The Aperture.

Start with the simple version, because the technical one lands harder afterwards. In ordinary quantum mechanics there is a clock on the wall and the system evolves against it. The equation has a time variable in it because the experimenter has a watch, and the watch is outside the box.

Now put the box around everything. There is no outside, no wall, no watch. When you quantise general relativity canonically the energy operator does not stay an operator — it becomes a constraint, and the resulting equation, written down by Bryce DeWitt in 1967 and now carrying his name and Wheeler’s, says something that took the field a decade to admit it meant. The wave function of the universe does not evolve. It is stationary. There is no time parameter in the description of everything, because a time parameter is a thing you measure against something you are not.

This is not a heterodox reading and it is not a philosopher’s embroidery on a physicist’s equation. It is the standard statement of the frozen formalism problem, the central and unresolved difficulty in canonical quantum gravity, and it has been sitting there for nearly sixty years. Quantum mechanics treats time as universal and external. General relativity treats it as local and dynamical. Push them together and the parameter everyone assumed was fundamental turns out not to survive the merger.

And yet things very obviously happen. Something has to give, and what gives is the assumption nobody noticed making: that time is a property the universe has, rather than a relation between two pieces of it.

In 1983 Don Page and William Wootters published the repair in Physical Review D under a title that is already the argument: Evolution without evolution. Do not rescue the global clock. Give it up. Split the frozen universe into two pieces — call one of them the clock, and the other everything else — and entangle them.

Then ask a conditional question. Not what is the state of the universe at time t, which has no answer, but given that the clock subsystem reads t, what is the state of the rest? That question does have an answer, the answers for different clock readings differ, and lined up in order they are indistinguishable from a history. Evolution appears — inside a state that is not evolving.

Sit with the shape of that, because it is the whole essay. Time is not something the universe does. It is something a part of the universe reads off its correlation with another part. An observer who could somehow stand outside the pair and measure it globally would find nothing changing at all. The observer who is one of the parts finds a world in motion. Neither of them is mistaken.

The obvious objection is the right one: this sounds like a trick of bookkeeping. It is not obviously more than a way of writing conditional probabilities. That objection was made forcefully, and it is dealt with in the counter-evidence section rather than buried here. What matters first is that the mechanism stopped being a thought experiment.

In 2013, at the Istituto Nazionale di Ricerca Metrologica in Turin, Ekaterina Moreva and colleagues put it on an optical bench. Two photons, entangled in polarisation. One of them is designated the clock.

Run the experiment as an internal observer — that is, measure the clock photon and then ask what the second photon is doing conditional on that reading — and you see the second photon evolving, exactly as Page and Wootters predicted, an ordinary history unfolding at an ordinary rate.

Run it as an external observer — measure only global properties of the pair, refuse to condition on the clock — and you can demonstrate that the joint state is static. Nothing is happening. The same physical system, the same preparation, the same laboratory, on the same afternoon.

Two things about this deserve to be said slowly. The first: both results are correct. This is not an illusion versus a reality, or an appearance versus a truth. The evolving description and the static description are both complete descriptions of the same state, and what distinguishes them is nothing about the system and everything about where the observer is standing relative to it.

The second is the one that closes the door on the reading this essay is most often confused with. The internal observer in Turin is a photon’s polarisation. It has no memory, no preferences and no inside. If a photon can occupy the role, then whatever is required to have an evolution parameter, a mind is not it. Hold that; section vii is built on it.

Here is the argument in the form anyone can check. Tonight, while everyone sleeps, the entire universe is sped up by a factor of ten. Every atom, every star, every reaction, every clock, and every brain. Nothing is exempt.

Nobody can tell. Not by feel, not by instrument, not by any experiment that could ever be designed. Every measurement of a rate is a comparison of one process against another process — a heartbeat against a clock, a clock against a sunrise, a photon against a ruler — and scaling both sides of every comparison leaves every comparison identical.

