Each voter scores every game 0–5 (via the grade labels). The two highest-scoring games advance to a runoff. In the runoff, each voter's ballot goes to whichever finalist they rated higher; the majority wins.
Ties and edge cases: Ties for 2nd place are broken by pairwise head-to-head among the tied candidates. A three-way tie in the runoff falls back to Copeland scoring.
Runoff flips the score winner
One game wins the score phase but loses head-to-head in the runoff — classic STAR behavior that pure score-based methods miss.
Every game that collected more "hard pass" votes than the least-vetoed game is disqualified first. The survivors then run through normal STAR (score phase picks two finalists, runoff picks the winner). A good middle ground between "crowd-pleaser wins" and "nobody is stuck playing something they hate."
A veto flips the winner
The un-vetoed STAR runner-up takes the win.
Every voter grades every game independently. The game with the highest median grade wins. MJ is known for resisting strategic voting — a single lone voter trying to push an outlier has almost no effect on the median.
Ties and edge cases: Ties are broken by the Balinski–Laraki distance: how many votes would need to move before the median changes. The median that is "further from the cliff" wins.
Polarizing winner
Catan gets 4 "excellent" and 3 "hard pass" — its median is "excellent" so MJ picks it, even though most methods pick the crowd-pleaser.
Same idea as IV·STAR, but the survivors run through Majority Judgment instead of STAR. Games with disproportionate "hard pass" votes are disqualified first; then the median-grade winner is picked from what remains.
A veto flips the winner
MJ's winner gets vetoed — IV·MJ promotes the runner-up instead.
Each voter ranks the games. The top of a ballot is worth N−1 points, second place N−2, and so on. Games with tied grades on the same ballot split the points evenly. Highest total points wins.
Rewards broad support
Borda favors the game most often ranked near the top — even if it rarely finishes first.
Each voter's top remaining choice gets their vote each round. The game with the fewest first-place votes is eliminated, and its voters' next preferences redistribute. Repeat until one game crosses 50%. Used in Australia and some US elections.
Ties and edge cases: Vote splits are tracked in exact rational arithmetic (scaled by LCM of 1..N) so multiple tied favorites on one ballot never cause float-precision rank flips.
Tennessee capital
The classic textbook scenario: Memphis wins IRV despite most voters preferring Nashville overall. Clear demonstration of IRV's 'center-squeeze' flaw.
Every pair of games is scored head-to-head by counting how many voters ranked one higher than the other. If one game beats every other in these pairwise matches, it wins. When there is no such "Condorcet winner," we fall back to Copeland scoring (most pairwise wins).
Ties and edge cases: If the pairwise comparisons form a rock-paper-scissors cycle (A beats B, B beats C, C beats A), the tally surfaces a "paradox" banner — the result is genuinely tied.
Rock-paper-scissors
A, B, and C form a cycle — no winner exists, and the paradox banner says so.
One voter (the most recent to submit a vote) decides the outcome. Included as a control: it is trivially strategic, ignores every other voter, and is the baseline against which every other method looks good.
One voter, no debate
When one person votes, they are by definition the dictator.