Skip to content

Voting 201 — reading results, comparisons, multi-winner intro

Audience: engaged learners, officials, the curious. Now we name things precisely, read a full result report, and compare methods honestly.

Prereq: Voting 101. · Up: Curriculum hub · Next: Voting 301.

201.1 — Reading the results (transparency)

  • Key terms: preference matrix, For / Equal Support / Against, Condorcet winner, summability, score distribution.
  • See it: the display-options example with the matrix and Condorcet blocks turned on. The auditable, precinct-summable heart of the count is the pairwise matrix.
  • How that matrix is built: Pairwise counting — every ballot is a tiny matrix — each ballot's pair verdicts add up to the election's preference matrix (on the canonical Ann/Bob/Cal ballots), the runoff as one cell of it, and why the whole thing sums by precinct.
  • When the count can't sum: Central tabulation — when every ballot must travel — the operational price of a non-summable count (IRV/STV): every ballot gathered in one place, a single point of failure, and a heavier audit — with the real incidents (Maine's courier runs, NYC 2021's test-ballot tally, Alameda 2022's misconfigured tabulator).
  • Two reports, one count: the same election appears both as BetterVoting's visual display and the LH engine's text report — why there are two, how they map, and the convert→validate→test pipeline: BetterVoting and the LH Engine — One Election, Two Reports.
  • The full audit report: the generated _tabulated.txt siblings carry the complete engine report (preference matrix + score distribution + the rounds). 101.4 showed the minimal on-screen view; reading the full report of those same elections (e.g. the _tabulated mirrors under Runoff Reversal) is the 201 skill — and why the on-screen report is minimal by default (don't overwhelm a beginner).
  • Worked walkthrough: How to Read a STAR Result Report — a full LH report annotated section by section, with a "which parts to show at 101 / 201 / 301" table.
  • Reading the runoff percentages: Two Denominators, One Winner — the runoff shown two ways (% of all voters vs % of those with a preference), and why the majority bar is half of the decided voters. The engine prints this summary by default (show_runoff_percent). Pairs with 301.4 (Equal Support).
  • A real election, end to end: What Makes the Best Pet? — a real 461-ballot BetterVoting STAR election, the actual export read section by section (scoring → runoff → runoff % → Condorcet), including the real-ballot difference between an abstention (blank) and an explicit-zero ballot.

201.2 — Edge cases & trust

201.3 — Nomenclature: RCV vs IRV vs RCV-IRV

  • Objective: stop conflating the ranked ballot (RCV) with one count (IRV).
  • Material: terminology basics · why the name is confusing.
  • The point: center squeeze / exhausted ballots are IRV-specific, not properties of all ranked methods (Ranked Robin isn't squeezed).

201.4 — STAR vs RCV-IRV (the honest comparison)

  • Key terms: exhausted ballots, center squeeze, wasted votes, the method scorecard.
  • See it: RCV-IRV vs STAR; any 4-candidate file prints the [Divergence from STAR] block automatically. The cross-method worked sets live in the method-comparisons sets.

201.5 — Multi-winner intro: Bloc STAR

  • Start plain: Electing more than one, simply — the majoritarian-vs-proportional fork before any machinery ("do you want the N best, or a body that mirrors the electorate?").
  • Objective: electing several seats with a majoritarian / at-large method.
  • Key terms: Bloc STAR, seats, at-large.
  • See it: Bloc STAR, 2 seats (in 02_STAR_Bloc/).
  • The point: a cohesive majority can sweep all seats — which is exactly what motivates 301's proportional methods.
  • Gentle committee intro (approval side): Electing a committee — making sure people have a voice — a counting-only walk through "most approved" vs "cover everyone" vs proportional, on Lackner & Skowron's steering-committee example.

201.6 — What are we optimizing for? (good winner, good method)

  • Objective: before comparing methods, name the competing ideals of a "good winner" and the criteria of a "good method" — and see there is no single ideal.
  • Key terms: consensus / Condorcet winner, majoritarian vs. utilitarian winner, VSE, strategy-resistance, summability, "a perfect system will never exist."
  • Pages: What makes a good winner? (four ideals; real Condorcet failures — Alaska '22, Burlington '09) and What makes a voting method good?.
  • The "experts reject RCV-IRV" claim, graded: Do the experts really think RCV-IRV is "bad"? — the pro-cardinal talking point (Smith / Brams / Quinn, VSE) taken seriously and checked.
  • Criteria side by side: Criteria at a glance — Approval / STAR / Ranked Robin / RCV-IRV pass/fail across every major criterion, each failure linked to a runnable election (with the "a table is a map, not a verdict" caveat up front).
  • "Condorcet extension" — the family, made simple: the Condorcet topic hub defines a Condorcet extension (elect the head-to-head winner when one exists; break cycles somehow) and — the calming fact — at three candidates the famous ones all collapse into one rule, so "which Condorcet method?" barely matters for a three-way race. The deep paradox theory (reinforcement, no-show) is 301.13–301.14.
  • The point: this is the map; the deeper theory (VSE math, Arrow / Gibbard–Satterthwaite, simulation models) is 301.

Up: Curriculum hub · Prev: Voting 101 · Next: Voting 301 — proportional, criteria, debate theory.