I’ve been trying to think more seriously about the entity that set off the LZ experiment. I was dismissive of the possibility of it being dark matter, perhaps unfairly so. On further reflection, I feel only fairly dismissive.
First, the formal significance that is claimed is 2.6σ. That seems to be a fair and fairly conservative statement. So it seems something did happen, the question is what. The odds of a 2.6σ event are about 1 in 200, so a fluke seems unlikely.
I can think of three generic possibilities:
- It’s a fluke.
- It’s dark matter (huzzah!)
- It’s some strange new particle-entity that is not dark matter.
I remain of the opinion that the most likely possibility is that is is a fluke. I know I just said that seems unlikely, but “seems” is doing a lot of work here. I think it is fair to say that known explanations are excluded at this confidence, but that leaves a lot of unknowns. Most unknowns turn out to be boring.
We’ve been down this road many times. The LHC has been running for many years now. It detected the Higgs particle, which was the last missing piece of the Standard Model. That’s a great success. It was hoped that the LHC would also reveal new physics beyond the Standard Model. That it has not done. But there have been many hints during its operation. These appear as glitches (deviations from expectation); there have been multiple glitches that appeared to be significant at the 2.something σ level. Each time such a glitch was reported, it resulted in a flurry of theoretical activity. Within days, the arxiv preprint server would be inundated with hundreds (literally hundreds) of preprints staking out a claim as to what the latest glitch might be. Each time more data were acquired, the glitch went away. So this seems like another such occasion, making (1), a fluke, seem like the least improbable outcome.
This is why the [arbitrary] threshold for discovery is set at 5σ (1 in 3.5 million). Flukes at the 2.something σ level happen too easily, too often. This phenomenon is not restricted to particle physics. The same thing happens in astronomy all the time. All the gorram time. It’s almost as if nature places more probability in the tail of the distribution for “weird” events than the Gaussian statistics we use say there should be.
Given this experience, finding a new particle outside the bounds of the Standard Model of Particle Physics seems a lot less likely than 1 in 200. Seems is again doing a lot of work here. How likely 1 in 200 hundred seems depends on your priors. If you’ve been working on this for a long time in the hopes of a positive result, these odds seem pretty good:
“I have been waiting for a positive result for, oh, my god, 40 years. So you can bet I’m very, very excited by the LZ results.”
If, like me, you’ve already convinced yourself that dark matter cannot explain the kinematic data for galaxies, then it is just another opportunity for the eyes to roll up inside the skull. Obviously we can’t both be right, which is why we do these experiments.
So what about the other possibilities, (2) and (3)? If the event is real, it is tempting to believe that it is dark matter. That’s what the experiment was designed to detect, after all, so that’s a natural first assumption. But just because you’re looking for something doesn’t guarantee that when you find something you find what you’ve been looking for.
As I’ve reviewed before, we’re way past the regime of what we expected to find. So it’d be great to find something, but there’s little reason at this point to expect that something to be what we set out to find. So it might be dark matter (2) or it might be something completely different (3). I’ve no way to weigh the odds of those possibilities because they both seem very unlikely.
For WIMP-like dark matter we at least have some expectations, and this event does not conform to those expectations. It is too high energy. For every event like this one, the detector should have already recorded lots and lots of lower energy events.
That doesn’t mean it isn’t dark matter (2), just that it isn’t what we expected. But we already missed out on what we were expecting, so maybe? Or maybe it is something unrelated (3), or nothing at all (1). One event does not help override whatever assumption bias we might have had about these possibilities to start.
To be convincing…
It occurs to me that I am likely to hold the putative detection of dark matter to a higher standard than other scientists. So what would I find convincing?
- First, the detection of a putative new particle has to cross the traditional 5σ threshold.
- Second, the same result must be obtained by independent experiments.
- Third, the new particle has to be demonstrated to have the right properties to be dark matter.
- Fourth, the new particle has to exist with the required mass density to be the dark matter.
- (Fifth, it should naturally explain the observed MONDian phenomenology.)
I put the fifth criterion in parentheses because it is a very high bar, and I can conceive of being persuaded to soften it. It is also more advanced tuition than most physicists have contemplated, so let’s start with the more basic requirements.
At present, none of the above criteria have been met by the new LZ result. There is an old claim to a significant detection by DAMA that satisfies the first criterion but was never independently reproduced, so fails the second criterion. Nobody outside the original experiment takes it seriously now for that reason. It is sobering to realize that while new and exciting, LZ has not yet progressed as far down this list as DAMA.
Supposing both the first and second requirements are met, then we have a new particle. That is certainly interesting, but by itself would not mean we’ve found dark matter. When we’re in the realm of imagining new particles, there is a vast landscape of possibilities. Only some features in that landscape could qualify as dark matter; there are other possibilities that could exist but have nothing whatever to do with dark matter. So the properties of the particle have to be right, and there is a long list to satisfy.
