Wednesday, October 12, 2016

What if dark matter is not a particle? The second wind of modified gravity.

Another year has passed and Vera Rubin was not awarded the Nobel Prize. She’s 88 and the prize can’t be awarded posthumously, so I can’t shake the impression the Royal Academy is waiting for her to die while they work off a backlog of condensed-matter breakthroughs.

Sure, nobody knows whether galaxies actually contain the weakly interacting and non-luminous particles we have come to call dark matter. And Fritz Zwicky was first to notice a cluster of galaxies which moved faster than the visible mass alone could account for – and the one to coin the term dark matter. But it was Rubin who pinned down the evidence that galaxies are systematically misbehaved by showing the rotational velocities of spiral galaxies don’t flatten out with distance from the galactic center – as if there was unseen extra mass in the galaxies. And Zwicky is dead anyway, so the Nobel committee doesn’t have to worry about him.

After Rubin’s discovery, many other observations confirmed that we were missing matter, and not only a little bit, but 80% of all matter in the universe. It’s there, but it’s not some stuff that we know. The fluctuations in the cosmic microwave background, gravitational lensing, the formation of large-scale structures in the universe – none of these would fit with the predictions of general relativity if there wasn’t additional matter to curve space-time. And if you go through all the particles in the standard model, none of them fits the bill. They’re either too light or too heavy or too strongly interacting or too unstable.

But once physicists had the standard model, every problem began to look like a particle, and so, beginning in the mid-1980s, dozens of experiments started to search for dark matter particles. So far, they haven’t found anything. No WIMPS, no axions, no wimpzillas, neutralinos, sterile neutrinos, or other things that would be good candidates for the missing matter.

This might not mean much. It might mean merely that the dark matter particles are even more weakly interacting than expected. It might mean that the particle types we’ve dealt with so far were too simple. Or maybe it means dark matter isn’t made of particles.

It’s an old idea, though one that never rose to popularity, that rather than adding new sources for gravity we could instead keep the known sources but modify the way they gravitate. And the more time passes without a dark matter particle caught in a detector, the more appealing this alternative starts to become. Maybe gravity doesn’t work the way Einstein taught us.

Modified gravity had an unfortunate start because its best known variant – Modified Newtonian Dynamics or MOND – is extremely unappealing from a theoretical point of view. It’s in contradiction with general relativity and that makes it a non-starter for most theorists. Meanwhile, however, there are variants of modified gravity which are compatible with general relativity.

The benefit of modifying gravity is that it offers an explanation for observations that particle dark matter has nothing to say about: Many galaxies show regularities in the way their stars’ motion is affected by dark matter. Clouds of dark particles that would collect in halos around galaxies can be flexibly adapted to match the observations of all observed galaxies. But dark matter particles are so flexible, that it’s difficult to reproduce regularities.

The best known of them is the Tully-Fisher relation, a correlation between the luminosity of a galaxy and the velocity of the outermost stars. Nobody has succeeded to explain this with particle dark matter, but modified gravity can explain it.

In a recent paper, a group of researchers from the United States offers a neat new way to quantify these regularities. They compare the gravitational acceleration that must be acting on stars in galaxies as inferred from observation (gobs) with the gravitational acceleration due to the observed stars and gas, ie baryonic matter (gbar). As expected, the observed gravitational acceleration is much larger than what the visible mass would lead one to expect. They are also, however, strongly correlated with each other (see figure below). It’s difficult to see how particle dark matter could cause this. (Though I would like to see how this plot looks for a ΛCDM simulation. I would still expect some correlation and would prefer not to judge its strength by gut feeling.)

Figure from arXiv:1609.05917 [astro-ph.GA] 


This isn’t so much new evidence as an improved way to quantify existing evidence for regularities in spiral galaxies. Lee Smolin, always quick on his feet, thinks he can explain this correlation with quantum gravity. I don’t quite share his optimism, but it’s arguably intriguing.

Modifying gravity however has its shortcomings. While it seems to work reasonably well on the level of galaxies, it’s hard to make it work for galaxy clusters too. Observations for example of the Bullet cluster (image below) seem to show that the visible mass can be at a different place than the gravitating mass. That’s straight-forward to explain with particle dark matter but difficult to make sense of with modified gravity.

The bullet cluster.
In red: estimated distribution of baryonic mass.
In blue: estimated distribution of gravitating mass, extracted from gravitational lensing.
Source: APOD.

The explanation I presently find most appealing is that dark matter is a type of particle whose dynamical equations sometimes mimic those of modified gravity. This option, pursued, among others, by Stefano Liberati and Justin Khoury, combines the benefits of both approaches without the disadvantages of either. There is, however, a lot of data in cosmology and it will take a long time to find out whether this idea can fit the observations as well – or better – than particle dark matter.

But regardless of what dark matter turns out to be, Rubin’s observations have given rise to one of the most active research areas in physics today. I hope that the Royal Academy eventually wakes up and honors her achievement.

88 comments:

Quentin Ruyant said...

Having a look at the history of science, one can observe that working theories are always conserved rather than revised in front of contradictory evidence. This is because it is always possible to form a local hypothesis in order to explain away an anomaly rather than abandoning the whole theory: typical examples are anomaly in the orbit of Uranus (explained away by Neptune) and mercury (tentatively explained away by Vulcan). Relativity was not designed to solve the anomaly with mercury, but for more theoretical purposes: unifying electromagnetism and mechanics. But eventually, it succeeded for Mercury too! The lesson is that conservatism in science is usually a good strategy, because it helps generate new successful hypothesis, but that it can fail at some point.

I can't help seeing an analogy with dark matter. We have an anomaly that we explain away with an hypothesis (dark matter). I'm not competent to know to what extent this hypothesis is successful in making new predictions, but from the outside, it seems closer to Vulcan than to Mercury. So I'd agree that building a new theory like MOND just to explain a particular anomaly is not necessarily the best strategy, and that new theories (in contrast with new local hypothesis inside a theory) should better be proposed to solve theoretical problems, such as quantum gravity. As relativity did. But still, the main theoretical difficulties in contemporary physics are linked to gravity, and the phenomena dark matter aims at solving are gravitational, so if I were to make a pronostic, on the basis of history of science, I'll bet that the anomaly will disappear when physicists come about with a new theory of gravity that they designed for different purposes. This is a non-specialist bet, of course.

Phillip Helbig said...

"Another year has passed and Vera Rubin was not awarded the Nobel Prize. She’s 88 and the prize can’t be awarded posthumously, so I can’t shake the impression the Royal Academy is waiting for her to die while they work off a backlog of condensed-matter breakthroughs."

I don't think that that is the reason. The Academy tends to be conservative. I think that if dark matter were actually detected, then Rubin would get at least a share of the prize. Rubin is associated with dark matter, while modified gravity is still in the running. Yes, there is a lot of other evidence for dark matter in many regimes. On the other hand, there are regimes where MOND works surprisingly well, and dark matter does not, at least not without many epicycles. This does not mean that MOND is right (it is almost certainly not, at least in the basic form), but it does mean that any dark-matter theory has to explain MOND phenomenology. In particular, Rubin's specialty of galactic rotation curves is something which does not naturally fall out of any dark-matter scenario.

Phillip Helbig said...

"But once physicists had the standard model, every problem began to look like a particle"

This is by far the best summary of those heady days:

http://astroweb.case.edu/ssm/mond/flowchart.html

Phillip Helbig said...

"Or maybe it means dark matter isn’t made of particles."

Or it is made of particles, but they are self-interacting and form macroscopic objects. You mentioned this a bit more than a year ago.

Phillip Helbig said...

"In a recent paper, a group of researchers from the United States offers a neat new way to quantify these regularities."

One of the authors is Stacy McGaugh, a leading MOND proponent. I find him more objective than many others on both sides of the debate. The paper you mention is mainly phenomenology, and it is a challenge to all theories to explain it.
javascript:void(0)

Unknown said...

The bullet cluster might deserve a Nobel prize: it added new important information. While galactic rotation curves don't add much to Zwicky

Phillip Helbig said...

"But regardless of what dark matter turns out to be, Rubin’s observations have given rise to one of the most active research areas in physics today. I hope that the Royal Academy eventually wakes up and honors her achievement."

Again, the fact that it has spawned a huge amount of research isn't enough for the Nobel Prize. Think about inflation: this idea has spawned even more research, but still no Nobel Prize. Maybe you can suggest that Rubin, like the inflationary aficionados, get the Breakthrough Prize. (This is much more money as well!)

Sabine Hossenfelder said...

Phillip,

Ah-ha, I didn't know that flowchart, hilarious!

Sabine Hossenfelder said...

