All,

Roel is right and the omission is mine. My list was of the per-check
record. The roll-up is a separate agreed layer and I left it off.

The freeze list, complete:

the per-check record with its five states and the cause rule
  the twelve rejection rows as consolidated
  the split between consistency rows and evidence rows
  carry-or-reference with digests
  the roll-up as Roel traced it: an aggregate only over declared and
    exercised checks, never emitted without its complete denominator,
    each completeness claim carrying the size of the population it
    counts over, and a claim over an empty population reported as not
    claimable
  the counter over carried against referenced, which sits on that
    roll-up rather than beside it

Freezing the record and not the aggregate leaves ungoverned the number a
report adds up from it, and that is the half a reader sees first.

Three +1s now for the freeze and for Evgenii editing. He asked for a yes on
the list in his introduction and did not get one. He has three.

Nicolás

Nicolás Rocchia
disensor.dev · San Francisco, Córdoba, Argentina