From a generic astronomical perspective, the fourth criterion is the most important. We have a pretty good idea of the mass density that dark matter needs to have in the solar neighborhood to explain Galactic kinematics. That’s what these experiments are trying to detect: the dark matter particles in our Galaxy’s dark matter halo that happen to be passing through just now. So for whatever we detect to qualify as the dark matter, there needs to be the right number of particles with the right mass to satisfy astronomical requirements.
This fourth criterion is a pretty high bar. It does not suffice merely to find a new particle, it’s gotta be the right kind of particle and these particles have to exist with the right density. That can be measured, but it requires lots of detections: enough to work out the properties of the particles and how many are whizzing by. This is part of the disconnect I had in the last post with people complaining about statistical significance: they were focused on the significance of this one event; I was worried about having enough detections to estimate the mass density, or at least see if it is anywhere near the right ballpark. I’m more concerned with the astrophysical relevance of a new particle than with the new particle itself.
This is a generic problem in the field of dark matter, being as it is at the interface of two different fields. Astronomers need dark matter; we don’t care so much what it is if it performs as required. Particle physicists heard we astronomers needed dark matter, so are more than happy to suggest possibilities, but are more interested in the particle itself than what it does for galaxies. I don’t say what it does for the universe because that’s a lower bar where any dynamically cold, non-baryonic mass will do. The cosmic density needs to be right, but there’s no way to measure the cosmic average, only the local density of the Galactic dark matter halo.
The highest bar is the fifth criterion, which I think warrants its own post.
@tritonstation When "nature places more probability in the tail of the distribution for “weird” events than the Gaussian statistics we use say there should be," it's called a log-normal distribution!
-It's a fluke.-
On seeing this post I dialed up a query on google: “Can any particle accelerator in the world reach the mass-energy of the new particle detected by the LZ experiment?” It detailed a number of accelerator facilities that could reach this threshold of 200 GeV (though, I thought it was 400 GeV). But I was surprised by the last entry of the AI overview: “While accelerators have enough energy to create particles in that mass range, dark matter candidates (if they exist) interact so weakly with normal matter that they would mostly pass straight through high-energy particle detectors without leaving a trace, making sensitive underground xenon vats necessary to spot rare natural scatters.” I thought they could just look for missing momentum in the LZ particle’s mass range (plus whatever kinetic energy is involved) following a collision. I hope that a particle physicist, reading your blog, will chime in to clarify the situation. I’d be very surprised if there wasn’t a stampede at particle accelerators all over the world to be the first to produce one of these particles (unless, maybe, the mass range of interest has already been thoroughly explored).
Indeed – the signature of such a particle having been created in a particle collider would be its absence in the check-sum of mass-energy. The putative DM particle would not be detected, but you could tell that something left without being detected. That was the original indication for neutrinos; it is just the same here.
Whether this particular candidate particle should already have been detected in this way, I don’t know: it is a good question. There is certainly enough energy at the LHC to create it, but the odds of doing so are small and would have to be worked out.
This reminds me of Littlewood’s Law of miracles. which goes something like this: if a miracle is something you notice which has a one in million probability, then we should each expect to see one miracle per month, since in a month’s time we notice about one million things.
Hi Stacy.
Thank you for this post on “What can’t it be?”
I am minded of the 9-11 attacks and Nate Silver’s analysis using Bayes Theorem in “The Signal and the Noise”. Before the first plane struck one of the twin towers, he put the chance of a terrorist attack at 1:200,000. After the first plane struck he changed that to 38%. When the second plane struck he updated the probability to 99.99%.
We need a second event in the Lux-Zenon detector, or an independent measurement.
I can look at the 1 in 200 chance this way: we’ve been looking in like 300 ways/times, so now the 1 in 200 chance is expected to turn up once.
“… dark matter cannot explain the kinematic data for galaxies …” What might be plausible theories for FUNDAMOND?
