The Worlds Still Split — But Gravity Decides When

Foundations and interpretation — Quantum-Geometric Correspondence series

The picture you've probably already met

If you've read any popular physics in the last decade — Sean Carroll's Something Deeply Hidden, most likely — you've met the Many-Worlds Interpretation of quantum mechanics. The pitch is bracingly simple. Quantum mechanics says a particle can be in a superposition, in two places at once. When you measure it, textbooks tell you the wavefunction "collapses" to one outcome. Many-Worlds throws the collapse away. Nothing collapses, ever. Instead, the universe branches: in one world the detector clicks left, in another it clicks right, and there is a copy of you in each, equally real, each convinced their outcome is the only one that happened.

It's an elegant idea precisely because it removes something. No mysterious collapse, no special role for observers, no line between the quantum and classical worlds. Just the Schrödinger equation, running forever, unitary and deterministic, splitting the world into ever more branches.

Our own framework — the Quantum-Geometric Correspondence, or QGC — is, on paper, a Many-Worlds theory. It keeps all three of Carroll's core commitments: the wavefunction is the complete description of reality, evolution is always unitary, and there is no collapse. If you asked "does QGC believe in wavefunction collapse?" the honest one-word answer is no.

And yet it does not sit where Carroll sits. Gravity moves it. This essay is about exactly how far, and in which direction — because the destination turns out to be a genuinely different place on the map, one that is empirically collapse-like while remaining a no-collapse theory. We call the stance objective branching, and getting to it takes four steps.

A note before we start, because it matters: the thing that moves us off Carroll's spot is not that gravity collapses the wavefunction. It's subtler and, I think, more interesting than that.

Step one: there is no "when" for the split

Start with the oddest difference, the one that sounds like philosophy but is really about the equations.

In Carroll's Many-Worlds, the wavefunction evolves in time. Worlds branch as events: at 3:00 the detector hadn't fired, at 3:01 it had, and somewhere in between the world split in two. The multiverse is a tree, and the tree grows toward the future.

QGC's foundational object is different. It's the Wheeler–DeWitt state, and it obeys a deceptively short equation:

That zero on the right-hand side is doing something violent. In ordinary quantum mechanics the Schrödinger equation has a time derivative in it — the state changes as $t$ ticks forward. The Wheeler–DeWitt equation has no time in it at all. The universe's total state just is; it doesn't evolve, because in a theory where geometry itself is quantum, there's no external clock to evolve against. Time is not fundamental in QGC. It's emergent — recovered, through a mechanism due to Page and Wootters, as a correlation between some subsystem you treat as a clock and everything else.

Here's what that does to branching. If there's no fundamental time, there's no fundamental moment at which a world splits. What Carroll would describe as "the world branching at 3:00" becomes, in QGC, a static correlation pattern frozen into a timeless block — a slice you read off along the emergent time direction, not a thing that happens. Carroll's multiverse is a growing tree. Ours is a finished crystal, and "branching" is the story you tell when you trace your finger along it.

This isn't wordplay. It changes what the question "when does the world split?" even means. And the answer to that question — once we get to step four — is the sharpest, most testable thing in the whole theory.

Step two: gravity solves the problem Many-Worlds usually dodges

Every honest account of Many-Worlds has a soft spot, and Carroll is candid about it. It's called the preferred basis problem, and it goes like this.

A superposition can be sliced up many different ways. The cat is "alive plus dead" — but mathematically it's equally true that the cat is in some bizarre combination like "(alive+dead) plus (alive−dead)." So why do the worlds split along the alive/dead line, the one that matches our experience, rather than along one of the infinitely many grotesque alternatives?

The standard Many-Worlds answer is decoherence: the environment is constantly "monitoring" the system — air molecules, stray photons — and that monitoring picks out a basis. Fine, but notice what this answer depends on. It depends on the environment. It's emergent, approximate, and relative to whatever happens to be doing the monitoring. There is, in canonical Many-Worlds, no fundamental fact about what the branches are. Push hard on "how many worlds are there?" and Carroll will tell you, correctly, that the question has no exact answer.

