homenowprojectswritingresearchabout

On this page

  • The question I started from
  • What I gave up
  • What survived
  • The one idea underneath
  • Where I think this leads
  • What would change my mind
  • How the work is done
  • The short version
  1. Home
  2. /Writing
  3. /What Gravity Does Not Do: The Big Picture So Far

What Gravity Does Not Do: The Big Picture So Far

November 18, 2026·12 min read
physicsquantum gravitygravitational decoherencetimeresearch notes
Quantum-Geometric CorrespondencePart 19 of 19
Previous

Why Gravity Alone Cannot Decohere a Superposition

On this page
  • The question I started from
  • What I gave up
  • What survived
  • The one idea underneath
  • Where I think this leads
  • What would change my mind
  • How the work is done
  • The short version

This post steps back. The papers on the research page each answer a narrow question, and from the outside they can look unrelated: pulsars, X-ray detectors, entangled microdiamonds, the absorption of gravitational waves. They are one line of thought. Here is the line, what it has established, what it has given up, and where I think it goes.

The question I started from

My intuition was always that gravity has something to do with time, and that this is the missing piece between quantum mechanics and general relativity.

Penrose gave that intuition a sharp form. Put a mass in a superposition of two places. Each place bends spacetime differently, so each branch has its own notion of time, and there is no preferred way to line the two up. Penrose argued that such a state is unstable and decays, at a rate set by the gravitational self-energy of the difference between the two mass distributions. Diósi reached the same formula from a different direction, as a noise model:

Γ=C EGℏ.\Gamma = C\,\frac{E_G}{\hbar}.Γ=CℏEG​​.

For a microgram of silica held a millimetre apart, this is about 65 picoseconds divided by the unknown constant C. It is one of the very few proposals where gravity and quantum mechanics meet in a number that an experiment could measure.

For a long time I tried to show that this rate follows from known physics. I wrote that quantum mechanics and general relativity are two descriptions of one thing, and that the rate was where the two descriptions touch. I extended the picture to cosmology: dark energy, galaxy rotation, the horizon as a clock.

What I gave up

Most of that did not survive being checked.

Three independent reviews, done blind, graded the unified picture as not established. The cosmology claims did not follow from the axioms they were supposed to follow from; one of them needed an axiom to fail. The constant C that I had argued was fixed turned out to be free. The claim that the rate is "derived" was true only of one static quantity, not of the rate.

So I retired the thesis. The name of the project stayed, the claim went. What was left was a smaller and better question: if the Diósi–Penrose rate were real, what would have to be true of the world? And is any of that already ruled out?

That change of question is the hinge of the whole programme. Trying to prove the rate produced essays. Trying to break it produced results.

What survived

Five results came out of the second question. They sit in four papers: two published as preprints and two working papers on this site. A fifth paper is being held back until the authors whose result it concerns have seen it, and is not described here.

1. Gravity as we know it does not produce the rate. A steady decay rate needs an environment that keeps fluctuating at arbitrarily low frequencies. The quantized gravitational field around a static mass does not do that. It takes a small, fixed toll on coherence, a few parts in a thousand for a microgram, and then stops; the loss is reversible. The same holds for any environment in its ground state, by a general argument: at zero temperature noise and absorption are the same quantity, and absorption vanishes at zero frequency. These statements are standard physics. Their consequence is that the rate cannot be a property of the vacuum.

2. A warm medium would produce the rate, and the double pulsar forbids it. An environment that is not in its ground state can supply the noise. But if it couples to mass the way any ordinary medium does, the noise that decoheres a body also drags a moving one. That is the fluctuation–dissipation theorem, the same reason a pollen grain that jitters in water also feels the water's viscosity. Two neutron stars orbiting each other would lose energy to that drag, and the orbit of the double pulsar PSR J0737−3039 is measured so well that there is no room for it. The bound on the medium's effective temperature comes out at more than 3.3×10¹⁶ kelvin, sixteen orders of magnitude above the value proposed for the published model of this kind, if the medium behaves at kilometre scales as it does in the laboratory. This is the paper Double-Pulsar Drag Bounds Dissipative Sources of Diósi–Penrose Decoherence.

What is left after these two steps is noise with no back-action at all: a rule added to physics, not derived from it, with two free numbers. Data allow it. One clean measurement of the coherence of a superposed microsphere would fix its strength.

3. If the rule is real, it cannot pass for quantum gravity in the entanglement test. The most discussed experiment in this field would let two small masses interact only through gravity and look for entanglement between them. Entanglement is taken as the signature that gravity is a quantum channel. The Diósi–Penrose model, in its consistent form, also entangles masses under some conditions, which blurs the test. I found the exact condition, a single geometric comparison, and showed that the entanglement the model produces is weak: for well-separated superpositions it is transient and a small fraction of what quantum gravity gives. An observation of the full effect still rules the model out. This is Entanglement from the Diósi–Penrose Model.

4. The best-developed theory of classical gravity fails an X-ray test. If gravity is fundamentally classical while matter is quantum, consistency forces a trade: the gravitational field must fluctuate, and matter must receive random kicks. Kicked charges radiate. Underground detectors built to look for dark matter see no such radiation, and for the covariant theories in which the noise is Gaussian the mismatch is a factor of roughly a hundred million, under assumptions the paper lists together with the ways out. This is Spontaneous X-rays Exclude Gaussian Covariant Classical–Quantum Gravity.

5. How matter absorbs a gravitational wave. The last piece looks like a change of subject and is not. Absorption is the other face of noise, so I worked out what quantum matter can take from a gravitational wave. The answer is a sum rule: the absorption summed over all frequencies, suitably weighted, is fixed by how the body's mass is spread out. For slow matter that is its moment of inertia. For a single relativistic particle it is the slope of its gravitational form factor, which measures the radius of its mass distribution. The object that does the absorbing turns out to be the energy flowing along the wave's own light rays. This is Graviton Absorption and the Slope of the Gravitational Form Factor.