Phillip,


Re Rubin: She observed regularities in spiral galaxies which cannot be explained by general relativity plus the known matter. This observation is now commonly attributed to particle dark matter. That this particle has not been detected and indeed might not exist is imo irrelevant to assess the impact of her achievement.

Take as example the 2011 Nobel Prize to Perlmutter, Schmidt, and Riess. They discovered not dark energy (or the cosmological constant), but evidence for the expansion of the universe, even though this expansion is now commonly attributed to dark energy. Same thing. The observation is relevant independent of whether a theoretical explanation has been satisfactorily settled.

Quentin Ruyant said...

I meant closer to Vulcan than to Neptune.

Bob Anderson said...

Sabine,

An interesting read as usual. I was however just a little lost by the sentence " But dark matter particles are flexible, that it’s difficult to reproduce regularities." ?

Also I think in the next paragraph "starts" should of course read "stars" .

Thanks for all your efforts to keep the rest of us up to date with Modern physics.

Bob

CGT said...

Hi, I have a basic question about GR regarding the difference between dark energy & dark matter as far as theory is concerned. When speaking of the Einstein equation, Is it the case that the contribution of dark matter is always included in the stress energy tensor (source term) and that dark energy is included in the cosmological constant term? If so, is this the main reason to distinguish between these two forms of 'darkness'? I ask because I don't normally read about dark energy being 'composed of particles' in the way dark matter is discussed phenomenologically. A source of dark energy, by analogy to non-zero vacuum values of the Higgs field in the standard model, seems more 'insubstantial' phenomenologically. So whatever physical mechanism contributes to a non-zero cosmological term would not also make a contribution to the stress energy tensor (source term) as well?

Sabine Hossenfelder said...

Unknown,

You are wrong. Zwicky's observation might have been some oddity. There were at that time lots of things out there in the universe which weren't well understood. Science is really driven by finding patterns and regularities and that's what Rubin did.

Sabine Hossenfelder said...

Phillip,

"Inflation" is not an observation. As I said, it clearly doesn't makes sense to hand out a prize for the discovery of particle dark matter, since that has not been confirmed. But the observation of rotational curves, and all their later implications, stands regardless.

Sabine Hossenfelder said...

Bob,

Thanks for letting me know, I've corrected these typos.

Sabine Hossenfelder said...

CGT,

That's a good question but one which requires a longer answer. I'll take this on in one of my future Dear Dr. B columns. Best,

B.

Phillip Helbig said...

"Ah-ha, I didn't know that flowchart, hilarious!"

I am a walking encyclopedia of funny stuff in astrophysics. :-)

Phillip Helbig said...

"Inflation" is not an observation.

True. I used this as a counterexample because it is something else which has stimulated a lot of research.

I'm not saying that I agree with the academy; I'm merely trying to find patterns and regularities in their behaviour. :-)

As I said, it clearly doesn't makes sense to hand out a prize for the discovery of particle dark matter, since that has not been confirmed. But the observation of rotational curves, and all their later implications, stands regardless.

I agree that is is an important observation, but somehow don't think that the academy sees it this way. Think of the photoelectric effect: the prize was awarded for its explanation, not for its discovery.

Take as example the 2011 Nobel Prize to Perlmutter, Schmidt, and Riess. They discovered not dark energy (or the cosmological constant), but evidence for the expansion of the universe, even though this expansion is now commonly attributed to dark energy. Same thing.

I agree, to a large extent. I would venture to say, though, that the cosmological constant, at least phenomenologically, is better understood than dark matter, in that one value fits all observations without any fudging. Nevertheless, I have heard criticism of the award for the 2011 prize precisely because it is "just" an observation. (Originally, Nobel wanted the prize to be awarded for an invention or a discovery. Whether an observation is a discovery is a matter of interpretation.)

Phillip Helbig said...

"Is it the case that the contribution of dark matter is always included in the stress energy tensor (source term) and that dark energy is included in the cosmological constant term? If so, is this the main reason to distinguish between these two forms of 'darkness'? I ask because I don't normally read about dark energy being 'composed of particles' in the way dark matter is discussed phenomenologically. A source of dark energy, by analogy to non-zero vacuum values of the Higgs field in the standard model, seems more 'insubstantial' phenomenologically. So whatever physical mechanism contributes to a non-zero cosmological term would not also make a contribution to the stress energy tensor (source term) as well?"

Short answer: To some extent, it doesn't matter which term you include it in (i.e. in the stress-energy term or as a separate term). Mathematically, that is, it doesn't matter. People who believe that the effect of the cosmological constant is due to some sort of "stuff" often include it in the stress-energy term, while those who think that a "bare" cosmological constant is the better option (at least until observations prove otherwise) like it in a separate term. Weinberg is famous for, among other things, using both: a pure cosmological constant with a large magnitude but negative sign, and a vacuum-energy (i.e. QM) contribution to the stress-energy tensor with almost the same magnitude, but positive. The net result is what we observe. He explains this almost-cancellation anthropically.

Unknown said...

Sabine, thanks for your insights in this excellent article. You bring important informations to us on the frontiers of physics!

Tom Andersen said...

That graph it its an observational astronomy graph - is the most important data I have seen in physics since the initial LIGO event.

Its basically the cuspy core dark matter problem only writ large. Not only do dense cores of DM not exist, it seems like there may be a maximum density of DM. My own entirely unpopular view is that DM turns into regular matter when some density is reached. I think that just because one MOND model works on one problem is not an indication that it is right - there are many MOND models - so parameter tuning will happen.

Uncle Al said...

" Tully-Fisher relation" Milgrom acceleration - angular momentum is trace non-conserved. Randomly oriented galaxies are globally conservative. Einstein-Cartan-Kibble-Sciama gravitation has no Equivalence Principle. Achiral spacetime curvature dominates gravitation. Chiral spacetime torsion is the literal footnote, then Noetherian leakage of Milgrom acceleration plus Sakharov baryogenesis conditions.

"Bullet Cluster" If spacetime deformation were knotted, then easily tied but not untied. Linearly polarized light (quasar) passing through "dark matter" is rotated vs. frequency. Geometric Eötvös experiments opposing enantiomorphic space groups single crystal P3(1)21 versus P3(2)21 alpha-quartz test masses are diagnostic. LOOK

http://www.sawyerllc.com/DataSheets/Quartz_Optical_Laser_Applications.pdf
Both hands of single crystal alpha-quartz as +X-region seed-free wave plate (WP) bar.

Sam Homiller said...

Hi Sabine,

Didn't Babcock also make similar measurements of the rotation curves in the Andromeda Galaxy around the same time as Zwicky's result?

http://adsabs.harvard.edu/full/1939LicOB..19...41B

Rubin surely deserves a ton of credit for improving these measurements, and perhaps for kick-starting the stream of further looking into dark matter as a possibility, I'm just curious as to why nobody mentions Babcock's name when discussing the lack of a prize for dark matter (although of course, he's also dead now, so he couldn't get a prize either).

Thanks!
Sam

Sabine Hossenfelder said...

Sam,

How do you measure systematic patterns in galactic rotation curves in one galaxy?

andrew said...

Go Vera Rubin!

One of the more interesting and little known efforts is by Alexandre Deur whose primary physics gig is doing QCD at Jefferson Labs. He's published two articles in peer reviewed journals (albeit little cited) that investigate whether dark matter phenomena and some dark energy phenomena could arise from a weak field approximation of the graviton self-interaction developed by analogy to self-interacting gluon interactions in QCD, which is discernible only in very massive systems which are not spherically symmetric. https://arxiv.org/abs/1407.7496 and https://arxiv.org/abs/0901.4005

For those who prefer the slideshow format: http://www.phys.virginia.edu/Files/fetch.asp?EXT=Seminars:2693:SlideShow

Deur's effort is notable because (1) it is the only modified gravity or dark matter theory to accurately predict that more spherical elliptical galaxies have less apparent dark matter than less spherical ones by a factor of four, (2) it explains why satellite galaxies tend to be in the plane of a spiral galaxy, (3) it shares all of the strengths of modified gravity theories but unlike MOND it naturally scales all of the way from dwarf galaxies to galactic clusters with the right order of magnitude for dark matter phenomena, (4) it is consistent with the bullet cluster, (5) it is theoretically well motivated from quantum gravity concepts and parallel interactions in QCD, (6) it explains all of dark matter and much or all of dark energy with only a single beyond the Standard Model particle, the widely hypothesized spin-2 graviton, without introducing any other new particles, fields or dimensions of space-time, (7) it needs just one physical constant (the gravitational force coupling constant) rather than the three needed in lambdaCDM (G, lambda and a CDM mass) or in Bekenstein's relativistic generalization of MOND call TeVeS (G, lamda and a dark matter acceleration constant), (8) it naturally explains the "coincidence problem" and the small size of the cosmological constant, (9) its strong gravitational field predictions are generally identical to GR because systems like black holes, neutron stars and the Big Bang are almost always close to spherically symmetric (except for stellar sized binary black holes which are much less massive than galaxies and galactic clusters and supermassive black holes and the Big Bang).