Let us assume that gravitational energy is conserved, & inertial mass-energy is equivalent to gravitational mass-energy. What might be the conceptual basis for FUNDAMOND string theory as opposed to dark matter particles such as sterile neutrinos, axions, or gravitinos? Why is the following wrong? The stringy multiverse consists of alternate universes embedded in a multiverse interstitium. During each Planck time interval, the quantum vacuum in our universe interacts with the multiverse interstitium. This hypothetical interaction creates 3 dimensions of Milgrom spin and 3 dimension of Guendelman-Guth spin in gravitons, which have quantum spin 2, according to Weinberg. The 2 additional types of spin do not change the Weinberg spin, but do create Milgrom inertia and Guendelman-Guth inertia. The 2 new types of inertia create 2 new types of spontaneous symmetry breaking (SSB) in string vibrations. The 2 new SSBs explain the dark matter phenomenon & the dark energy phenomenon. In the standard form of Einstein’s field equations, both the –1/2 & the Λ represent fundamental symmetries for string vibrations. In the standard form of Einstein’s field equations, replace the –1/2 by –1/2 + FUNDAMOND-data-function & replace the Λ by Λ + dark-energy-data-function. The mathematical basis for the preceding scenario consists of 2 new worldsheet scalar functions, as indicated by Guendelman in:
Guendelman, E. I. “Dynamical string tension theories with target space scale invariance SSB and restoration.” The European Physical Journal C 85, no. 3 (2025): 276.
https://link.springer.com/article/10.1140/epjc/s10052-025-13966-9
https://arxiv.org/abs/2104.08875
One of the assumptions made by LZ scientists and many of the scores or hundreds of preprints trying to hitch this data point to a theory is that there is just one kind of DM particle. If so, since the inferred DM flux from Milky Way dynamics can be estimated moderately specifically, you know the flux of DM mass through the detector and its approximate velocity, and so the data point reflects a combination of DM particle mass and its cross-section of interaction with a Xenon nucleus.
But, if you abandon that assumption, and if other types of DM particles are too low in mass to be detected by LZ, then all bets are off.
One way this could be dark matter is if there are multiple types of dark matter. This could be a very rare, heavy species of dark matter, the moral equivalent of a uranium atom in interstellar space, and might actually have a cross-section of interaction comparable to a neutrino (a normal weak force interaction).
Most dark matter, the moral equivalent of a hydrogen atom in interstellar space, might have a similar weak force coupling, but a much lower mass (perhaps the 10^-22 eV/c^2 boson suggested by fuzzy dark matter advocated which just happens to be roughly the average mass-energy of a graviton) and doesn’t show up at LZ because it is far below its detection threshold which is in the ballpark of 0.6 GeV. Instead, it would be far below the neutrino threshold for detection since it would be much less massive than neutrinos are while having the same weak force charge. This kind of dark matter would have distributions (due to a large Compton wavelength) more similar to inferred dark matter particle halo distributions.
The core-cusp problem pretty much insures that the bulk of DM must either have a mass of 10 keV or less (warm dark matter or lighter types of DM particles), or a very particular strength of self-interaction that is mostly ruled out by observational constraints.
I’m not a fan of dark matter particle theories for the reasons you identify. But if this were a DM particle, that is the interpretation that makes the most sense to me.
Honestly, however, your point about non-Gaussian statistics is probably the most compelling, along with the fact that the look elsewhere effect should really consider all of the direct dark matter detection experiments that could detect a particle like this one and not just LZ. Those to factors would reduce the anomaly from 2.6 global significance to perhaps 2.2 sigma or less.
Another contender for a no-dark-matter-particle solution to kinematic evidence for extra gravitational mass in astronomical systems is Deur’s Self Interacting (SI) gravity hypothesis, which you have covered extensively at your “Dispatches From Turtle Island” blog. Over the last two or three days I was reading your “Deur’s Work On Gravity” subsection in that blog and Deur’s power point presentation, which nicely condenses his model. A few years ago I was intensely interested in his work, seeing it as a very elegant solution to the cosmological kinematic data, versus the Concordance DM model, most astrophysicists adhere to.
But then I began to have second thoughts. In his paradigm the planar geometry of spiral galaxies reduces the gravitational field from 3D to 2D, hence the 1/r force law in the outer galactic regions. For galaxy clusters the individual galaxies in the cluster act like point masses, so the gravitational field lines, with the added self-interaction property, contract into relatively tight bundles between galaxies reducing the geometry of the gravitational coupling from 3D to 1D. The resultant enhanced gravity force between galaxies mimics the presence of extra hidden mass. But what worried me was the total sum of gravitons is still the same for these structures as before (assuming no dark matter involved). So why would the gravitational field beyond the perimeters of these structures show a higher gravitational potential as indicated by lensing arcs, since I assume that potential reflects the total number of virtual gravitons available to curve the spacetime adjacent to these astronomical structures? I’m probably missing something here.
Consider one of the references quoted by the LZ experimenters.
According to Douglas Clowe, Marusa Bradac, Anthony H. Gonzalez, Maxim Markevitch, Scott W. Randall, Christine Jones, & Dennis Zaritsky, “An 8-sigma significance spatial offset of the center of the total mass from the center of the baryonic mass peaks cannot be explained with an alteration of the gravitational force law, and thus proves that the majority of the matter in the system is unseen.”
