BV2263 — an approved write-in discards ordinary ballots

Baseline capture — recorded before any fix. Everything under Actual result is what BetterVoting does today (2026-08-02, production). The point of this page is to make the fix verifiable: when the change lands, re-run these steps and the Expected after the fix section is the assertion.

Sheet row still needed. BV2263 was allocated from the block after the highest id observed across the repos (BV2262) — it is not yet a row in the test-case sheet. Paste-ready: BV2263-2267-sheet-rows.tsv. If it collides with an existing row, renumber here.

Purpose

Show that approving a write-in causes ordinary ballots to be dropped as abstentions, by putting the same seven ballots through two races that differ only in whether Cedar is a write-in or an official candidate. Filed upstream as #1470.

Master data

Election	43jp39 · results
Method	STAR
Voter access	open (anyone with the link)
Public results	on

Ballots

Ann  Ben  Cedar
  4    4    (not written in)   x4
  4    2    3
  0    3    5
  1    2    0

Race 1 — Cedar is an approved WRITE-IN
Race 2 — Cedar is an OFFICIAL candidate (blank on the first four ballots)

Steps

1. Open the results page.
2. Compare the reported voter count and winner of race 1 against race 2.

Both races are on the same ballot paper, so every voter voted in both, identically.

Actual result — today

	Race 1 — write-in	Race 2 — official
Voters reported	3	7
Abstentions	4	0
Scores	Cedar 8, Ben 7, Ann 5	Ben 23, Ann 21, Cedar 8
Winner	Cedar	Ben
Tie-break	none	none

Verify without logging in:

curl -s https://bettervoting.com/API/ElectionResult/43jp39 | jq '.results[] | {tally: .summaryData.nTallyVotes, abstentions: .summaryData.nAbstentions, winner: .elected[0].name}'

What is wrong

Four voters gave both official candidates four stars and are reported as having abstained. Their ballots contribute no score and no pairwise preference, so a candidate exactly one voter scored above zero wins the race. The two races are functionally identical and disagree on both the electorate and the winner.

Expected after the fix

Race 1 reports 7 voters and elects Ben, matching race 2. The four 4,4 ballots are counted, each expressing a strict preference for both official candidates over Cedar.

The 43jp39 election is closed, so these numbers are frozen and can be diffed against a fixed build.

Status 2026-08-20: the fix is written and verified locally — branch fix/1470-write-in-abstention-normalization, commit c2fc5bd8 in clone bv-1470, parked under the 🚦 PR freeze (details on the issue page). The patched tabulator reproduces exactly this table for race 1: tally 7, abstentions 0, Ben 23 / Ann 21 / Cedar 8, Ben elected. Once the fix deploys, this page's curl is the acceptance check.

Related

#1470 · issues/1470-writein-abstention-discards-ballots.md