I wanted to make a few assorted comments about the OPERA experiment’s painful climb-down, and about yesterday’s widespread response to it, which bothered me a lot. You may want to read my initial post from yesterday, and also my attempt to sharpen the main question OPERA left unanswered in my second post. [ALSO: look ahead to the next post, in which many of the confusions that were still present at the time of this post were resolved.]
Over the past day I’ve learned enough to be
pretty convinced (but not certain) that the situation that we are in is case (e) [or a version of case (d)] as described in yesterday’s post: that probably the previous OPERA experimental data is tainted and we can draw no conclusions from it at all. It’s not that they found a problem that shifts their data so that it is consistent with Einstein’s relativity [case (b) from yesterday], and they can say that the neutrinos travel as expected. (Press reports that said so are just wrong.) It’s that they found a problem that means their data from last year can’t be interpreted at all… at least, not at the moment, and maybe not ever. If true, this would indeed mean that there is no longer any data from OPERA that can be used to measure neutrino speeds to good accuracy, and we’re back where we were before OPERA ran in the first place: with no reason to think there’s anything amiss with Einstein’s relativity equations. As for OPERA, the only way forward is to rerun the experiment (apparently in March-April-May.)
[The NY Times article that appeared today (which attributes OPERA, a non-CERN experiment, to CERN; what has journalism come to these days?) has some additional details, but if you read it carefully, those details don’t change anything written in this post. See the first comment at the very end of this post.]
Ok, some comments.
First: I think it is important for people to understand that the last couple of days have been pretty ordinary days in the particle physics world. Most physicists I know said “oh,” when they heard the news that problems had been identified at OPERA, spent a few minutes chatting about it, and went back to work. One of my famous senior colleagues, whom I was emailing for an unrelated reason, ended his email this way: “Thanks! I hope that things are going well for you. And, apparently, one less neutrino puzzle to worry about ….” (The ellipsis was his.)
Why so blasé? Almost all of my colleagues had put very, very low odds on the OPERA experiment’s result being correct — not because the people who did the measurement were considered incompetent or stupid, but because (a) doing experimental physics is very challenging; (b) this particular result was especially implausible; and (as everyone in the field knows) (c) most experiments with a shocking result turn out to be wrong, though it can take months or years to find the mistake.
The press (and therefore the public) never really quite understands how rare it is for claims of extraordinary discoveries to stand the test of time. And that’s why although the press makes such a big deal out of a first announcement of an exciting result, in the scientific community it is often only the second, or even third, confirming experiment (sometimes done by the same people, sometimes by others) which turns a raised eyebrow into a raised level of interest.
Second: I am astonished and dismayed at the level of vitriol and scorn heaped publicly on the OPERA folks for having… for having … for having done what?
We know there is a problem with a wire. But I don’t know yet what it is, or why they failed to find it. I certainly don’t know why it would make the neutrinos appear to arrive early in the same way in both OPERA-1 and OPERA-2; a random problem wouldn’t obviously do that. I don’t know if they simply overlooked the problem, or if it was particularly difficult to find for some reason. “Faulty wire” makes for good headlines, but do you know what it means? I don’t.
Anyone who has tried to find an intermittent problem in an electrical circuit knows it can be maddeningly difficult. Read this comment I received yesterday, from someone who obviously has a lot more experience than I have:
“One thing people outside the field may not realize is that in even in a relatively simple particle physics experiment, the data acquisition system can easily use hundreds or thousands of cables (electrical and fiber optic), of many different lengths running from many different locations to many other different locations. A bad cable, or a bad connector, or a mislocated cable can be the cause of endless grief; sometimes it causes an obvious problem that clearly points back to the cable or connection in question, but sometimes it can just cause subtle timing effects that are monstrously difficult to diagnose. Even if a bad cable or connection is found it’s not always clear what effect it might have had on previously collected data, and it may not be possible to correct the data to account for it (other than by blowing up the systematic error to the point where any result is more or less meaningless.) I see no CYA conspiracy here, just scientists doing their job carefully after finding the sort of problems you try hard to avoid in the first place but which happen to the best of us at times.”
Usually (not always) when an experiment fails, it does so because something went awry that wasn’t obvious for some reason. Let’s wait to find out what the case was this time, before insulting people.
