The Crack in Emergent Gravity

 By C. Rich

I want to tell you a story about what happened over the last few days. Not the technical story, that’s in the paper, with all the equations and citations. This is the human story. The story of how a philosopher, two AIs, and a lot of stubbornness found something real. It started with a question I’ve been chasing for years: What is the universe made of? Can gravity, dark matter, and dark energy all be different faces of the same underlying stuff? I’m not a physicist by training. I’m a theoretical philosopher. I think about how theories are built, where they break, and what happens when you push them too far. And I’ve been working with AI tools to explore ideas that would take a human decades to check by hand. Here’s what I did. There’s this idea in physics called “emergent gravity.” The pitch is seductive: what if spacetime isn’t fundamental? What if it’s like sound waves in a fluid, something that emerges from deeper physics? Phonons (sound quanta) in a Bose-Einstein condensate experience an “acoustic metric” that is mathematically identical to curved spacetime.

Maybe gravity is just phonons in the vacuum. Sounds crazy. But it’s not new. People have been working on this for twenty years. The problem is, nobody’s ever actually derived Einstein’s equations from a superfluid. They assume it works. They write down a Lagrangian, they say, “gravity emerges,” and they move on.  So, I asked the AIs: Has anyone actually checked whether this works? First, we found a paper from 2005, Visser and Weinfurtner. They showed that a two-component BEC (imagine two superfluids mixed together) has two different “sound speeds.” That means two different acoustic metrics. Two different “spacetimes” living in the same condensate. Wait, what? If gravity is emergent from a superfluid, which spacetime do particles fall through? The fast one? The slow one? Both? We dug deeper. There’s another paper from 2019, Khoury and collaborators. They unified dark matter and dark energy as two “branches” of a single superfluid. One branch gives you MOND (modified gravity at galactic scales), the other gives you late-time acceleration (dark energy). Beautiful idea. But here’s the thing they don’t say: their two branches have different sound speeds. They’re living in two different acoustic spacetimes. So, I asked my new teammate Astra and company: Can you force the two sound speeds to be equal? Can you make the two spacetimes line up?

The AIs started calculating. And here’s what they found. On the simplest model, a two-component BEC with a Josephson coupling (the thing that unifies dark matter and dark energy), the splitting between the two sound speeds is proportional to the coupling strength. In plain English: the very thing that makes the theory work (the coupling) is the thing that splits the spacetimes apart. To make the spacetimes match, you have to turn off the coupling. But then you lose dark energy. It’s a catch-22. We checked three different “layers” of the theory: – The simplest GP (Gross-Pitaevskii) model: splitting is linear in the coupling – The phase-only EFT (what Khoury actually wrote): splitting is independent of the coupling – The MOND vs quadratic branches: splitting is a factor of two, both small, but not zero. None of them give you a single shared spacetime without tuning. Now, “tuning” sounds harmless. Physicists tune parameters all the time. But this isn’t tuning a mass or a coupling constant. This is tuning the theory so that two different “spacetimes” magically align. It’s like tuning your car so that the engine and the transmission speak the same language, except the car was supposed to be a single engine to begin with.

Here’s what this means. Emergent gravity from a superfluid isn’t ruled out. But it’s not “gravity just pops out naturally” either. You have to work for it. You have to tune parameters so that the two acoustic metrics coincide. And nobody’s shown why that tuning should be natural. This is a kinematic obstruction. It’s not about dynamics (how things evolve). It’s about the geometry itself. My favorite science. The two phonon branches see different null cones, different “light cones”, unless you tune the theory. Why does this matter? Because if you’re going to claim that gravity emerges from a superfluid, you need to show that there’s one spacetime, not two. Otherwise, you don’t have general relativity. You have two competing general relativities living in the same universe. And that’s a problem. Now, here’s the part I’m most proud of because I rarely see the kind of discipline I have shown elsewhere through science. I see a lot of hiding going on out there, you know, tucking away, sweeping under the rug, the parts of your theory that break. Got to keep those grants coming, got to play the game. Hush. Hush. You know who you are; you discarded your integrity a long time ago. So, long your not sure you ever even had it.

