Strategy with STAR voting and IRV

clay shentrup
3 min readMay 21, 2020

--

With multi-round voting methods, there are two fundamental types of strategy: compromise and pushover.

Compromise means helping to advance someone you like less but who has a greater chance of beating a competitor you like even less in the next round. Think of it like voting Democrat instead of Green so the Democrat makes it to the runoff with the Republican and has a better chance of winning than the Green would have. (My aunt in Iowa did this when she voted for Joe Biden in the 2020 Democratic primary even though she preferred Elizabeth Warren, since she thought Biden would have a better chance to beat GOP nominee Donald Trump.)

Pushover means doing the opposite—helping advance a weak candidate against a candidate you prefer. E.g. a Republican voting Green because he’s sure the Republican will make it to the runoff, but the race is close between the Green and Democrat — and the Green polls worse against the Republican (due to capturing less of the center).

Instant Runoff Voting (the most well known form of ranked voting, and thus widely referred to generically as “ranked choice voting”) is vulnerable to both. You can imagine my aunt ranking Biden as her first choice with IRV, even though Warren was her sincere favorite. Here’s a great video explanation by Andy Jennings, a co-founder of the Center for Election Science who did his math PhD thesis on voting methods.

What do these strategies look like with STAR voting? (Note this same logic applies to a primary/general system where the first round is conducted via score voting, or even approval voting.)

Compromise

You prefer Green over Democrat over Republican. The total scores favor Green over Democrat over Republican, but the ranked preferences favor Democrat to Republican, and Republican to Green. Republican (your least favorite) is going to beat Green, but if you give Democrat a 5 and Green a 0, then you might switch the top two finalists to Democrat and Republican instead of Green and Republican, which will cause Democrat to win instead of Republican. You get your #2 instead of your #3.

Pushover

You prefer Republican over Democrat over Green. The total scores favor Republican over Democrat over Green, but the ranked preferences favor Democrat to Republican, and Republican to Green. Democrat (your second favorite) is going to beat Republican, but if you give Green a 5 and Democrat a 0, then you might switch the top two finalists to Republican and Green instead of Republican and Democrat, which will cause Republican to win instead of Democrat. You get your #1 instead of your #2.

Plausibility

The difference with STAR voting is that these strategies are virtually impossible, because the scoring process obviates the vote splitting problem that causes these issues to be plausible with IRV.

Compromise with IRV involves trying advance a candidate who is broadly strong but narrowly weak due to vote splitting (e.g. a moderate Dem being “squeezed” by a progressive entrant such as a Green) instead of a candidate who is narrowly strong but broadly weak due to vote splitting (e.g. a progressive such as a Green getting a bigger slice of the electorate than the moderate Dem, despite being weaker in a head-to-head contest against the conservative opposition).

Pushover with IRV involves the opposite: trying to advance a candidate who is narrowly strong but broadly weak due to vote splitting (e.g. the Green) instead of a candidate who is broadly strong but narrowly weak due to vote splitting (e.g. the moderate Dem).

But this vote splitting issue is only a factor with IRV because it is effectively just a round-by-round version of plurality voting, where you can only vote for one candidate at a time. This vote splitting is what produces the disparity between narrow and broad support. STAR voting eliminates this problem by allowing voters to score each candidate independently and simultaneously. (And as a reminder, this is true when using score voting or even simple approval voting in a two round process, like the approval voting open primary with top-two general used in St Louis.) As such, it is no great surprise that STAR voting outperforms IRV with any mixture of honest or tactical voters according to voter satisfaction efficiency.

A graph of voter satisfaction efficiency by Dr. Jameson Quinn, Harvard stats PhD and former board member with the Center for Election Science

--

--

clay shentrup
clay shentrup

Written by clay shentrup

advocate of score voting and approval voting. software engineer.

No responses yet