Deur's work on the problem to date has not looked at cosmology in any depth and unlike most GR works does not devote much analysis to mathematical rigor and mathematical consistency. It makes the kind of approximations that phenomenologists are prone to make. I've corresponded with him and he's had to put the project on the back burner for the day job at Jefferson Labs, but it is a far more fascinating effort than purely phenomenological and rather Byzantine work by Moffat at the Perimeter Institute, and probably overlaps with massive graviton theories and f(R, T) theories of more orthodox GR investigators, as those theories also examine the self-interaction of the gravitational field.

Andrew Thomas said...

That graph showing correlation between acceleration and observable mass is very interesting.

If it wasn't for that blinking Bullet cluster result, which strongly suggests particles. I wonder if someone might check that Bullet cluster result sometime, to see if we can really trust it.

andrew said...

One of the more successful recent efforts to reproduce the baryonic Tully-Fischer relation with CDM models is L.V. Sales, et al., "The low-mass end of the baryonic Tully-Fisher relation" (February 5, 2016). https://arxiv.org/abs/1602.02155 As this paper explains (citations omitted):

"[T]he literature is littered with failed attempts to reproduce the Tully-Fisher relation in a cold dark matter-dominated universe. Direct galaxy formation simulations,for example, have for many years consistently produced galaxies so massive and compact that their rotation curves were steeply declining and, generally, a poor match to observation. Even semi-analytic models, where galaxy masses and sizes can be adjusted to match observation, have had difficulty reproducing the Tully-Fisher relation, typically predicting velocities at given mass that are significantly higher than observed unless somewhat arbitrary adjustments are made to the response of the dark halo."

The paper's simulation usually a model that has sixteen parameters carefully "calibrated to match the observed galaxy stellar mass function and the sizes of galaxies at z = 0" and "chosen to resemble the surroundings of the Local Group of Galaxies", however, and still struggles to reproduce the one parameter fits of the MOND toy-model from three decades ago.

Much of the improvement over prior models has come from efforts to incorporate feedback between baryonic and dark matter into the models, although generally in a manner than is more ad hoc than it is firmly rooted in rigorous theory or empirical observations of the feedback processes in action.

One of the more intractable problems with simulations based upon a dark matter particle model that has been pointed out, for example, in Alyson M. Brooks, Charlotte R. Christensen, "Bulge Formation via Mergers in Cosmological Simulations" (12 Nov 2015) https://arxiv.org/abs/1511.04095 is that their galaxy and mass assembly model dramatically understates the proportion of spiral galaxies in the real world which are bulgeless, which is an inherent difficulty with the process by which dark matter and baryonic matter proportions are correlated in dark matter particle models which are not a problem for modified gravity models. They note that "we also demonstrate that it is very difficult for current stellar feedback models to reproduce the small bulges observed in more massive disk galaxies like the Milky Way. We argue that feedback models need to be improved, or an additional source of feedback such as AGN is necessary to generate the required outflows."

Brooks has also written a couple of even handed review articles evaluating current dark matter particle theories in light of the evidence: Alyson Brooks, "Re-Examining Astrophysical Constraints on the Dark Matter Model" (July 28, 2014). https://arxiv.org/abs/1407.7544 and was a co-author of a paper finding that two leading dark matter particle theories have very similar predictions. https://arxiv.org/abs/1501.00497

Bill said...

Thank you for another interesting post, Dr. Hossenfelder.

I've always found it strange that the energy-momentum tensor T for electromagnetism is traceless, while the Einstein tensor G is not. In some variants of modified gravity theory, the traceless Cotton tensor replaces the Einstein tensor, leading some to believe that perhaps a conformally invariant theory of gravity might explain dark matter and dark energy without having to resort to new particles and/or hypothetical scalar-vector-tensor-spinor fields. Mannheim, Kazanas and others have explored this possibility, sadly without much serious attention.

Meanwhile, dark matter particles have not been found despite heroic efforts, and dark energy continues to be viewed as a kind of mysterious aether field. Perhaps the particles simply don't exist, while dark energy is simply a consequence of the cosmological constant. Why isn't modified gravity being taken more seriously?

johnmn3 said...

Thanks for breaking it down, like usual, Sabine. Since I'm in an easier position to "go with my gut," I reckon dark matter is just another 'god of the gaps.'

Tam Hunt said...

Interesting piece Sabine, thanks. I'd love to hear your thoughts on the possibility that GR is simply wrong. As you write: "none of these would fit with the predictions of general relativity," seemingly acknowledging my point from your other recent post that the various patches of dark matter, dark energy and inflation are simply fixes to ensure that GR works. Without these fixes GR fails. So maybe GR is the problem, as you suggest with your ruminations about MOND and other approaches. I think it's high time that we remove GR and SR from the pedestals of indubitability in modern physics.

piein skee said...

The fact is Dark Matter halos do not explain why radial velocities are invariant away from the core anyway. All it can do is show one way to kinda sorta get outer star orbital speeds happening too fast. The fact it's a constant velocity at all radial points outward gets put down to a coincidence. So more fine tuning.
how will modified gravity explain the large scale structure which totally depends on dark matter. The filament structures. What about the Dwarf Galaxies having massively more dark matter kg for kg...like 5 or 10 times more (because they need it to explain why they stay intact at all. Or the fact that their problem - the Dwarfs - is the opposite to spirals....with stars orbiting at far too slow velocities. The should just straighten out and go their own way right out of the galaxy. I mean, invisible matter that can be cut to fit any anomaly. What makes me cackle is hearing these people bemoan the dearth of major anomalies that contradict the standard model of cosmology or whatever, to excuse themselves of the mess frontier physics and cosmology has become on their watch.

There has been more anomalies in the past 20 - 30 years than at any time in the history of Science. More, and bigger. Vastly bigger. Do you know what drove those geniuses of yesterday year, to tear up 100 yeara of settled physics and declare a catastrophe (ultra violet) that kicked off an intellectual journey of discovery that left modern physics in its wake? A lightbulb the filament of lightbulbs, which had newly been invented. Not that it lit up, but that it lit up the way that it did. Sure lucky no one was brilliant enough to come up with invisible electrons that are just there, balancing the inequalities real nice. What was it Stalin said? "If there is an anomaly there is a problem. If There is no anomaly.... there is no problem.

piein skee said...

"How do you measure systematic patterns in galactic rotation curves in one galaxy?"

photometry...relative redshift between orbitals at different distances from the core. I should think.

Kimmo Rouvari said...

Hi Bee,

Do you know any physicist wondering how all of the interactions and dark matter/energy could be fitted together? (hypothesis: they emerge from the same principles)

Sabine Hossenfelder said...

piein,

We're having a little bit of a misunderstanding there. I was saying that Rubin's achievement was to show that galaxies are systematically odd. For that she measured rotation curves of several dozen galaxies (I think 60 or so). My question how do you show this from one galaxy's rotation curve was a rhetorical question. Yes, there are several measurements involved in getting one rotation curve. That wasn't my point. It's still only one galaxy. Hope that clarifies it.

Sabine Hossenfelder said...

Kimmo,

Yes, I've seen various proposals of that sort, but none of them is remotely plausible.

Sabine Hossenfelder said...

Tam,

Well, we know that SR and GR aren't "simply wrong." They work dramatically well. I don't know what you mean. There are loads of people who are considering that they fail in some limits. But so far nothing speaks for that.

Alexey Fedotov said...

Dear Sabine,

can you please write how modified gravity can be compatible with general relativity? I thought that GR is complete theory, without free parameters (except of possible lambda term).

Sabine Hossenfelder said...

Alexey,

I mean it's possible to modify GR so that it remains generally covariant, which Newtonian gravity isn't. The Einstein-Hilbert action of GR has two free parameters, Newton's constant and the cosmological constant. In modified gravity, you have more free parameters, depending how how many fields and terms you add.

Tam Hunt said...

Hi Sabine, you write in this piece above that many observations "don't fit with the predictions of general relativity" so I'm not seeing how that statement fits with your comment at 1:10 AM that SR and GR "work dramatically well." It seems that one of your statements is inaccurate, no? Surely SR and GR do work in many situations but there is a growing body of empirical anomalies that neither theory can explain (which is why, of course, MOND and the other approaches you survey are still being discussed) and this is what I'm suggesting should be leading to a long-overdue re-examination of these pillars of modern physics.

