An Analysis of Ranked Voting in the 2020 Elections

Ranked-choice voting is an election method that executes automatic runoffs by eliminating the least desired candidate if no and until one candidate achieves more than 50% of the tabulated votes.  Voters provide a list of candidates in alternative order so that if the method eliminates their candidate, it knows to which candidate to transfer their votes.

In 2020 Maine held, for the first time, ranked-choice voting elections for all of its candidates to federal office.  Maine had passed ranked-choice voting in 2016 and used it in 2018 to elect only the U.S. Representative for Congress District 2.  San Francisco has been running ranked-choice voting elections since 2004.  Alaska recently passed ballot measure 2 creating an open, non-partisan, plurality-style primary for election to state executive and state and national legislative offices from which the top four candidates would proceed to a ranked-choice voting style primary. How well has ranked-choice voting lived up to its promises?

The first question is, “What had ranked-choice voting promised to do?” By looking at the ballot measures for Maine[i] (2016, question 5) and San Francisco[ii] (2002, proposition A), Alaska (2020, measure 2) as well as fairvote.org’s website (the most prominent proponent of ranked-choice voting in the United States), and the Hewlett Foundation[iii] (a San Francisco based ranked-choice voting promoter), ranked-choice voting promised mainly four things:

1. It can save money as the runoff election is embedded in the process.
2. For voters, offering more choice. Ranked choice voting opens up a potential pathway for third parties, increasing greater third and independent participation in elections, and also tends to increase voter turnout.
3. It produces more civil campaign dynamics, as candidates are incentivized to avoid alienating voters from whom they might pick up second choice votes, and thus more broadly reflecting the values of the electorate
4. Outcomes in RCV reflect the preferences of a majority of citizens and reflect the core democratic principle of majority rule, mitigating the likelihood that a candidate who is disapproved by a majority of voters will get elected

And how well has it done this?  I’ll let you award the final scores (0 “completely failed” to 9 “completely succeeded”).

Promise #1 (no live runoff elections).  Regarding the first point, Maine never required a “50% majority” like San Francisco, and so it never had any “expensive, low- turnout” election runoffs with which to contend.  Certainly, however, San Francisco has avoided the cost and the pain of runoff elections since the implementation of ranked-choice voting.  Clearly, ranked-choice voting delivered as expected. Alaska’s measure 2 also repealed specific special runoff elections, so it should experience this benefit as well.

How would you score ranked-choice voting’s on its deliver on promise #1:___

Promise #2a (more choices).  Looking at the offering of “more choices,” it is assumed that this indicates some alternative to Republican or Democratic candidates: “other” candidates.  Did elections using RCV offer more “other” candidates than previous plurality elections, and do we see a change in the number of “other” candidates being elected?

Maine used ranked-choice voting to determine the outcome of four elections, U.S. President, one U.S. Senator and two U.S. Representatives.  I won’t look at the position of president as, since plurality favors either a Republican or Democrat, it would not make sense for Maine voters to vote for a third or independent candidate even if they wanted to.  For the other three races I looked at how many “choices” other than Republican and Democrat appeared for the people of Maine. (Ranked-choice voting election races appear in blue tinted areas.)

Year	Other choices. (Senator)	Winner	Other choices. (Rep 1)	Winner	Other choices. (Rep 2)	Winner
2020 (RCV)	2	Republican	0	Democrat	0	Democrat
2018	1	Independent	1	Democrat	2	Democrat
2016	no race	no race	1	Democrat	0	Republican
2014	0	Republican	1	Democrat	1	Republican
2012	4	Independent	0	Democrat	0	Democrat

Maine offers only an extremely small dataset at this point, but certainly, it has yet to show indications of more choices than previously enjoyed by Maine voters.

What about San Francisco?

