The Planck paper with results on dust in the BICEP2 patch of sky is now out, see here. I’m sure experts will weigh in soon and I’ll link to such discussions, but my non-expert take is that Planck is saying that what BICEP2 saw is likely just dust. See section 6 of the paper, especially figure 9 which appears to show that BICEP2’s claimed value of r=.2 is just what you’d expect from dust.
Update: More details from Natalie Wolchover and Sean Carroll.
Looks like Scientific American will have to pulp this month’s magazine, with its Lawrence Krauss cover story about how BICEP2 is experimental evidence for quantum gravity and the multiverse.
Update: For more press coverage, see Nature, New Scientist, BBC News, The Guardian, the Washington Post, the Daily Mail and the New York Times.
The best explanation for all this that I’ve seen of course is from a blogger, Sesh Nadathur at Blank on the Map.
Update: Jester has a sensible take on this fiasco here. It now seems that release of the full Planck polarization results has been pushed back from October to “late November”, just before the early December conference planned long ago to discuss the results. The joint analysis of BICEP2/Planck data that will show if there’s any evidence of something besides dust is supposed to be released at the same time.
Sean Carroll weighs in here:
I never quite grasped the full genius of the band known as Kansas, until now.
“Dust In The Wind”
I close my eyes only for a moment, and inflation’s gone
My Nobel dreams pass before my eyes, a curiosity
Dust in the wind, all inflation is is dust in the wind
Same old song, just a drop of hype in the CMB
Inflation crumbles to the ground, though we refuse to see
Dust in the wind, all inflation is is dust in the wind
Now, don’t hang on, nothing lasts forever but the earth and sky
It slips away, and all your money won’t another primordial wave buy
Dust in the Linde, all inflation is is dust in the wind (all inflation is is dust in the wind)
Dust in the wind (inflation is dust in the Linde), inflation is dust in the wind (the wind)
Verbatim reader comment at Carroll’s blog, under the reader name George Efstathiou [who looks to my eye to be either the biggest cheese at Planck or close to it]:
“As a member of the Planck Science Team, I would urge caution concerning the interpretation. What we are saying is that polarised [sic] dust emission in the BICEP2 field is high. But it may be that there is something left in the BICEP2 signal that can be attributed to gravitational waves. We need to cross-correlated [sic] the Planck maps with the BICEP2 maps. This analysis is underway.”
BICEP2 was definitely way out of line on a) the meaning and value of the Planck slide, b) using Planck data before Planck itself was ready to publish, c) extrapolating from something they didn’t understand, d) the intensity of its readings, and e) the reliability of its assertions, and that seems to me like an awful lot of basic stuff to be wrong on. But it’d be wrong as well to leap entirely in the opposite direction.
This may well follow the members of the BICEP2 team thru the remainder of their careers (for some, still new), which means it’d be nice for accuracy.
ATLAS and CMS hated the preliminaty combination of their Higgs search results by outsiders. They frequently told everyone not to buy any of this until the full, cross-correlated analysis was ready. and when it was, the result was virtually indistinguishable from the preliminary combinations.
I’m quite sure it will be the same here. Just imagine: would Planck announce now that the BICEP2 signal is dead for sure, what would they tell the taxpayer why they are doing the crosscorelation analysis anyway?
Since this was trumpeted as a prediction of inflation the failure to observe this prediction would deal a huge blow to the theory, right?
Or is that not how science works anymore?
I also remember this being trumpeted as a victory for the multiverse…
I think it’s completely accurate at this point to say that BICEP2 has provided zero evidence for primordial gravitational waves, instead is seeing pretty much exactly the expected dust signal.
This may change in the future, based on Planck data, new BICEP2 data, and a joint analysis of the two data sets (although seeing a significant signal this way doesn’t appear very likely), but that’s a separate issue. I don’t think it’s fair to use this possibility to try and evade the implications of the bad science that BICEP2 has done, promoted by press conference, and gotten on the front pages of prominent newspapers and magazines.