Sabine Hossenfelder said...

Tam,

No, I did not write that "many observations don't fit with the predictions of general relativity." I don't know where you got that from. Indeed general relativity in form of the concordance model fits almost all existing observations. There are just a few oddities that it has trouble with, and these are the ones I was discussing here. General relativity works extremely well, and this is the reason why physicists do not like the idea of modifying it - it is very hard to do without ruining the ability of the theory to fit the huge amount of other data that it does work with so well. Consequently, they prefer to add particle dark matter to the standard model. Best,

B.

Shantanu said...

Note that if you just look at galactic observations, there is a much more prosaic explanation of no dark matter.
See
https://arxiv.org/abs/1510.05534
Would be interested to here everyone's thoughts on this.

Tam Hunt said...

Hi Sabine, what I'm referring to is your statement in the article: "The fluctuations in the cosmic microwave background, gravitational lensing, the formation of large-scale structures in the universe – none of these would fit with the predictions of general relativity if there wasn’t additional matter to curve space-time." So you seem to be suggesting quite clearly that dark matter was proposed specifically to save general relativity from anomalous observations, no? And if so, how can you also state that GR works dramatically well? It would seem that these statements are in conflict. And ditto, of course, with dark energy and inflation which were also proposed in order to allow GR to explain the observations.

Sabine Hossenfelder said...

Tam,

You are misunderstanding this. Particle dark matter is a source-term to general relativity. You do not modify general relativity. You add a certain type of matter. And that works dramatically well. As I wrote. Same for dark energy. I don't know what you think is in conflict with what.

Sabine Hossenfelder said...

Everyone:

I'm getting a lots of comments to this post in which people propose their own theories for what dark matter might be. You are wasting your time. I do not approve such comments. If you want to discuss your own theory of something, please do so elsewhere.

Phillip Helbig said...

"Note that if you just look at galactic observations, there is a much more prosaic explanation of no dark matter.
See
https://arxiv.org/abs/1510.05534
Would be interested to here everyone's thoughts on this."


Even assuming that this paper is 100% accurate, the problem is "if you just look at galactic observations" (by which I think that you mean "if you look at just galactic observations"). Any valid theory has to explain all observations. Even MOND supporters admit that on cosmological scales the agreement between the standard LCDM model involving dark matter and observations is very, very, good, and there is no obvious way to avoid dark matter here. (Proponents of the standard model have been somewhat less apt to admit that there are problems at galactic scales; this doesn't mean that MOND is true, but it does mean that any theory has to explain the observations which MOND explains easily and naturally (many of which did not exist when MOND was formulated).)

Phillip Helbig said...

"You are misunderstanding this. Particle dark matter is a source-term to general relativity. You do not modify general relativity. You add a certain type of matter. And that works dramatically well. As I wrote. Same for dark energy. I don't know what you think is in conflict with what."

Indeed. This is a common criticism, and completely off the mark. GR is a theory where space-time tells matter how to move and matter tells space-time how to curve. The theory has absolutely nothing to say about the reason for existence, composition, number density, or other properties of the source terms. Saying that GR is incomplete for that reason is like saying that Linnaeus's binomial classification scheme is faulty because it did not predict the existence of gorillas.

Unfortunately, "but we don't know what they [dark matter and dark energy] are" is an argument sometimes used by supporters of MOND. (Fortunately, they still have a strong case even without this argument.) To me, it would be much more surprising if the stuff we are made of is the dominant stuff of the universe. (Expecting this because "we should be typical" isn't valid here, since requirements for life might be highly atypical. Similarly, most baryons are in stars or gas, but neither is a good place for life to form.)

Saying that dark matter and/or dark energy are some sort of epicycles to save the appearances of GR is completely wrong. It implies the expectation that a theory should predict, from first principles, the complete composition of the universe. No theory does that. (I'm not saying that it is logically impossible, just that no rival theory to GR does anything like this. Even MOND does not predict the density and other properties of baryons.)

Shantanu said...

Phillip, in astrophysics there is a confirmation bias (many papers on this subject in astro-ph). Not saying that all the deductions from LSS and CMB from lambda CDM are wrong. But I am pretty sure these fits are influenced to some extent by previous results.

Sabine Hossenfelder said...

Shantanu,

Well, there isn't only a confirmation bias in astrophysics... But since you mention it, let me add the following interesting observation. If there's an observational result that doesn't speak for CDM, then it gets blamed one some misunderstood stellar or intergalactic medium astrophysics. If there's otoh an observational result that might be a signal for CDM, then the idea that it might be some misunderstood "normal" astrophysical process gets ignored and everyone pulls out their favorite dark matter particle to fit the signal. (Which then goes away.)

Koenraad Van Spaendonck said...

@Phillip

I couldn't really agree here :

"Saying that dark matter and/or dark energy are some sort of epicycles to save the appearances of GR is completely wrong. It implies the expectation that a theory should predict, from first principles, the complete composition of the universe. No theory does that. (I'm not saying that it is logically impossible, just that no rival theory to GR does anything like this. Even MOND does not predict the density and other properties of baryons.)"

It's not as if GR was some sort of adding epicycles to Newtonian gravity, it actually improved the formula's from first principles.
It is not unthinkable that GR can be generalized in a similar manor, to improve very long distance gravitational behaviour.

No theory indeed will predict the whole universe from first principles , because we are just human beings, not omniscient beings, but history shows that theories of gravity, of the universe do indeed get generalized, it just takes several centuries. What's a hundred years of validity of GR in the scope of thousands of years to come. There is little chance it will be the final chapter.

Best, Koenraad

Shantanu said...

Sabine: I completely agree

Andreas Maier said...

I'm very sceptical about that paper. It's main evidence is shown in a log-log plot. And log-log plots are notoriously difficult to interpret reliably. Very often one sees correlations, which actually don't exist. I would like to see a linear plot of the data and I hope the referees will do so, too.

Tam Hunt said...

Hi Sabine and Philip, I disagree with your logic and am surprised that you’d take this position. As you yourself write very clearly in your piece DM was proposed specifically because GR couldn’t explain the observations about lensing, CMB and large-scale structure (and again ditto for dark energy and inflation for other observations). DM was proposed, with still no evidence of actual DM particles other than the original anomalous observations that led to its being proposed even forty or so years later, in order to fit the data with the theory.

It is undeniable that without the DM modification GR does NOT explain these observations. So you are saying that “no, GR is correct still even though we’ve now included a number of massive changes to the composition of our universe, with limited to no evidence otherwise for these changes, in order to make GR work.” I fail to see how these changes can be interpreted in any other way than as attempts to save GR from falsification.

DM is just one of an infinite number of possible suggestions to explain the anomalous observations. And surely you’d agree that it is highly ad hoc and unsavory to preserve a theory (GR in this case) by simply positing an unknown and still undetected substance that just must be there in order to save the theory. And when we have galaxies now being discovered that require literally 99% DM in order for GR to be able to explain their rotations wouldn't you agree there’s something awry here? At what point do ad hoc fixes render the underlying theory untenable? I’d hope that a 99% ad hoc fix would be that point!

Sabine Hossenfelder said...

Tam,

All we're saying is that particle dark matter is NOT a modification of general relativity. It isn't, it works, and no amount of your complaints will change anything about this.

"And surely you’d agree that it is highly ad hoc and unsavory to preserve a theory (GR in this case) by simply positing an unknown and still undetected substance that just must be there in order to save the theory."

No, I don't agree.

akidbelle said...

Dear Sabine,

with a rather external eye, I would say that the fashion of the century is to pile-up fields to explain data within existing frameworks. The difference between gravity and particles physics is that the latter has a head-start in precision and can kill ideas one by one (provided gozillions dollars and decades to the next test). I think the bottom line requiring that nobody looks elsewhere is known as first principles. If so, the next theory is required to be more or less a cut and paste of the old one. Is that right?

Let us be a bit aggressive: If right, this process is the very opposite of research, because everyone already knows what remains to be found - not only its theoretical form. (I agree physics is about solving problems and theories cannot come out of thin air.)

Thanks,
J.

Tam Hunt said...

Hi Sabine, so you are okay with an argument that requires 99% of an unknown and undetected substance to explain particular galactic rotations within the prevailing theory? At what point do you find such arguments untenable? When physicists like yourself refuse to address these basic points it begins to look like groupthink has a good hold on physics today because to outsiders like myself (I'm coming at these issues as a philosopher of physics) it looks pretty absurd.

Sabine Hossenfelder said...

Tam,

Yes, I am perfectly ok with that. Dark energy and dark matter both are very simple explanations in terms of the degrees of freedom. They are simple yet they describe our observations. (Except for the shortcomings which I mentioned above.) It's this simplicity that what makes them successful theories. How much of dark matter there is is entirely irrelevant for what the theoretical success is concerned - the percentage is just a number that is extracted from the data.

