Extraction Coverage Report

San Francisco, California, 2024 General Election

Audit Engine version:'v2.32.X'

Run Date: 2026-10-23 22:57:09.289844 Copyright (c) Citizens' Oversight


Return to Master Report Index


Reconciles the extraction task list (full_bif) against what extractvote actually produced. A real gap ('missing' below, or an incomplete chunk table) needs investigation; 'excluded by profile' and an out-of-scope style count do not -- they're profiled_bif's own selection and the style-scope filter (roismap/include_styles_with_contest) working as intended.

Ballot sheet reconciliation

full_bif's ballot_id keys one physical ballot sheet, not a voter's whole (up to 4-sheet) completed ballot.

bucket sheets
total ballot sheets (full_bif) 133422
profiled (included) sheets 34541
with votes.csv record 34534
with a logged exception (no votes row) 7
with votes row and exception 1031
excluded by profile 98881
missing 0

Style scope coverage

'Out of scope' ballots are real ballots whose style has no roismap entry for this job's extraction scope -- they cannot extract regardless of BMD/non-BMD routing, and are expected to appear as exceptions, not as an unexplained gap.

in_scope distinct_styles ballots
in scope (mapped) 110 33439
out of scope (unmapped) 330 99983

is_bmd chunk-vs-corrected mismatch coverage

No is_bmd correction table found for this job -- nothing to report.

Chunk completeness

'complete' is 1 when at least as many chunks were built as the task list has -- a 'built' stage status alone does not confirm this (see RUNBOOK's extractvote gate). The task list is full_bif's own chunk count ('total_chunks'), except when use_profiled_bif is TRUE, where it's 'profiled_chunks' instead -- entire full_bif chunks with no selected ballots never get a profiled_bif chunk, so full_bif's count overstates what extraction could ever build.

table chunks_built total_chunks profiled_chunks complete
marks 1335 1335 1335 1
votes 1335 1335 1335 1
pmvs 1335 1335 1335 1
wins 1335 1335 1335 1