Skip to content

Instantly share code, notes, and snippets.

@pcantrell
Created January 2, 2017 02:49
Show Gist options
  • Save pcantrell/a6f130f06f66bb159cac91f9bd4df8fe to your computer and use it in GitHub Desktop.
Save pcantrell/a6f130f06f66bb159cac91f9bd4df8fe to your computer and use it in GitHub Desktop.

Suppose we have a country of 10 mutually hostile religious factions. Each is around 10% of the population; the largest group is the Fooists, who are 12%.

Suppose each relgious faction nominates a candidate whose platform is “outlaw all other religions except mine.”

Suppose that there is also an 11th candidate, the tolerant one, whose platform is “allow freedom of religion.”

The country is so balkanized that every citizen’s preferences are as follows:

  1. My religion’s candidate
  2. The tolerant one
  3. I don’t care

If everybody votes for their honest top choice, what happens with instant runoff (a.k.a. ranked choice) voting? The tolerant one is everybody’s second choice, a clear consensus candidate. However, because they are everybody’s second choice, they are immediately eliminated.

It’s therefore in everybody’s interest to lie and list the tolerant one as their #1 choice. This means that the voting public hears the message “our system is broken; you need to lie on the ballot in order to win.” And how do the actual results come out? It’s a mess.


This is an extreme example, designed to highlight how absurdly IRV can degenerate, but the same effect can play out more subtly with even just three parties.

IRV prevents spoilers only when third parties are tiny challengers. If a third party ever actually takes off, the system breaks down. It’s almost as bad as plurality — and more confusing. In particular, it shares with plurality the property that it gives terrible results if everyone votes honestly.

The Center for Election Science delves into this in more detail, with historical examples.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment