For those of you who read the news reports about OPERA, and its potentially (not) superluminal neutrinos, on Thursday or on Friday morning, and stopped following after that, I have news for you: almost everything that appeared in the press up to that point was wrong in some important details. Thanks to my readers and their comments and detective work, we’ve collectively managed to figure out much more clearly what’s actually going on. I put up a relevant post Thursday morning and another Thursday afternoon, but I especially recommend Friday morning’s post (and comments) and Friday afternoon’s post (and comments). I really emphasize the value of the comments; I have some very well-informed and insightful readers who contributed a great deal. You can read this summary post first, and then go back to the older posts and read through the earlier viewpoints and the detailed commentary. [The science press has caught up, though; here's an accurate article from 2/27 in Nature.]
Most press reports on Wednesday, Thursday and Friday boiled down to this statement: “The OPERA folks found a loose wire, and when they fixed it their timing shifted by 60 nanoseconds [billionths of a second], bringing neutrino speeds right back to where they were supposed to be.” That’s certainly what the original Science Insider article implied, from which many articles took their cue. This is illustrated in the Figure below (labeled (b) to be consistent with a figure from an earlier post. ) The original OPERA result — that neutrinos arrived 60 nanoseconds before they were expected to — is shown as (a).
But this statement is completely wrong.
In fact the OPERA press release made clear that there were two problems (a problematic fiber-optic cable and a miscalibrated oscillator), causing shifts in opposite directions, and mentioned that a re-run of the experiment would be necessary. Still, most press articles seemed to give this lip service, and assume the correct reading of the situation was that the fiber was the main source of the problem, and that a re-run of the experiment was just pro forma. They mostly stuck with the simplistic idea that the OPERA people found a mistake and now everything agrees nicely with Einstein. A few, such as the New York Times, did a somewhat better job. But they still missed key points.
So what is the real story? Probably in the end everything will indeed agree with Einstein, but that is not known to be the situation right now. Instead things at OPERA are in a much greater degree of disarray. The two problems they identified are pretty big, potentially as big as the effect they measured or bigger. This is illustrated in (d/e) of the Figure. They know roughly how big they could be. But they have no idea how big they actually were during the experiment. Not yet anyway — they’re working on it, through indirect methods that they haven’t described to us. In the case of the fiber they have made clear that the effect could be up to 100 nanoseconds, but might be smaller. And they have implied that the problem with the oscillator could be tens of nanoseconds, though they haven’t been precise about that. The sum of the two problems together is therefore completely uncertain; it could be big or small, positive or negative. It could explain the shift they found, but not necessarily. They hope to figure this out to at least some degree, and be able to tell us more in a few weeks.
In short: the two problems together might give a shift of 60 nanoseconds in the right direction and explain their original unexpected result, but this is not yet known to be the case. For all we know right now, there could be another experimental problem which is the real cause of the 60 nanosecond shift. The mystery of the 60 nanosecond shift is not resolved.
Added for clarity and emphasis: That said, there is currently no credible evidence that neutrinos travel faster than light.
We also learned a lot more, mainly from a German newspaper article that readers translated for us (translation in the first comment on this post), about this “loose wire”, or much more accurately, problematic optical cable. It wasn’t loose. But it had a job to do, and how well it did its job turned out to be very sensitive to exactly how it was oriented. Its job was to bring a timing signal — a periodic flash, basically — into a box that converted the flash into an electronic signal, which went from there into OPERA’s main clock (which is the one that had a miscalibrated oscillator.) What the OPERA folks discovered — and this wasn’t obvious, apparently, and took some “detective work” to find — is that if this cable wasn’t oriented just right, not all the light would get into the box, and that would cause the box to fire its electronic signal a bit late — tens of nanoseconds too late. [Thanks to expert readers who left comments over the weekend, we've learned a great deal more about how this works, although there's still more to be understood.] And this would make the expected time for the neutrinos later than it was supposed to be, meaning the neutrinos would appear to arrive early. But they don’t know how the fiber was oriented during the experiment — they didn’t know about this problem, so they weren’t monitoring the fiber, and it may have shifted around over the three years of the experiment — so they don’t know how big this effect was while the experiment was running.
In short, this was not just some crappy loose wire. It was quite a bit more subtle than that, and much harder to find. And it is very hard now to diagnose its effects.
Now is that the whole story? Maybe not. As I emphasized in an update to Friday afternoon’s post (and as three of my readers also pointed out), there is something deeply mysterious going on. Both of the problems that OPERA has uncovered would be expected to be unstable.
- The oscillator was at some point measured to be calibrated; now it is not properly calibrated. What happened? When did this change? Before the experiment started? After it ended? During it?
- And any delay caused by the optical fiber was very sensitive to how the fiber was oriented. But was the effect of the fiber constant throughout the entire experiment?!? It is hard to believe that the orientation of the fiber didn’t change at all over the period of three years (2009, 2010, 2011) during which the original version of the OPERA experiment was carried out. Simply jostling it during maintenance would have been enough to cause a timing shift.
Naively, then, we would have expected that if the fiber and/or clock had something to do with the problem, the result obtained by OPERA would not have been the same across all three years (but it was!), and/or that the version of the experiment done during two weeks in October and November of 2011, using much shorter neutrino pulses, would have given a different result from the original one (but it did not!)
So what’s going on? Is there some good reason that the fiber, or the fiber and the clock together, would give a stable 60 nanosecond shift across three years? Or are the fiber and the clock not actually responsible, perhaps because their effects were actually smaller than feared? Could some other large and still unknown problem in fact be responsible for causing a stable 60 nanosecond shift?
Unless I’m totally missing the point, these mysteries must all be obvious to the OPERA folks and must be bothering them a lot.
No matter how well the OPERA experimenters are able to estimate the size of the combined effect of these two problems, the only way to fully resolve these mysteries is to fix the known problems and re-run the experiment (in May). If OPERA then finds no shift remaining, they can conclude that they probably found the cause of their anomalous result. If there is still a shift, then either there’s some other unknown problem that’s responsible or they’re still measuring apparently-faster-than-light neutrinos. But other experiments (BOREXINO, ICARUS and LVD) will be making their own measurements at the same time, so they’ll either get the same shift (suggesting a real physics effect) or no shift (suggesting OPERA has an additional problem) or perhaps all different shifts (which would suggest this measurement is even harder than it already appears!)
So I think we’ll be talking about OPERA at least through May — and don’t be surprised if new problems turn up with the measurements made by the other experiments, delaying the results, so perhaps this will continue all summer before it is resolved. But, umm… I’m told that Warren Buffet suggests you not invest in stocks that depend on Einstein being overthrown in 2012.