Third: Many people in the particle physics community feel (and felt from the beginning) that OPERA should have been more patient and done far more checks of the result before they released it. Generally, I’m one of them, but a problem for OPERA was that their result had to do with Einstein, and anything having to do with Einstein gets blown up into the biggest-thing-ever by both press and public. Einstein’s an icon of our modern age, like Newton to the previous one; anything that could unseat his equations and concepts, even in a small way, is big news. And of course if OPERA were correct, it would actually be huge news. So I don’t know if OPERA could have avoided the situation that they got themselves into. [CERN, which isn’t even the host lab for OPERA and has no affiliation or membership in it, is another matter; why they got themselves involved I have no idea.] In addition, OPERA’s result was announced to the world not by OPERA itself, but in a leak by a blogger; so the situation got away from them even at the start. I doubt they could have kept it secret for the months it took to find the wire problem. They would have had to work very hard to downplay the result; I wish they had made a great effort to do that, but I don’t know if it would have helped in the end.
Finally, I would like to remind you what it means to be a scientist, and human. Let me quote something that I said in my very first post after the OPERA paper appeared (you may enjoy reading the whole post, to see how it looks in hindsight). I wrote:
A last remark for the night: think about what it is like to be an experimentalist making a revolutionary statement of this magnitude. Talk about sticking your neck out! This result either means a Nobel Prize or international embarrassment — perhaps even ridicule if a serious mistake was made; there’s no middle ground. The combination of excitement, hope, and terror must be unlike anything most of us will ever experience. I cannot imagine how any of them have slept for days; I cannot imagine that they will sleep well for months, until a second experiment reports, “We have measured the speed of neutrinos, and we confirm…”
They’re suffering plenty. No one needs to make it any worse by being cruel, especially before we know the whole story.
25 thoughts on “Everybody’s a Critic”
Regarding OPERA, please look at the following NYT article (hopefully it is reliable):
The article’s seems to be based on an interview with the leading OPERA physicist Autiero. It says, that the OPERA team checked the fiber-cable connection with “dimmer light pulses” and found a 60ns error, “that could bring the neutrinos’ speed back under the speed of light” (this seems to be in agreement with the “Science” report).
They also report, without details, about another problem that might increase the speed of neutrinos (this seems to be in agreement with the CERN press release).
Well, I don’t think you are reading the article carefully. This is how mistakes get made; the scientist says one thing; the press person writes another; and the public reads a third.
Read what it precisely said:
“The second potential error is in the fiber-optic cabling that carried the GPS data five miles to the underground detector. The investigation discovered that for dimmer light pulses, the circuit receiving the data introduced delay — up to 60 billionths of a second — that could bring the neutrinos’ speed back under the speed of light.”
The words “up to” are crucial, and they probably reflect a rephrasing of what Auterio said, and possibly a mis-phrasing. Because no scientist would say it that way. That’s a journalist restating something. I don’t know what Auterio said. He clearly did NOT say “we found a 60 nanosecond error”, because the reporter wouldn’t have said “up to” in that case. He might have said “whose size we don’t know but which could have been as large as 60 nanoseconds” or “which introduced random errors which reached as high as 60 nanoseconds” but certainly something more complex and detailed than “up to.”
And the most important words here are “dimmer light pulses” — which means that it probably wasn’t so easy a problem to find, since they probably checked the cable with standard light pulses and saw no issue, and only later realized there was a subtle failure mode with the dimmer pulses.
And then it says:
“But Dr. Autiero said the issues it identified did not conclusively prove or disprove the findings. ‘We are not sure of the state of this connection in the past,’ he said.”
This is indeed consistent with what I called case (d) or (e) in yesterday’s post. So there’s nothing in this article that would change anything I wrote in today’s post; I still don’t think we have a complete story here. The data is tainted, but the identified problem is only a suspected cause, and not a known cause.
For all we know, the real cause still hasn’t been found. And Dr. Auterio implies that some people in the experiment think it is possible that this “faulty wire” was working fine when the experiment was running.
I also recall OPERA being extremely careful in their first statement. There was no sensationalism in there; rather they more or less stated their own disbelief and invited everyone to help confirm or falsify the result. From what I have heard from them, I don’t blame them.
I disagree with this statement:
“This result either means a Nobel Prize or international embarrassment ”
If they would have left out this sentence in their abstract (and paper)
“This anomaly corresponds to a relative difference of the muon neutrino velocity with respect to the speed of light (v-c)/c = (2.37 \pm 0.32 (stat.) (sys.)) \times10-5. ”
Instead, if they just referred to this result as a timing anomaly and refused to publicly accept an interpretation of this as neutrino velocity, they would come off as having made a minor mistake in a very difficult measurement. Only the NewScientist would have picked this up and there would be relatively little egg on their face.
If the result was to be confirmed in the future, they would still get 100% of the credit, go to Stockholm and be on Charlie Rose speaking about their breakthrough moment when they realized Einstein was wrong. They aimed their discussion purposefully wider than the narrow technical community that was necessary. This is the appalling thing.
