Astronomers excited by absolutely nothing

There's a very nice spoof article in which a claimed detection of a dark galaxy is rewritten as "absolutely nothing". This would actually be more appropriate for the paper below, in which the authors spend their entire time analysing a hole, i.e. absolutely nothing.

I'm being facetious. It's a very interesting paper, and though there are some parts I would quibble with, I'm surprised it hasn't been accepted despite being submitted back in November. The authors analyse a stellar stream around the Milky Way using data from the shiny new Gaia telescope, which gives super-awesome position, distance, and proper motion measurements (that's movement across the sky, as opposed to motion along the line of sight). This particular stream has a couple of odd gaps in it, which they claim are best explained by collisions with the long sought-after dark matter halos. So exactly what I'm interested in, just on a smaller scale.

They describe their preliminary investigations in paper I. The stream is clearly visible in a standard colour-magnitude plot and also in a plot of proper motions and positions. It also has a distinct "spur" extending north from one of the gaps, running roughly parallel to the main body of the stream. They identify all the stars in this stream and its spur by drawing polygons in the different plots, which as far as I can tell they do by eye. This clearly works, though it wouldn't be surprising if there were some stars included which aren't really in the stream. So the gaps may look like they're just lower-density regions with fewer stars present, but they could well be really empty gaps.

As far as I can tell they either don't use or don't have distance measurements. I don't know why this is. They are able to fit the orbit of the stream though, and identify its likely progenitor star cluster.

In the first paper they briefly speculate that a dark matter halo could have disrupted the stream to create the gaps. In this paper they go further, trying to constrain the nature of the perturber. They don't really consider other explanations much, though they do at least consider whether the gaps could have formed simply as part of the normal disruption of a cluster. They say no, the resulting stream in that case should be smooth.

Their model is a test particle model of a cluster ejecting stars as it orbits in a gravitational potential mimicking the Milky Way, with another massive particle used for the perturber (which is given different masses, sizes, encounter times, distances and velocities). Since the stars are test particles, which experience the effects of gravity but are themselves massless, they're ejected from the cluster artificially, so this isn't a self-consistent model of the stream formation. That's fine, because that's not the goal : they're more interested in the disruption of the stream, which test particles are perfectly adequate for. It means their models are computationally cheap, but of limited predictive power.

They run a bunch of simulations to explore the possible parameter space, though I'm not sure how many (at least 25) or how many test particles they use. I'm also not sure why a full n-body simulation would be so prohibitive : the progenitor cluster only has a mass of 70,000 Suns, so the particle number doesn't need to be crazily high. Anyway, their tests define the broad regions of parameter space of the perturber. They also allow them to predict the line of sight motions of the stars, which could be tested with future data. You can watch an example of one of their simulations below :

More images here, but there's not much information to explain what's going on. The movie shows the stream as it would appear on the sky from the centre of the Galaxy. I don't know if that strong, stream-wide warp is due to the encounter or simply its orbit around the Milky Way.

I'm a bit surprised that an encounter with a massive compact object can leave such neat gaps (or at least depleted regions). Based on similar simulations that I've run, I would have expected the stream to be much more distorted near the impact points, rather than stars being so neatly removed without bothering their neighbours much. It would have been nice to see what the simulation looks like from above; maybe things do look more distorted from other angles.

The authors seem pretty convinced by the dark matter halo explanation for the perturber. I definitely want to agree with them, and there's nothing obviously wrong with the analysis, but I think they're a rather too confident in their conclusions. They have quite a protracted discussion on whether it could be due to a black hole, which feels forced and unnecessarily exotic, and a lot of speculation about what they could do in the future. Sure, it might help constrain dark matter models, but don't count your chickens and all that.

Their constraints on the perturber mean that it is compatible - in terms of size and mass - with dark matter halos predicted in CDM simulations, but only just. It would have been nice to have more discussion on this point, and in particular to try and test what would happen to the stream if it encountered more typical dark halos. This should be easy since the simulations should be computationally dirt cheap. There's also not too much discussion of context either, i.e. how many similar streams are known, do any others have gaps, etc. But it's definitely an interesting result, and with lots of scope for further analysis and tests, it's also one that could potentially get a lot more interesting in the future.

The Spur and the Gap in GD-1: Dynamical evidence for a dark substructure in the Milky Way halo

We present a model for the interaction of the GD-1 stellar stream with a massive perturber that naturally explains many of the observed stream features, including a gap and an off-stream spur of stars. The model involves an impulse by a fast encounter, after which the stream grows a loop of stars at different orbital energies.

