Worth recording alongside the Administrative Review Tribunal because it is designed on a different principle, and the difference is instructive.

The Board is non-adversarial: DVA is not represented, the proceeding is not a contest between parties, and the Board can obtain evidence itself. That design exists because the applicants are veterans, frequently unwell, frequently unrepresented, and the alternative is a person with post-traumatic stress disorder cross-examined about their service.

This database's general enthusiasm for removing review stages does not extend here. A review body that is cheap for the applicant to use, non-adversarial, and specialist is a good thing to have between a decision-maker and a court, and its existence is not evidence of duplication.

The question this database can legitimately ask is the same one it asks of merits review generally: what proportion of decisions are changed, for what reasons, and what would have to change upstream for those matters not to arise. For the NDIS that number is published and it is 64 per cent. For veterans' claims the equivalent figure is not published in comparable form, which is itself the finding.