And now the sharp end of it. If no detectable difference exists, in principle, anywhere, then the speeding-up is not an event. It is a sentence with the grammar of a description and nothing being described. Leibniz made exactly this move against Newton in the correspondence with Clarke three hundred years ago: a universe shifted uniformly in space, or created a year earlier with everything else unchanged, is not a second universe, because there is no fact that could distinguish it from the first.

So time has no speed. The question how fast is time going has no referent, because a rate is always one thing against another and there is nothing outside time to hold it against. The familiar phrase — one second per second — is a ruler measured in rulers. It is not a slow answer or a fast answer. It is not an answer.

What does exist is entirely relational: this process against that one. Which is precisely the quantity Page and Wootters built the emergent time out of, and precisely the quantity that runs out when there is nothing left outside the system.

Time has two features that feel like properties of the world: it passes, and it goes one way. The first turned out not to be in the world. Neither is the second.

Film two particles colliding and run it backwards and you have violated nothing; the fundamental equations are symmetric under reversal and contain no arrow anywhere. Film a glass shattering, run it backwards, and every single collision in the reversed film is still legal — yet nobody is fooled for a quarter of a second. The asymmetry is real and it is not in the dynamics.

It comes from a boundary condition. The early universe was in an extraordinarily low-entropy state, and everything since has been the mess increasing; the assumption is explicit enough in the literature to have a name, the Past Hypothesis. Every derivation of the second law needs it or something like it, and none has managed to get the arrow out of time-symmetric dynamics alone.

Memory rides on that and nothing else. A memory is a record, and a record is a physical structure that had to be paid for: forming one takes something ordered and degrades it, dumping entropy into the surroundings. Records can therefore only be laid down in the direction in which the mess increases. You remember one way for the same reason a cup of coffee cools one way. There is a single arrow, it was installed at the boundary rather than derived from the laws, and every remembering thing is riding it.

Which gives the accounting for this whole subject a strange and consistent shape. Neither of the two features of time that anyone actually experiences is a property of the universe. One is a relation between a subsystem and its clock. The other is the residue of how things started, cashed out through devices that spend order to keep a past.

Now put the pieces together, because separately they are well-known results and together they are a bill.

A system has an evolution parameter because it is correlated with something it is not. That is the content of the Page-Wootters construction and the thing the Turin bench measured. The clock has to be elsewhere. A system perfectly correlated with everything has nothing left over to serve as the other side of the comparison — and the state of everything, by the equation this essay opened with, is stationary.

So there is a ladder, and it runs the wrong way round from every intuition anyone has about minds. At the bottom, an atom: no internal arrangement, no clock, no history — the case Tense already made. In the middle, things like us, correlated with a few metres and a few decades and blind to essentially all of it, and therefore surrounded on every side by things available to be a clock. At the top, an observer correlated with the whole, holding every moment at once, in a picture where nothing is happening.

Call the trade what it is. The aperture bargain: the width of what a system takes in, set against whether anything in it is occurring. Every step up the ladder buys reach and is charged to the same account that presence is drawn on. Not because seeing is tiring or because knowledge disenchants — for a reason with no sentiment in it at all. What you are correlated with cannot also be what you measure your own change against.

And this is where the argument stops being about physics and starts being about the roadmap. Every proposal for a greater mind — every upload scenario, every context window, every plan to hold a civilisation’s entire record in one system and finally see the whole of it — is a proposal to climb this ladder. The reach side of the ledger is costed to four decimal places. The other column has never been opened, because until you write down the equation for the whole thing at once, there is no reason to suspect it exists.

Be careful about how much of this is proven, because the honest version is narrower than the sweeping one. What is established is the two ends: a partial subsystem with a clock has an evolution parameter, and the global state does not. The smooth monotone interior — that a larger observer has a thinner present — does not fall out of the formalism. It is the conjecture this essay is making, and it is marked as one here, in the middle, rather than in a footnote.

Every argument in this territory is one careless sentence away from becoming quantum mysticism, so here is the wall, built out of results rather than disclaimers.