Manufactured crises

A very nice piece by Ethan Siegel, on which I will offer just a few thoughts.

The big problem that many non-scientists (and even some scientists) will never realize is this: you can always contort your theoretical ideas to force them to be viable, and consistent with what's been observed. That's why the key, for any theory, is to make robust predictions ahead of time: before the critical observation or measurement is performed. This way, you can be certain you're testing your theory, rather than tinkering with parameters after-the-fact.

I have to add that sometimes you don't know exactly what it is your theory predicts ahead of time. Most ideas grow in response to one specific problem, rather than being forced on you by multiple lines of evidence. It's only later that you get to test the idea against areas that were never considered during its inception : there are implicit assumptions you weren't aware of. So most ideas that work well for that one single issue have to be rejected because they are fatally contradicted by other arguments. This means it's very hard to avoid tinkering with the model later on; this can be entirely legitimate, but things quickly and inevitably get messy.

The key to cutting through all this is to focus on the key aspect of the theory, to try and test as much and directly as possible the most fundamental component of it, and not confuse this with more nuanced aspects like the particular value of one parameter.

For example, the standard model of cosmology depends on the notion that most of the mass in the Universe is undetectable. It's not at all implicitly obvious from this lone statement what this applies for galaxy formation. Can you tell me, simply based on this and nothing else, how galaxies assemble and evolve ? No, you can't. And thus we're sometimes told that observations like downsizing, the missing satellite problem, the Tully Fisher relation etc. are all evidence against dark matter, whereas in fact it's more likely that they're only evidence - if at all - against highly specific aspects of the galaxy formation model. That's a very different prospect.

(Writing up a more detailed piece as to how the problems with dark matter are related to implicit assumptions has been on my to-do list for a while. I'll get around to it eventually.)

That's the key that's so often overlooked: you have to examine the full suite of evidence in evaluating the success or failure of your theory or framework. Sure, you can always find individual observations that pose a difficulty for your theory to explain, but that doesn't mean you can just replace it with something that does successfully explain that one observation. You have to account for everything, plus the new observation, plus new phenomena that have not yet been observed... That's why practically every working scientist considers these proposed alternatives to be mere sandboxing, rather than a serious challenge to the mainstream consensus.

Not that there's anything wrong with spitballing - luck plays a key role in which ideas survive and which have to be cast aside. A good idea requires intelligence, but no-one is so intelligent that they can see straight away that their idea solves all the problems. Speculation and considering alternatives are fundamentally good things and to be encouraged. The difficulty comes not from freewheeling, playful speculation, but when people stop playing around; when they refuse to acknowledge that their idea doesn't work, or, equally, that someone else's idea does work.
Saying, "I've just had an idea" is great.
Saying, "I'd like to see what the community thinks of this, here's a paper" is great too.
But saying, "this idea clearly shows I'm a genius and everyone else is stupid", mistaking legitimate, necessary speculation for actual advancement, is not nice, and we'd be wholly better off without it.

On the other hand, we might not be able to avoid it. The consensus is only the consensus because it survives repeated attacks such as these without persuading more than a handful of people that it's wrong. While you can't decide truth by majority, the problem is that no-one really seems to have much of a better idea as to how you do establish truth (an immensely difficult philosophical problem). And a consensus without alternatives is no consensus at all, so having to endure a handful of intelligent people who doubt evidence that seems to everyone else to be incontrovertible, well, perhaps that's just something we have to live with.

But we mustn't forget or throw out the existing successes of General Relativity, the expanding Universe, the Big Bang, dark matter, dark energy, or inflation. Going beyond our current theories includes — as a mandatory requirement — encompassing and reproducing their triumphs. Until a robust alternative can reach that threshold, all pronouncements of "big problems" with the prevailing paradigm should be treated for what they are: ideologically-driven diatribes without the requisite scientific merit to back them up.

Yes. And it's far easier to amass interest if you say,
"ALL OF SCIENCE IS WRONG AND THEY'RE ALL JUST A BIG BUNCH OF ARSES" 
than if you say,
"ALL OF SCIENCE PROVISIONALLY CORRECT BUT MORE TESTING IS NEEDED FOR FURTHER INCREMENTAL IMPROVEMENTS, LIKELY LEADING TO AN EVENTUAL PARADIGM CHANGE AT SOME POINT, AS IS THE ENTIRELY CORRECT AND PROPER METHOD OF ORDINARY SCIENTIFIC ADVANCEMENT."

