I am mostly finished with a post following on from what would be required to make a convincing dark matter detection that explains why it will never happen. I’ve been nearly done for several weeks, but haven’t had time to complete it. Time carries on, and during that time a story appeared in Quanta magazine about observations claiming to be the final nail in the coffin for MOND, all the usual talking points we’ve covered here before, blah blah blah. But since it was in the news, I felt obliged to hop on social media (ug) and push back against the tidal wave of confirmation bias. So I’m just gonna reproduce that thread here, maybe adding further commentary if I can find some copious spare time.
First, some context: we’re talking about ultradiffuse galaxies in the vicinity of NGC 1052. These objects are abnormally large for their luminosity, so very diffuse. (Very low surface brightness in my 1996 terminology, ultradiffuse in that of van Dokkum et al. 2015.) I’m not a fan of describing them as “ghostly,” but low surface brightness galaxies have been described as ghostly before. Maybe “ghostly” is not entirely misplaced here, as one possibility is that they have been tidally whacked out of equilibrium and are slowly expanding towards an ultimate dissolution, fading like ghosts into the intergalactic graveyard.
But that’s getting ahead of the story. Let’s cut to the thread:
Story out in Quanta about the diffuse galaxies DF2, DF4, & DF9. It conveys the message that these objects are the nail in the coffin for MOND. Is it so?
www.quantamagazine.org/triplet-of-g…
Triplet of ‘Ghostly’ Galaxies Hints at What Dark Matter Is — and Isn’t | Quanta Magazine
www.quantamagazine.org/triplet-of-ghostly-galaxies-hints-at-what-dark-matter-is-and-isnt-20261014/
52111
Oct 14 at 2026 at 1:31 PM
When DF2 was first reported, it was portrayed as being fatal for MOND. And so it appeared at first glance, with an observed velocity dispersion of 3 km/s where MOND predicted 20 km/s for an isolate object. Pretty bad! But a casual glance showed that it wasn’t an isolated object, so the EFE matters.
A more careful analysis yields a MOND prediction of 13.4 +4.8/-3.7 km/s. Contemporary reanalyses of the data estimated an observed velocity dispersion close to 12 km/s. That sounds more like corroboration than refutation: arxiv.org/abs/1804.04167
MOND and the dynamics of NGC1052-DF2
Lots of observational resources have been applied since then; I won’t attempt to review it all here. The new paper offers two independent measurements, which are 6.3+3.1/-3.6 and 9.2+3.8/-4.5 km/s. These are both consistent with 10 km/s, which is within the 1 sigma uncertainty of the prediction.
So MOND is falsified at one sigma.
That’s assuming these systems are in equilibrium, which is not obvious. We have to invoke some pretty dramatic non-equilibrium events to strip them of dark matter (in the conventional picture, galaxies cannot form without dark matter so we have to get rid of it. Difficult but not impossible.)
To be fair, we should worry about non-equilibrium situations in MOND as well. Fortunately, simulations have been done that illustrate what happens to a dwarf galaxy that gets tidally disrupted by a close passage with a giant galaxy. arxiv.org/abs/astro-ph…
Dwarf satellite galaxies in the modified dynamics
arxiv.org/abs/astro-ph/0005194
The velocity dispersion departs from the equilibrium prediction and the scale size of the dwarf slowly expands. This actually provides a natural explanation for the large half-light radii of these DF dwarfs. tritonstation.com/2025/09/15/n…
Non-equilibrium dynamics in galaxies that appear to lack dark matter: ultradiffuse galaxies
I’ve no idea if this is what happened in these systems as I have not+ been around several Gyr to watch it happen. But there are certainly giant galaxies in the neighborhood, so it’s no less plausible than the bullet collision hypothesis in the DM context.
DF2 has consistently been portrayed as fatal for MOND despite consistently being shown not to be a problem. I’ve given up writing formal papers showing this, as nobody cares. They just hear what they want to hear, which is that MOND is wrong. Indeed, after the 2018& incident I swore off doing so.
It’s sad, but there is no point in correcting the scientific record while the relevant scientific community suffers a catastrophic failure of objectivity.
Strong words? Are you a scientist working in the field? Have you heard of DF2? If so, did you know its interpretation was in dispute? Or only that it falsifies MOND?
If you have heard of DF2, have you also heard of Crater 2? It has a tiny velocity dispersion (2.7 km/s) that makes no sense in LCDM but was correctly predicted a priori by MOND. arxiv.org/abs/1610.06189
MOND Prediction for the Velocity Dispersion of the `Feeble Giant’ Crater II
Crater 2 and DF2 are similar in a number of respects. Why should we obsess about one while ignoring the other? That’d be an example of the failure of objectivity.
