As promised, the folks at NEIU have posted the video of my discussion with Scott Dodelson last week, so here you go:
I am in the midst of writing a related post on cosmic tensions, so hopefully I can post that soon as well.
25 thoughts on “NEIU debate: dark matter or modified gravity”
Comments are closed.
I thought it was ironic that Scott started out by emphasizing how important it is to be humble, since cosmology is millennia in the making, but followed that immediately with his reasoning that everyone who had previously tried to solve it “had gotten it horribly wrong”.
Indeed. No worries that future folks will say the same of us!
Professor Dodelson assumes that, after quantum averaging, Einstein’s field equations are 100% correct. Professor McGaugh raises the question, “Is it possible to falsify the existence of invisible mass?” Is MOND inertia more likely than dark matter particles? Are MOND’s empirical successes a confirmation of string theory (with some new hypotheses)? Is Professor Milgrom the Kepler of contemporary cosmology? Is MOND inertia the key to understanding string theory? Is MOND inertia a stringy reaction to dark energy? Is dark energy a stringy reaction to MOND inertia?
Let us assume that gravitational energy is conserved and all gravitons have spin 2. Consider the following: “MOND as manifestation of modified inertia” by Mordehai Milgrom, 2023, https://arxiv.org/abs/2310.14334 . How plausible are the following hypotheses? Nature consists of 1 copy of Einsteinian spacetime and 3 copies of virtual spacetime combined with string vibrations. Nature’s strings continually vibrate and continually undergo virtual curling-up and virtual uncurling of the 3 copies of virtual spacetime. The virtual curling-up creates MOND inertia & mass-energy inertia and thus explains MOND’s (approximately) successful predictions. The virtual uncurling creates dark energy & quantum mass-energy with virtual particles, starts with stringy unified bosons/fermions, and has three fundamental phases that explain the 3 generations of fermions.
If MOND inertia is physically real, then Einstein’s field equations might be modified as follows: Replace G(μ,ν) + Λg(μ,ν) = κT(μ,ν)
by G(μ,ν) + Λg(μ,ν) = κT(μ,ν) + M(T(μ,ν)) , where the function M(T(μ,ν)) represents the MOND-data-function as a function of the energy-momentum tensor, and
G(μ,ν) = R(μ,ν) – 1/2 g(μ,ν) R .
In the MOND regime, put M(T(μ,ν)) = – ((3.9±.5) * 10^– 5) * g(μ,ν) R . Claim: in the MOND regime, this is approximately correct in terms of empirical data. Are the preceding ideas wrong?
Stacy, how frustrating was that “debate”. I assume it is routine for you, but if this is standard way the dark matter community approaches this issue, then it seems to me to be an indictment of their intellectual and scientific approach. To summarize Scott’s argument: The MOND results for galaxies are “not impressive”, and the “blind” LCDM fits to cosmological predictions are so marvelous (57 Sigma!!), that there really is no need to have a serious discussion. In what universe are the galaxy MOND predictions “not impressive” and, at the least, something that LCDM must address, as opposed to Scott’s statement that LCDM makes no predictions about galaxies, so who cares!
I do have a some specific questions that I hope you can maybe address in a future post. Can you provide some, at least, qualitative information about how “impressive” are the “57 Sigma” blind fits to CMB? First, how many adjustable parameters are there: clearly dark matter, and, I assume, dark energy. You mentioned the lithium problem – is that, in effect, an adjustable parameter? And deuterium, could it, in truth, also be considered adjustable. For example, do changes in deuterium (or lithium) abundance, within the experimental uncertainty, significantly change the CMB fit? I assume there are other adjustable parameters in the model – could you elaborate. Secondly, how “blind”, really, are the fits. Scott showed a slide that he implied showed “hundreds” of blind LCDM predictions. How independent are these predictions? Are they novel, or just another variation of the CMB fit? Thirdly, could you comment on the “57 Sigma”? Finally, as a major support of the LCDM, Scott showed the recent CMB fits extended to small angular scales, which I assume are “blind”, in the sense that it did not require any further parameter adjustment. Is this extended fit, in your opinion, a good test of LCDM?
Lastly, I gather from several of your posts that you are not overly impressed by the Skordis and Zlosnik fit to the CMB using an extension of MOND. Could you elaborate on this? Is it because Skordis and Zlosnik are, in effect, simply adding a contribution that is equivalent to dark matter and is not, in your view, a natural extension of MOND?
That’s a lot.