Disagreement between scientists is necessary. But the survivorship-biased narrative that's given to the public is just stupid. If a million monkeys have a million ideas, some of them will be better than others. If one of them gets lucky and has a good idea, that doesn't mean that monkey was anything other than a lucky monkey. Not even if said monkey was resolutely convinced of their idea and dismissed all the other monkeys as being too stupid to see it, because all the other monkeys probably said the exact same thing.

That doesn't means that some monkeys aren't more intelligent than others either. But there will be far fewer clever monkeys than stupid ones, so it's entirely possible that one of the stupid ones hits on a good idea before one of the clever ones. That does not mean we should encourage the stupid monkeys to believe the system is against them and they might be a misunderstood genius, because it would be vastly more productive to just encourage them to become a more intelligent monkey who studies hard. The other monkeys will listen to them, if they have something interesting to say.

Cosmology's Only Big Problems Are Manufactured Misunderstandings

If you keep up with the latest science news, you're probably familiar with a large number of controversies concerning the nature of the Universe itself. Dark matter, thought to outweigh normal atomic matter by a 5-to-1 ratio, could be unnecessary, and replaced by a modification to our law of gravity.

Is it a bird ? No, it's a plane

It's been more than a year since we heard much about the radial acceleration relation, and for good reason. When it was discovered it was hailed as a stupendous breakthrough... by its discoverers. Everyone else just sort of said, "sorry, what ?" or at best, "hmm, yes, that's a bit strange".

To recap, this is the discovery that the expected acceleration predicted by the visible material in a galaxy correlates extremely well with its observed acceleration. Because most of the mass of a galaxy is invisible dark matter, there isn't a direct one-to-one relation between the two. What seemed interesting to some people was the fact there was a correlation at all. It seemed weird that if dark matter was really so dominant that you should be able to use the visible matter to predict what it was doing.

Other people just shrugged their shoulders and said, "we see that in our simulations all the time". Later on they were able to show why it happened : it's a perfectly natural result of selection effects. You can't form galaxies in dark matter halos of any old mass, and the standard model makes very specific predictions for the distribution of dark matter and thus the resulting acceleration. To oversimplify (see first link for details), in effect the acceleration of baryons depends on the acceleration from the dark matter, rather than it being the other way around (which would be really weird). You can find a lengthy write-up of the whole sorry business here, including why lots of supporters of modified gravity got very cross.

While the main form of the relation appears to already have been entirely explained within the standard model, the last vestige for an interesting possibility was the stronger scatter at very low accelerations. It wasn't really clear if this was simply due to bad data, stronger intrinsic random variations, or if the slope of the relation hadn't been correctly measured. One intriguing effect noticed by the modified gravity camp was that if you accounted for the acceleration from the bigger parents of satellite galaxies, the deviant points seemed to move back to the standard relation, albeit with more scatter than the rest. But this scatter also appeared to be explicable in standard models simply because the scatter is intrinsically higher at lower accelerations.

Everyone agreed that the key place to look to settle the matter was the low acceleration regime. That's what this latest paper does, using very high quality data so that any deviation can't be attributed to measurement errors. And they do find that the deviation is real... but it's not random. Instead, there's a third parameter at work - the distance of each point from the centre of the galaxy. What's so far been plotted as a 2D relation with high scatter is, in fact, a 3D relation with very low scatter. What we've been seeing isn't a simple linear relation after all, it's actually a plane. And this is exactly consistent with the varying distribution of dark matter in different galaxy types.

Although the results don't exactly rule it out, there's just no need to invoke different physics to explain this relation. With the main relation was predicted by modified gravity theories, this particular aspect was predicted by dark matter. This won't be the final nail in the coffin (it will be interesting to see how the modified gravity community respond to this), but to my mind it further reinforces the uselessness of this relation as a way to distinguish modified gravity theories from dark matter. In my view, modifying gravity remains an unnecessarily drastic step that raises far more questions than it solves. What's so bad about dark matter anyway ?

The Radial Acceleration Relation (RAR): Crucial Cases of Dwarf Disks and Low-sur

The Radial Acceleration Relation (RAR): Crucial Cases of Dwarf Disks and Low-surface-brightness Galaxies