This is a perfectly good example of normal science: a group makes claims, they are checked and found to be incorrect. What’s not normal is a massive publicity campaign for an incorrect result, and the open question is what those responsible will now do to inform the public of what has happened. “Science communicators” often are very interested in communicating over-hyped news of a supposed great advance in science, much less interested in explaining that this was a mistake. Some questions about what happens next:
1. Will the New York Times match their front page story “Space Ripples Reveal Big Bang’s Smoking Gun” with a new front page story “Sorry, these guys had it completely wrong?” Or will they bury it in the specialized “Science” section tomorrow with some sort of mealy-mouthed headline like the BBC’s today that BICEP just “underestimated” a problem?
2. Will Scientific American in the next few months put out a magazine cover saying “Our October magazine cover was nonsense”?
3. Will the BICEP2 team withdraw their PRL paper?
4. Will Linde/Guth/Starobinsky return their May Kavli Prize, which was awarded with the explanation
“More evidence was provided earlier this year by an experiment at the South Pole called BICEP2 which, however, awaits confirmation by independent data. BICEP2 detected swirls in the polarisation of the CMB that are believed to be caused by the gravity waves spawned during inflation, as predicted by Alexei Starobinsky.” see
Maybe things are different in the experimental world, but based on a long career of watching hype about strings, susy, extra dimensions never matched by public explanations of what went wrong, I’m not so optimistic.
So, I’m curious, has anyone done some sort of investigation of when “science by press conference” exploded? And why? Is it the money involved? There’s always been pretty big money in physics, no? As a mathematician, it’s not something I’m exposed to really, big NSF grants in math wouldn’t even cover lunch in most physics grants. Off the top of my head, the first big press conference announcement of an incorrect result I can think of was the cold fusion thing, but there must be some before that, no? Particle physics always got pretty good press coverage that I remember, but you wouldn’t see announcements until things had at least been accepted for publication that I remember, or close.
The difference with math I think is that there’s just a lot more public interest in claims to have discovered a limitless source of energy, or seen back to the first gizillionth of a second of the Big Bang than there is in, for example, having proved that zeroes of a certain function lie on a certain line in the complex plane.
Unlike some other people, I don’t actually have any problem with the idea of holding a press conference when you publicly release results for which there’s a lot of public interest. You’ll need to deal with the press, and that’s one way to do it. Keeping such a result under wraps as it goes through a refereeing process is not so easily done, and putting out a preprint on the arXiv is a good way to allow your colleagues to evaluate your claims. But when you do this, you’ll have to deal with press interest.
In the BICEP2 case, there would have been no problem at all if they put out a paper and press conference saying “we have a solid observation of non-zero B-mode polarization, have to wait on Planck to know if it’s dust or not”. Instead they convinced themselves they’d made a huge discovery and decided to throw long, making dramatic claims (5 sigma effect!) to the press. Question now is how the mess they created by doing this will get cleaned up (or will it?).
Peter, just on your point 3 above: the actual PRL version of the BICEP paper contains rather more careful statements about the possibility of foreground contamination and the need for follow-up checks than the gung-ho pre-print version from March.
Re the other points: concluding at this stage that a measurable gravitational wave signal does not exist would be to make a similar mistake as BICEP but in the opposite direction. Of course this does appear to be the more likely outcome, but it isn’t certain yet.
You write, “concluding at this stage that a measurable gravitational wave signal does not exist would be to make a similar mistake as BICEP but in the opposite direction.”
Were Peter to host a major hour-long press conference and hire a film crew and visit Andre Linde’s house to announce the news of the failed BICEP2 results to him, and then film him crying instead of celebrating, then, perhaps, Peter would be doing the same thing in the opposite direction, but only if he were also scraping data from a powerpoint presentation to justify the results.
I toally agree that we need a discussion regarding how such discoveries should be communicated to the public and that scientist needs to be careful when presenting their findings. However, I think you being unfair in the rest of the critique you make. I have some comments regarding the retorical questions you presented in the comment above:
Point 2): “Will Scientific American in the next few months put out a magazine cover saying “Our October magazine cover was nonsense”?”
If you would take time to read the first line of article you would see in the first paragraph that the author talkes about what this *would imply* if it was true and not claiming that inflationary B-modes are a fact: ” *If* the recent discovery of gravitational waves emanating from the early universe holds up under scrutiny, it will…”. What can possibly be wrong with this?
Point 3) “Will the BICEP2 team withdraw their PRL paper?”.
Why should they retract the paper? This shows little understanding about how the experimental science works. If you would take time to read the paper (and the critique of it) then you would find that the main part of their analysis has stood the test of time. They do find a B-mode signal (and Planck confirms this, but it turns out that it’s likely to be dust). They also use the best availiable information about polarized dust availiable at the time and this seemed to suggest that it’s small. However, I do think that they should be critiqued for claiming a too significant (7sigma) detection when the dust issue was as uncertain as it was, but this is not enough to merit retraction in my opinion.
Point 4): Again, Kavli says *awaits confirmation* and note that they do not get the prize just for the B modes, but for the idea of inflationary cosmology. It is still a strong consensus that inflation (or something like it) have happened and we do have other evidence for it. B modes is just the missing piece of the puzzle that would settle this for good. People coming up with an idea that has stood the test of time (40 years) and whose work has lead to a lot of development in the field deserves a prize (with or without B modes).
The BICEP team arguably made two different mistakes: one to do with the actual science (the over-optimistic interpretation of foreground models) and one to do with the presentation of their result to the press. I was referring to the first one rather than the second — and in any case, I’m not convinced that having the scientific process played out in spotlights actually does the reputation of science (as opposed to the reputations of some individual scientists) much harm.
Really, there is NOTHING more interesting that whether the zeros of that function have real part 1/2 🙂
I don’t understand the technicalities at all, but can some clearer conclusions be obtained by repeating the experiment at multiple frequencies to try to separate the dust signal from a purported primordial signal?
Does this type of experiment at least have the potential to exclude (or confirm) some versions of inflation?
Sesh (and Winther),
Upon better consideration, I agree that the BICEP2 PRL as published probably doesn’t need retraction. I do think though that in this case the referees likely saved them from themselves, because their paper as submitted would be a much better candidate for retraction.
I don’t think anyone is claiming that now we know a measurable primordial signal doesn’t exist, and that such a thing is not contributing to the BICEP2 data. However, we do know that if it is in their data, they now have nothing at all that would justify such a claim. Maybe Planck will help them out to get some such evidence, but that’s speculation about the future (and I suspect most people would now give long odds against it).
The SciAm cover has nothing about “if this holds up under scrutiny”. More seriously, I think publishing in September a cover story about this, even with the “if this holds up under scrutiny” caveat, was irresponsible, given that since May it had become widely known that this was NOT holding up under scrutiny. To be fair, since it’s behind a paywall, I haven’t read the whole piece. Perhaps later on it explains the dust issue as a serious concern.
Do you really think that the award of the Kavli prize this particular year was not influenced by the BICEP2 claims to have experimentally vindicated inflation?
Not a physicist,
Yes, that’s what many experimental groups are now trying to do.
As discussed here at various times, the problem with inflation is that it’s only a rather general idea, which you can implement in many ways, and its seems you can get any value for r that you want, from BICEP2’s .2 to hopelessly small. Steinhardt argues vigorously that because of this it’s not science, but that argument is best left for another time.
Yes, inflation models can predict “r” values over a wide range, depending on model details and basic assumptions. If it is all just dust in the interstellar wind, it means only that inflation with such large values of “r” has been ruled out, again.
And again, inflation does NOT imply a multiverse, which is just one type of inflationary model, the Linde eternal inflating universe, with nucleating bubbles (of which we are in one). And this is NOT the same as the string “multiverse,” which is multiple disjoint realities, or something along those lines.
Inflation is an idea historically prior to and logically independent of string theory. The two have nothing to do with each other.
Sesh and Winther,
A New York Times story by Dennis Overbye on this is now out
He quotes Paul Steinhardt as saying that the BICEP2 PRL should be retracted.
To be more concrete, is it anticipated that in the next few years, some values of r could be experimentally excluded, at least painting inflation into a corner (of parameter space).
Not a physicist,
In some sense the BICEP2 result that they see what is expected from dust already provides a limit of r <.2 or something. The joint Planck/BICEP2 result in a few months will likely put a more stringent limit. Planned other experiments will take the limit down further, I don't know what the ultimate limits are. None of this though I think could be described as painting r into a corner, since an appropriate metric is probably in terms of the exponent: there are various versions of inflationary theories with very small predictions for r.
Peter, there is no doubt that the prize was influenced by the BICEP2 results, but the work they got the prize for has been so important for modern cosmology that it hardly matters: they would/should have gotten it anyway at some point. You can compare it to Higgs getting a prestigious prize before the Higgs was found – nobody would protest (and imo inflation is to cosmology what the Higgs is to particle physics : we know something like it has to be there, the question is just what exact form it takes). Importantly: it’s not the Nobel prize – that would be something I too would protest against.
This is still not over, but if it turns out to be just dust then I would still not agree that it is as damaging as you think. As you write yourself, this is a good example on how science is supposed to work and cases like this will also make sure that future experiments will be much more careful with what they claim (and how they present it to the media) which is a good thing.
IIRC if the Bicep-2 results were correct, it kills Starobinsky’s 1980 model.
But I agree in general that he and Linde as well as Guth were awarded prizes for their pioneering contributons to inflation. Its somewhat unfortunate that Demos Kazanas never gets any recognition for his work (even though the Bicep-2 team is one of the few experimental groups to have cited his 1980 paper).
Hi Peter, I had a quick look at Larry’s article in Sci Am, and I don’t think any pulping or revisions will be required. The piece is laced with phrases such as “The observation, if confirmed”….and other sensible caveats.
I think it’s important to emphasize that the Planck result goes some way towards suggesting that the BICEP 2 observation may be dust – this is not at all the same as establishing that the presence of primordial grav. waves in the CMB is ruled out!
The Krauss cover story does indeed contain caveats such as “…the result (or observation), if confirmed.” Normally, this sort of caveat means “if the result is replicated, or produced by some other experiment.” But given that the background noise is now shown to be sufficient to swamp any possible CMB signal, it seems to me that there is as yet no BICEP observation or result to be confirmed. Rather there is a claim, which is now shown to be unwarranted.
I meant foreground noise.
When I finish an article with a hefty dose of caveats, I am most often angry because my time has been wasted. I have never written such and I ask what is the motivation of those who do. Well, such an article is another publication on one’s record and if it is in a well-read magazine then that is all the better. It adds to the fifteen minutes of fame.
Professor Woit’s questions, answered (some of them, anyway; two):
1. a) No. And b) No. Instead, the NY Times published a fairly meaty, non-mealy-mouthed Denis Overbye story that, per the usual Grey Lady etiquette, appeared on top of page 1 of the SCIENCE section, and has moved down during the day with subsequent editions (The headline itself still appears, but it’s fading like a tropical sunset.)
2. If there’s anyone in the blogosphere who can speak authoritatively to SciAm’s historical position on the unraveling of string theory, it’d be you. Has SciAm EVER backed down on string? Or on any cover page article? If the answer to both is ‘No’, I should think the answer would apply here.
I agree with you citing Seth Nadathur for ‘best explanation’; but also maybe funniest. His take on the paper (not to leave out the paper itself, pretty mealy-mouthed in its own right), that the sheer size of this galaxy neuters BICEP2 technology for use on even larger scales, sparks the image of potential practical application in the opposite direction: detecting unusually high accumulations of dust at very small scales – useful for cleaners of our largest, most byzantine and ornate mansions and other large buildings (tho likely beyond the pocketbook of most in the middle class, who, like the poors, unable to afford the tens of millions in money and dozens of volunteer experts, will continue to be stuck with eye-balling it on their own).
My questions were kind of rhetorical… As expected, Dennis Overbye of the NYT wrote a good survey of the situation, but yes, I’m guessing it will appear tomorrow not on the front page, but in an inside section. Someone much better informed than me about the ways of the publicity world recently told me that, despite things moving online, positioning in the print paper is still considered the final say on what is important and what isn’t.
SciAm has had critical articles about string theory and SUSY over the years. I’ve never seen anything though like this current situation though: the highest profile result in years in the highest profile subject is dramatically announced, within a couple months it’s clear there are problems with it, a few months later SciAm decides to put this on its cover anyway. I don’t see how sprinkling in a few “if confirmed” helps the matter.
Let’s all calm down for a second.
(1) Dust does not explain the entire shape of the curve that the BICEP2 and KECK experiments measured. If you look at Fig2 of the new Planck paper, the slope of the polarization vs. multipole for dust looks nothing like the curve for gravitational waves for low values of multipole.