QGC gives a different kind of answer, and it's the theory's cleanest win. Gravity couples to mass and energy — to where the stuff is. And gravity is the one environment you can never switch off, never shield, never pump out of your vacuum chamber. So the monitoring that picks the basis is always, universally, the gravitational monitoring of position. The worlds split along the where-is-the-mass line because gravity forces them to.

In QGC, the preferred basis isn't a contingent fact about your apparatus. It's forced by the one interaction every lump of matter is subject to. Gravity selects position, everywhere, for everything.

This is a stronger claim than Many-Worlds usually dares to make, and — crucially — it's testable, because it predicts exactly which superpositions decohere and how fast. More on that in a moment.

Step three: the other worlds are other spacetimes

In Carroll's most austere formulation, there's nothing but the wavefunction — an abstract vector in an abstract space, from which even the appearance of three-dimensional space is supposed to emerge. The branches, in that view, are patterns in that vector. Ghostly, mathematical.

QGC agrees that geometry is emergent rather than fundamental, but it fills in a concrete middle step that makes the branches far less ghostly. When matter goes into superposition, it drags spacetime along with it. A mass on the left curves spacetime one way; the same mass on the right curves it another. So the full state isn't just "matter here plus matter there" — it's

where the $g$'s are two genuinely different geometries — two different spacetimes. Schrödinger's cat isn't two arrangements of atoms against one fixed backdrop. It's two universes with different shapes, superposed.

And this is what makes the branching stick. Two different spacetimes are spectacularly distinguishable — far more so than two arrangements of atoms — so once the geometries differ, the branches can't re-interfere. The permanence of the split, which in Many-Worlds is a matter of "well, in practice it never comes back," is in QGC tied to the difference between two spacetimes and, through that, to the arrow of time itself.

There's an irony worth naming. This makes QGC the less radical cousin of Carroll's program. His "Mad-Dog Everettianism" dissolves space entirely into abstract quantum structure. QGC keeps a real, if emergent, geometry in the picture — and puts it to work as the thing that picks the branches. We hold onto more of the spacetime furniture, not less.

Step four: the split happens on a schedule — and the schedule is a collapse law

Now the punchline, and the reason this essay exists.

In canonical Many-Worlds, there's no particular rate to branching. Decoherence is generically fast, it depends on the environment, and nobody hands you a formula for how long a given superposition lasts. QGC does hand you that formula. A mass $M$ spread over a distance $d$ decoheres — the world "splits" — on a timescale

Plug in a microgram-sized grain separated by a millimeter and you get about 1.6 nanoseconds. Plug in a single molecule and you get longer than the age of the universe — which is exactly why molecules stay quantum and dust grains don't. "When does the world split?" stops being a philosophical shrug and becomes a number you can compute and, in principle, measure.

Here's the twist that puts QGC in genuinely new territory. That rate — the rate at which our unitary, no-collapse theory says the worlds branch — is numerically identical to the rate at which a completely different family of theories says the wavefunction physically collapses.

Those are the objective-collapse theories of Diósi and Penrose. They are not Many-Worlds theories at all. They say quantum mechanics is wrong, that there's a real, random, physical collapse, that only one world survives and the others are destroyed. And they predict a collapse rate of $GM^2/(\hbar d)$ — the same expression, to the letter, as our branching rate.

So here is where gravity has quietly carried us. QGC is a Many-Worlds theory whose branching is governed by the very law a single-world collapse theory would use to kill the extra worlds. For any object big enough to see, the two stories make identical predictions. You cannot do an experiment on a dust grain that tells you whether the other branches are quietly still out there (our view) or were violently destroyed (Penrose's view).

