Constructing a Theory of Everything

Just what is a fundamental physics theory supposed to do?

For most of the history of physics, the answer has been to predict the future. Take the state of a system now and then use the theory to calculate its later state. But a quiet revolution is underway, one that could open the door to answering some of the biggest outstanding questions in physics and beyond.

Constructor theory offers a radical new approach to tackling foundational questions. In its modern form, it has largely been developed by Oxford physicists David Deutsch and Chiara Marletto, not as a theory of what happens, but of what can and cannot happen. That's a subtle, yet profound shift in perspective. "It's meant to be the most fundamental way of phrasing everything we have in physics—like a theory of everything, ultimately," says Vlatko Vedral, who joined Deutsch and Marletto a little later to investigate the framework.

Instead of starting from initial conditions and laws of motion, as physics traditionally does, constructor theory starts from statements about which tasks are possible and which are impossible for constructors to perform. It serves as a meta-theory that then treats everything else—including the old-style laws of motion—as something that should emerge from those more fundamental statements.

Marletto, Deutsch and Vedral were invited to author a new paper, "Tests of Constructor Theory," to launch FQxI's forthcoming Foundational Review Series. In the paper, the trio argues that re-framing physics in terms of 'constructors' could help us in the quest to unite quantum theory and gravity, understand thermodynamics and even the evolution of life. Crucially it has also produced a growing list of genuine, checkable experimental proposals. Some have already been run; others are queued up for the next generation of quantum hardware to test.

Self-copying Machines

Although Deutsch proposed the modern version of constructor theory in 2012, the lineage of constructor theory is far older. "Constructor theory itself goes back to John von Neumann," Deutsch says, referring to the twentieth-century Hungarian pioneer of quantum theory, computer science and many other fields. "He wasn't trying to make a new foundation for physics, but trying to understand how it is that living processes can be instantiated in physical systems."

Von Neumann formalised this idea in what's now called the replicator-vehicle architecture—a self-copying machine paired with its own blueprint. Von Neumann called his self-reproducing machines "constructors," which is where the theory's name comes from.

In its modern formulation, a constructor is any system that can perform a specific physical task while remaining able to perform it again. A task here just means a transformation from one physical state to another. Deutsch defines the constructor more precisely as something that "takes an input, and produces some output and possibly some waste." But crucially nothing else.

"Rather than acknowledge that we may be not able to address certain questions because it requires a detailed knowledge of initial conditions, which we don't really have, we should be thinking more in terms of principles," says Vedral.

In a way, physicists are already used to this approach; they have been invoking it implicitly for over a century. "When you have a principle like the conservation of energy, it's a statement that puts constraints on what can happen in an actual experiment," says Marletto. Theories are discarded if they don't adhere to energy conservation laws or the sacred Second Law of Thermodynamics.

Constructor theory's possible/impossible statements work similarly in that they're falsifiable in exactly the way conservation laws are, even though no single experiment confirms or denies the principle directly.

Testing Gravity's Quantumness

While this all sounds great, is there reason to believe that constructor theory is correct? Even if it is, does reframing physics in this way actually help us understand the world better?

The trio's answer is a resounding yes to both questions. Subjecting constructor theory to experimental tests could have far-reaching consequences, they argue, including making inroads into the seemingly intractable problem of uniting quantum with gravity. "Chiara and I started discussing this a long time ago, probably 2014 or 2015," Vedral recalls. By then it had long been established that two particles in the lab can become entangled, in such a way that measurements performed on one influence its twin. Often the twin particles are created together in the lab to generate entanglement or they are manipulated with laser pulses to entangle them. But Vedral had an intriguing alternative idea. "I thought it would be interesting to come up with a situation where entanglement is generated between material objects, but the only force that's allowed to act between them is gravity," Vedral recalls. This discussion eventually led to one of the most promising tests of constructor theory, in the form of the proposed Bose-Marletto-Vedral (BMV) experiment, which asks whether gravity can quantum entangle two masses (see walkthrough below).

The setup is, in effect, a quantum version of the classic eighteenth-century Cavendish experiment that first measured gravity's strength, using a torsion balance: two small lead spheres were attached to the end of a wooden rod, which was suspended by a thin wire, while two large weights were positioned nearby, attracting the small spheres gravitationally and causing the wire to twist.

In the BMV experiment, the masses are dramatically smaller, which means quantum effects come into play. "We are talking about masses that are pretty close to the mass of a biological cell," Vedral says. One such quantum effect is superposition, which allows each mass to be held in a two locations at the same time—something no one has ever done before.

A big mystery about gravity is whether is can be described as a quantum force, like electromagnetism and the strong and weak nuclear forces. The team hopes that the BMV experiments would reveal if gravity does indeed act in a superposition, as any quantum force ought to do. Or could it be that somehow gravity collapses these superpositions and still acts in a classical way? "If entanglement appears it would really signal that there must be something quantum in gravity," Vedral says.

