How we rate hype
Every rating on this site comes from the table below. We record which signals are present and quote the evidence; the arithmetic is fixed in code. You can check our working on any story page.
Negative points push a story toward Low hype, positive toward High. A total of -3 or below is Low; 3 or above is High; anything between is Medium.
This is a rubric, not a verdict on anyone's integrity. A High rating means the claim outran the evidence available to us at the time - not that the underlying work is bad.
| Points | Signal | What it means |
|---|---|---|
| -2 | Peer-reviewed publication exists | A journal version exists, not just a preprint or a blog post. |
| -2 | Independently replicated | A group unaffiliated with the authors has reproduced the central result. |
| -2 | Reports logical, not physical, qubits | Headline count is error-corrected logical qubits rather than raw physical qubits. |
| -2 | Headline does not exceed the paper | The press claim is within what the accompanying paper actually states. |
| -1 | Standard benchmark used | Randomized benchmarking, quantum volume, CLOPS or similar - not a bespoke metric. |
| -1 | Shipped, not a roadmap milestone | Hardware or software customers can actually access today. |
| +1 | Public company, announcement precedes earnings | Timing alone is not evidence of bad faith, but it is worth flagging. |
| +2 | "Quantum advantage" or "supremacy" in the headline | Claims that have repeatedly been walked back by later classical algorithms. |
| +2 | No error rates disclosed | A qubit count without gate and readout fidelities is close to meaningless. |
| +2 | Weak classical baseline | Compared against an unoptimized classical method, or none at all. |