We call this objective branching, and it's a real third position, distinct from both neighbors:

• It's not canonical Many-Worlds, because our branching isn't a vague, environment-relative, rate-free affair. It's objective (gravity forces it), universal (nothing escapes gravity), geometric (the branches are different spacetimes), and it runs on a definite clock.
• It's not objective collapse, because nothing actually collapses. The evolution stays unitary; the other worlds are not destroyed.

QGC takes the Many-Worlds skeleton and equips it with an objective, gravitational branching law so rigid that the many-worlds reading and the single-world reading become experimentally identical — and then it does something unusual. It declines to choose between them. Whether the un-selected spacetimes are really out there is, for any macroscopic object, a question no experiment can answer, and QGC treats that honestly as an open question rather than pretending to have settled it.

That agnosticism — not collapse — is the "different point of view." Gravitational decoherence doesn't collapse the wavefunction. It does something stranger: it makes the collapse and no-collapse stories converge so completely that the difference between them stops being physics.

Where we're honest that Carroll has more to say

It would be dishonest to end on a victory lap, because there's one central question where our theory offers less than Carroll's, not more.

That question is the Born rule — the recipe that says the probability of an outcome is the amplitude squared, the $|c|^2$ at the heart of all quantum prediction. Nobody has ever derived it from more basic principles; every version of quantum mechanics simply assumes it. Carroll, to his credit, has a genuine proposal — the "self-locating uncertainty" argument he developed with Charles Sebens, which tries to show that a rational person who knows they've just branched, but doesn't yet know which branch they're in, should place $|c|^2$ odds on each outcome.

We went looking, carefully, for whether any of QGC's distinctive machinery — the emergent time, the gravitational decoherence, the holographic bounds, the Wheeler–DeWitt structure — could do better. The answer was a clean, disappointing no. Emergent time gives probabilities of the wrong mathematical form. Decoherence tells you when and in which basis the worlds split, never with what probability. And the Wheeler–DeWitt framework actually makes the "measure problem" harder, not easier, because the natural geometry on its space of possibilities isn't the right kind to define probabilities on at all.

So on the deepest puzzle in the foundations of quantum mechanics, QGC inherits the mystery unchanged and adds nothing. On this one, Carroll is ahead. We think it's important to say so plainly rather than let the theory's genuine strengths elsewhere paper over the gap.

What's at stake, and how we'll know

The beauty of objective branching is that it isn't only philosophy. Step four rests on a specific, still-unproven piece of the theory — the claim that gravity's branching runs at the $G^1$ (Diósi–Penrose) rate rather than the much slower $G^2$ rate that ordinary quantum field theory would predict. Those two rates differ by a staggering factor of about 10³⁵. One of them is right.

And we can find out. Levitated-optomechanics experiments — tiny particles held motionless in laser traps, in deep vacuum, near absolute zero — are climbing toward the mass and separation scales where this becomes measurable. The decisive experiment is roughly a picogram-scale superposition held coherent for about a second, and the plausible window for it is 2028–2035.

If those experiments see coherence surviving to the slow $G^2$ timescale, then gravity is "just another quantum field," the objective branching story dissolves, and QGC's distinctive content is wrong. If they see it die on the fast $G^1$ timescale, then gravity really does impose an objective, universal law on when the worlds split — and the strange convergence between many-worlds and collapse becomes a fact about nature rather than a curiosity about equations.

Either way, we'll have learned something profound about the oldest puzzle in quantum mechanics: not by arguing about interpretations, but by measuring when a dust grain stops being in two places at once.

A plain-language companion to the Quantum-Geometric Correspondence programme's four-axis contrast with Everettian quantum mechanics and its "objective-branching" positioning. The decoherence rate and its experimental target are developed in the foundations paper; the "no leverage on the Born rule" finding is a deliberate null result. As always: none of this is confirmed. The theory's virtue is not that it's proven but that it's sharp — it stakes its distinctive claim on one number an experiment this decade can measure.