election_title: "Majority Judgment 101 — the highest median grade wins"

scenario_description: |-
  The simplest grade election. Three candidates, five voters, and the six-word
  common language Balinski and Laraki propose: To Reject, Poor, Acceptable, Good,
  Very Good, Excellent. Majority Judgment ignores totals and averages entirely -
  it lines each candidate's grades up in order and takes the MIDDLE one. Highest
  median wins.
  Bruno wins on a median of Very Good. The candidate to watch is Alice: she
  collects the two loudest grades on the whole ballot (two Excellents, more
  enthusiasm than anyone else gets) and still finishes second, because her median
  is only Good. That is the median doing exactly the job it was chosen for -
  nobody can lift a candidate by grading them harder, only by being one more
  voter who ranks them at or above that middle grade.
  V5 left Alice UNGRADED rather than grading her. Under both grade procedures an
  ungraded candidate takes the bottom of the scale, so that blank counts as To
  Reject; the report says so and names the cell. The convention looks like
  bookkeeping and is not - it is the entire mechanism of the truncation paradox
  (Felsenthal's Examples 24 and 28).
  The mean agrees with the median here (Bruno 3.2, Alice 2.8, Cleo 2.4), which is
  the honest 101 picture: the two usually pick the same candidate. Where they come
  apart is the interesting part - see 07_Concepts/voting_paradoxes/majority_judgment.md.
  This is a grade-ballot file, not an LH election file: the grades are words, which
  fit neither the engine's numeric 0-5 ballot nor BetterVoting, and MJ's tie-break
  exists in neither. Count it with
  tools_adam/pref_voting_tabulation_engine/grade_methods_report.py, which computes
  the median and the tie-break from scratch and cross-checks both against
  pref_voting's majority_judgement.

grade_method: MajorityJudgment
grade_scale: "To Reject|Poor|Acceptable|Good|Very Good|Excellent"

# Titles for the drawn ballots — the grade block is transposed (a row per
# candidate), so a voter's note can't ride on their row the way a `#` comment
# titles a score ballot. Read only by build_style_ballot_images.py.
voter_notes:
  V1: "An Alice enthusiast — Excellent"
  V2: "The same again — Alice's second Excellent"
  V3: "Bruno Very Good; Alice merely Good"
  V4: "Alice down to Poor; Bruno still Good"
  V5: "Left Alice ungraded — that counts as To Reject"

grades: |-
  ,V1,V2,V3,V4,V5
  Alice,Excellent,Excellent,Good,Poor,
  Bruno,Very Good,Very Good,Very Good,Good,Poor
  Cleo,Good,Good,Acceptable,Acceptable,Acceptable

expected_winners:
  - Bruno

# file: mj_101_c3_b5.yaml