The one idea underneath

Every one of these results is the same statement applied in a different place: you cannot have noise for free.

In physics, a thing that shakes other things can also be shaken by them. Fluctuation and dissipation are tied together by positivity and by conservation laws, and the tie is quantitative. If gravity, or a medium, or a new field delivers random kicks to matter, then it also absorbs, drags, or radiates, and one of those consequences shows up somewhere we have already looked with great precision.

  • Ask for noise from the vacuum, and the tie says there is none at zero frequency.
  • Ask for noise from a medium, and the tie gives drag, and the pulsars say no.
  • Ask for noise in a classical gravitational field, and the tie gives X-rays, and the detectors say no.
  • Ask how much a body can take from a gravitational wave, and the tie gives a fixed budget set by its mass distribution.

What remains standing is exactly the option that cuts the tie: noise with no partner, postulated. That is what a collapse model is. The programme has not refuted it. It has shown what believing it costs: a new law of nature with no back-action, and a preferred frame somewhere in the law or the state.

I think of the work so far as a map of what gravity does not do to quantum matter. It does not decohere it at a steady rate. It does not drag it. It takes almost nothing from it in a gravitational wave. Each "does not" is now a number.

Where I think this leads

Here I move from results to a bet. Nothing in this section is established.

I do not think the collapse rule is true. I think gravity is quantum, that geometry can be in superposition like anything else, and that nothing collapses. We see one geometry because we are inside one branch. On that view the Diósi–Penrose rate in nature is zero, and the collapse models are what physicists call foils: precise wrong answers that tell you what the right answer has to avoid.

That leaves Penrose's original argument without its conclusion, and I find that more interesting than the conclusion ever was. He was right that two branches of a superposed mass have two different flows of time and no canonical way to compare them. If the state does not decay, then that mismatch is simply a fact about the world, and the question becomes what it means. What does a clock read when the mass that sets its rate is in two places? What does "at the same time" mean across branches?

My leaning is that time is not fundamental. At bottom there are correlations between physical quantities, and we call one of them a clock. The flow of time and its direction are not in the laws; they come from the fact that any observer has incomplete information, and from the fact that some systems keep records. I am close to Rovelli here, and to Barbour on the claim that the laws supply a moment of greatest uniformity from which arrows of time point away in both directions. I also take seriously Smolin's warning that timelessness may be an artefact of applying to the whole universe a method built for its parts.

Underneath that is the idea I trust most in the whole subject: that entanglement and geometry are the same thing seen two ways. Known physics already says a version of this. The entropy of a region of space has a geometric part, the area of its boundary, and requiring that entropy to be stationary gives Einstein's equations, up to the cosmological constant. If that is the right foundation, then time should come out of entanglement too, as the flow that a quantum state defines on the things an observer can measure.

So the programme's next question is where gravity meets time in a form that can be tested, which means clocks. Not the cosmological clock of the universe, where the experts are far ahead of me, but small ones: a clock in a superposition of heights, two clocks whose rates depend on each other's position, the loss of clock coherence in a gravitational field. The tools built for the collapse question, exact response functions, sum rules, bounds that follow from positivity, are the right tools for that.

What would change my mind

A picture is only worth holding if something could break it.

  • A measured decay rate of the Diósi–Penrose form. If a superposed microsphere lost coherence at a rate proportional to its gravitational self-energy, with other causes excluded, I would be wrong about collapse, and the results above say what would come with it: noise without back-action and a preferred frame.
  • No entanglement where quantum gravity predicts it. If two masses coupled only by gravity stayed unentangled in an experiment sensitive enough to see the predicted effect, gravity would not be an ordinary quantum channel.
  • Spontaneous radiation. A reproducible excess of X-rays from ordinary matter with no known source would reopen classical gravity.
  • A clock result that cannot be reproduced by unitary quantum mechanics in curved spacetime.

None of these has happened. Several are within reach of experiments now being built.

How the work is done

One rule runs through all of it: a claim is kept only if it survives an attempt to destroy it.

Every result is first written as a derivation and attacked before any paper exists. Every number in a paper is recomputed from its stated formula by code that runs with every change. Every reference is checked against the publisher's record. Papers are read cold before they go anywhere. When a claim fails, it is withdrawn in public, and the essays on this site carry their corrections at the top.

This week the three finished papers were audited again from scratch, by a reading that saw only the PDFs. All the main results held. Each paper had one wrong secondary statement, and all three have been corrected. I would rather publish the correction than the confidence.

I do not expect to out-calculate the people who do this for a living. The aim is to ask questions that sit between fields, where a definite calculation is possible and has not been done, and to be strict about what the answer supports.

The short version

I set out to show that gravity ends quantum superpositions. The evidence says it does not: not through the vacuum, not through a medium, not through a classical field of the kind proposed. What is left is a bare postulate that experiments can test and that I expect to fail.

If it fails, the link between gravity and time that started all this is still there, without a collapse to resolve it. Understanding that link, what time is when geometry itself can be in two states at once, is what the programme is now for.

Share this post

MS
Marc Sperzel

Builder and independent researcher. MSci Physics, King's College London. Writing about quantum mechanics, gravity, and information theory.

GitHubORCIDEmail

Stay updated

Get notified when I publish new writing or research.

Related posts

The Diósi–Penrose Rate as a Thermal Fluctuation

October 5, 2026

Why Gravity Alone Cannot Decohere a Superposition

October 5, 2026

A Quantum Clock Is Not a Decoherence Law

September 6, 2026