one owner per claim

Two registers had grown to cover the same ground. One was meant to hold only claims that cut across the whole estate; over time it had acquired role definitions and storage rules that were true of exactly one system, because that was where the person writing happened to be.

Duplication of this kind feels safe. Both documents say the same thing today, so a reader who opens either gets the same answer. That is precisely the trap: two copies are a fork with a delay on it. The day one is corrected they disagree, and nothing in either document can detect it.

The resolution was contraction rather than copying. Each register keeps only the claims it is the sole owner of. Where a claim applies to one system, it lives in that system's register, and the cross-cutting one keeps a pointer naming where the rest of the story is. The storage principles genuinely shared by everything stayed; the storage detail specific to one system left, and stopped being read by people it did not apply to.

The cost is reading effort, and it is not trivial: a reader of the cross-cutting register now has to follow a reference to get the whole answer, which is the thing a single source of truth was supposed to spare them. The boundary also needs re-arguing every time a claim begins to apply to both.

What it proves is that they will trade convenience of reading for the property that a claim has one owner, and that they treat a duplicated fact as a defect rather than as reassurance.