Tail assignments deliver the donor's joint vector verbatim (fixes the recipient-value overwrite) - #570
Merged
Merged
Conversation
The review of the register proposal (populace#570 round 1) found the real defect underneath the concentration measurements: the recipient- candidate merge (row.update(candidate.to_dict())) let the recipient's EXISTING tax-unit values replace the selected donor's joint vector for columns held at tax-unit grain — production replay showed 4,923 of 15,228 assignments disagreeing with their donors and 99.7% of intended unrecaptured-1250 donor mass lost ($11.53B -> $33.72M), while reconciliation self-confirmed because it derives expectations from the same overwritten assignments. The selected donor's joint vector now wins after the merge, and a construction-level assertion re-derives every assignment column from the donor frame keyed by donor_source_id — a leak fails at the source, independent of downstream reconciliation. The donor-keyed fidelity test binds donor -> assignment -> manifest -> materialized frame per column, and fails against the pre-fix code (mutation-checked). The register mechanism from round 1 is withdrawn per review: the unrecaptured-1250 'improvement' it rationalized was recipient-value record splitting, and collectibles concentration is 67.2% inherited from pre-existing base rows — a whole-field waiver would have masked both. Concentration is re-adjudicated on the rebuilt frame with real donor vectors arriving, through the unchanged fail-closed checks. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
MaxGhenis
force-pushed
the
cg-tail-concentration-register
branch
from
July 29, 2026 02:05
09e34dc to
94a026e
Compare
MaxGhenis
changed the title
This was referenced Jul 29, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Round 2 — the PR pivoted per review
Round 1 proposed a reviewed concentration-exclusion register. The review refuted it by finding the real defect underneath the measurements: the recipient-candidate merge (
row.update(candidate.to_dict())) let the recipient's existing tax-unit values replace the selected donor's joint vector for columns held at tax-unit grain. Production replay: 4,923 of 15,228 assignments disagreed with their donors; 99.7% of intended unrecaptured-1250 donor mass was lost ($11.53B → $33.72M) — and reconciliation self-confirmed because it derives expectations from the same overwritten assignments. The register would have waived a concentration produced by a broken transfer (its unrecap 'improvement' was recipient-value record splitting; collectibles is 67.2% inherited from base rows).The fix
_JOINT_VECTOR_COLUMNSfield re-overlaid from the donor row.donor_source_id; a leak fails at the source, independent of downstream reconciliation.🤖 Generated with Claude Code