The Pre-Registered Predictions

Before the machines are built, we publish specific numbers that the hardware will either confirm or refute, dated and frozen so we cannot move them later.

Predictions written before the test

A pre-registered prediction is a specific claim, published and dated before the measurement that could disprove it, so it cannot be adjusted after the fact to match whatever the hardware shows. This is the strongest form of scientific honesty available to a design-and-simulation program, and we hold ourselves to it.

What we have put on record

The frozen breeder design point is itself a set of predictions: a spherical tokamak reaching Q_sci 3.424 and fusion power 88.7 MW at plasma current 9.86 MA and peak field 16.84 T (8 T on-axis), with negative triangularity delta -0.30, producing roughly 4 kg/yr tritium-class product and about 1.97 kg/yr helium-3 plus 14 MeV neutrons. First-of-a-kind first tritium around 2030 is the event that begins testing these against measurement.

The gates are predictions too

The burner (Aegis / MetroVolt) gates are pre-registered as well, as predictions of difficulty: a plug coil overstressed 3-3.9x at the design bore, an operating regime 166-830x beyond any device, He-3 demand about 400x domestic supply per commercial unit, and availability of roughly 0.86-0.995 against a Tier III bar of 0.99982. If future work resolves any of these, that resolution will be measured against the number we published, not a softened one.

Pre-registration only works if the prediction is fixed. That is why it depends on the frozen-design methodology and change control: the numbers were set, dated, and made public before the tests that will judge them.