The BMV experiment provides a cunning way to tease out how gravity and quantum effects interact. But it's not immediately clear what any of this has to do with constructor theory.

The subtlety comes in what the constructor-theoretic framing buys you if entanglement is observed. Without constructor theory, theories of gravity that predict no entanglement would have to be ruled out one by one. Instead, Marletto has used constructor theory to define classical information processing purely in terms of which tasks are possible or impossible if the system obeys classical rules. In other words, it isn't tied to any specific model of gravity. This 'General Witness Theorem' allows a much stronger claim: no theory in which the interaction between the two masses is classical—known or yet to be invented—can produce entanglement. It makes the argument that gravity is non-classical watertight against every possible classical alternative, not just the ones physicists have already thought to write down.

The price, as Vedral notes, is that this generality cuts the other way too. "You can rule out a lot of theories, you can confirm a lot of theories as well," he says. "However, you cannot discriminate between them." In other words, the General Witness Theorem tells you classical-vs-quantum, not which quantum theory of gravity is correct. But even that would be a big breakthrough.

Fabrizio Placentini, from the National Metrology Institute of Italy, agrees that the theory's far-reaching approach—usually seen as a strength—could be a potential problem. "When dealing with such a broad framework, the initial difficulty is restricting ourselves to a scenario with verifiable predictions," he says. "My belief is that, currently, the theoretical investigations on constructor theory are still on way too abstract ground to provide specific predictions in actually realisable tests."

Measurable Signatures

It is true that a full-strength quantum Cavendish-like experiment is yet to run. It needs masses held in superposition for longer—and to be better isolated from noise—than today's technology allows. But there has been encouraging progress. For instance, physicist Jonathan Jones, also at Oxford, has run a proof-of-principle version, which shows that the signatures constructor theory predicts are measurable in principle, in a quantum computing set up that uses nuclear magnetic resonance qubits (preprint).

There are other tests of constructor theory brewing. A second front concerns thermodynamics and the mystery of the arrow of time. The puzzle to be solved is that the underlying laws of physics are time-reversible, yet we see irreversible processes everywhere. Marletto's resolution rests on what it means for a constructor to work in a cycle— it must end up back in exactly the same state it started in, ready to perform the same task again. That's a much stricter requirement than simply asking whether a process is allowed by the equations of motion.

Marletto illustrates this with a glass of water. "A constructor can warm the glass by mechanical means only, for example stirring," she says. It ends the cycle unchanged, ready to stir again. But run the same physics backward and there's no equivalent device. "No entity can work in a cycle, cool down the water by stirring and have no other side effects," Marletto says.

Reversing the dynamics is mathematically permitted, but building a constructor that reliably brings it about is not. That gap is the whole point. A forward constructor existing, plus time-reversal symmetry in the underlying laws, doesn't imply a backward one exists too. The asymmetry lives in what's constructible, not in the laws themselves. This matters because it changes the status of the second law itself. The usual view treats it as a statistical tendency—one that is true on average, but not exact, since nothing rules out a freak fluctuation running the other way.

Deutsch rejects that statistical framing. "There is a sort of prejudice in physics that thermodynamics can only be approximate, it can't be exact," he says. "But in constructor theory, you can in fact state exact analogues of the second law." Because the law is now a statement about what's constructible rather than what's statistically likely, it holds with the same exactness as the conservation of energy.

This isn't just theoretical. "We actually have an experiment that was performed a few years ago by a group in Turin that demonstrated this directly," Marletto says (preprint).

Reconstructing Life's Origins

The most ambitious—and least experimentally mature—branch of Constructor Theory goes back to its origins in replication and explores what physics requires for life to exist at all.

Marletto argues von Neumann's replicator-vehicle architecture isn't just sufficient for life, but necessary. "You will find it like that anywhere in the universe, if there is life anywhere," she says. Life on worlds as diverse as Earth, Saturn's moon Enceladus or a distant exoplanet, should all play by the same constructor-theoretic rules.

It's clear that constructor theory has far-reaching potential. What makes this moment interesting is not that the theory has been vindicated. After all, none of its central, most novel claims have been confirmed or refuted yet. What's new is that a framework once criticised as too abstract to engage with experiment now has a concrete set of testable proposals, several already run in simplified form.

The BMV gravity test feels like a genuine fault line. A positive result would be one of the few clean experimental handles physics has ever had on quantum gravity. Whether or not constructor theory itself survives in its current form, with its distinction between what's possible and impossible, it is now asking questions that evidence-based physics might have answers to.

Lead image: David Deutsch, Chiara Marletto, Vlatko Vedral. Banner art and BMV walkthrough credit: ©FQxI/Gabriel Fitzpatrick (2026). BMV walkthrough based on art from the Conjecture Institute.