Year	Other choices (all races)	Winner (all races)
2020	2	Democrat
2018	0	Democrat
2016	0	Democrat
2014	0	Democrat
2012	0	Democrat
2010	0	Democrat
2008	0	Democrat
2006	1	Democrat
2004	3	Democrat
2002	3	Democrat
2000	3	Democrat
1998	0	Democrat
1996	1	Democrat

Remembering that San Francisco first used ranked-choice voting in 2004, we can see that before 2004 (data on the city of San Francisco’s website only goes back to 1996) in the races for state senate and state house held (1996-2002: a total of four elections), a total of seven “other” candidates appeared as choices for the voters.  After the institution of ranked-choice voting (2004-2020: a total of nine elections), six “other” candidates appeared as choices for voters.  In fact, for six elections in a row, under ranked-choice voting, no options beyond Republican or Democrat were offered to voters.

San Francisco provides enough data points to indicate that ranked-choice voting does not appear to supports ranked-choice voting’s promise of “offering more choices.”  Additionally, as no candidates other than Republicans or Democrats have been elected to these offices since their implementation, it does not appear to support the promise that it  “opens up a potential pathway for third parties.”

Promise #2c (more voter turnout).  There is a final portion to Promise #2: an increase in voter turnout.  This has two components, 1) did voter turnout increase after ranked-choice voting began, and 2) was the increase due to ranked-choice voting.  Based on voter turnout data for both the U.S. and San Francisco, we learn this:

We see that San Francisco has always had more voter turnout than the rest of the nation and it is increasing starting from 2002. However, the same is true of the United States as a whole, and the changes in voter turnout appear to reflect more on the country as a whole than on just San Francisco or the introduction of ranked-choice voting.

How would you score ranked-choice voting’s on its deliver on promise #2:_____

Promise #3 (more civil campaigns).  Promise #3 is hard to judge as we lack objective criteria of what “civil” might mean, so we must examine subjective data. Here I provide some pertinent quotes from San Francisco:

“In theory, ranked-choice voting also discourages nasty campaign tactics, in part because candidates vie for second-place positions from their opponents’ supporters.  We’ve seen this play out in a number of ways during this mayoral race. In a statement of friendship and to woo one another’s supporters, Jane Kim and Mark Leno [candidates for San Francisco Mayor, 2018] have each co-endorsed the other as a second choice, sending out flyers and ads asking voters to choose both of them in the election. Kim [independent] and Leno [independent] have also bragged about their second-place endorsements from Democratic clubs across the city, recognizing the value in being someone’s No. 2 vote.

“And while vitriol has certainly erupted between the Leno and Kim campaigns and that of their opponent, London Breed [independent], the race has been relatively free of candidate-bashing between Leno and Kim.” [iv]

“To ‘game’ the system in a simple plurality-winner election, the basic strategy involves mobilizing your base while trying to tear down competing candidates. This involves lots of scorched-earth negative campaigning.

“To ‘game’ the system in a ranked-choice voting election, the basic strategy is to try to appeal broadly and say, I’d like to be your first choice, but if I can’t be your first choice, I’d like to be your second choice. In ranked-choice voting, we’d expect soft alliances among candidates who realize that they’re both stronger through coalition building than they are by law-of-the-jungle campaigning . . .  Ranked-choice voting generally improves civility of campaigning. Voters in cities with ranked-choice voting report that campaigns became less negative after ranked-choice systems went into place. Cities with ranked-choice voting also have higher voter turnout.”[v]

San Francisco appears to give the system a pass, with more cooperation and civility occurring between like-minded candidates (Kim and Leno, but not Breed).  Maine, however, did not seem to act in a more “civil” fashion, or even a different one.

Cary Weston, a partner at the Bangor marketing firm of Sutherland Weston, said about the fact that spending on political advertising in Maine was tracking to set a record high in the 2020 campaign, “It’s loud, it’s obnoxious, it’s everywhere. I really don’t think anyone’s going to notice because it’s so loud and obnoxious anyway.”

The Bangor Daily News noted that, “The [U.S. Senate] race between Collins and Democratic nominee Sara Gideon, which also included independents Lisa Savage and Max Linn, shattered state spending records, with more than $200 million flowing into Maine through the candidates and outside groups, nine times more than any other political contest in the state’s history.”[vi]

How would you score ranked-choice voting’s on its deliver on promise #3:_____

Promise #4 leaves us at a complete loss. It’s not a question of whether or not ranked-choice voting does or does not deliver the “preferences of the majority,” it’s a question of whether or not ranked-choice voting can identify “the preferences of the majority.”