I do think this attitude is typical: the CMB is fit so well by LCDM, it must be true. I am not myself so impressed with this, having explored such fits & predictions. Some free parameters provide more freedom than others; without dark matter only a very narrow range of power spectra are possible but with it pretty much any pattern of oscillations can be fit. Moreover, the oscillations physically connect all the points that measure them, so it overstates the case to say (as many people do) that there are thousands of independent points. Measurements at different observed angular scales yes, physically independent no. Once you’ve fit the first three peaks there is not much wiggle room left. A little, but there isn’t lots of new information there short of something that breaks the whole picture a lot more than MOND does.
Anastasiia tried to ask about the deuterium/lithium/second peak issue, and I’ll write more about that in an upcoming post. Basically, LCDM missed its prediction for the second peak and broke BBN. Everything after that is a fitting exercise, and LCDM has enough free parameters to fit any physically plausible power spectrum. Most cosmologists never appreciated this, and have apparently gaslit themselves into believing this was all predicted when the fit to each iteration of the data was a pretty bad predictor of the next iteration.
I think the Skordis & Zlosnik work is very important for showing that LCDM is not the only way to fit the CMB data. The argument that LCDM fits so well is undermined if something else fits equally well. Apparently this isn’t widely known, because if can’t be possible, can it? so why would they pay attention.
So AeST is an incredibly important advance in this regard. Whether it is the final theory is another matter. Maybe it is, maybe it is just one step in that direction, maybe we’ve only just cracked open a Pandora’s box of possibilities.
Hi Stacy. Thank you for announcing the NEIU Debate on “Dark matter or modified gravity” – I was lucky enough to sit in on this.
My main takeaway was Scott Dodelson’s statement that cosmology with the LCDM model essentially says nothing about structures smaller than 5 Mpc (in today’s Universe). So cosmology, for him, makes no attempt at explaining galaxies or galaxy clusters. On the other hand, LCDM is good at explaining BBN (Big Bang Nucleosynthesis), CMB (Cosmic Microwave Background), and the formation of large scale structures. I note, in passing, that cold dark matter (CDM) is a critical component of LCDM cosmology – if there is no dark matter, then LCDM is dead.
MOND provides empirical relations that enable us to predict (not just explain) the behaviour of galaxies and galaxy clusters, and importantly it does this without any dark matter. However, at the moment MOND does not claim to be a theory of cosmology. So MOND is good at the stuff smaller than 5 Mpc (the galaxies and galaxy clusters) but stays away from the bigger stuff.
I’m afraid that the success of LCDM means the cosmologists will continue ignoring MOND and the physicists will continue searching for dark matter particles.
My current problem with cosmology and dark matter lies with the BAO (Baryonic Acoustic Oscillations). The first peak in the CMB corresponds to the distance the baryon-photon fluid travelled before recombination (nothing to do with dark matter). This also corresponds to the observed peak in galaxy separations from galaxy surveys. If every galaxy forms inside a dark matter halo, then what caused dark matter to clump at the same distance scale as the baryonic matter? – they should be completely unrelated, especially as dark matter outweighs baryonic matter 5-to-1.
Well, that’s why I say the evidence is incommensurate – https://arxiv.org/abs/1404.7525
It isn’t quite as simple as bigger or smaller than 5 Mpc. MOND is better at emptying out the voids on larger scales than that. It also predicts earlier structure formation and the appearance of very large scale structures, as observed. One has to ignore these observations to sustain the insistence the LCDM is better for all things large scale.
I’m less concerned about that than this weird obsession with simulations, as if that were the only way to address smaller scales. By this line of reasoning, it should be impossible to predict anything that happens in the solar system, as Kepler’s laws can’t be resolved by cosmological simulations. That sounds less like a scientific argument to me and more like a lame excuse. Why does MOND get any prediction right?
If the LCDM model fails at the galaxy level—as clearly demonstrated by the Webb Space Telescope’s findings of massive galaxies in the early universe—it should be obvious that LCDM also fails at higher levels of hierarchical structure since these larger structures are formed from galaxies.
Contrary to what many theoreticians may lead you to believe, phenomenological models like MOND are not the exception but the rule in many-body problems. This is true whether one is considering models that describe atomic nuclei with high atomic numbers or models that describe protein molecules.
The so-called fundamental theories are, de facto, applicable only to very simple systems. As soon as a system involves a large number of discrete components—and sometimes even when there are only a few dozen—these “fundamental” theories become inadequate, and some form of phenomenological model is required. Obviously General Relativity is not an exception.