I don't think you're qualified to yell groupthink - you can leave that to me ;)

Best,

B.

piein skee said...

"I was saying that Rubin's achievement was to show that galaxies are systematically odd. For that she measured rotation curves of several dozen galaxies (I think 60 or so). "

Sure, but all the fundamental challenges, breakthroughs, breaking new ground, and so on, happen as part of a single galaxy study. Publishing a single galaxy study with those results automatically triggers one or more larger scale surveys of multiple galaxies, applying the standardized sequences thrashed out in the initial single galaxy study, as mechanically as possible, is feasibly well within the capacity of the generic survey template. So in context of the reader-question, about the real or possible significance of this parallel work at the level of a single galaxy, it could in answer be very significant indeed. Because it's reasonable to suppose both of these people were doing single galaxy studies on exactly the same terms at more or less the same time. It's also sensible to suppose both settled on Andromeda in the event, because it's well oriented for redshift photometry involving relative differential of the stars and structure away from the core (i.e. it's increasingly problematic toward both 'flat-on' and 'edge-on' galaxy orientations to our vantage point)

That level of coincidence certainly makes it reason to question whether and what extent one of them took the ideas and work of the other. If it was her that's one way to explain how she came to be in a position to kick off the larger survey in advance of him. That wouldn't be a case of 'scooping' it would be straight plagiarism, for the simple reason all the major accomplishments happen in the single galaxy survey.

You should consider looking into the situation, because there's enough information in play there, that if you published a controversial findings, you'd very plausibly enjoy the satisfaction of it feeding to a lasting revision...which'd elevate your status to that of a scientist who has made her mark. Almost no-one in your generation has accomplished that.

Quentin Ruyant said...

@Tam, Sabine, Phillip

Just to clarify the discussion a bit, I suggest we differentiate:

- the theoretical framework (e.g. newtonian mechanics, relativity, quantum field theory), i.e. the "core" of the theory

- the facts and models produced within the theory (ideal pendulum, solar system model, superconductivity, atom of hydrogen, cosmological models...).

Facts and models are mediators between theory and experience. They are *applied* to a domain of experience. They are what can be directly confirmed or disconfirmed if they correspond or not to experimental data.

Theories, in contrast, are never strictly refuted by experience, because they're more like tools (laws formakism etc.) to produce models. If a model doesn't work, you can always adjust your assumptions and come with a new model that better fit the data. But theories are still indirectly confirmed by their capacity to produce a growing amount of confirmed facts and models that produce new, unexpected predictions. They are indirectly disconfirmed by facing a growing amount of anomalies that require ad-hoc models/hypothesis with growing complexity. Often these anomalies are later explained by a brand new framework, and then only we realise that they were really anomalies.

I suggest (taking inspiration from Lakatos) that an hypothesis is ad-hoc when it does not produce new predictions that are independent of what they were designed to explain. So novel predictions is really the central criteria to assess if a theory is successful. The prediction of Neptune is an impressive example of the success of newtonian physics for example.
Neptune was an ad-hoc hypothesis until it was observed.

Would you agree with this picture?

If so, then Sabine is right to say that general relativity is not particularly disconfirmed by galaxy rotation curves and the like. One can still produce models where it works perfectly. There's nothing bad in this: it's how science has always worked. Produce new hypothesis to explain data, then try to confirm them independently.

But Tam's (and MOND defender's) worry can be expressed as follows: what it takes to make it still work is not very fruitful, in that it does not produce novel predictions: we never observed dark matter independently of the phenomena it is designed to explain. It could happen in the future (in particle accelerators) but we don't have it yet. Meanwhile, dark matter is still, in the sense explained above, an "ad-hoc" hypothesis. (I'm ready to hear arguments to the contrary, because I'm not a specialist).

So there's no reason to reject dark matter, because producing this kind of hypothesis is what scientists always did, often with great success, but one can worry that in this case the success is not coming very fast and that it might never come... Until a new theory (a new framework) explains all these "anomalies". Which also happened in the history of science (we tend to forget failures, but they did happen!).



Sabine Hossenfelder said...

Quentin,

No, you have it backwards - a theory is what maps a model (a math-thing) to observation (data). See here for explanation.

Having said that, you can define 'ad hoc' to mean anything you want, but what you write is not the way the word is presently used. It's mostly used, I'd say, to mean "unmotivated" in the sense of making an hypothesis that doesn't follow from any deeper principles. Actually Lee made some distinction to this extent in the paper I mentioned above. An ad hoc hypothesis can totally be predictive. An example may be the gyromagnetic moment of the electron. In the first phenomenological model (Goudsmit and Uhlenbeck) they made the ad hoc assumption that it's two - and that is highly predictive for atomic spectra (for example). The factor two was later derived from Dirac's equation, so then it was no longer ad hoc.

I am not sure exactly what the problem is here. I think it should be clear from my post and from my comments that I am not sold on particle dark matter. Nevertheless the concordance model is the currently best explanation for our observations and it requires no modification of gravity.

As I mentioned I quite like the idea that dark matter is a fluid that mimics modified gravity. Also, I don't think the difference between modified gravity and extra matter is very sharp. In both cases you add extra stuff to the Lagrangian. The difference is just what stuff and how does it couple. It's actually more a community-divide than a mathematical divide I'd say.

Of course there's no reason to 'reject dark matter'. Look, we already know that whatever explains our observations can be described to very good accuracy as cold dark matter. That is, just to make sure we're on the same table, actually just some kind of pressureless, weakly interacting, fluid-stuff. Even if we find some better explanation at some point - one that describes more data better without adding more parameters - we already know that it must almost act like CDM in many regards. Best,

B.

N said...

I am all for MOND. If true, things could again become interesting.
As for Rubin's Nobel, it would take you or me or someone to go out there with a FTL drive and bring a basket of that dark stuff to the Stockholm Academy.
An maybe even that would not suffice.

Quentin Ruyant said...

(thanks for your answer but I can't see my previous comment in the thread?)

Sorry but no... Models are the mediators between theory and experience when it comes to *empirical confrontation*. You don't need Newtonian laws of motion to relate a model of the solar system to our observations: everything is in the model. But you need a model to check the predictions of Newtonian physics in the solar system. Models (not theories) are applied to a specific domain of experience.

What you're saying in the post you mention is just that theories are used in *model construction* but theories are not sufficient (and not even necessary!) to build models of phenomena from our observations. Models come with assumptions that are not in the theory, nor in our observations (Neptune was postulated). In some disciplines, you have models without a real theory (population growth, ...), and historically, models can function as a heuristic tool to derive laws and build a theory from the phenomena, not the other way around (Models of the atom are examples). So there is no sense in which theories would be mediators.

Regarding ad-hoc I quote Wikipedia: "In science and philosophy, ad hoc means the addition of extraneous hypotheses to a theory to save it from being falsified. Ad hoc hypotheses compensate for anomalies not anticipated by the theory in its unmodified form."
This seems correct to me, and this is exactly what I meant in my previous comment, not something I made up. These hypotheses converts into specific models. Now if the hypothesis is confirmed by *new* predictions it is no more an ad hoc hypothesis, but a confirmed one. You would agree that Neptune *was* an ad hoc hypothesis in the sense of Wikipedia before it was observed, and that it is *not* an ad hoc hypothesis today because we observed it.

You don't think the difference between dark matter and MOND is very sharp, but it is: in one case, the theory is modified. In the other, it's the models. This has implications: for example, following MOND, it is not conceivable that a galaxy would have no dark matter in it and follow "normal" rotation curves, while following dark matter, it is possible in principle.

Sabine Hossenfelder said...

Quentin,

I am not interested in discussing the meaning of words. You can use the words 'model' and 'theory' and 'ad hoc' in whichever way you want. I have merely told you that the way you use them is not the way they are presently being used by practicing scientists, and you should keep that in mind.

You continue to talk about MOND. Please note that I'm not talking about MOND. I'm talking about modified gravity.

As to your statement what is and isn't conceivable, I don't know where you get that certainty from. Show me a structure formation simulation with particle dark matter where you end up with a galaxy that doesn't contain dark matter. I would guess that the chances of this happening are basically zero. And for what modified gravity is concerned, as I said previously, these theories have various fields in them, each of which needs their own initial value. Consequently there must be various possible galactic models, and in particular the outcome of structure formation should also path-dependent. Unfortunately, I can only do guesswork on this account because for all I can tell there isn't any analysis of this. And why is that? Well, that's where Tam can start yelling groupthink if he wants.

I hope this explains why I said the boundaries between the two aren't all that sharp. Best,

B.

