Coverage

Gaps 6

1. 769 of 1,264 sources are the operator describing itself

   Only 59 are independent. A company's own site is evidence of what it says, not of whether it is true. This is the biggest weakness in the dataset, and it is structural: operator sites are what exists publicly, and independent checks mean peer-reviewed papers and regulator dockets, one record at a time.

   What would close it: Peer-reviewed sources per claim. NIF is the worked example — press said 3.15 MJ, the published paper said 3.1 ± 0.16.

2. 59 of 733 figures verified

   Verified means the source was opened in a verification run and confirmed the figure. Everything else is recorded faithfully from its source but has not been independently checked by us.

   What would close it: Verification passes.

3. No organisation record is verified as a whole

   256 of 477 have their name, website and summary checked against the archived page instead. That check does not confirm classification (sector, approach, fuel, country, founding year), because nothing on a company's own homepage shows that we classified it correctly.

   What would close it: A domain agent reviewing classification against evidence, not more automation.

4. 19 of 36 registered metrics are in use

   The other 17 are modelled and empty; they are listed below. Triple product has 27 figures, and 1 is promoted to achieved, because the same-plasma condition is hard to confirm from public sources.

   What would close it: Primary-source capture per device. This is the work that would make the benchmark better than Wikipedia rather than equal to it.

5. 350 of 350 vocabulary terms have a written definition

   Undefined terms are never published in the glossary — a label the site cannot explain is a classification it has not earned. Every term currently attached to a record is defined.

   What would close it: Writing them. Domain terms need a domain author, not a template.

6. 66 of 307 facilities hold no figure at all

   49 are fusion devices and 17 are fission reactors. Where the source is the IAEA survey, its general-information table carries only type, status and ownership, with no radius, field or pulse length. The emptiness is what the source says, not a failure of extraction.

   Facilities with no figure, by recorded approach

   Approach	Facilities
   Magnetic mirror	9
   Field-reversed configuration	7
   Inertial confinement	6
   Fission, reactor type not established	4
   Spherical tokamak	4
   Z-pinch	4
   Tokamak	4
   Sodium fast reactor	3
   Pressurised water reactor	3
   Magnetic confinement	3
   Stellarator	3
   Not classified	2
   Water-cooled	2
   Reversed-field pinch	2
   Magnetized target	2
   Fusion, approach not established	2
   Fluoride salt-cooled high-temperature reactor	1
   Boiling water reactor	1
   Molten salt reactor	1
   Projectile-driven	1
   Levitated dipole	1
   Inertial electrostatic confinement	1

   What would close it: Operator publications per device. Absence here is honest, not lazy.

Source independence

Every source, by how far it stands from the claim it supports.

Metrics modelled but empty 17

Metrics in the registry with no figure behind them. Each is a specific, nameable gap.

• Capsule gain
• Engineering gain (Q_eng)
• Lawson parameter (nτ_E)
• Neutron yield per shot
• Repetition rate
• Peak field on coil
• Tritium breeding ratio
• HTS tape critical current
• Fuel enrichment
• Refuelling interval
• Core outlet temperature
• Capacity factor
• Overnight capital cost
• Levelised cost of energy
• Headcount
• Cumulative funding raised
• Patent families

Totals