“A direct empirical proof of the existence of dark matter”, 2006
https://arxiv.org/abs/astro-ph/0608407
However, Clowe et al. only considered a few alternate theories of gravity — there might be an infinite number of possibilities for gravitational lensing alternatives for various FUNDAMOND string theories and other quantum gravitational theories. Therefore, the claim of “8-sigma significance” is highly questionable.
Clowe et al were assuming we see all the baryons; we already knew when they wrote that that this was not the case in clusters in MOND.
See Hernandez (https://arxiv.org/abs/2604.10811) and Famaey (https://arxiv.org/abs/2605.10022).
Another good analogy is found in the search for extra-terrestrial life – especially on Mars. Nothing in our experience to date has panned out, but there have been many OMGs – from the canal, to the first look at Martian chemistry, to the deep dives into polar water. Life on Mars would be a great find, but it will take more than one molecule of a suspicious protein to declare victory.
An important aspect is the X-files “I want to believe” syndrome. Many people remain convinced that some new form of particle dark matter exists; it’s just a matter of finding it. Every hint of a detection is colored by that rose-tinted perspective.
Very good analogy
Looking for life elsewhere, the bar is set very high. It has to be absolutely certain, and it’s hard to be, simply because of the situation. Even if we return a sample it can get contaminated, and just the risk of that can make results uncertain by definition. So in general there are a lot of discoveries that might or might not be life – they don’t achieve enough certainty to do anything, but they don’t get falsified either. They just wait in a superposition of states.
In the media these make a lot of OMGs, and they soon seem to fade away, giving the impression that this is like the search for DM. But it’s not the same – for one example the Murchison meteorite contained some of the building blocks of life: amino acids, and all five nucleobases necessary to form DNA and RNA. Isotope measurements showed they formed before reaching Earth. When you add up the other clues as well, it looks almost inevitable that there’s life out there, which is why astrobiology is one of the fastest growing fields there has ever been.
It is possible that dark matter will turn out to be families of ordinary protons and electrons bound in super compact stable configurations, spectrally degenerate with H-like ions – which would explain non-detection. these would be lab synthesizable and testable. It can be shown analytically that these can replicate MOND-like regularities (cuspless cores, slope 4 BTFR, a_0 normalization, causal reason for disk-halo conspiracy) without modifying gravity. Given that this hypothesis would require both an extension to QED (point-particle assumptions prohibit such states – only hydrogen is allowed), AND a modification to standard BBN (these would be baryonic particles, which means the baryon-photon ratio would be off by a factor of ~ 6), given severe THEORETICAL priors imposed, the only way it survives is empirically – if these configurations can be made and detected in a lab, then their existence will require a theory re-write. The math works and the physics is testable.
Baryonic states are possible so long as they don’t participate in BBN; can these “super compact stable configurations” be formed before that?
Even if they are, I don’t see how this reproduces MOND.
They could plausibly be formed before that – or would need an alternative model to explain light element abundances – deuterium is the main problem. As for MOND reproduction – I have two papers (one submitted to MNRAS, one a week away from MNRAS submission) that details how a total matter distribution law (dark + visible) could recover the MOND regularities without modifying gravity. the first paper stays phenomenological – it specifies the properties of the DM candidate required to generate the mass distribution law and tests predictions (not fits) on SPARC, LITTLE THINGS and also vertical tests vs MOND predictions. LCDM does mostly fits, not predictions. It achieves parity with MOND radially and separates from QUMOND vertically. the second paper proposes a lab-testable specific DM candidate family that delivers the phenomenology from Paper 1.
Is this stuff collisionless? It doesn’t sound like it.
Initially collisional briefly upon formation, subsequently collisionless. It is a family of particles with very structured properties – ionization and excitation energies, interaction radii, masses. lab-testable
“… the new particle has to be demonstrated to have the right properties to be dark matter.”
If dark matter particles are not modified gravitons, there needs to be some surprising explanation of why the dark matter particles track, in a MONDian way, gravitational geodesics predicted by general relativity (GR) for the huge number of galaxies in which MOND is approximately successful.
According to Prof. Milgrom, “In comparison with GR, it can be argued that the amount of data confronted successfully with MOND by now far exceeds those for GR even today. But from the point of view of theoretical development and understanding of fundamentals, MOND is not yet on par with GR.”
“MOND vs dark matter in light of historical parallels” by Mordehai Milgrom, 2020
https://arxiv.org/pdf/1910.04368
“It’s almost as if nature places more probability in the tail of the distribution for “weird” events than the Gaussian statistics we use say there should be.”
Sounds like we need a MOGS theory here: modified Gaussian statistics that explains observed discrepancies in the low-probability regime… 🙂