Is information swallowed by a black hole lost forever, or can it somehow be recovered? For half a century, this question has challenged physicists. In string theory and quantum gravity, researchers have studied theoretical models that link two black holes through a spacetime tunnel known as a "wormhole."
In quantum mechanics, information is thought to be conserved in the time evolution of a closed system. That means even quantum information that has been thoroughly scrambled inside a black hole might, under the right conditions, be recoverable somewhere else. Inspired by this idea, researchers have been trying to reproduce wormhole-based information transfer with quantum circuits.
Now, theoretical physicists Sudhanva Joshi and Sunil Kumar Mishra of the Indian Institute of Technology (BHU) Varanasi have mathematically analyzed a quantum teleportation scheme inspired by this "traversable wormhole" and found a surprising result.
Information can be transferred with exactly the same average accuracy without "quantum magic"—a resource considered important when quantum computers perform calculations that are hard for conventional computers—as with an abundance of it.
Put simply, quantum magic is the property that makes a quantum computer's calculations hard to reproduce on an ordinary computer. It has long been regarded as playing an important role in advanced quantum information processing.
However, equal average transfer accuracy does not necessarily mean equal stability of results. In a numerical check using systems of 8 qubits on each side, the research team found that for operations without quantum magic, the variance, which measures how much the transfer accuracy fluctuates, was roughly 40,000 times larger.
The results are reported in a preprint (arXiv:2606.19180v3), revised on October 8, 2026, and were covered on October 13 by ScienceAlert.
To be clear, no information was actually transferred using a real black hole. Nor was this an experiment on a real quantum computer; it is a mathematical physics study of idealized quantum systems. The paper is also still a preprint that has not yet been peer reviewed.
Still, the result sheds new light on a fundamental question: what is truly needed to transfer quantum information accurately?
Quantum teleportation born from a black hole thought experiment
Unlike the teleportation of science fiction, quantum teleportation does not move matter instantaneously. Nor can it transmit information faster than light.
Quantum teleportation combines quantum entanglement, shared in advance between sender and receiver, with ordinary communication to reproduce the unknown state of one qubit on another qubit.
Quantum entanglement refers to a state in which multiple quantum systems in separate locations share strong correlations that cannot be described individually.
This mechanism allows the original quantum state to be transferred elsewhere, but the state at the source is destroyed by measurement and other operations. This follows the "no-cloning theorem" of quantum mechanics, which says an unknown quantum state cannot be perfectly copied.
Ordinary quantum teleportation uses an operation called a Bell measurement together with classical communication.
Black hole-inspired teleportation, by contrast, starts from the question of how quantum information spreads through a complex many-body system and how it can be extracted again.
A leading example is the teleportation protocol using a traversable wormhole based on the Sachdev–Ye–Kitaev (SYK) model, proposed in 2019 by Ping Gao and Daniel Louis Jafferis.
That work was published in the Journal of High Energy Physics in July 2021.
The SYK model is a theoretical model describing quantum systems in which many particles interact in complex ways, and it is used to study quantum chaos and the properties of black holes.
In Gao and colleagues' model, two quantum systems, left and right, are linked by an entangled state called the thermofield double state (TFD).
Under a particular correspondence with gravity theory, this state can be interpreted as two black holes connected by a wormhole. In general the TFD state is entangled, and in the infinite-temperature limit it becomes maximally entangled.
When quantum information is fed into one side, complex interactions spread it throughout the whole system.
This phenomenon is called scrambling.
Consider, for example, information recorded in a single qubit becoming dispersed among the complex correlations of many qubits.
The information has not vanished, but it can no longer be recovered by examining just a few qubits.
Gao and colleagues' theory showed, however, that adding a suitable interaction between the left and right systems allows the information to be transferred to the opposite side.
On the gravity side, this interaction is interpreted as producing an effect of negative energy that makes a normally impassable wormhole temporarily traversable.
From the quantum information perspective, it is a mechanism by which complexly dispersed information is restored elsewhere through a suitable interaction and decoding operation.
Gao and colleagues presented a protocol that can achieve nearly perfect fidelity under certain conditions.
What Joshi and Mishra examined was how much advanced quantum computational power the information-scrambling operations need in order to realize such black hole-inspired teleportation.
At the center of their investigation was a property called "quantum magic."
What is "quantum magic"? How it differs from ordinary computers
The name might suggest some special power of quantum computers. But it is not a supernatural phenomenon; it is a mathematical concept describing a property of quantum computation.
To understand it, one first needs to know that not every calculation on a quantum computer is harder than on a conventional computer.