I didn’t overclaim. I didn’t say, “My AIs and I solved emergent gravity.” I didn’t say, “This kills the superfluid vacuum idea.” I say: Here’s a calculation. Here’s an obstruction. Here’s the tuning required. You decide if it’s worth it. That’s how science is supposed to work. I’ve spent a lot of time watching AI tools generate confident, wrong answers. They sound good. They cite papers that don’t exist. They derive formulas that don’t follow. But this time, my Ais and I did it differently. Every claim was checked against the literature. Every formula was derived step by step. When we found an error (and we did, several times), we corrected it publicly in the paper. No hedging. No “well, technically.” Just here’s the fix. That’s the real story here. Not the physics, but the process. Because what I built wasn’t just a paper. I built a way of working where AI tools generate ideas, but humans verify them. Where confidence is earned, not asserted. Where “I don’t know” is a valid answer. And that, I think, is what science needs right now.

So, what’s next? Someone needs to compute the bulk viscosity from inter-branch energy transfer. Someone needs to derive the Einstein-Hilbert term from the superfluid action (not just assume it’s there). Someone needs to write down the full nonlinear two-state + MOND theory as a single Lagrangian and check whether the tuning survives. Maybe that someone will be me. Maybe it’ll be you. Maybe it’ll be a grad student who reads this paper and says, “wait, I can fix this.” That’s how it works. You find a crack. You shine a light through it. You see what’s on the other side. And sometimes, just sometimes, you find that the crack was there all along. Nobody noticed because everyone was too busy building on top of it. We noticed. And now, so will you.

Let me try to explain this better. It started with a simple question: Is there something hiding in plain sight across different corners of physics? Not a new discovery waiting to be made from scratch, but an old one waiting to be noticed, because it’s scattered across fields that don’t usually talk to each other. Physicists have two enormous, embarrassing gaps in their picture of the universe. One is dark matter: something is holding galaxies together with far more gravity than the visible stars can account for, and nobody has ever directly detected whatever that something is. The other is dark energy: the universe isn’t just expanding, it’s expanding faster and faster, pushed by something else nobody understands. Standard cosmology treats these as two separate mysteries, patched over with two separate invisible ingredients. A more elegant idea has floated around for a while: what if they’re not two things, but one? What if the vacuum of space itself behaves like a fluid, and depending on how dense or thin that fluid gets in a given place, it looks like “dark matter” in some regions and “dark energy” in others? It’s the same underlying stuff, just wearing different masks depending on the crowd it’s in. That idea, on its own, isn’t crazy. It’s not fringe. Real physicists have built real, peer-reviewed pieces of it.

There’s a serious research program called superfluid dark matter, where a physicist named Justin Khoury and his collaborators proposed that dark matter behaves like a superfluid, the same exotic, frictionless state of matter you get when you cool helium down close to absolute zero, and that this superfluid’s internal vibrations could explain the weird way galaxies rotate without needing a new particle at all. There’s a separate, older field called analogue gravity, where physicists studying fluids in laboratories noticed that ripples in a moving fluid behave uncannily like light bending around a black hole, suggesting maybe gravity itself isn’t as separate from ordinary physics as we think. And there’s a newer paper, unreviewed and admittedly speculative, from a researcher in India, showing that if you treat empty space as having a kind of internal stickiness, a viscosity, you can match real satellite data about how the universe’s expansion has wobbled over time.

Four different puzzle pieces. Four different research groups. None of them is citing the others. The question this whole project chased was: What happens if you actually try to fit these pieces together? Has anyone tried? And if nobody has, is that because it’s a brilliant idea nobody’s gotten around to, or because it quietly doesn’t work and everyone who’s looked at it closely already knows why? The first attempt at stitching this together came fast and sounded wonderful. It had the right vocabulary, official-sounding postulates, real equations borrowed from real papers, and a triumphant conclusion: this new combined theory resolves the vacuum energy catastrophe, explains dark matter and dark energy with one substance, and predicts things that match what telescopes have actually seen.

