Choose-One (Plurality) — the simplest count there is

Mark one box. Count the marks. Most marks wins. That's the whole method — and this folder is about what it can and cannot do, using an election small enough to fit on a napkin.

Choose-One (also Plurality, First-Past-The-Post, or just "the ballot you already know") is the method almost every reader grew up with. It earns its place: it is the easiest to explain, the easiest to hand-count, and nobody needs a lesson to fill it out. Everything else on this page is about the price of that simplicity.

---

▶ Live on BetterVoting: vote · results ↗ (election q2rkfm, Test ID BV2257).

The silly election

Five coworkers are picking lunch. Three options. One box each.

                   Sushi  Tacos  Pizza
  Sushi-lover        X      -      -
  Sushi-lover        X      -      -
  Taco-lover         -      X      -
  Taco-lover         -      X      -
  Pizza-fan          -      -      X

  Count the marks:  Sushi 2 · Tacos 2 · Pizza 1

And that is the entire count. Sushi 2, Tacos 2 — a dead tie, and the ballots have nothing left to say. There is no second round to run, no second preference to look at, no head-to-head to check, because none of that was ever collected. A choose-one ballot holds exactly one bit per voter, and both bits are spent.

So the winner is decided by the pre-published lot order:

 A 2-way tie for first: Sushi, Tacos — 2 mark(s) each.
   Counting the marks is all a choose-one ballot can do, so the ballots cannot break it;
   the pre-published lot order decides: ['Sushi', 'Tacos', 'Pizza'].

[Lot-decided tie — rare]
  ⚠ The result here was set by lot, not by the votes.

Winner — Choose-One / Plurality Voting Method (single winner)
 Sushi   (2 of 5 marks, by lot)

Five people, one lunch, and the answer came out of a hat.

Run it: reader page · lunch_choose_one_dead_tie.yaml

"Out of a hat" is literal — and the two engines disagree by design

This is the rare case in this library with no deterministic winner, and the two tabulators handle it differently:

	how the tie is broken	winner
This repo (LH engine)	the case's pre-published lot order Sushi > Tacos > Pizza	Sushi, every run
BetterVoting	at random — the frozen export literally records tieBreakType: "random"	Sushi this time; a re-run could say Tacos

Neither is wrong. A tied choose-one election has no answer inside the ballots, so something outside them has to choose, and the only real question is whether that something was published before the vote. It's why real choose-one jurisdictions have coin-toss and drawn-lot statutes on the books — they are not folklore, they are the documented consequence of a one-bit ballot.

The voter who could have fixed it

Look again at the Pizza-fan. Theirs is the only ballot that could break the tie — they are the one person in the room with no stake in Sushi-vs-Tacos, so their opinion is exactly the tiebreaker the election needs.

The ballot gave them no way to say it. Their single mark went to Pizza, who cannot win, and about the two who can win they were never asked. That isn't a bad voter or a bad strategy; it is the ballot refusing to carry the information.

The same five people, on a ballot that asks more

These are the canonical team-lunch voters — same people, same opinions, no strategy anywhere. Hand them a 5-star ballot and they say considerably more:

Voter	Sushi	Tacos	Pizza
Sushi-lover	5	0	3
Sushi-lover	5	0	3
Taco-lover	0	5	3
Taco-lover	0	5	3
Pizza-fan	3	1	5

Now the Pizza-fan's "Sushi over Tacos" is on the ballot, and so is everyone's quiet "…Pizza is fine, actually" — four voters rate it a 3. STAR elects Pizza, the one lunch nobody objects to, and no lot is needed. Same electorate, same opinions: the entire difference is how much the ballot let them say.

That's the honest comparison to draw here — not "Choose-One is stupid." It counted its ballots perfectly. It just wasn't given much to count.

What to take away

• Choose-One is not wrong, it is thin. One bit per voter. Every method downstream of that ballot has the same one bit to read.
• Ties are structural, not bad luck. With no preference data there is nothing to break a tie with, so small elections land on the lot more often than you'd guess.
• The fix is the ballot, not the arithmetic. Ask voters for order, or for strength, and the same people resolve the same election themselves.
• Publish the lot order in advance. If the ballots can't decide, something has to — and it must be a rule set before the vote, not a coin found afterwards.

Related

• Plurality — the topic page — the family, the terms, the failure modes
• Minority winner — the other classic Choose-One story: 34% wins while two-thirds wanted someone else
• Same matrix, different plurality — three electorates that look identical to Choose-One
• Multi-member plurality — Block Voting, Limited Voting, SNTV
• The traditional voting style — what happens when voters bring the choose-one habit to a 5-star ballot

---

Up: 06_Other — other methods & engines

file: README.md