Quentin Ruyant said...

Sabine,

Yes this clarifies in what sense the difference is not sharp, thank you.


About models: I'm taking about models used and referred to by scientists, not anything else. Cosmological models, models of our solar system, models of atom, models of population growth... It's the models that are compared to experimental data, not the theory directly. It makes no sense to say that the theory is a mediator between experience and models.

Sabine Hossenfelder said...

Quentin,

Yes, you compare the model to data. How do you compare it? You use a theory... Look, the model is a collection of mathematical definitions and equations. You need a theory to tell you how these are identified with observations. Example: General relativity is a theory. The concordance model is a model which is mapped to observation by use of the theory. The standard model is a model which is mapped by qft to observation. Etc. Best,

B.

Tam Hunt said...

Hi Sabine, when a theory can explain anything -- due to various fixes added each time an anomaly arises -- how can that theory ever be falsified? Or are we now in the "post-falsificationism" era? That itself would be a great topic for a blog since luminaries like Sean Carroll have surprisingly argued against falsificationism.

Also, I guess I'm a bit confused now on your position. The point of your blog (your title even) is that MOND is now getting a new look because of ongoing empirical challenges to GR, which you enumerate briefly in your piece. You write about how the particle-based notion of DM may be wrong and how MOND and MOND-like approaches are being re-examined b/c of the difficulties in finding particle-based DM candidates in the many experiments that have indeed failed. So, again, how is GR an empirically adequate theory, that works "dramatically well," as you write above, when the entire discussion about DM and MOND has been prompted by GR's empirical failures, as you yourself write?

You write: "The fluctuations in the cosmic microwave background, gravitational lensing, the formation of large-scale structures in the universe – none of these would fit with the predictions of general relativity if there wasn’t additional matter to curve space-time." And you add: "Maybe gravity doesn’t work the way Einstein taught us." Then you add, with respect to the particle DM hypothesis explaining the Tully-Fisher relationship: "It’s difficult to see how particle dark matter could cause this."

So, again, what exactly are we disagreeing about? I'm suggesting that we have numerous observational challenges to GR, which you explicitly state you agree with. And then you suggest that I'm wrong when I suggest that maybe it's time we re-think GR. Something isn't connecting here obviously :)

And as for my qualifications to suggest groupthink, kind of like you writing about philosophy of physics in previous blogs? ;) I don't care much for academic boundaries, actually. I figure "have brain, will travel." With a decent grasp of logic and language, plus the wide availability of peer-reviewed literature online, in readily available books and with libraries when necessary, anyone who cares to can and should explore fields that aren't technically their own. I'm technically a renewable energy lawyer (but with a background in science) as my day job but I've read widely for a few decades now, and published in philosophy and other fields for a while now. Anyway, as always, I appreciate the stimulating dialogue.

Tam Hunt said...

Quentin, I'd suggest that after forty years of not finding dark matter this ad hoc hypothesis should probably be considered disproved and we should move on to alternative approaches.

Quentin Ruyant said...

One makes assumptions about a domain. One applies theoretical laws to these assumptions. This results in a mathematical structure that represents the domain. You call "model" the assumptions, and I (and to my knowledge most philosophers discussing scientific models) call "model" the resulting structure.

That makes no practical difference once a background theory is assumed. You claim that your use, not mine, corresponds to that of practising scientists. Although scientific use needs not be as precise as philosophical use in these matters, I doubt it, because I think that the background theory is always implicit when a scientist exposes a model: as I view it, the standard model describes particles as fields with specific characteristics, so QFT laws are implicit and full part of the standard model. I would also say that according to scientists, the classical and quantum harmonic oscillator are different models, for example, although they're built from similar assumptions, because they take place in different theories.

I might be wrong--after all, you're the practising scientist-- but still, I'd like to see evidence for your claim: for example, an academic article where a scientist would expose assumptions as a "model" without assuming any particular background theory (or even better, being willing to apply the "same model" in different theoretical frameworks).

In any case, the resulting mathematical structure is what I meant by "model" in my previous comments.

Sabine Hossenfelder said...

Quentin,

You don't normally see different theories applied to the same model because that doesn't make any sense from a practical point of view. You only see this sometimes as mistakes that get buried in the history of science. (Eg Einstein trying to figure out how to extract observables from the Riemann-tensor.)

What you do in practice instead is develop a model (or more often several) within the context of a theory to explain observations. It then only makes sense to apply different theories to the same model if the theories lead to more or less the same outcome. You see this for example in the foundations of quantum mechanics, eg the double-slit experiment. Same model, yet there are n>3 different theories to map it to data (all with the same outcome). Here too you have a case where a theory failed: Schroedinger originally tried to interpret the wave-function for the electron as a density-distribution. (Right model, wrong theory.)

I have never met a physicist who uses the word 'domain' in the sense that you explain it. (I only know it from the context of domain walls.) I know some physicists who call 'framework' what I call 'theory' and who call 'theory' what I call 'model' and what they call 'model' I would more specifically call a 'phenomenological model', which you might call 'ad hoc'. Anyways, so much about the verbal confusion. Actually if you have a reference on the way you use the words I would be interested to look at this just to avoid this problem in the future.

Sabine Hossenfelder said...

Tam,

Ok, good, I think I owe you an explanation, or at least an attempt at.

First let me say that I get a lot of comments from hobby physicists who insist that GR is all wrong and their great theory is right (I don't post these, but trust me they're plenty). Anyone who comes at me with comments of this kind is by default assumed a crank. If you want to have a reasoned conversation with a physicist I would generally recommend you don't start with complaining that GR is wrong. So my apologies for the misclassification.

Second, GR is on all accounts a dramatically successful theory. The issues that I mention above are anomalies. All theories constantly have to struggle with anomalies. The big question is how seriously do you think they are? What I have, maybe not very successfully, tried to convey is that the current consensus is that they're not very serious. They're some puzzles. Most in the field think that we'll figure these out sooner or later. The cusp-core problem might be an issue of misunderstood energy flows. The dwarf galaxy issue really might not be one. The alignment of the satellites might merely be coincidence. The regularities in the galaxy rotation curves which I wrote about here are imo the toughest issue. I don't know any good explanation in terms of the concordance model.

The problems with modified gravity otoh are much more serious. I don't know of any good demonstration that these models reproduce the biggest achievements of the concordance model, structure formation, cmb, bbn etc. This isn't to say that it can't be done. Just that for all I know it hasn't been done. Consequently, the currently best model is still the concordance model. And the concordance model's theory is general relativity (unmodified).

Third, as to falsifiction. Strictly speaking one can't falsify any theory because it can always be fudged. I like to say we don't falsify theories, we implausify them. To implausify one theory you have to come up with a better one. And there's no better theory than GR right now.

Fourth, sure I agree that we should rethink GR. I write papers about that kind of thing.

Fifth, as to groupthink. I'm happy to hear that a philosopher has an interest in sociology because it's a great frustration of mine if philosophers of science neglect the influence of sociology. I've learned it isn't called groupthink it's called communal reinforcement. And, yes, I am worried about this.

Best,

B.

Sabine Hossenfelder said...

Quentin,

PS: It occurs to me that a lot of qm textbook examples do what you ask for. You can do a calculation, eg for a scattering process, either classically or quantum mechanically. Same model, different theory.

Quentin Ruyant said...

@Sabine

This https://www.amazon.com/Scientific-Philosophy-Science-Daniela-Bailer-Jones/dp/082296273X

Or this for a formal approach http://plato.stanford.edu/entries/models-science/#ModThe


Quentin Ruyant said...

As for "domain" it might be a bad translation of my native French.

S said...

There was a time the majority of physics community taught there should be an ether, but they didn´t find any.
Then someone came and told us, there is no ether but special relativity.
Nowadays the majority of the physics community is teaching there should be dark matter, but they do not find any.
I´am sure someone will come to tell us, there is no dark matter but a (maybe slightly) modified theory of gravitation (compared to GR).

… and a comment to the bullet-cluster picture. Did you check which kind of theory they used to calculate the amount of gravitational lensing? If they used GR, then it comes not as a surprise to me, that the red and blue regions do not match. If they would have used the (currently not known) new theory of gravitation, presumably they would have found perfect matching.

Phillip Helbig said...

"The problems with modified gravity otoh are much more serious. I don't know of any good demonstration that these models reproduce the biggest achievements of the concordance model, structure formation, cmb, bbn etc. This isn't to say that it can't be done. Just that for all I know it hasn't been done. Consequently, the currently best model is still the concordance model. And the concordance model's theory is general relativity (unmodified)."

