No voting system can be perfectly fair. Kenneth Arrow proved that when there are 3+ candidates, it's mathematically impossible to satisfy all reasonable fairness criteria. This earned him the 1972 Nobel Prize in Economics.
Arrow identified four seemingly reasonable requirements for any voting system:
The system must work for ANY possible combination of voter preferences.
If ALL voters prefer A to B, then society must prefer A to B.
The ranking of A vs B should only depend on how voters rank A and B, not other candidates.
No single voter should always determine the outcome regardless of others.
It is mathematically impossible for any voting system to satisfy all four criteria when there are 3+ candidates. Every system must violate at least one!
Choose a scenario and voting method to see which criteria get violated:
Before Arrow's theorem, the Marquis de Condorcet discovered a disturbing possibility: collective preferences can be cyclic, even when individual preferences are not.
Imagine three voters ranking three candidates:
| Voter 1 | Voter 2 | Voter 3 |
|---|---|---|
| π > π > π | π > π > π | π > π > π |
Now let's count pairwise preferences:
Like rock-paper-scissors, there's no clear winner! Every candidate loses to one other candidate. This is called a Condorcet cycle.
Violates IIA: Adding a third candidate can change the winner between the original two (spoiler effect). Bush vs Gore vs Nader in 2000.
Violates IIA: The ranking of A vs B can change when a "clone" of C enters the race, even if nobody changes their A-vs-B preference.
Violates IIA and Monotonicity: A candidate can lose by gaining more support! Also suffers from spoiler effects.
Satisfies IIA and Pareto, but obviously violates Non-Dictatorship. Arrow proved this is the ONLY system that satisfies the other three!
Arrow's theorem isn't just mathematical curiosityβit has profound implications:
While we can't satisfy all criteria, we can make trade-offs:
Possible escapes: