CryptoArticle is online

Google’s Willow Chip Surpasses Classical Limits in a Quantum Coordination Game

數字匠人
Google’s Willow Chip Surpasses Classical Limits in a Quantum Coordination Game

Table of Contents




You might want to know



  • How did Google’s Willow processor beat the classical limit in the Mermin-Peres magic square game?

  • What do the larger 71-qubit and 105-qubit tests show—and what do they not prove?



Main Topic


Google reported that its Willow quantum processor won a mathematical coordination game 98.3% of the time, exceeding the 88.9% ceiling for classical strategies. The findings appeared in a study published in Science on Thursday. The result concerns quantum contextuality: the idea that the outcome of a measurement can depend on which other measurements are made alongside it.


In the researchers’ words, “you cannot assign fixed, pre-existing values to everything you might measure.” That distinction is central to understanding the experiment. It describes a feature of quantum theory and a particular kind of game, rather than showing that a quantum processor can already perform every task faster or better than a classical computer.


The 98.3% result came from a test using four qubits; it was not the outcome of the separate 105-qubit benchmark. The larger experiment addressed a different, more limited mathematical task. Neither result establishes that Google has a commercial quantum computer or a fault-tolerant machine.



The first test used the Mermin-Peres magic square game. In this game, two players must fill intersecting parts of a three-by-three grid without communicating with one another. Their answers must satisfy rules for the relevant rows and columns, even though those rules conflict in a way that makes the task challenging. The players must also agree on the value in the square where their assigned parts intersect.


If the players rely on a prearranged classical strategy, they can succeed on no more than eight of nine assignments, equivalent to 88.9%. Quantum theory allows entangled qubits to win every round in principle. In the experiment, Willow won 98.3% of 900,000 trials, according to the researchers. They wrote that “our qubits coordinated better than any pre-written table allows.”


The measured result fell short of the theoretical quantum ideal. The researchers attributed this difference to unavoidable noise in the physical hardware. In practice, quantum processors are affected by errors, so the behavior of a real device does not perfectly match the predictions for an idealized system.



Google also measured contextuality using a numerical score. A model in which all outcomes are predetermined can score at most 4, while quantum theory allows a score of 6. Willow registered 5.590, with a reported uncertainty of 0.008. This measurement provides another way to examine whether the processor’s observed behavior is consistent with contextual quantum effects.


These scores should be understood in the context of the test they measure. They do not provide a general rating of the processor’s usefulness, nor do they directly demonstrate that it can solve practical problems better than classical computers. Instead, they indicate how the device performed on specific experiments designed to probe quantum behavior.



Scaling up, with caveats


Google’s other experiments involved more qubits, but tested different challenges. In a separate GHZ parity game, as many as 71 entangled qubits had to produce coordinated binary answers. The performance was compared with the strongest classical winning rate, which approaches 50% as the number of players increases. Google said its processor remained above that classical limit throughout the tests, even as hardware noise accumulated.


The researchers also verified genuine entanglement across 45 qubits. Entanglement is a quantum relationship between systems that cannot be fully described by treating each system independently. Demonstrating it across multiple qubits is relevant to understanding how quantum devices behave as their size increases, although it does not by itself establish reliable, large-scale computation.


In another experiment, the hidden linear function task, a shallow quantum circuit had to produce a binary string that satisfied mathematical constraints. Willow reached roughly 60% success using 105 qubits. This benchmark relied on a conjectured classical bound, rather than a proven limit that rules out every possible classical approach. It should therefore be distinguished from the four-qubit game, whose comparison is with the stated 88.9% classical ceiling.


The 105-qubit task was a shallow mathematical circuit, not a demonstration of a fault-tolerant quantum computer. The largest instance still failed in about four out of 10 runs. That outcome illustrates why qubit count alone is not enough to establish practical capability: reliability, error correction, circuit depth, and the difficulty of the task all matter.



Google says these experiments can reveal small, systematic errors that conventional randomized tests may fail to detect. The company described contextuality testing as “a sensitive new method to catch subtle hardware errors.” If developed further, such tests could help engineers evaluate circuits intended for areas such as chemistry and materials research.


The work follows Google’s October report of a 40-fold reduction in errors when preparing magic states. Magic states are a resource needed for fault-tolerant quantum computing. The earlier result and the new experiments address different parts of the broader challenge: improving device operations and testing whether hardware displays the expected quantum behavior.


Fault tolerance matters because useful quantum computations would need to remain reliable despite the errors that affect physical qubits. Researchers generally expect this to require methods for detecting and correcting errors, along with substantial improvements in how quantum systems are controlled. A result showing a quantum advantage on a specific game is therefore an important research finding, but it is not equivalent to demonstrating a broadly useful, error-corrected computer.


The implications for Bitcoin and other digital systems also require careful framing. A future fault-tolerant quantum computer could threaten digital signatures, which help secure transactions and verify ownership. However, the Willow experiments described here do not show that such a machine exists or that current signatures can be broken. The 105-qubit benchmark was a shallow mathematical task, and its performance was imperfect.


More broadly, the findings show how researchers can use carefully designed games and mathematical tests to compare quantum behavior with classical strategies. The strongest conclusions are specific: Willow exceeded a defined classical ceiling in a four-qubit coordination game, achieved a contextuality score of 5.590, and completed separate tests involving 71 and 105 qubits. The results are evidence of progress in experimental quantum computing, while the limits of each benchmark remain essential to interpreting that progress.



Key Insights Table



































Aspect Description
Magic square game Willow used four qubits and won 98.3% of 900,000 trials, above the 88.9% classical ceiling.
Contextuality score Willow scored 5.590, with a reported uncertainty of 0.008; predetermined-outcome models score at most 4, while quantum theory allows 6.
GHZ parity game A separate experiment involved as many as 71 entangled qubits and remained above the strongest classical winning rate, which approaches 50% with more players.
Hidden linear function task A shallow circuit reached roughly 60% success with 105 qubits; the comparison uses a conjectured classical bound.
Practical limits The 105-qubit result does not establish fault tolerance, and the largest instance failed in about four out of 10 runs.
Potential research value Contextuality tests may expose subtle hardware errors and support evaluation of circuits for chemistry and materials research.


Afterwards...


The next steps for quantum computing research include improving qubit reliability, developing effective error-correction methods, and determining how experiments can scale without losing control of noise. Researchers will also need to clarify which benchmark comparisons are proven and which depend on conjectures, so that results can be assessed consistently.


Further work on contextuality tests could help engineers identify systematic hardware errors that ordinary randomized checks miss. At the same time, practical applications in chemistry and materials research will require circuits that are not only large enough to model meaningful problems, but also dependable enough to produce useful answers repeatedly.


Willow’s results are best viewed as a focused advance in measuring and testing quantum behavior—not as proof that fault-tolerant computing has arrived. Continued progress will depend on connecting these experimental demonstrations to reliable computation, while maintaining a clear distinction between a quantum advantage on a defined task and a general-purpose technological breakthrough.

Last edited at:2026/11/22

數字匠人

Author

數字匠人

Idle Passerby

A easy-going and enjoys walking in the scenery of life. He is also an observer and sees the world with a calm heart.