Back in January 2015, I attended a conference called "Beyond LambdaCDM", which was attended by many people working on modified gravity, some working on MOND, and some "mainstream" physicists who believe that LambdaCDM is essentially correct. (Note that talking too loudly about Lambda 25 years ago was almost enough to get one classified as a crank. Fortunately, science is self-correcting and this type of arrogance is gone. Check out the story of the crying student in Dennis Overbye's Lonely Hearts of the Cosmos.) So much for the claim that non-mainstream ideas are not seriously considered by the community. People who meet in Norway in the winter to discuss physics are very serious. :-)

After a comment after one of the talks where the advantages of some rival theory had been touted and problems glossed over, George Efstathiou set the bar really low. All that is required is to explain currently available data as well as LambdaCDM (i.e. not even requiring falsifiable predictions) and "I will give you a job". As far as I know, George hasn't hired anyone as a result of this challenge. :-)

Phillip Helbig said...

The discussion has cleared up some misconceptions, but one I think still remains. Dark matter is not some sort of ad-hoc idea to save GR. If that were true, then it means that our expectation is that everything which has gravity is somehow detectable by humans with their current technology. It would also imply that Uranus was an ad-hoc modification to Newtonian gravity to save a theory which should be sent to the scrap heap.

Yes, something similar happened later with Vulcan (there is a nice new book on this, The Hunt for Vulcan by Thomas Levenson (recommended; my review will appear in the February issue of The Observatory)) and, in this case, the answer was modified gravity and not dark matter. But not because no-one found Vulcan after n years, where n is some number determined by how long some pundit wants to wait before giving up. Modified gravity (i.e. GR) won out because it was a good theory, made testable predictions, was elegant, etc. And remember how long it took from theory to observation in the case of the neutrino, and there we even knew where to look and how many to expect.

piein skee said...

Quentin says "In any case, the resulting mathematical structure is what I meant by "model" in my precious comments."
'
So is the 'resulting mathematical structure' (henceforth 'RMS') the Field Equations when the subject is Einstein's GR? Is that in fact the resolved instance of of RMS when the subject really is GR? For you personally?

Or is the resolved identity of 'RMS' context sensitive? For example, does it matter what the physicist is saying the assumptions are?

p.s. I've no objections..just curious that's all..

Quentin Ruyant said...

@piein skee

To me:
- the field equations are the content of the theory.
- a model is their implementation in a particular geometry, i.e. solutions to these equations: for example, Friedman's model of the universe proposed in the 1920's, or contemporary models of the universe.

I put the distinction theory/model exactly where the distinction equations/solutions is.  This a one-to-many relation: equations can have different solutions, theories have several possible models, which correspond, intuitively, to several "possible worlds" (several ways the world could be if the equations are true). This is a way to understand why models, not theories, are directly confronted to experience: a theory can survive empirical failures because it has many models, one of which might still work.

By the way, this is exactly how logicians conceive of models in the tradition of Tarski's model theory. Tarski formalised theory-model relations (when do math structures satisfy logical statements?) in a set-theoretic+formal-language approach, which is widely used in works on the foundations of mathematics, and this was applied to scientific models in philosophy of science in the 1960's by Suppe and Suppes, who thought that Tarski's conception was analogous to how scientists conceive of physical models "satisfying" theories. These are very useful formal tools for conceptual analysis (just to give an example, "necessarily" can be interpreted as "true in all models" in this account, and we retrieve Krikpe's semantic for modal logic)...

So yes I think that models are context sensitive in a sense.
Contrarily to field equations, particular models rest on assumptions about what the universe is like as a matter of contingent fact: it's initial states, etc. (It seems that Sabine equate "model" with these assumptions alone but I'm still sceptical that they can be framed in a theory independent way: to me, the best way to expose our assumptions about the universe is to expose the resulting structure, as Friedman did. But let's not quibble on terminology).

I hope it clarifies.


@Phillip

It seems that you understand "ad-hoc" to mean: something that couldn't be verified in any way, but still posited to explain anomalies. Following your understanding, I agree that Neptune was not ad-hoc and that dark matter is not necessarily ad-hoc. Now I understand it in a looser way that would admit that an hypothesis is "temporarily ad-hoc" until it is independently confirmed. Let's talk about "unconfirmed hypothesis" if you prefer.

I completely agree that the amount of time an hypothesis remains unconfirmed is not decisive. Perhaps the amount of such unconfirmed hypothesis, and the resulting complexity of the theory, is more relevant to a theory being "implausified", as Sabine puts it. I leave to the specialists this kind of evaluation in the case of GR.

Now I'd say that, at least, having persistent unconfirmed hypothesis, despite our efforts, is an *indicator* that perhaps we are confronted to anomalies that will be explained away by a future theory of gravity. As we know that we need a new theory of gravity anyway, for purely theoretical reasons (problems with quantum mechanics) I find this conjecture quite plausible: it's not as if appealing to a new theory of gravity was "ad hoc" (;-)), and I think that it's how proponents of modified gravity such as Smolin view it too. But that's just a bet of course!

Phillip Helbig said...

"It seems that you understand "ad-hoc" to mean: something that couldn't be verified in any way, but still posited to explain anomalies."

I am using it the way it should be used, namely an ad-hoc explanation is one specially constructed to answer some question without any relation to anything else. Wikipedia says

"In science and philosophy, ad hoc means the addition of extraneous hypotheses to a theory to save it from being falsified. Ad hoc hypotheses compensate for anomalies not anticipated by the theory in its unmodified form.

Scientists are often skeptical of scientific theories that rely on frequent, unsupported adjustments to sustain them. Ad hoc hypotheses are often characteristic of pseudoscientific subjects such as homeopathy."

This is a good example. When scientists say that homeopathy can't work because (apart from the fact that it fails the tests normally required to prove the effectiveness of medicines) the solutions are so diluted that there is only an infinitesimal chance of there being even one molecule of the substance in the solution, homeopathic quacks claim that "water has memory" or "it's quantum" or whatever, in other words, the hypothesis has no supporting evidence other than the specific problem it was created to solve.

This does not mean that it cannot be verified. For example, an ad-hoc hypothesis is that planets move because angels push them. In principle, one could fly to the planet and look for the angel. Finding him would verify the hypothesis. In principle, one could falsify it by not finding him, but then another ad-hoc hypothesis of invisible angels would solve the problem. And so on.

So, Neptune (not Uranus as I too hastily wrote) wasn't ad hoc. Planets were known to exist, and so positing the existence of an additional planet was not ad-hoc. Yes, it was motivated by the observations, but that doesn't make it ad-hoc (even though essentially all ad-hoc hypotheses are motivated by observations---the converse isn't true). Not long before, Herschel had discovered Uranus to be a planet, so it was clear that there could be previously unknown planets. I see dark matter in a similar way. We see something which cannot be explained by visible matter, but invisible matter would explain it, so this is a good hypothesis. Being invisible is not ad-hoc; it has to be, otherwise we would already know about it, and there would be no puzzle.

Similar to the story of non-baryonic dark matter is that of baryonic dark matter. Not all baryons emit electromagnetic radiation, but there is indirect evidence of them. When calculating the amount required by the CMB and by big-bang nucleosynthesis, finding that the numbers agree boosts confidence in the hypothesis.

Dark matter is difficult to falsify for the same reason that one is not required to prove one's innocence in court. By the same token, it could in principle be proven, just by finding it, like a criminal investigation could find the proverbial or perhaps literal smoking gun.

Claiming that dark matter is ad hoc is tantamount to claiming that all components of the universe have to be detectable with technology in existence at the time the observations motivating dark matter were made. It also wasn't proposed to save the theory because, as Sabine wrote, GR says absolutely nothing about the sources of the gravitational field. If some theory had predicted both the law of gravity and various amounts of matter, then one found evidence for another type of matter, one could see that as ad hoc because it extends the original theory specifically to address a shortcoming. But GR doesn't say anything about baryons or any other sources of the gravitational field, so the fact that some were known when it was formulated and others were not is no big deal.


Quentin Ruyant said...

Philip

If you think ad-hoc hypothesis can later be verified, I'm using ad-hoc in the same way. My worry is just that there's no difference of nature between postulating a new unseen planet, or a new undetected form of matter, and postulating angels or water memory. The only difference is one of degree: in one case, we have a very well confirmed theory that fails to account for some specific phenomena, and we postulate something that allow us to keep the theory working. In the other case, the whole theory rests on such postulations. But the nature and role of these postulates (saving the theory in front of contradictory observations) is exactly the same.

