RAR !

The sound made by dinosaurs and also the Radial Acceleration Relation, which I sometimes incorrectly call the MDAR (Mass Discrepancy Acceleration Relation). Whatever you call it, it's a relation between the expected and measured acceleration of stars and gas in galaxies. There's a really neat correlation between the two, which is thought to be odd because the dark matter ought to be dominant and not correlated with or controlled by the visible matter (which is only a small fraction of the total mass).

I did a pretty thorough write-up of this when it was really in vogue. You'll want to read that one for the details I'll reference below, otherwise I'll just give a very superficial overview here. Skipping to the final sentence :

"What I suspect will happen is that we'll see a few more papers on this over the next year before everyone becomes horribly disillusioned, gives up and goes home."

Which is pretty much exactly what happened. The "discovery" of this "law" was hailed with some strong rhetoric and met with some vehement arguments because only the "discoverers" thought it was significant, pointing, they thought, to a flaw in the standard model of gravity. There are of course a bunch of disagreements in the field of galaxy dynamics, in part because comparing observations and simulations is bloody complicated. The neat thing about this relation was that it reduced things as much as possible back to basic physics, so that ought to avoid some of the uncertainties.

But within days of the first paper, others showed that this already happens in standard simulations. The initial claims that this result would be very difficult to reproduce with dark matter (especially by strong advocates of MOdified Newtonian Dynamics, who seemed to think it was impossible) were swiftly and decisively refuted. Later, other groups were able to explain exactly why this relationship arises in the standard model, showing that this is indeed entirely natural. The idea that the dark matter and normal matter should be uncorrelated is too naive : in essence, it's not that the normal matter controls the dark matter, but it's the other way around. Hence there's a correlation between the two. That's grossly oversimplifying, mind you.

The authors of this new paper appear to have conceded that essential point. Initially the relationship itself was taken as evidence in favour of MOND and against CDM, but the authors quote the scaling relations shown that can reproduce the RAR without much in the way of criticism, except for a few token statements that not all simulations show this. By and large, I commend their change of stance on this (but see below).

However, now the focus of attention shifts. While massive galaxies all lie on a very narrow RAR with low scatter, that's not the case for the dwarfs. It's already been demonstrated that dwarfs have much more scatter and possibly follow a slightly different RAR than massive objects. The scatter is noticeably worse for objects with lower quality observations, so some of that is undoubtedly just due to measurement errors (lower mass galaxies have fewer stars so it's harder to get exact measurements, and also they're not as dominated by rotation as more massive disc galaxies - hence they have more intrinsic variation of motions). But for the dwarfs which do have really good data, the stronger scatter and deviation seems to be real, albeit weaker than the faintest dwarfs with the poorer data quality.

Unfortunately for the MOND crew, simulations have already shown that this too happens in standard CDM simulations. They reference the work showing this ([31], see also blog post) but don't seem to comment on this aspect of the result. So here they appear to be trying to use this deviation as a way to test between MOND and CDM without acknowledging that this has already been shown not to be the case. They use their own CDM simulations and show there's an increase in scatter but not a change in the relation, claiming that the observed change in the slope points to an important discrepancy between theory and observation. Well, it might, potentially. But again, they don't comment on the earlier result which did show a change in slope, and there's not really much acknowledgement that the worst of the difference appears to be due to observational limitations. There's very little (if anything) that's new here, and as usual, the statement, "This result could hint towards a scenario in which there is no DM and the law of gravity needs to be modified along the lines of MOND" is technically correct, woefully misleading, and even this weak statement doesn't seem well-justified by the evidence.

"Villains who twirl their moustaches..." and all that. Though there are some fair points here, personally I've become convinced that MOND is just silly. It's true that a lot of CDM people are also biased, but the MONDian attempt to present a narrative of an oppressed group of dissenters is fundamentally flawed. You might not see this from any one paper, but overall that's exactly what they're doing. This isn't the only example of MONDers hearing but ignoring arguments. I know this for a fact because I inadvertently gave a lecture explaining why planes of satellites aren't real with one of the researchers in that field in attendance. His subsequent lecture notes ignore my arguments completely (admittedly I was rather rude about the whole thing, so - credit where credit is due - kudos to him for remaining civil).

In short, this deviation doesn't appear to be at all useful as any kind of test for MOND versus CDM. Although it's often stated that MOND predicts this relation, it'd sure be nice if some MOND group would please actually show this on the main plot - which again is not done here. Anyway there are just too main observational uncertainties.

It's an interesting question as to what would constitute a really good way to differentiate between the two theories. This approach might work if we had very high quality data of the 3D motions of the stars in satellite galaxies. But we also need a version of MOND (or equivalent) that's compatible with GR and can do all the other necessary predications. Despite decades of effort, we don't have such a theory. So for now I stand by labelling it as just silly. There just doesn't seem to be any need for the damn thing.

https://arxiv.org/abs/1712.04448