Or how about And XXVIII? You know that one? That was the golden test we proposed in arxiv.org/abs/1301.0822
Andromeda Dwarfs in Light of MOND
That came out in favor of MOND – as did all of the rest* of the dwarf satellites of Andromeda. arxiv.org/abs/1308.5894
Andromeda Dwarfs in Light of MOND. II. Testing Prior Predictions
Does MOND have problems? You bet! But it has lots and lots of predictive successes that people seem to be oblivious to. Many of these are predictions that cannot even be made in the context of dark matter. DF2 is a good example: we didn’t anticipate it was DM-free, we inferred that from the data.
If instead DF2 had a large velocity dispersion (like DF44), then we would happily report that it had a surplus of dark matter. There’s no prediction to test; it’s just whatever the observations dictate.
MOND makes actual predictions that mostly come true. This being astronomy, there are always some dodgy cases. So DM appears to be better in those cases by virtue of not making a testable prediction. Does that make it better science?
There still remains no satisfactory explanation in terms of dark matter for MOND’s predictive success. If you can come up with one, then perhaps dark matter can be viable. Without that, just complaining about what MOND gets wrong without engaging with what it gets right is intellectually dishonest.
+Apparently I omitted the “not” in the Bluesky thread. I may be old, but I’m not that old.
&Before that time, I was extremely conscientious about examining claims to falsify MOND. I’ve done this for literally hundreds of cases. The recurring theme had been that these claims are greatly exaggerated. Not all, of course, but more often that not, it was simply a matter of doing the analysis properly and including the errors. The current example is a good one: the claim is that these ultradiffuse galaxies have no dark matter (or MONDian mass discrepancy). The data are indeed consistent with that. But there are uncertainties in all data, and here they encompass MOND’s predictions as well.
At any rate, I no longer see correcting the mistakes of others as a constructive use of my time.
*Except And V. That one has a somewhat too-high velocity dispersion for MOND. If I were to follow the template offered by many examples in the literature, I should loudly advertise how this one case falsifies MOND while downplaying how MOND correctly predicted the velocity dispersions of the other three dozen dwarf satellites of Andromeda.
Here are the predicted and observed velocity dispersions for the dwarf satellites of Andromeda from McGaugh & Milgrom 2013a:
Note that And IV is blank here because it turned out not to be a satellite of Andromeda as initially thought, but rather a background object. This was anticipated by MOND and later confirmed, so was in effect another successful prediction. (I do not remember the story behind the other blank cases offhand.)
This plot is from the follow-up paper, McGaugh & Milgrom 2013b. Note that these were almost all a priori predictions. (Some of the above were too, but observers were doing the follow-up work on newly discovered satellites so fast that these numbers appeared in the literature before we could finish writing the prediction paper. Not that it matters, as the predictions of MOND are not flexible beyond some range in the stellar mass-to-light ratio, so it predicts the same thing whether the answer is known or not.) And XXVIII was a golden case because it was far enough from Andromeda to be in the isolated rather than EFE regime, greatly reducing the uncertainty in the prediction (since it only depends on observations of the dwarf and not of the EFE from M31).
Note the last line of the caption: we know there are cases that are out of equilibrium. There are certainly other systems in the universe that are out of equilibrium, like the tidal dwarfs for which we also found an apparent lack of dark matter. We haven’t been shouting that these falsify MOND (as we had hoped to be able to do! that’s why we did the experiment!) because we recognized that there was no way the equilibrium predictions could apply. Perhaps that is also the case for the ultradiffuse dwarfs discussed above.
There have been a few more dwarfs discovered since these plots were made. So far, these have been consistent with MOND. I include the plots from the cited papers because that’s what is handy and few people seems to be aware of this work. If a prediction is published in the literature but no one ever reads it, was a prediction even made?
I was surprised by the mild and ambiguous tone of your response, as quoted in the Quanta article. Why are not you as explicit there about how biased and unscientific their interpretation is?
First off, I was talking to the reporter in ignorance of what the final story would be. I had no reason to expect a hit piece, nor could I expect to be quoted fairly if that were what it was.
Second, there is no correct level of flame. A response that seems appropriate to you here and now would seem way over the top to many scientists, no matter how well justified.
This discussion made me aware, as I was not previously, of the dramatic asymmetry in MOND vs DM theories. Astrophysicists seem to have no qualms about accepting some bizarre scenario that strips out all the DM – and then see this as evidence for DM! One might imagine that finding a galaxy without DM might give them some doubts. Of course, in MOND, this flexibility is absent. This dramatically illustrates a fundamental difference in the two theories. A difference that I do not think was emphasized enough, for example, in the recent post by Jim Schombert on the philosophy of science issues raised by these two theories . I would have hoped that Quanta could, at least, have pointed it out.