That's why I referred to Lakatos in a previous comment. According to Lakatos, there's only a difference of degree between science and pseudoscience. There are successful research programmes that keep on producing new confirmed facts, and there are degenerating programmes that only produce more and more unconfirmed facts to "save" the theory. The crux, according to Lakatos, is whether ad-hoc hypothesis are able to produce new successful predictions. Neptune was ad-hoc when formulated because the only goal of this hypothesis was to keep Newtonian theory working, despite prima facie refutation. But it produced new predictions: that if we look at this specific place in the sky we'd find a planet, and that's what happened! That's how the theory produced a new fact. To the contrary, water memory has never been confirmed independently and homeopathy never produced any new facts, it only grows with more unconfirmed hypothesis. But the ad-hoc nature of hypotheses is not different in both cases.

You can say that Neptune was not proposed to save the theory because Newtonian physics doesn't specify how many planets there must be. Fine. But I don't really see the difference with homeopathy. Newtonian physics *would* fail if not with Neptune. GR *would* fail if not with dark matter. Homeopathy *would* fail if not with water memory.
The problem with homeopathy is that it only has ad-hoc hypothesis supporting it, and no independent evidence, but that doesn't mean that GR is completely free from ad-hoc hypothesis.


You say "Claiming that dark matter is ad hoc is tantamount to claiming that all components of the universe have to be detectable with technology in existence at the time the observations motivating dark matter were made.".

No, it's tantamount to claiming that GR would fail without dark matter, and that dark matter has no independent confirmation apart from the fact that GR would fail without it.

Compare : "claiming that water memory is ad hoc is tantamount to claiming that all features of water have to be detectable with technology in existence at the time the hypothesis was made". No it's not.

Koenraad Van Spaendonck said...

@Sabine

Suppose dark matter does exist.

What could be the scientific reasons why it is undected or undetectable today ?
Is it as simple as saying that whatever we shoot at it, doesn't bounce back ? (Doesn't interact, such as no WIMPS at LUX etc)
Could it mean that the very conceptual principles of detection as we know them, are not the right way to go ?


And what means of detection do scientists propose to detect it in the future ?

I suppose this begins with hypotheses on what dark matter might consists of, otherwise it would be hard to devise a means of detection for it.

Current efforts : link :
http://m.phys.org/news/2016-07-scientists-invisible-dark.html

Extra question:
In this article it is said :
"Dark matter is everywhere. Hundreds of millions of dark matter particles pass through Earth every second, Gaitskell said. But the problem is they are "just crazy weak" and they zip through Earth as if it barely exists, he said."

Why is it assumed that giant amounts of dark matter particles are passing through the earth ?

Best, Koenraad

Quentin Ruyant said...

We're at the beginning of 20th century. Scientists recently confirmed general relativity by observing how the light from distant stars is deflected by the sun during a solar eclipse. Imagine a cranky physicist comes and says: "wait, no, Newtonian physics is still valid but there must be some transparent medium around the sun that deflects light. It makes perfect sense: we already know that light can be deflected when traveling through medium with different refraction indices. It's not as if I was suggesting sundering crazy. No need to accept general relativity."

Would you agree this transparent medium hypothesis is ad-hoc?

Now imagine the very same situation, where we observe that light is deflected by the sun, except that Einstein is never born and nobody knows about general relativity. Scientists gather and come up with the following statement: "we have new evidence of a transparent medium around the sun that deflects light. We suspect that its mass is quite low, although it could explain mercury's problematic trajectory as well. We will search for new evidence that this medium is present in distant galaxies as well by trying to observe similar lensing effects."
And sure they will find it, and they will propose different models. But they'll never find independent evidence for the existence of this medium, apart from lensing effects (and mercury's trajectory). Now physicists are still bothered by the incompatibility between Newtonian physics and Maxwell electromagnetism, but apart from a few dissidents, they don't think that both problems are related, and they're still working on it anyway.
They dismiss proposals to modify Newtonian gravitation: the transparent medium hypothesis works well.

Would you agree that the hypotheses of a transparent medium in both scenario are the same? And that the fact that Einstein was born is a mere contingency? Then if the hypothesis was ad-hoc in the first scenario, why shouldn't it be in the second? And what differentiates the second scenario from what's happening right now with dark matter?

Tam Hunt said...

Hi Sabine, thanks for that detailed follow up. It sounds like we are generally on the same page after all! I would very much like to see a future blog on “implausification” vs. falsification, particularly since falsification almost always gets rhetorical support by scientists while in practice it seems, as you suggest, to generally be ignored. Perhaps physicists should discuss more explicitly what it would take to falsify, or at least implausify, theories as a general matter.

I’m also curious why you and other physicists think that the observational anomalies that led to DM, DE and inflation are not that serious. From my perspective it seems that the standard model and GR, which forms its core, have been distorted beyond recognition in order to accommodate these major observational anomalies. It can certainly be a fun game (and a good source of funding, of course) to come up with plausible fixes for these massive and ongoing observational anomalies, while preserving GR, but at what point should we conclude that such fixes (like DM, DR, inflation, etc.) aren’t any longer plausible? As I suggested before, when 99% DM is required to explain the observed rotational velocity of particular galaxies, and when the standard model suggests that 96% of the universe (including DM and DR) is not baryonic matter, that we are firmly in the range of implausibility.

I’ve also noticed a trend toward circularity and tautology in many areas of research that are cited as support for things like DM (this is not a problem only in physics; it seems to be a problem in almost all research fields). When we trace back the various lines of evidence we often find that these supposedly independent lines of evidence actually rest themselves on the assumed truth of the argument to be proved and thus cannot be evidence for that argument (this is the definition of circularity and tautology).

Last, since I haven’t reviewed your work in detail and don’t know if you’ve already suggested this elsewhere, what form would an alternative approach to GR/gravity take? You suggest in this blog that you’re sympathetic to modified gravity approaches that are compatible with GR, but how can they be compatible with GR if they’re modified gravity? Isn’t modifying gravity by definition not compatible with GR?

Phillip Helbig said...

"My worry is just that there's no difference of nature between postulating a new unseen planet, or a new undetected form of matter, and postulating angels or water memory. The only difference is one of degree: in one case, we have a very well confirmed theory that fails to account for some specific phenomena, and we postulate something that allow us to keep the theory working."

The basic difference of opinion here is that I don't think that the theory fails just because the sources are not known. There are many reasons why the comparison with homeopathy is not valid. One is that when homeopathy was devised people didn't know about molecules and assumed that water was infinitely divisible, or at least much more than could be done in a laboratory.

"Why is it assumed that giant amounts of dark matter particles are passing through the earth ?"

The conclusion follows from certain assumptions: dark matter is in free particles with a certain mass and so on. Pretty much the same is true of neutrinos: millions passing through you, and you don't even notice.


GR says how matter affects space-time and vice versa. How does invisible matter cause anyone to think that GR is somehow failing? I think my comparison with Linnaeus is good: does the fact that his binomial classification not predict gorillas prove that it is intrinsically flawed? Also, water memory makes no predictions other than the ones it was designed for. Dark matter does make predictions. The problem is that there are different hypotheses, so we have to rule them out one by one. This is happening. Neutrinos as hot dark matter don't work because it doesn't jibe with structure formation. Solar-mass black holes don't work because they aren't seen in microlensing surveys, and so on.

Sabine Hossenfelder said...

Tam,

Indeed, there's unfortunately a lot of tautology and circularity in the literature. Though the topics that would come to my mind are primarily inflation, quantum foundations and, to some extent, particle physics - not cosmology. What did you have in mind?

I am sorry for the badly phrased explanation of how modified gravity is compatible with GR. I already explained this in a comment above, I meant it respects general coordinate invariance (which the Newtonian limit doesn't.). This means you at least have a chance to reproduce the achievements of GR in some limits.

As to observational anomaly... I'm not sure what observational anomaly you think leads to inflation. As to the observations that lead to the introduction of dm and de. Well, the thing is that they're not serious enough. You see, you can fit all the observations with two parameters added to the sources. And that fits a *lot* of data. For what a theoretical explanation is concerned, this is an excellent one. And it's been hard to improve on.

Best,

B.

Shantanu said...

Just fyi, people interested in this (from scientific and sociological point of view) can also read this very nice thread on cosmologist Peter Coles' blog (where yours truly as well as Phillip Helbig) commented extensively.
https://telescoper.wordpress.com/2013/07/27/no-more-ripples/

Yves said...

Hi Sabine,

"It’s difficult to see how particle dark matter could cause this. (Though I would like to see how this plot looks for a ΛCDM simulation)"
Maybe this one: https://arxiv.org/abs/1610.06183v1

Sabine Hossenfelder said...

Yves,

Yes, thanks, I'd seen the paper (and already read it). If I find the time, I'll briefly summarize this later.

Yves said...

S. McGaugh has already reacted to this paper: https://tritonstation.wordpress.com/2016/10/21/la-fin-de-quoi/