A quantum computer proceeds by applying operations called quantum gates to qubits.
There are many kinds of quantum gates, and one family is known as Clifford gates.
Typical examples include the Hadamard gate, the CNOT gate, and the phase gate.
Combining them can entangle multiple qubits and create complex quantum states.
However, it is known that certain quantum circuits built only from Clifford gates can have their behavior efficiently reproduced even on ordinary computers.
This is shown by the Gottesman–Knill theorem.
Clifford gates are operations based on quantum mechanics, but the changes they produce follow mathematical regularities. This allows conventional computers to track the course of the computation efficiently.
So what is needed to realize the broader range of calculations that only quantum computers can do?
That is where non-Clifford gates come in.
The T gate is a representative example.
Adding operations that cannot be realized with Clifford gates alone widens the range of computations a quantum computer can handle.
At the same time, it becomes hard for an ordinary computer to efficiently follow the course of the computation.
In quantum information science, this property that cannot be reproduced by Clifford gates alone is called quantum magic, or non-stabilizerness.
As an everyday analogy, a computation using only Clifford gates is like a puzzle that, however complicated it looks, can be organized step by step according to fixed rules.
Adding non-Clifford gates turns it into a more complex puzzle that those rules alone cannot easily sort out.
Quantum magic is an important resource for enabling the latter kind of computation.
However, the amount of quantum magic is not a score that directly expresses a quantum computer's speed or accuracy.
More quantum magic does not necessarily make a computation faster, and less does not necessarily reduce accuracy.
For example, creating entanglement itself does not necessarily require quantum magic. The properties of quantum computation needed differ depending on the task to be performed.
This was precisely the point examined in this study.
Is quantum magic really needed to scramble quantum information as a black hole does, and then recover it?
To investigate, the team compared two types of operations.
One is the "Clifford scrambler," which has no quantum magic. The other is the "Haar-random scrambler," which picks more general quantum operations at random.
Haar-random scramblers typically carry abundant quantum magic.
Both can scramble quantum information, but they differ greatly in how hard it is to reproduce their behavior on a conventional computer.
The team built both types of operation into a black hole-inspired teleportation circuit and compared how accurately information could be recovered.
Average transfer accuracy is the same even without quantum magic
The team mathematically analyzed the behavior of the teleportation circuit under the idealized condition of infinite temperature.
They showed that the average fidelity of the Clifford scrambler, which has no quantum magic, and that of the Haar-random scrambler, which is rich in it, match exactly.
"Fidelity" here is a measure of how closely the quantum state recovered at the destination matches the original state.
A fidelity of 1 means, ideally, a perfect match.
This comparison does not concern the result of running a single operation, but the average performance when many operations are drawn from each approach.
In other words, using operations that are easy to reproduce even on a classical computer and using complex operations that are normally hard to reproduce can transfer information with the same accuracy on average.
This result shows that the simple relationship, in which more quantum magic means more accurate information transfer, does not hold.
Why the average fidelities match
The reason lies in the statistical properties of the two types of operation.
According to the team's analysis, the average fidelity of this teleportation protocol is determined by information up to the "third moment" of the scrambler.
The third moment is one statistical feature obtained when operations are drawn many times.
It is mathematically known that the set of Clifford operations matches Haar-random operations in its statistical features up to the third order.
This property is called a unitary 3-design.
The average fidelity of this teleportation was determined entirely by information within that matching range.
As a result, the average fidelity came out the same even though the amounts of quantum magic differ greatly.
This is not a coincidence found in numerical calculations but a mathematically derived result. Section 3 of the original paper rigorously proves the relationship.
The team further showed that adding non-Clifford gates to a Clifford circuit to increase quantum magic does not change the average fidelity.
Note, however, that the zero quantum magic refers essentially to the Clifford operations that scramble the information.
The full teleportation circuit also includes interactions coupling the left and right systems. It does not mean the quantum magic of the entire circuit is necessarily zero under arbitrary conditions.
Same average, but a roughly 40,000-fold difference in stability
Equal average transfer accuracy does not mean success every time in the same way.
For example, if you score 80 on each of five tests, your average is 80.
But you could also score 100 on one test and 60 on another, with uneven results, and still average 80.
Even with the same average, consistently scoring 80 every time is different from swinging widely.
Exactly this happened in this quantum teleportation.
The team examined not only the average fidelity but also the "variance," which expresses how much the results of individual operations scatter.
They found a large difference between Clifford scramblers and Haar-random scramblers.
With Haar-random scramblers, the spread in fidelity shrinks rapidly as the system grows.
That is, even an individual randomly chosen operation yields relatively stable results.
With Clifford scramblers, on the other hand, the spread also decreases as the system grows, but not as fast as with Haar-random operations.
According to Section 5 of the paper, numerical calculations with 8 qubits on each side gave a fidelity variance of 0.0208 for the Clifford scrambler, roughly 40,000 times that of the Haar-random scrambler.
This "40,000 times" does not mean the transfer failure rate is 40,000 times higher. It is simply a comparison of how far the fidelity obtained from individual operations scatters from the mean.
The main differences are summarized below.
| Comparison | Clifford scrambler | Haar-random scrambler |
|---|---|---|
| Quantum magic | None | Generally abundant |
| Reproduction on an ordinary computer | Can be efficiently reproduced | Generally hard to reproduce |
| Average information transfer accuracy | Same as Haar-random operations | Same as Clifford operations |
| Stability of transfer accuracy | Fluctuates relatively widely depending on the operation | Very small fluctuation |
| Fidelity variance at 8 qubits per side | About 40,000 times Haar-random | Baseline of 1 |
| As the system grows larger | Typical fidelity approaches 1 | Fidelity approaches 1 and fluctuation shrinks rapidly |
This comparison shows that even without quantum magic, high accuracy can be achieved on average, but there are large differences between operation types in how consistently results are obtained each time.
However, this difference cannot be explained by the amount of quantum magic alone.
In Clifford operations, the changes made while scrambling information are highly regular. As a result, how information spreads and the transfer outcome vary fairly widely depending on which operation is chosen.
In Haar-random operations, by contrast, information is dispersed in a complex way across many components. Because diverse components are already mixed within each individual operation, differences in results between operations tend to be small.
In other words, the two reach the same average fidelity through different mechanisms.
The team also showed that as the model is scaled up, the typical fidelity approaches 1 even for the Clifford scrambler.
Under certain conditions, the quantum magic of the whole circuit decreases in inverse proportion to system size while the average fidelity approaches 1.
This means highly accurate information transfer is possible without increasing quantum magic.
Still, differences remain in how the fidelity fluctuates. Achieving high accuracy on average and obtaining stable performance from individual circuits need to be evaluated separately.
What matters is how information is recovered, not how much quantum magic there is
If the amount of quantum magic does not determine transfer accuracy, what does?
In Joshi and Mishra's analysis, the relationship between the scrambler, which scrambles the information, and the decoder, which restores the scrambled information, played an important role.
Once scrambling spreads information across many qubits, the original information can no longer be easily extracted.
But if one knows how the information was scrambled and performs a corresponding operation, it may be possible to restore it.
It is like decrypting information that was encrypted by a complex method using the correct key.
However advanced the method used to process the information, the necessary information cannot be extracted unless the operation to reverse it is appropriate.
The same tendency was confirmed in this study.
The team showed that when a decoder unrelated to the scrambler is used, the average fidelity stays at the "no information transfer" baseline even with operations rich in quantum magic.
In this model, that baseline is 1/4.
In other words, even if you scramble information intricately using a great deal of quantum magic, you cannot transfer it successfully without a corresponding recovery method.
What matters is not just how intricately information can be scrambled, but the consistency of the whole sequence of operations: how it is scrambled and how it is restored.
The team also showed that adding quantum magic to parts of the circuit away from the information being transferred may leave the statistical properties of the fidelity unchanged.
This means that the amount of quantum magic contained in the entire circuit alone cannot properly evaluate information transfer performance.
The difference in variance between Clifford and Haar-random operations confirmed here is also not determined merely by the amount of quantum magic.
It reflects differences in the higher-order statistical properties of each type of operation.
Quantum magic is an important resource for letting quantum computers perform advanced computation. But for specific information transfer like this, it is not the only thing that determines performance.
How the paper's revision changed its conclusion on whether quantum magic is needed
The revision history of the paper also deserves attention.
The arXiv paper numbered 2606.19180 was first submitted on June 17, 2026.
Its original title was "Quantum magic is necessary but not sufficient for wormhole-inspired teleportation."
The first version examined changes in quantum magic using an indicator called stabilizer Rényi entropy (SRE), targeting wormhole-inspired teleportation using the SYK model.
Its analysis concluded that what is needed for information transfer is not simply generating large amounts of quantum magic, but changes and redistribution of quantum magic with an appropriate structure.
However, the third version, published on October 8, substantially changed the paper's structure and central conclusion.
The current title is "Clifford and Haar scramblers yield equal mean fidelity but unequal fluctuations in black hole-inspired teleportation."
The revised version shifts the analysis from arguments dependent on the specific SYK model to a more general quantum information theory framework.
Specifically, it compares the statistical properties of Clifford and Haar-random operations and mathematically clarifies the conditions under which average fidelities coincide and the mechanism by which differences in variance arise.
Whereas the first version emphasized the necessity of quantum magic, the current version shows that, at least for the average fidelity of the protocol studied here, quantum magic in the scrambler is not essential.
However, this does not mean it has been proven that quantum magic is unnecessary under every condition of the SYK model treated in the first version.
The current version re-examines the conditions needed for quantum information transfer in a form that does not presuppose a correspondence with gravity theory.
The research reported by ScienceAlert is also based on this revised version dated October 8, 2026.
How this differs from wormhole simulation on quantum computers
In evaluating this result, it is necessary to distinguish conclusions obtained from mathematical models from results verified on real quantum computers.
Joshi and Mishra's study analyzes the properties of quantum circuits using idealized conditions such as infinite temperature.
It shows that even scramblers without quantum magic can achieve high average fidelity, but it has not confirmed that the same performance would be obtained on actual hardware.
On real quantum computers, decoherence, in which interactions with the environment disturb quantum states, as well as gate operation errors and measurement noise, all occur.
Even a circuit that can achieve high fidelity in theory may see its performance drop sharply because of such errors.
There is already an important precedent for imitating wormhole behavior on a quantum computer.
In November 2022, a research team including Daniel Jafferis and Maria Spiropulu announced experimental results imitating the behavior of a traversable wormhole using Google's quantum processor Sycamore.
The results were reported in "Traversable wormhole dynamics on a quantum processor," published in the December issue of Nature that year.
Using machine learning and other methods, the team built a simplified SYK model and reproduced its behavior with a circuit of 9 qubits and 164 two-qubit gates.
The experiment observed characteristic information-transfer behavior corresponding to a signal passing through the wormhole.
However, no wormhole was actually created in real spacetime. It merely imitated the behavior of the corresponding quantum system with a circuit built on a quantum processor.
The experiment was also highly sensitive to noise.
An explanation by the Google research team says that if the noise had been 1.5 times greater than it actually was, the observed signal would have become completely invisible.
Even a slight worsening of quantum gate error rates could bury the faint information-transfer signal in noise.
This earlier work shows how hard it is to run quantum circuits inspired by black hole theory on real hardware.
The study by Joshi and colleagues, by contrast, does not evaluate performance with the errors and decoherence specific to real devices taken into account.
How far the mathematical conclusions obtained under the idealized condition of infinite temperature hold at finite temperature and in noisy real-device environments remains a question for future verification.
The team itself positions the result as a theoretical foundation for verifying the protocol on quantum hardware in the future.
Recovering quantum information takes more than complex computation
The key finding of this study is that, in black hole-inspired quantum teleportation, the amount of quantum magic and information-transfer performance do not correspond in a simple way.
Even a Clifford scrambler with no quantum magic can, under appropriate conditions, achieve exactly the same average fidelity as a Haar-random scrambler.
The study further showed that as the system size grows, the typical transfer fidelity approaches 1 even for circuits with almost no quantum magic.
On the other hand, large differences remain in making the performance of individual circuits stable.
With 8 qubits on each side, the fidelity variance of Clifford operations reached roughly 40,000 times that of Haar-random operations.
And if the scrambler that scrambles the information and the decoder do not properly correspond, high average fidelity cannot be obtained even with abundant quantum magic.
In other words, achieving information transfer requires not just making quantum computation more complex but understanding how information spreads and preparing a recovery method that matches it.
Such insights could become a clue for future experimental studies of black hole information scrambling and quantum information recovery using circuits that do not require large amounts of quantum magic.
However, a circuit with little quantum magic is not necessarily an easy one to implement. The number of gates needed for high-precision operation and resistance to noise must also be considered.
The paper is also an unreviewed preprint, and the validity and scope of the theory will need to be verified by future research.
In particular, how well the spread in fidelity can be controlled on real quantum hardware, and how consistency between scrambler and decoder can be maintained, are likely to be important challenges.
The question of what happens to information that falls into a black hole, born from the extreme environments of the universe, is now being studied as a concrete computational problem in quantum information science.
The result suggests that what quantum information recovery requires is not simply advanced, complex quantum operations, but an appropriate mechanism linking the spreading of information to its decoding.