis using DDM2 shifts the central value of r from 0.2 down to 0.16. Now we know that their best model underestimates the foreground, so we know the significance must go down further. By how much, I don't know. Various rumors place the significance of the corrected signal between 0 and 2 sigma.

A corrected signal between 0 and 2 sigma with the old error bar of +0.7/-0.5 would imply r between 0 and 0.1 with the new analysis and would have a central value consistent with Planck.  END UPDATE.

Off topic but also notable: one of a number of final reports from Fermilab's D0 experiment has reached the following conclusion:

We present an overview of the measurements of the like-sign dimuon charge asymmetry by the DO Collaboration at the Fermilab Tevatron proton-antiproton Collider. The results differ from the Standard Model prediction of CP violation in mixing and interference of B^0 and B^0_s by 3.6 standard deviations.

Evidence at the 3.6 sigma level from a reputable particle physics experiment of beyond the Standard Model phenomena is indeed notable, and confirms previous published results from 2010 and 2011. (Jester discussed the similar but slightly stronger 2011 results.)

But, given recent concern over the declining accuracy of the Run-II muon detectors in the D0 experiment over time, the low power of Fermilab relative to the Large Hadron Collider (LHC)'s experiments (ATLAS, CMS, LHCb), and the apparent lack of clear replication of this result by the other experiment at Fermilab (CDF) or the B-factories (Belle and BaBar), one should not rush to assume that this result is correct.

Notably, the D0 paper whose pre-print came out today does not discuss the replication of this result, or lack thereof, by other experiments.  According to a power point presentation by D0 related to the previous result, CDF saw something, but only at the 2 sigma level.  A result in this level could seem notable when it isn't, for example, simply because one or two important sources of systemic error were considered but underestimated in magnitude.  Early LHCb results strongly constrain the magnitude and nature of the anomaly reported by D0.  BaBar experiment results also fail to strongly confirm this result and the other "B-Factory" at Belle has also not produced dramatic confirmations of this result.  A possible flaw in the Standard Model expectation calculation used by D0 has also been identified.  More recent LHCb results and B-factory results are consistent with the Standard Model expectation.  So, the result from D0 most likely represents experimental error, a statistical fluke, or a problem in modeling the expectation, rather than new physics.

It is not immediately obvious to me what kind of beyond the Standard Model theories would predict the level of dimuon charge asymmetry observed by D0 without observing other phenomena that have not been observed at Tevatron or the LHC.