From a formal mathematical perspective, it has already been shown that irreducibility is the norm rather than the exception in complex systems. In other words, the number of configurations that cannot be reduced to first principles far exceeds those that can.
References.
Is independence an exception?
C. Calude, H. Jürgensen, M. Zimand. Applied Mathematics and Computation Volume 66, Issue 1, November 1994, Pages 63-76.
I don’t understand the comfort by which it was agreed that relativistic effects are of no consequence in resolving cosmic tensions. Isn’t cosmology necessarily relativistic?
Why be more comfortable with a universe dominated by imaginary energy and imaginary mass than one where not fully understood relativistic effects dominate cosmology?
I don’t think anyone is saying that.
Hi Stacy, thanks for sharing this. I was wondering about the purported proof of dark matter halos provided by galaxy-galaxy lensing, according to Prof. Dodelson (51:10). Has MOND anything to say about that? I guess “vanilla” MOND, being non-relativistic, can’t really address a phenomenon such as gravitational lensing? But are there extensions that do, and if so, would those be able to account for the phenomena observed? Or is this a genuine problem for MOND?
Lensing is in good agreement with MOND. It wasn’t always, but it is now in better agreement with MOND than with dark matter. There is some disagreement about interpretation; see my comments on the same point arising from my discussion with Simon White last year: https://tritonstation.com/2024/01/08/discussion-of-dark-matter-and-modified-gravity/
My understanding of the situation is, MOND is successful for galaxies and galaxy clusters but does not even try to say anything about the CMB. On the other hand LCDM is successful in fitting the CMB but only give statistical predictions about galaxies, and no predictions for anything smaller. At the scale of galaxies and galaxy clusters where both say something, they’re (kind of) compatible with the observation and thus with each other. (I’m sure people will come up with yet another tweaked CDM to accommodate the too luminous too early galaxies.)
Then I don’t even see the conflict here. Having two models for the same object, that both work to some extent, is a good thing. I think the issue in “dark matter or modified gravity” is not necessarily with either of the theories, but with the “or”.
By the way, how many sigmas did people have for confirming Newtonian mechanics before the fateful year 1905?
It is an overstatement to say that LCDM doesn’t make predictions on galaxy scales. Those predictions were made, failed, insisted upon, failed some more, were backed away from, and now some people claim not be able to predict anything and shouldn’t be expected to do so while others insist that they can explain everything perfectly well, if only after we’ve told them what the need to explain, and even then they pick and choose which parts they address. There are, for example, many claims to explain the radial acceleration relation, but exactly zero of these provide a satisfactory understanding of why the baryons (and the baryons alone) appear to be the source of the gravity in galaxies. Most don’t even attempt to address this fundamental aspect of the problem, because to do so forces one to realize that invoking dark matter halos makes no sense.
In one point I heard Scott Dodelson saying something like “It must be satisfying to be able to observe a galaxy and make predictions, but I dont think its so important for science” (or something like that)
I find that statement, well, not his best point, predictive power and falcifiability are some of the cornerstones of scientific method (at least if you are not into religions like superdeterminism etc)
Its not even related to MOND, being able predict galaxy rotation curves from “still photos”, means that rotation curves are related ONLY to baryons, which means that DM is either tightly related to baryonic structure or it irrelevant to the rotation.
You have a better developed physical intuition than many practicing physicists.
Apparently one thing I failed to emphasize strongly enough is that dark matter really cannot explain the kinematic observations. It does so only if we hold it to the lowest possible standard, in which case sure, whatever discrepancy is observed must be that invisible stuff. But there are outright contradictions in the dark matter interpretation that its advocates don’t appear to be aware of and certainly don’t engage with.
Great discussion, which to me you won without trying too hard. You say ‘how do we get out of the rut?’ Part of the rut is that in many places either theory might (at a stretch) cover the data. Do you think the places where differences between the theories really stick out could be game changing, letting us pin things down? (Wide binaries looked like they might, but the data analysis isn’t good enough yet.) It seems that some required properties of DM might do it – because of cluster collision data, the general view is that DM is collisionless. If so, would that rule out missing baryons?
I’ve been reading the Oehm/Kroupa paper (33.40 in the video), which seems to have taken this collisionless aspect further, and falsified the kind of DM that has the expected properties of matter. Again, we don’t observe the result of any friction. But if it’s collisionless there as well (as is consistent with the small-scale medium in PSG, which although not matter, is what is taken to be DM), do you think a hybrid theory of that kind, with unexpected properties for ‘DM’ could be possible?