This has precedent: Pioneer didn’t go around saying “Einstein is wrong” Gravity isn’t behaving as a 1/r^2 force. They reported an unexplained anomalous acceleration. No embarrassment for science.
I think that would have just made them look bad. In the Pioneer context, it would be like saying that the spacecraft were at location x when they should have been at x’ without mentioning that anomalous acceleration. The OPERA paper stressed that further investigation was needed and did not speculate on any implication of the result, if it turned out to be correct. The press conference was scheduled after the news had leaked onto the Internet. Even in interviews, they’ve repeatedly stressed that it’s likely some experimental error. That this whole thing was blown so much out of proportion, is more a fault of the media, IMO, than OPERA.
Jay — While I have some sympathy for this point of view, I don’t think, in the end, that I agree.
I agree that sentence should not have been in the paper and certainly not in the abstract. I also think they were very aggressive to say they had 6 sigma evidence (which required adding all systematic errors in quadrature.)
First, you should remember that the blogs [Dorigo’s blog I think] broke the story, and blew it out of proportion from the start. That’s why there was so much media attention when OPERA announced its result — there had been three days of rumors running around the world. I don’t know that the removal of that sentence would have changed how the blogs covered the story, and therefore how the media covered the story. I think it was out of OPERA’s control. No one in the press read that sentence; they read the blogs. And a crack in Einstein’s theory is too good a story for the media not to want to sell it.
Second, I believe that the OPERA people, to their credit, were careful to state their lack of confidence every time they spoke to the media. The people who were not careful were always other irresponsible people who were being interviewed about the OPERA result, or the press who wrote about it, who often left out their interviewees’ caution.
Third, you can argue (and I would) that the short-pulse version of OPERA should have been done before the original result was released. But after it was done, its success removed a long list of possible problems. In fact, the very types of problems which they are talking about now — drift or instability in electronics — is one of the main risks that the second short-pulse version of OPERA, done over two weeks, should have removed from the first long-pulse version, done over three years. I am sure that, had I been a member of the experiment, I would have voted to go public after the short-pulse version gave the same answer as the long-pulse version. And then the same level of hype would likely have occurred, just starting two months later than it actually did.
Still… I agree the whole thing was badly mismanaged. The minimal press release put out on Wednesday, which left out far too many details, did not help matters. And CERN, which should have stayed out of this, now is heavily damaged in the public eye, for absolutely no reason whatsoever.
It might help to check the fiber with a polarization scrambler on the transmitter and a polarimeter on the receiver.
I used to demonstrate a 100ps delay change in a small spool of fiber by gently exhaling on it from a few feet away. Its delay is temperature sensitive and is also subject to sudden shifts as the polarization mode changes. Communication systems recover the clock from the data at the receiver and do not depend on the delay through the fiber.
Fascinating that the “authoritative” New York Times’ story was wrong in saying OPERA is a CERN experiment and that it was CERN that made the checks. How much reliance can then be placed on the rest of the piece?
I agree, science journalism is atrocious these days, and only seems to be getting worse. I only hope that the pendulum swings back in the other direction before too long. It’s gotten to the point where I don’t even trust supposed direct quotes from scientists in news articles (I myself have been misquoted on a number of occasions).
Anyway, thanks for writing these series of articles and setting things straight about the latest OPERA announcement. It seems there is a lot of misinformation out there, some being spread by people who frankly should know better.
I for one, am happy to wait a few more months for more data with these problems corrected. I also hope that the other independent experiments that were planned to test these results are still going to go forward.
First of all, thanks for the detailed account of the facts. I enjoy reading your blog very much.
While reading your analysis of what is correct and what is not I felt that I should point out some thougts. Newspaper articles contain errors, they do so very often. Most of them try to make a normal story interesting, some try to make a big story where there was nome at all. News about faster than light neutrinos is a good story, I my view that is no problem. But a news report must not be confused with hard science where error bars and systematic errors are very inportant. So the not any more faster than light neutrinos is also a story and the explanation wih the cable seems plausible. It will not sell as good as the first, but not bad. What I want to point out is that scientists must understand that news is mainly entertainment. Any celebrity can confirm that truth is not what sells best. Most people know this and do not take everything as serious. I for example started to be interested in the topic and spent many hours reading about quantum physics. Thus I learned the detailes behind the story. I suggest to see the positive side in news. It makes much more ppl talk about physics which helps the overall quest for knowledge even when some facts are not pwerfectly correct in order to make a good story. (Note that I dislike journalism making fun of scientists, but there is obviously a market for such a thing)
Thanks for the clarifications.