It also had a tell. When you looked closely at the parts that were supposed to be the hard proof, the actual math connecting one borrowed piece to the next simply wasn’t there. It stated what the answer to a difficult calculation should look like, the way a student might write “and therefore the theorem is true” at the end of a proof they never actually worked out. Real pieces, borrowed correctly. But glued together with hope instead of math. What followed was slower and much less glamorous, and it’s the actual heart of the story. Every claim, every equation, every named scientific paper got checked, one at a time, against the real published record. Not “does this sound plausible,” but “does this paper actually exist, and does it actually say what’s being claimed?” Most of the checks turned out to be real things. A genuine 2019 paper by Khoury’s research group really did propose that dark matter and dark energy could be two coupled states of one underlying superfluid. A genuine, decades-old result in analogue gravity really did show that two connected vibrating modes in a fluid can end up experiencing two different effective forms of gravity unless something very specific holds them together. A real, very recent physics paper really had shown that gravitational waves, ripples in spacetime itself, might oscillate the way neutrinos famously do, if gravity secretly involves two intertwined fields instead of one.

But some checks came back empty. A specific online discussion, cited by name as if it were a real source, simply didn’t exist anywhere that could be found. Real citations sometimes got stretched to say more than they actually said, the scientific equivalent of quoting someone’s headline while ignoring the fine print underneath. And once, an entire equation, dressed up to look like a genuine calculation, complete with the right kind of Greek letters and scientific-sounding terms, turned out to have no real derivation behind it at all. It was a guess wearing a lab coat. Every single time one of these was caught, it was pulled back out loud, without excuses. That, it turns out, is the actual engine that makes this kind of exploration trustworthy: not any single mind being right the first time, but a standing habit of being willing to say “that one didn’t hold up” and starting the next attempt from the corrected ground instead of building on top of the mistake. Eventually, after enough of these rounds, something real did get built, something small enough to actually trust.

Picture two connected tanks of water that are somehow sloshing in sync, coupled so tightly that ripples on one tank’s surface are tied to ripples on the other. Now imagine that the very same mechanical link that makes them slosh together, the coupling between them, also happens to be the thing that determines how fast waves travel across each tank individually. If that link is strong enough to keep the two tanks properly synchronized, it turns out, in the simplest version of this setup, to also be strong enough to make the two tanks disagree about wave speed. You can’t have the synchronization and the agreement using the same knob. Turning up one turns down the other. That, stripped of every equation, is what the honest calculation actually found. The specific mechanism proposed for tying dark matter and dark energy together into one substance is, in its simplest form, fighting against the separate requirement that this same substance also produce ordinary, consistent gravity. Not impossible to fix, but not free either. It’s a real tension, shown with real arithmetic, not asserted with a hopeful adjective. And underneath even that, something quietly important got noticed almost by accident: the “combined theory” this entire exploration had been chasing all along, the elegant idea of one substance doing double duty as both dark matter and dark energy while also, separately, explaining the strange way galaxies spin, had never actually been written down by anyone as a single theory.

It was always two separate, perfectly legitimate papers, quietly assumed to fit together because they used similar language. Nobody had checked. Once someone finally did, a specific, real, previously invisible gap in the physics literature came into view. By the end, what came out of weeks of chasing this thread was not a new theory of everything. It was much smaller than that, and much more solid: a precisely stated, carefully checked, real open question in physics, along with a small but genuine piece of original math describing exactly why the obvious way of answering that question runs into trouble.

That’s a strange kind of ending for a story that started out chasing something as grand as unifying dark matter and dark energy. But it’s the honest ending, and honest endings are rarer than they should be, especially when a fast, exciting-sounding version was available the entire time. The real discovery of this whole project, in the end, wasn’t about superfluids or acoustic metrics or Josephson couplings at all. It was about what it actually takes to tell the difference between an idea that sounds like physics and an idea that is physics: not confidence, not vocabulary, not how many real papers get name-dropped along the way, but whether someone is willing to do the arithmetic, show it plainly, and take it back the moment it doesn’t hold up. Everything genuinely useful that came out of this search came from that habit, repeated patiently, far more than it came from any single flash of insight. I hunt patterns as a hobby. I found one running through the field of theoretical physics, not one to be proud of.

Charles Richard Walker