News update: Reviewers claim CMB analysis was flawed
Did scientists make enough allowance for space dust? Phil Dooley reports.
Harvard astrophysicist John Kovac’s announcement in March that his team had discovered gravitational waves rocked the world. Not only was it a confirmation of Einstein’s 1916 theory of General Relativity, it was the “smoking gun for inflation”, a 1980 refinement of the Big Bang theory. But in releasing the news ahead of peer review and publication the scientists may have jumped the smoking gun and detected nothing but space dust.
“The effect on the world was a shockwave, from the halls of government to school children around the world,” Paul Steinhardt, director of the Centre for Theoretical Physics at Princeton told a special seminar at the university in May. “It reminds us that we bear a certain responsibility to get it right.”
At the same meeting, Raphael Flauger, a researcher at Princeton's Institute for Advanced Study, having delved into the data collected by Kovac’s Antarctic-based telescope, BICEP2 (Background Imaging of Cosmic Extragalactic Polarisation 2), revealed what appeared to be a devil in the detail. The BICEP2 team claimed to have detected the imprint of gravitational waves in the ancient polarised radiation left over from the Big Bang – the so-called cosmic microwave background (CMB). Those imprints took the form of swirls known as b-modes. But Flauger suggested those swirls may be caused by space dust looping the loop in the Milky Way’s magnetic field. The dust – tiny particles ejected from supernovae and other material – can also polarise light and create swirly patterns.
The BICEP2 team was well aware that space dust could interfere with their results. To deal with it, the they used six different models to predict the patterns it could produce. They then subtracted the six patterns, one at a time, from the raw data, and compared the results. In theory, the signals left over would expose the longed-for gravity waves.
The problem was getting the right data on the cosmic dust in the first place. The Planck space telescope, which mapped the CMB from space and also looked at the effects of space dust - was the best source but its data had not yet been formally released. So the BICEP2 team got the next best thing. The Planck team had presented their preliminary results on a slide at a conference. The BICEP2 team extracted that preliminary data – “a valid approach”, according to Flauger. “To make progress, you have to either wait – which no one wants to do right now – or one has to be creative. And that’s what the BICEP2 team did.”
'It’s peer review, in a totally new and fascinating way.'
Unfortunately, says Flauger, the margin for error in doing so was larger than BICEP2 scientists realised. So the team didn’t correctly calculate the uncertainty in the data. In other words, foreground dust could make a much bigger contribution to the pattern of swirls than the BICEP2 team accounted for.
Nevertheless the BICEP2 team stands by its data. Jamie Bock, from Caltech, a principal investigator on the BICEP2 team, points out that the correction for the dust did not hinge solely on the preliminary Planck data. “We presented four lines of evidence about dust. People can attack each one of those lines of reasoning – and they have – but we say that together, the data disfavour a dust interpretation.”
University of Melbourne astrophysicist Katie Mack followed the debate as it erupted across social media. “It’s peer review, in a totally new and fascinating way,” she says. “They got the result out fast, and let the whole community evaluate it. I am really enjoying being able to have access to all the information and see experts sharing their thoughts and expertise on blog posts and so on.”
Bock says the BICEP2 team debated whether to open their results up to widespread scrutiny. “We expected comment, which is part of the reason we did it, opening it up to public review.” The work has also been submitted to a peer-reviewed journal, and subsequently modified based on reviewers’ comments, he adds.
While other areas of science frown on public dissemination of results prior to publication in a journal, physicists in the 1990s pioneered the idea of uploading their unpublished papers for community comment. However in this case, with the stakes so high – no less than proving the existence of gravitational waves and bolstering inflation theory – and the reputations of dozens of astrophysicists on the line, the usually polite tenor of the science blogosphere has turned strident.
“I have heard a few people disparaging the BICEP team because people got really, really excited and now they feel they were misled,” says Mack “I don’t feel that way. I hope their result is solid, but I am less confident than I was.”