However, while writing a reply to your comments on the NYT article, I found a recent FAZ interview with Caren Hagner, the leader of the German research group associated with OPERA., who participated in the search for errors. She gave much more details:
Queried after the first error, she responds, that by detective work they have luckily found out, that it makes a difference as to how the cable is screwed. Depending on the position of the connector, the transformed electronic signal can be delayed. They have found out, that if the cable only slightly deviates from the ideal position, only a part of the signal arrives in the box. Depending on the signal level of the input signal, the delay can amount up to one hundred nanoseconds.
However, they don’t know, how inclined the screw and thus the temporal delay actually was during our measurements last year. In the moment, they can only qualitatively estimate the order of magnitude. Though they work to calculate the effect of the connector from other data of the detector. However, this will need some time. The connector actually could be the cause why they have measured superluminal neutrinos.
And queried after the second error, she responds, that this correction would make the neutrinos even faster. In the moment it appears as if those two effects could collectively just explain the previously measured deviation of 60 nanoseconds. Then one possibly would have exactly the arrival time of neutrinos, that one would expect when they are moving with the speed of light.
And queried after the precise quantity of error, she responds, that the analyses are still ongoing, in some weeks they can quantify the measurement errors. But in any case they have to re-measure with a new neutrino beam, this is quite clear.
It seems, that we have (possibly) a time-of-flight correction of up to 100ns by the first error, which is (possibly) changed in the opposite direction by the second error, which (possibly) gives the measured 60ns. To be sure, they must further evaluate the data and use a new neutrino beam in May.
THANK YOU!! That is what we have been looking for. As soon as I get a free moment I’ll update the post.
” Second: I am astonished and dismayed at the level of vitriol and scorn heaped publicly on the OPERA folks for having… for having … for having done what? ”
I fully agree with Prof. Strassler on this.
It always makes me sad too to see how scornfully particle physicists (mostly theorists but sometimes experimentalists too) are sometimes treated in the comment sections of other blogs these days by unthoughtful people :-(. These bright scientist, who are of course still humans and can make mistakes as anybody else, do not deserve this!
I at least want to see this particle physics business going well and I am happy if they get enough funding for their research because it is my favorite spectator sport and I am so damn curious about the fundamental questions they try to answer 😀
I`m sure this nice site here helps in getting more people interested and makes them appreciate and understand a bit what particle physicists are doing 🙂
I agree entirely that OPERA acted appropriately. I think it’s actually quite wonderful that their results were taken seriously, even if it was with the general assumption that there was probably a flaw. The fact that something was openly published that apparently contradicted something as fundamental as Einstein’s work is very reassuring: orthodoxy can be challenged by new evidence.
I feel a lot of sympathy for the OPERA folks: intermittent faults just in domestic appliances can be intensely annoying…
The new evidence and the old experiments only confirm that Einstein is absolutely wrong.
Study, please, deeply the whole Blog:
Please, any question let me know.
Lucy Haye Ph. D.
Great article Matt, I just have a quick question. You keep insisting CERN has no affiliation with the experiment and it’s confusing why they’re connecting themselves with it, but the Wikipedia page for OPERA describes it as a collaboration between CERN and LNGS, and the OPERA home page mentions the neutrinos are produced by CERN. I know Wikipedia probably isn’t a reliable source, but doesn’t the fact that CERN produces the neutrinos make them part of the experiment?
Yes, I have been thinking this requires clarification.
It is quite common in particle physics that a laboratory produces a particle beam, but a particular group of people who may have nothing to do with the laboratory builds a particle detector, sometimes very far away, to analyze what happens when that beam hits something.
Now the question is: what is the experiment? Is it the particle detector? Or is it the beam and the particle detector?
Obviously CERN has to approve of any use of its particle beams, so yes, it approved of OPERA. And OPERA can’t operate without CERN’s beam. But OPERA is built and operated (and the neutrino-speed measurement was planned and carried out) by OPERA, which includes people who are from various participating institutions: universities and laboratories from all over the world. CERN is not among them. CERN personnel who built and operated the particle beam are not signatories on the OPERA paper. No CERN personnel were involved in checking the experiment for errors except as requested by OPERA personnel who needed to make sure they had the timing right on the CERN end. All of the problems found at OPERA so far are on the OPERA end of things, in the Gran Sasso lab in Italy. And if the OPERA measurement had been correct, almost all of the glory would have gone to OPERA and its personnel, with a nod to CERN for providing a high-quality beam, and a nod to the Gran Sasso lab for hosting the OPERA deteector.
In other words, it is an issue of distance from and responsibility for the actual work. If the problem were with the neutrino beams, then CERN would have to take the blame. But as far as I can tell the problems are all in the OPERA detector area, and those have nothing to do with CERN at all.