Ranked-choice voting cannot determine what “the preferences of the majority” of citizens are as ranked-choice voting never asks this from the citizens.  To illustrate:

Assume an election between four candidates ends with Candidate 1 getting 60% of the votes, Candidate 2 getting 40% of the votes, and the other two getting 0% due to elimination.  Ranked-choice voting would say that the preference of the majority is for Candidate 1. However, the only way we know to check this is to ask what each voter thinks about each candidate.  Ranked-choice voting does not do that; instead, the method has voters submit an ordering of alternatives (which it calls a rank) on a ballot.  For our example election, our voters submitted their listings of alternatives like this:

voters	Candidate 1	Candidate 2	Candidate 3	Candidate 4
A	1	4	3	2
B	1	4	2	3
C	4	1	4	2
D	4	1	3	2
E	3	4	1	2
total score	2	2	1	0

In the initial round, Candidate 1 has 40% of the vote, Candidate 2 has 40%, and Candidate 3 has %10 of the vote.  Candidate 4 has 0% of the vote and is eliminated.  As Candidate 4 was not any voters’ first choice, no votes are transferred to the other three candidates.  Candidate 3 is the next to be eliminated and voter E, which had placed Candidate 3 as their top choice, now has their vote transferred to Candidate 1.  Candidate 1 has won with 60% of the vote.  The rankings, as used by ranked-choice voting, do not tell us the preferences of the people.

If we ask the voters to provide their preferences by scoring the candidates on a scale of 0 to 9, we learn this:

voters	Candidate 1	Candidate 2	Candidate 3	Candidate 4
A	9	0	2	8
B	9	0	6	6
C	0	9	4	8
D	0	9	5	8
E	2	0	9	7
total score	20	18	26	37

Now, and only now, after voters have scored all the candidates, do we learn which candidate the voters prefer, Candidate 4.  Voters prefer Candidate 4 almost twice as much as Candidate 1.  So, ranked-choice voting did not provide the “preferences of the majority.”

The idea that we must find a “50% majority” is myth, not math.  It comes from plurality voting where voters provide the system with the minimum amount of information.  As no one knows what the voters really want, the only way to ensure that the most desired option is identified is to establish a 50% majority requirement.

In plurality it could be, but isn’t necessarily true, that 60% of people prefer either Candidate A or B to Candidate C.  But as those voters divide their vote, 30% to Candidate A and 30% to Candidate B, Candidate C wins.  This vote splitting, leading to the election of a least desired candidate, is what ranked-choice voting is supposed to fix.  However, the problem with ranked-choice voting is that it determines who wins by eliminating the Candidate with the fewest top scores.  As shown above, this could lead to the elimination of the most desired candidate.  True, votes aren’t split, but they also don’t go to the most desired option.  The “most desired “has no set relationship to the “most ranked #1.”  That spurious condition used by ranked-choice voting is only accurate when 100% of the voters chose the same candidate as their top choice.

How would you score ranked-choice voting’s on its deliver on promise #4:_____

How did ranked-choice voting measure up?  Did it deliver on its promises to advocates who worked so hard to get it passed?

Ranked-choice voting ends election runoffs, which were San Francisco’s top problem and the reason they implemented it. San Francisco did not implement ranked-choice voting to create more room for third parties, nor to provide more choices to voters, nor to ensure the “desires of the majority” are delivered.

Maine did not have a problem with runoff elections. Maine thought that, by implementing ranked-choice voting, it could (as described by the Maine League of Women Voters):

• Restore majority rule and results in the election of consensus candidates who are more broadly support by voters.
• Eliminate vote splitting and the need for strategic voting, so voters never have to vote for the lesser of two evils when there is another candidate they really like.
• Encourage greater civility in campaigns, as candidates must reach beyond narrow bases of support to build majority coalitions, knowing that negative campaigning can backfire when voters have the power to rank more than one candidate.
• Work just like actual runoff elections without the cost and delay. Ranked Choice Voting is the most cost-effective and efficient way to conduct a runoff, and it is the only reform that allows overseas and absentee voters to fully participate.

