The radical idea that space-time remembers could upend cosmology todayheadline

If you wave your hand in front of your face, you won’t notice anything particularly interesting. Perhaps a gentle waft of air against your cheek – that’s about it. No epiphany. No major sign that anything out of the ordinary has occurred. And yet it is often when we look below the surface of the everyday that we find the extraordinary.

I believe that when we sweep a hand through the air – or indeed when anything moves at all – space-time is imprinted with a memory of what happened. The change resulting from this motion may be far too subtle for us ever to discern, but on the scales of the wider cosmos, space-time’s memory is crucial. In fact, I would go further. I have come to believe that space-time isn’t the kind of empty nothingness most of us think it is, but instead, at a fundamental level, it is made of stored information.

That might all sound rather bold. It certainly recasts our view of the canvas on which reality plays out. But over the past couple of years, I have bounced this idea around my brain – and inside a quantum computer – testing its limits. That has led me beyond a reformulation of space-time, and therefore the force of gravity, to grapple with the other forces of nature too. It has also helped tackle a key problem in quantum computing – and there are glimmers of bigger breakthroughs on the horizon. So let me tell you how I think space-time really works.

These ideas first began to germinate in my mind about 15 years ago, when I was working as a consulting engineer and studying part time for a PhD in machine learning. By day, I travelled all over Europe to visit various companies, often fixing physical machines; by night, I was in a more abstract realm of computers and information processing. It was somewhere between these two worlds, in the weeds of fundamental physics, that I found something interesting.

Before we dive in, I need to say a little about our best fully fledged theory of space, Albert Einstein’s general relativity, and why it is both brilliant and incomplete. In essence, it says that space-time is like a stretchy sheet that is deformed by anything with mass. The resulting curves in space-time create the force of gravity.

Einstein’s theory works wonderfully, but it doesn’t gel with the other great bastion of modern physics, quantum theory. The problem is that the two ideas start off with conflicting assumptions about the nature of reality. In particular, general relativity envisages a smooth space-time, whereas quantum theory says that matter and energy come only in discrete chunks. The most common view among physicists today is that we must find a way to bring gravity into the quantum fold – which means building a theory of quantum gravity.

But back to my own story. Thanks to my work on machine learning, I had begun to think more expansively about how information is stored in brains and computers, and to wonder: what actually is information? That led me to study the physics of quantum information, which insists that information is a physically real thing that can’t be created or destroyed. Imagine tossing a book into a fire. You may not be able to read it any more, but quantum mechanics says the smoke and ash still contain the information, albeit scrambled and dispersed.

All this brought me to another problem that turns out to be crucial to this story. It is known as the black hole information paradox. According to general relativity, anything falling into a black hole crosses the event horizon and disappears from view. We also know that black holes evaporate exceedingly slowly into nothing – and this suggests that the information contained in anything that falls into them vanishes. Except, no: quantum theory insists information can’t be destroyed. We have a paradox.

Space-time’s memory cells

As I mused on this puzzle on planes and trains between my consulting jobs, I started to think we might have missed something about the way space-time stores information. To understand my idea, you first need to know that I assume from the start that space-time isn’t a smooth, continuous fabric, as it is in general relativity, but is instead made of extremely small, discrete cells, like an invisible grid at the deepest level of reality. This isn’t an entirely new idea in itself: many hypotheses that imagine gravity as a quantum force assume space-time is granular. But I build on this by describing how each of these space-time cells can act like a memory unit.

This is, admittedly, a bizarre thought. We are used to information being stored in physical objects with variable properties. Neurons in our brains fire or stay silent; charge builds up and dissipates in computer bits. How can empty space hold information when there is nothing “inside it” to change? The key is to realise that modern physics describes all particles and forces as excitations in quantum fields – mathematical structures that span space and time. Space-time itself is, in principle, no different, and each of my cells of space-time would have a quantum state that can change. Imagine it as like a tiny dial or switch. There is also a more emergent kind of quantum information at play that describes the relationship of each cell to the others – this isn’t held in any one cell, but in the sprawling network of relationships between them.

This is where we return to black holes. When something moves through space-time, it should subtly change the state of all those tiny dials in the space-time cells it interacts with. It is as if space has been imprinted with a memory. And I began to suspect this might offer a way out of the black hole information paradox. Because here is the thing: even when a black hole finally evaporates, its imprint on the space that surrounded it remains. Information doesn’t vanish after all – it has been written somewhere we hadn’t thought to look.

It took me many years to arrive at my solution to this problem – and I didn’t do it alone. These ideas were shaped by long days and nights of conversations with Valerii Vinokur, Eike Marx, Reuben Brasher and Jeff Titus, all colleagues at Terra Quantum, the quantum computing company where I now work. In 2024, my colleagues and I published a paper that describes what we call the imprint operator, a collection of mathematical functions that sets out how information can be imprinted in this way. We also showed theoretically that this mechanism allows space-time to store the information that falls into a black hole.