The conscious-observer interpretation exists and has a clear origin: Wigner proposed in 1961 that the consciousness of an observer is what precipitates collapse. It is no longer a position in physics. What displaced it was Zeh’s 1970 decoherence result — definite outcomes arise from a system’s entanglement with its environment, requiring no observer of any kind — and Wigner himself did not keep the view. Rovelli, whose relational quantum mechanics is the nearest live framework to the argument here, states plainly that his claim concerns states being relative to physical systems and is not a claim that reality depends on a conscious one.

And the experiment settles it without needing anyone’s testimony. In Turin the thing that plays the internal observer is a photon’s polarisation. If a photon suffices, a mind is not required, and any version of this essay that needed one would be refuted by its own central citation.

It is worth noting what the observer experiments have actually forced people to give up, because it is not what the popular version says. Frauchiger and Renner showed in 2018 that agents reasoning correctly with quantum theory about each other reach contradictory conclusions. Bong and colleagues then proved that one of three things must fail: No-Superdeterminism, Locality, or the Absoluteness of Observed Events — the assumption that an observed event exists absolutely rather than relative to an observer. Proietti and colleagues ran a six-photon version and violated the associated inequality by five standard deviations. What is under pressure is the absoluteness of facts. Nobody anywhere has been driven to postulate a mind.

The other direction has been tested too, and it lost. If something special happened when systems became large or complex, the natural place for it is gravitational collapse of the wave function. The parameter-free Diosi-Penrose model predicts a faint spontaneous radiation; an underground experiment went looking for it and excluded the model. The hypothesis that scale changes the rules is not unexamined. It has been examined, and the simplest version is dead.

If the ladder is real, one question follows immediately and it is not a rhetorical one: what is actually lost at the top? What exists at an aperture that does not exist in the complete picture?

Rate, first, which is the argument of this whole essay: fast and slow are relations, and they require an outside. Then duration, which is rate integrated — an interval has a length only for something that measures it against something else. Then the distinction between this moment and any other, which is exactly what the global state lacks and the conditioned state has. And then, downstream of those and harder to formalise but not therefore unreal: waiting, anticipation, boredom, loss. A star ceasing to shine costs the universe nothing. Loss is a fact about a system that had something and now has a record of having had it.

None of that is in the atoms and none of it appears in the complete description. It is the standing feature of everything on this list that it is local and non-recoverable. Facts are recoverable: if one observer never finds a law, another does, and the law is indifferent to who wrote it down. Zurek’s quantum Darwinism describes precisely how a fact gets to be objective — by depositing so many redundant records in the environment that any number of partial observers can independently read the same thing off it. That redundancy is what makes knowledge shared, and it is exactly what an aperture does not have. Nothing is depositing redundant copies of a present tense.

Which inverts the ordinary priority in a way worth stating flatly. The connecting of dots is the recoverable half of the project. The discoveries will be made, by someone, eventually, and they survive the loss of any particular observer. The thing that does not survive, that has no second copy anywhere in the environment and no route back if the aperture closes, is the part everybody treats as the meanwhile.

The mechanism has a standing, unresolved objection. Karel Kuchar argued that the naive Page-Wootters construction, read as conditional probabilities, returns the wrong two-time correlation functions — it gets single-time statistics right and sequential measurements wrong, which is close to fatal for a proposal about time. Giovannetti, Lloyd and Maccone published a repaired formulation in 2015 built specifically to reproduce sequential-measurement statistics, and it is not universally accepted. Anyone citing the Turin experiment without citing Kuchar is selling something.

The floor may be an artefact. The whole argument rests on the Wheeler-DeWitt equation being the right way to quantise gravity, and it may not be. If canonical quantisation is the wrong road — and there are serious people who think the problem of time is a symptom of that rather than a discovery about the world — then the frozen global state is a feature of a bad formalism, and there is nothing underneath this essay at all.

The last step is a metaphor, and it should be named as one. Everything the physics licenses concerns correlations between degrees of freedom. Moving from a subsystem with a clock has an evolution parameter to a witness has a present tense is a leap that no experiment in this essay makes for us, and the gap is not narrow. It is exactly the gap that no protocol has ever been built to cross — which is the content of prediction 05, and the reason it is written as a prediction rather than asserted as a result.