timoteo -> RE: Dark matter (Nov. 6 2017 8:02:06)
|
I hadn't read about that one. So I searched and found the preprint at https://arxiv.org/abs/1504.01527 A preprint is a paper that hasn't been published yet, but is made available to others so they can see details of results quickly without having to wait 6 months or more until after the peer review and publication. So the caveat is that this is not the exact paper that was published - there could have been corrections or additions in response to the peer review or editors' feedback. Sometimes preprints never make it to publication, if for example the authors withdraw the paper for some reason or peer review finds significant problems that can't be corrected. I read the preprint. They seem to have a significant signal. Five standard deviations means that this is almost certainly NOT a statistical fluctuation. Everyone publishes signals of five sigma - it's the two-three sigma signals that get you in trouble by being wrong more than you expect :-) The size of the signal raises issues because this sort of experiment has been done thousands of times - even when it's not the subject of a paper, experimentalists will do things like measure the angular correlation of decay products just to see if there are problems with their detectors. You measure everything you can, especially well-known quantities, then you check and recheck, to give you confidence that you have accounted for all the systematic errors inherent in measurement. Only then do you have some confidence that anything new you find is real. That's why it's odd that a signal of this magnitude hasn't been seen before. Even then, experiments find "new" stuff all the time. Every experimentalist has a story about someone who got burned by publishing or almost publishing a wrong result. But that's getting less likely these days with larger experiments and everyone acting as a check on everyone else's enthusiasm to rush to publication. Back when experiments were three guys at a particle accelerator, things were different. I call this a "cold fusion" moment - that was also a big signal that was totally wrong - it couldn't be reproduced at all by anyone. I was there when the cold fusion paper was first presented at a conference (it was announced way beforehand, via press conference, not even a preprint - they wanted to be sure to secure the patent rights ...). At least one of my colleagues went out and bought stock in palladium (essential for the result), on speculation that it was a real result. He lost a lot of money ... So on this one I'm saving my judgement until it's reproduced. I've seen hundreds and hundreds of tantalizing results. I've been on an experiment where we found "new" stuff that we didn't publish because it just wasn't a large enough signal or because we couldn't see the same thing in a completely different dataset. I've seen many papers that hint at the detection of supersymmetry, or at the detection of a fifth force, or at the discovery of a new particle, or all sorts of other things. If they're real, someone else will be able to measure the same thing, especially for an experiment like this where they don't need an extreme setup (for example, don't need a multi-million dollar experiment at the world's largest particle accelerator ...) Correlation of e+e- pairs from an essentially at-rest system is an experiment I did in one of my undergraduate physics lab courses. I'm not saying this experiment was at that level, just saying reproducing this result will not be hard to do if it's real. Even if it's reproduced, there's really no theoretical interpretation yet. Sure, there are some preprints I found that suggest this signal could be caused by this or by that, but the problem with a lot of these theories is they can only *analyze* and say whether something is consistent or not, they are no good at *predicting* what the experiment should see. To me, they always read like artificial constructs needed to fit the fact, but not motivated by anything. (A totally made-up example would be "well, if there were 12 types of Higgs bosons, then we could conceivable see this sort of result.") A theory that interests me would have to provide some means of testing, not just analysis. It would say, if this theory is true, then it would explain the results AND you should also see xxx is a different experiment. This experiment was set up to try to detect dark matter. They concluded that they had evidence for a light gauge boson, which is a force-carrying particle, hence evidence for a 5th force like you said. I'll be interested in seeing whether it can be confirmed. It's been a year, so either no one is trying (I doubt that, given the potential importance of the find) or they haven't managed to reproduce it yet and are still trying, checking and rechecking everything - you have to be a lot more careful when disproving something ...
|
|
|
|