MAGIC's observation of Gamma Ray bursts

A couple of days ago, the MAGIC collaboration posted a paper on the arXiv:

Probing Quantum Gravity using Photons from a Mkn 501 Flare Observed by MAGIC

which I didn't mean to comment on, but it seems it has caused quite an astonishing amount of discussion. The SciAm blog sees hints of a breakdown of General Relativity, Lubos Motl expresses his skepticism, Slashdot proclaims it is a test of String Theory, which then causes Buffy Woit to fight against the living dead predictions, thereby unfortunately claiming to be in agreement with Lubos, who returns the favor with a second post denying any agreement with "the critique of science", fading out in a characteristic rant about 'aggressive imbeciles'. Even the more reasonable article by Chris Lee on Ars Technica which cautions that quantum gravity has probably not "made a sudden leap forward" proclaims that "this is the first real data against which such a theory can be tested", the caution of which however leads the SciAm bloggers to nag that those "who have been loudly complaining about the lack of [observations that probe string theory] have gone silent."

This is the summary, status Sunday evening. In case I loose one or the other reader here, let me give you my opinion up front: there is no experimental evidence for quantum gravity, and none of the present theories can be tested with currently available data. If you're looking for sensations, you're on the wrong blog. I suggest you try one of those mentioned above instead.

As a prerequisite, the issue is an observation of gamma ray flares from the active galaxy Markarian 501. The data was taken in summer 2005, the most characteristic event being on July 9th. The peak energy of the flare was around a TeV, it lasts about 2 minutes. The emitted spectrum spreads over some orders of magnitudes. I'm not an astrophysicist and details elude me, so please don't ask me in which constellation Mkn501 can be found, or what the status is about understanding AGNs.

The point is that if one knew the emitted signal, brightness and spectrum, measuring what arrives on earth would allow one to conclude what happened to the signal on the way. If the signal propagates through vacuum, General Relativity (GR) tells you what to expect. There is a redshift, brightness drops, but these effects are well known and computable within GR. If what happened to a photon during travel would depend in some unusual way on the energy of the emitted photon, it would distort the shape of the signal. This is called dispersion. If a family has to get through a crowd of many people, the small children will be the first to arrive, and then have to wait for their parents who have a harder time pushing their way through. Such, what originally started as a localized group arrives with a time delay between various parts. A similar effect happens to an emitted signal of photons if there is a dispersion.

Problem is, for the case at hand the emitted signal is not so very well known. However, Likelihood analysis allows to examine how high the chances a signal is distorted in a certain way for different frequency regions, and for the observed flare the MAGIC collaboration concludes that its shape "is unlikely, and consequently that the time shift of 4 ± 1 min between the highest and the lowest energies is more probable." [astro-ph/70702008, p. 13]. The time of this delay would be about the duration of the signal itself.

In their new paper the MAGICians write one can circumvent knowing the true shape of the time profile at the source by basically taking the result and assuming a parameter dependent blurring has taken place that distorts the signal differently in different energy regimes. If one reconstructs the original signal with that assumption, one can examine it for certain features, and ask for the parameter that optimizes that feature. If I understand it correctly, they have performed analyses with respect to two such features. The one is to "maximize the total energy in the most active part of the flare" and the other is "to optimize the sharpness of the transformed signal".

In both cases it is found that the signal can be optimized if one reconstructs it with a certain assumed dispersion relation that depends on a parameter. The best fit for the parameter is not for exactly zero which would correspond to no dispersion, aka standard propagation. It is interesting though that both analysis with respect to different features seem to fit together. There comes the chi-square fit, et voila, one has a signal at 95% CL. That is to say the detected signal would be more probable if there had been a dispersion. It's not striking how much one can improve it, but one can.

But anyhow, now the clue is to call the fitting parameter M_QG and claim it is an effect of quantum gravity (QG) , since "QG effects [...] cause the fabric of space-time to fluctuate on the Planck time and distance scales. It has also been widely suggested that this ‘foaming’ of space-time might be reflected in modifications of the propagation of energetic particles, namely dispersive effects due to a nontrivial refractive index induced by the QG fluctuations in the space-time foam and that one is "probing the Planck mass range for the first time". Depending on your preferences you might then argue this is either a prediction of string theory, or contradicting predictions of string theory. Or a prediction of LQG, or not a prediction of LQG.


What is even worse than the advertisement in the paper is the echo it has caused, which is more attention that I think is appropriate. What about all those hard working people who don't wrap their results into easily digestible catchy phrases? Won't they wonder what all their effort is good for if they see one gets much larger attention with things like this?

Yes, there are motivations, and indications that within certain scenarios one can have such a dispersion. I myself find this very exciting, but as the authors state themselves "The calculation of such effects is beyond the scope of current theoretical methods". Also, as Lubos has pointed out, the same collaboration had a preprint already in February about the gamma ray emission of the same source Mkn501, in which they attributed observed spectral features to different effects, and remarked "A somewhat more speculative issue that blazar emission permits to explore concerns non-conventional physics. Energy-dependent arrival times are predicted by several models of Quantum Gravity, [...]". Even given that the signal is confirmed by sufficient other data, and proves to be statistically significant, the most likely explanation I'd think would still be our lack of understanding of AGNs.

The problem I have with that paper is not the analysis itself (which I wouldn't have found remarkable enough to mention here) but that it is clearly sold as a made up sensation. The whole introduction doesn't belong there. If they want to analyze their data with respect to dispersion, then they should do that, and not take several further leaps of faith towards quantum gravity. If at all, then this belongs into the discussion, but neither in the abstract, nor in the introduction, nor the conclusion of a scientific publication - the first paper e.g. is much better balanced. All that they have found is with unconvincing confidence a signal can be 'optimized' if a dispersion relation had a parameter that happens to be somewhere by the Planck scale. That itself has nothing to do with quantizing gravity, neither with a "Break down of General Relativity" as the SciAm blog put it.

However, even if there was a dispersion, it wouldn't be clear at all that this has anything to do with quantum gravity. If the signal undergoes a dispersion, then the signal undergoes dispersion. Full stop. Everything else is speculation, and should clearly be called such. The question what an effective description has to do with the potentially underlying fundamental theory I call the 'Inverse Problem' - for more details, please see my previous post on Phenomenological Quantum Gravity.

To sum up: 1) There is the question whether the signal is significant 2) Even if it was, it would most likely be something that we haven't understood about astrophysical processes 3) Even if that was outruled there would be the question whether it had something to do with quantum gravity.

Bottomline: If one adds a parameter one can fit curves better. If the original fit was good, the parameter will be small. (Another example for this sort of analysis you find here.)

Now you can blame those who are conservative and would attribute a potential signal to old-fashioned physics to be boring and pessimistic, but the only thing you get from hyping non-results is more people wasting time with alleged 'effects' and made-up 'predictions'. Yes, I do think scientific publications should be conservative and cautious, especially if they claim to have evidence for new physics. Reading papers and follow-up articles with exaggerated claims about testing quantum gravity annoys me considerably. It's like strawberry yogurt where the fineprint says 'no fruit'.