However, 2020 showed us that ranked-choice voting does not show evidence of doing any of this (with the exception of no election runoffs.)

• It does not elect consensus candidates as it can’t identify them.
• It doesn’t eliminate vote splitting as it can’t properly identify voter preferences. And seeing that both San Francisco and Maine still almost exclusively elect Republican’s and Democrats, ranked-choice voting, if it doesn’t require voters to choose the lesser of two evils, it certainly requires them to take the lesser of two evils.
• It shows little evidence of greater civility (maybe in San Francisco).
• It does ensure no election runoffs. (which Maine didn’t do anyway, so . . . ? Also, I’m not sure what the LWV meant by “only reform that allows overseas and absentee voters to fully participate.”)

The citizens of Alaska followed the hopes of Maine in passing Measure 2, a desire that the will of the majority over the minority be implemented and more ability for independent and third party participation. Does the evidence indicate that ranked-choice voting will deliver this to Alaska?

As more and more states consider ranked-choice voting (too late for you, Alaska), those advocating for it, spending their time and hard-earned money to campaign for it, need to ask why are they doing it? Everything in the list of the League of Women Voters is met by score voting methods.  Score voting methods (every voter scores every candidate, high score wins (just like you scored ranked-choice voting on its promises); i.e. Approval Voting or STAR voting) do identify and elect the most desired candidate, they do eliminate vote splitting, they do require candidates to reach across the political spectrum to all voters, and they don’t require expensive runoffs either.

To all of those who have worked so hard to improve your state and our nation by getting ranked-choice voting adopted, I salute you and applaud your hard work.  I only regret that all of that effort now appears to have been misdirected.

But it’s not over.  We needed to test ranked-choice voting, and now we have. Many, unlike those who embraced the untested ranked-choice voting have shied away from score voting, calling it untested.  That may be true, but if we have to choose between suffering a deadly disease, or taking a cure we now know doesn’t work, or taking a cure that appears in the lab to work but hasn’t been fully tested, which do you want to take?  All of those who supported ranked-choice voting showed that they wanted to take a chance.  Now we need them to do that again.  It’s time for score voting.

I leave you with this quote from the Bangor Daily News, which now must live with ranked-choice voting in Maine, “After some worry and a long legal fight, ranked-choice voting made no difference in the outcomes of Maine elections.”[vii]

Sources:

[i] https://www.maine.gov/sos/cec/elec/upcoming/citizensguide2016.pdf

[ii] http://www.smartvoter.org/2002/03/05/ca/sf/prop/A/

[iii] https://hewlett.org/historical-assessment-of-the-first-state-wide-passage-of-ranked-choice-voting-in-maine

[iv] Bishari, N. (2018, May 31). One, Two, Three: The Ins and Outs of RCV. Retrieved November 01, 2020, from https://www.sfweekly.com/news/one-two-three-the-ins-and-outs-of-rcv/

[v] Drutman, L. (2018, May 14). All politicians “game” the system. The question is how? Retrieved November 01, 2020, from https://www.vox.com/polyarchy/2018/5/14/17352208/ranked-choice-voting-san-francisco

[vi] Piper, J. (2020, December 30). The record spending in Maine’s 2020 Senate race could be a sign of things to come. Retrieved February 09, 2021, from https://bangordailynews.com/2020/12/30/politics/the-record-spending-in-maines-2020-senate-race-could-be-a-sign-of-things-to-come/

[vii] Andrews C, Shepherd, M, Piper, J  (2020, Nov 5) Maine’s experiment with ranked-choice voting was muted in 2020, Retrieved February 09,2021 from https://bangordailynews.com/2020/11/05/politics/daily-brief/maines-experiment-with-ranked-choice-voting-was-muted-in-2020/