Culling out cases of high uncertainty, the newly-created method can make accurate assessments (when it makes them), but current experiments already have listed down areas of imminent improvement, such as increased resolution, increased penalties for errors (the black art of adjusting regularisation terms), etc.