Cluster collision velocities require that the dominant mass be collisionless. That excludes gas, but not all baryons. Just need them to be in compact form.
As for hybrid models, it is indeed difficult to construct one that satisfies all aspects of the problem. I don’t want to say it isn’t possible, but I don’t see how this approach can replicate things like the external field effect.
Thanks, so you can have missing baryons in cluster collisions. But in the MW halo, if there’s DM there it’s non-baryonic, and according to that paper, non-matter (that would fit with non-detectable directly).
I’d say the EFE comes more naturally to PSG than to some theories. For the field to be weak enough to make the transition to the MOND regime, whatever the cause, all combined fields together have to reach g = a0. You can try to do that kind of field combining with curvature, and it’s very possible with a fixed GR field, but harder to shift around. in PSG the field is an ‘object’ in its own right, made of small-scale waves, as well as its role as a gravity field. So it can go through changes, get distorted out of shape (when the waves reach a certain point), blend with other fields easily, and is generally much more MOND-friendly. If the cause of MOND is that the waves change to a new state involving self-interaction, then it would make sense if all combined fields have to reach that point, rather than just one particular field.
The hybrid idea is that this is at work at the galaxy scale, ie. MOND. At larger scales an excess builds up, DM. The same medium does both. At the moment PSG satisfies all aspects of the problem in a conceptual way, but only some in a mathematical way. Of course, what people want is all aspects dealt with in a mathematical way, but perhaps that’s trying to run before we can walk.
Conventionally, we need both baryonic and non-baryonic dark matter in the halos of galaxies like the Milky Way. The dominant mass is the non-baryonic cold dark matter, but we know what the cosmic baryon fraction should be, and galaxies come up short of that check sum. So the undetected baryons needed to make it up to the cosmic baryon fraction are lurking in the halo in some unseen form, or have been ejected entirely out into the intergalactic medium.
Stacy, I noticed that in the debate, you mentioned Pavel Kroupa’s argument about Chandrasekhar dynamical friction ruling out dark matter with 5 sigma confidence (and I noticed that you stressed it’s Kroupa’s claim, not yours). I haven’t seen you mention it before, and I’ve seen that Kroupa leans very heavily on it as a falsification of dark matter. So I was curious to hear your thoughts about it. Do you find it to be a convincing argument or not so much?
Thanks for your exceptionally great blog. You’re a model scientist and communicator.
I wanted to make the case, impossible as it is in 20 minutes, that there is good reason to doubt dark matter can work irrespective of MOND. Kroupa has insisted that the absence of the expected amount of dynamical friction is an outright falsification, so it seemed worthwhile saying that I wasn’t the only person who saw problems with dark matter. One doesn’t consider something like MOND without first coming to doubt dark matter, but how do we get there?
I do find it to be a convincing argument. If you attempt to reproduce all aspects of the orbit, there simply is no solution. However, one does have to make some inevitable assumptions: that the dark matter halo is NFW as it should be, and that the Clouds are not on first infall. The argument holds but becomes weaker if you relax those well-motivated assumptions. Advocates of dark matter have always taken that path when pressed – there was a widespread stir when the large proper motion of the LMC was first discovered, as that in itself was in effect impossible for a bound orbit. So, despite the fact that the Clouds appear, in the form of the Magellanic Stream, to be orbiting the Milky Way, the story quickly changed to be that the LMC was not orbiting at all, but on first infall. This in itself is unlikely: we live at a special moment when the LMC happens to be near the pericenter of its hyperbolic flyby, but it isn’t outrageously unlikely. One still has to explain the Stream, which was initially thought to be made by tidal stripping, which requires at least one previous pass through the pericenter of a bound orbit. Or not – it could be that the gas in the trailing stream has been stripped by ram pressure during passage through the hot gas of the MW corona. That can’t explain the leading arm, but it could be that the apparent gas there is not related, it just happens to be there (looking at the data, that’s not as silly as it sounds). It would be good to see stars in the stream, and while there are some in the trailing stream, none have yet been found in the leading stream. THAT came as a surprise to me, and made me give more credence to the ram pressure scenario and less to tidal stripping.
All that said, it is still an unusual situation. The question is whether it is still 5 sigma after relaxing those assumptions. If not, then people forget it was ever a problem. (There are lots of problems that just get forgotten without really getting solved.)
Thanks. btw, after ‘the same medium does both’, I meant to say: this would seem a lot less likely without the mathematical evidence for the medium’s existence.