My collaborators and I began to refer to this idea as the quantum memory matrix (QMM) framework, and we quickly realised it extends beyond gravity. If space-time truly has a memory-like structure, then it should be able to store information from any of the four fundamental forces of nature. Apart from gravity, these are electromagnetism, which governs the physics of light, charged particles and much more, and the weak and strong nuclear forces, which rule over the goings-on inside atoms.

We found that, while the original imprint operator works well for gravity, extending it to describe the strong and weak forces required a more generalised version – not a replacement, but a refinement that accommodates the additional physics those forces involve. And in March, we broadened the framework to include electromagnetism too. All four fundamental forces fit into this unified picture. Each interacts locally with space-time. Each leaves a trace behind.

The fact that QMM can handle all four fundamental forces offers encouragement that this idea might have some real insight. What I like is its power and simplicity. We aren’t postulating new hypothetical particles or unseen dimensions, we are simply taking what we already know about quantum information and packaging it in a new structure. Still, it is a bold idea and it is fair to say that the physicists I have talked to about it have a few critiques. Some question the very notion of space-time having a memory – what is being remembered and how? Others wonder how we would ever test this idea. Still others feel it is just a twist on existing ideas from quantum gravity and doesn’t add anything truly new.

Tests in a quantum computer

It certainly does add something new, and we will get to that. But first I want to tackle the question of testing this idea. The best way to discover whether space-time holds information would be to try extracting it. That may sound like a wild notion, but we already have machines that can read and write quantum information – we call them quantum computers. Our existing quantum computers deal with quantum systems like atoms. Accessing smaller scales tends to require more energy, and getting down to the cells of space-time, which are vastly smaller than atoms, would require a particle accelerator capable of reaching energies a trillion times beyond what’s possible today.

Not something we’re going to pull off any time soon, then. Still, such a test can at least be simulated in an existing quantum computer – and since I work at a quantum technology company, that is exactly what my collaborators and I recently did. We began by taking a qubit, the quantum equivalent of a computer bit, in a known starting state and letting it evolve over time. This evolution was designed to simulate the way a cell of space-time would be imprinted with information as quantum fields wash over it. The question was: could our imprint operator accurately describe the qubit’s evolution?

To test this, we measured the state of the qubit after it had evolved and then applied a reverse version of the imprint operator to see if this would describe the original state. We found that it did indeed do so, with an accuracy of about 90 per cent. This wasn’t just a theoretical toy model. The imprint and retrieval protocols were grounded in QMM’s mathematical structure and translated directly into executable quantum circuits, validating the idea that memory-like behaviour is physically modellable.

You might be tempted to think this is all just meaningless simulation. But the point about simulations in quantum computers is that they involve real quantum states. The fact that the imprint operator works so well in a quantum computer is a strong hint that it could work for cells of space-time too.

As a bonus, our imprint operator turns out to have a practical use. One big problem with modern quantum computers is that information can’t be copied without introducing small errors, and as machines get larger – some machines now have thousands of qubits – these errors mount up and become a serious headache. Last year, Google Quantum AI and Google DeepMind demonstrated a way to clean up errors using artificial intelligence. But our imprint operator offers an alternative. Because it reads and writes data to qubits with such high accuracy, we found that combining our imprinting scheme with standard error-correction techniques reduced errors significantly – by as much as 35 per cent in some configurations – and allowed us to use up to 40 per cent fewer qubits for the same performance. To me, this is another subtle indicator that our QMM framework is on to something.

Dark matter as information

I mentioned earlier that QMM gives us something truly new, so let me now explain what I had in mind. Remember that the curvature of space-time in general relativity is influenced by mass and energy. In our framework, there is an extra ingredient that should also contribute to that curvature: the weight of information woven into space-time.

Astronomers already know that the gravity of many galaxies seems to be stronger than would be expected based on their mass and rate of rotation alone. Lacking an explanation, they have invented a substance called dark matter to account for the difference. However, no one knows what it might be. But perhaps my collaborators and I have stumbled upon the answer: could dark matter be information, stored across space-time in a way that generates gravitational pull? I think so. When we have run calculations to compare the theoretical gravitational effect of information and the observed effects of dark matter, the numbers more or less match.

One thing I remain curious about is just how good space-time’s memory is. In other words, how far back in history does it reach? My suspicion is that the whole of cosmic history is, in some sense, baked into space. After all, we know information cannot be destroyed. Admittedly, this isn’t something I can yet claim with any confidence. But I certainly have a much clearer vision of information’s role in the cosmos than I once did. I started this journey years ago with a question I couldn’t resolve, but now I am beginning to get solid answers – not just hand waving.