(2) The main, indirect evidence for inflation comes from the shape of the density fluctuations vs. wavelength in the universe. (see https://www.astro.virginia.edu/class/whittle/astr553/Topic16/t16_galaxy_power_spectrum.gif)
Inflation can explain the slope of the density fluctuations at long wavelength as well as their Gaussian nature. I know of no other model that can accurately explain the shape of the curve at long wavelengths and the Gaussian nature of the fluctuations.
The theory of inflation is no better or worse off today than it was before the new Planck paper.
I think it’s more accurate to say that inflation is no better or worse off today than in early March, before the BICEP2 announcement. The argument by BICEP2 that there was too little dust to account for their observations, so it had to be primordial is killed off by the Planck paper, returning things to the pre-BICEP2 situation.
I’m sure you’ll love this take on the story then…
The extensive caveats that are described in the SciAm paper remind me of reading papers in my field, control systems theory, in the early ’90s. They would start off by saying that “control design technique X is well suited to guarantee stability and performance metrics in the face of nonlinearities, noise, time delays, structured and unstructured model uncertainties, etc.” The next sentence would be “In this paper, we will restrict our focus to linear, time invariant single-input-single-output systems with no noise, no time delays, perfect sensors, perfect actuation, and no unmodeled uncertainties.” As somebody else said, if you’ve caveated everything, what’s the point anymore?
As far as I can tell, no one is arguing that this set of measurements amounts to a refutation of inflation as a modeling framework. What this data seems to indicate is that the dust+backgrounds (noise-only) model is a better model for the B-mode polarization BICEP2 saw than GW+dust+backgrounds (signal+noise). Sesh delineates this chain of inference really nicely at the bottom of his post Peter links to in update2.
This doesn’t mean there isn’t a signal buried in there somewhere, just that the simpler model is better at matching the data with fewer free parameters. At some point the measurements will get so precise that either the signal is found or cosmologist will have use more complex theoretical models, but as has been said ad-nauseum, we aren’t at there yet.
My opinion is that Bicep2 manipulated their simulations to gain worldwide popularity and in my opinion Bicep2 should apologize. 7 sigma is the maximized significance level of r different from zero when setting the dust to zero but, why Bicep2 had to invent a fictitious scenario by setting that dust to zero and publiciting their results with worldwide propaganda?.
Planck’s preliminary results say that the r = 0.2 could be explained only by dust. Planck must refine their analysis; so let’s see how this story ends.
PS: I have been discussing with Lubos Motl about this issue: he believes that I am an incoherent conspiracy theorist that enjoys attacking science. But he cannot explain why: from the r diffrerent from zero at 7 sigmas with Bicep2, we are now talking of a confidence of only 2.5 sigmas with Planck (and we will see which is the confidence in future analysis).
Today the joint analysis of BICEP2/Planck dust was done by a Chinese group arXiv:1409.7025. They got a new stringent limit of r (0.083).
Maybe SciAm has taken action. Since upgrading to ios8, their ipad app doesn’t work…
As a follow-up, the new version of the SciAm iPad app works with ios8, so I’ve finally been able to read the Krauss article. It actually mentions the Planck results and includes a supporting figure:
“Alas, as of this writing, the situation remains unsettled.”
“Recently the Planck satellite revealed that such dust could be more prevalent than previously thought.”
“The BICEP2 team stands by its estimates – but it now admits that it cannot rule out a dust explanation.”
Maybe Krauss (or SciAm?) got some insider info ahead of Planck publication? Anyway, the article ends up as an uneasy hybrid of theory and experiment, but maybe stands as a report from “the front lines”.
The problems with the BICEP2 result were known back in May, and I’d guess the SciAm reference is to other results about dust that came out earlier this year that reinforced evidence of a problem. What’s surprising here is that SciAm still thought it a good idea to put this kind of thing on the cover, several months after it became clear there was a problem with it.
Does PRL lose its credit?
Now the value of r significantly drops down and is consistent with Planck. However, there are several papers in which some theoretical models were supposed to solve the tension of BICEP2 and Planck. How will PRL deal with these papers?
Re SciAm, I concede…
Will you be participating in Festival Albertine’s “exploration of mathematical styles”?
I just read about that in the NYT this morning, hope to attend.