CERN operates the Large Hadron Collider [LHC] too. ATLAS and CMS, the two big detector collaborations, are actually hosted at CERN itself, along with LHCb and ALICE and a few smaller experiments. Of course all of those collaborations are international too; many labs and universities are involved. And although CERN’s involvement in those experiments is far greater than in OPERA, CERN bears the dominant responsibility for the proton beams and the proton-proton collisions; responsibility for the detectors is much more widely spread.
Today’s journalism seems to be more about “first” than about “correct.” Perhaps this is an effect of instant media. (I puzzle why it’s so critical to have results from U.S. Presidential Elections within 24 hours, when the law still imposes archaic delays enacted to allow for ground travel of the electors.)
I believe heartily in free and uncensored speech, that admits falsehoods and even hate speech, as long as the speech is attributed to the source. Speech and source may then be evaluated together, and over time, the value of each will become clearer. I believe this is important for both journalism and science.
I recall a news interview about a cryptographic artwork called Kryptos, releasing a hint from the artist for the amateur cryptanalysts who study his work. Carefully parsed by an informed reader, it was plain that the sense of the journalist was not the sense of the artist, about what was being “quoted.” A detail was out of place in a way that made it puzzling to reconstruct the error.
The bottom line is that this is why experiments must be reproducible. Whatever someone recorded about their observation, it is likely to be perceived differently by a later student, and misunderstandings can then be cleared up by further experiment.
Professor Strassler, thanks again for giving us the inside story.
29cm/8km = 3.625 * 10^-5
Rañada-Milgrom effect: Replace the -1/2 in the standard form of Einstein’s field equations by -1/2 + sqrt((60±10)/4) * 10^-5 .
Think about the OPERA neutrino anomaly in these terms:
(1) One or more experimental errors invalidate the anomaly.
(2) Quantum field theory is slightly wrong.
(3) General relativity theory (GRT) is slightly wrong
Is alternative (3) a winner? GRT is slightly wrong, Newtonian gravitational theory is slightly wrong, and the OPERA team has confirmed the presence of dark matter (or its Milgrom-equivalent) in GPS timing.
Overwhelming evidence shows that Milgrom is the Kepler of modern cosmology. The Lambda Cold Dark Matter (LCDM) model is not entirely correct.
I quote Prof. Dr. Pavel Kroupa from a (Nov. 1, 2011) e-mail,
“My criticism is not based on me not liking dark matter, but is a result of rigorous hypothesis testing such that, from a strictly logical and scientific point of view, LCDM is definitely not a viable model of cosmological reality. I do not write such statements because I do not like LCDM and its ingredients, but because every test I have been involved with falsifies LCDM. At the same time, the tests of MOND we performed were done on the same footing as the LCDM tests. The MOND tests yield consistency so far. I am not more “fond” of MOND or any other alternative, but the scientific evidence and the logical conclusions cannot be avoided. And it is true, I must concede, that MOND has an inherent beauty which must be pointing at a deeper description of space time and possibly associated quantum mechanical effects which we do not yet understand (compare with Kepler laws and the later Newtonian dynamics).”
Hi Matt, et al. I thought you might be interested in some first hand experience with difficulties with fiber optics and their effects. Some time ago my little research group was interested in determining the output spectrum of a tunable diode laser we were working with. We obtained an optical spectrum analyzer. The input was a screw-in connector that led to a multi-mode fiber in the OSA. We had to construct an input fiber from the free space mode of the laser into the OSA. This turned out to be much more difficult than expected. The fiber has to be cut at both ends very precisely. It has to be inserted into the end connectors very precisely. The input beam had to be launched into the fiber optic very very precisely. The input cable had to be screwed in perfectly – not too tight, not too loose. If anything was awry, the result was a hash of noise in the OSA. After much effort, everything was copacetic, the noise abated and we got a clean output spectrum of the laser, and could determine the tuning current desired.
It seems that unwanted modes of the original laser output were introduced by these faulty connections, although how this translates to a reduction of apparent time-of-flight is not at all clear. The appearance of noise in a fiber optic connection is clear evidence of connection. I read somewhere that some spurious noise at OPERA was simply filtered out.
One comment discussed the complexity of experiments like these and the number of possible points of failure. Can it be that some experiments are just too big to ever get a precise result?
Thanks, Jeff. I don’t think experiments like these are too big for a precise result. Smaller experiments have often failed too. And even with these problems, it is remarkable that OPERA got the same answer for long pulses over years as for short pulses over days. I think it is quite likely that a precise result will be obtained this year (and will confirm that the universal speed limit remains intact.)
Comments are closed.