One of the big issues in EA is figuring out how to improve institutional decision making. In California, the state narrowly avoided a situation where an incumbent governor with support from 49% of the voters would be replaced by a candidate that only 25% of voters voted for. The replacement candidate was even considered extreme relative to his state’s party’s typical candidates. This was clearly a value alignment failure.
In the linked September 17th, 2021 San Francisco Chronicle op-ed, I discuss how California (and other group entities) can improve value alignment. Below, I quote some selections from the op-ed, but readers should refer to the original article for more detail:
“In the U.S. context, there are four major election methods for electing leaders to political office: plurality voting, top two, ranked choice and approval voting.”
“A final option is approval voting, where voters mark each candidate they approve of, and whichever candidate has the highest approval rating wins the election. Unlike plurality voting, it overcomes vote splitting. Unlike top two, it is cheaper and requires only one ballot. Unlike ranked choice, it is simple, and the winning candidate will never have a lower approval rating than a losing candidate.
“During the 2016 Republican presidential nomination race, the Wall Street Journal and NBC News conducted almost monthly polls of nationally representative samples of Republican caucus/primary voters through Hart Research Associates/Public Opinion Strategies from March 2015 through April 2016.”
“In the 10 polls conducted by the Wall Street Journal and NBC News, Donald Trump was never the Republican candidate with the most supporters. This suggests that during primaries, Trump was never the Republican candidate with the highest approval rating.… [But] Trump was consistently the plurality voting winner, even in the polls where he had the fewest supporters, because the many candidates (over 15 initially) caused vote splitting.”
How to Avoid California Recall Disasters (SF Chronicle Op-Ed)
One of the big issues in EA is figuring out how to improve institutional decision making. In California, the state narrowly avoided a situation where an incumbent governor with support from 49% of the voters would be replaced by a candidate that only 25% of voters voted for. The replacement candidate was even considered extreme relative to his state’s party’s typical candidates. This was clearly a value alignment failure.
In the linked September 17th, 2021 San Francisco Chronicle op-ed, I discuss how California (and other group entities) can improve value alignment. Below, I quote some selections from the op-ed, but readers should refer to the original article for more detail:
“In the U.S. context, there are four major election methods for electing leaders to political office: plurality voting, top two, ranked choice and approval voting.”
“The problem with ranked choice is that it is complicated and can elect a candidate that is preferred by fewer voters than a losing candidate. When voters see this, it delegitimizes the election, and voters rebel against the system. This happened in the 2009 Burlington, Vt., mayoral election. Specifically, a majority of voters preferred Andy Montroll over Bob Kiss, but Kiss won the election anyway. Following Montroll’s loss, Burlington voted to replace ranked choice. Similarly, Sunnyvale; Aspen, Colo.; and many other American cities have adopted and then ditched ranked choice.”
“A final option is approval voting, where voters mark each candidate they approve of, and whichever candidate has the highest approval rating wins the election. Unlike plurality voting, it overcomes vote splitting. Unlike top two, it is cheaper and requires only one ballot. Unlike ranked choice, it is simple, and the winning candidate will never have a lower approval rating than a losing candidate.
Approval voting was adopted in St. Louis in 2020 and in Fargo, N.D., in 2018. In the 2017 St. Louis mayoral Democratic primary (which was effectively the real election, given St. Louis is a Democratic stronghold), there were seven candidates in a plurality voting contest. Lyda Krewson won with 32% of the vote while Tishaura Jones came in second with 30%. In 2020, under approval voting, Jones had the highest approval rating (57%) of the four candidates and went on to become mayor.
Polling in Fargo and St. Louis showed that it is broadly popular. In Fargo, where the approval voting initiative passed with 64% of the vote, polls showed that 71% of voters found it easy, and 62% liked it. In St. Louis, where the proposition passed with 68% of the vote, polls demonstrated diverse voter support for approval voting, including 75% (Democrats), 60% (Republicans), 72% (Independents), 79% (African Americans) and 71% (whites).”
“During the 2016 Republican presidential nomination race, the Wall Street Journal and NBC News conducted almost monthly polls of nationally representative samples of Republican caucus/primary voters through Hart Research Associates/Public Opinion Strategies from March 2015 through April 2016.”
“In the 10 polls conducted by the Wall Street Journal and NBC News, Donald Trump was never the Republican candidate with the most supporters. This suggests that during primaries, Trump was never the Republican candidate with the highest approval rating.… [But] Trump was consistently the plurality voting winner, even in the polls where he had the fewest supporters, because the many candidates (over 15 initially) caused vote splitting.”
“Social choice scholars (the technical name for election scientists) will tell you that due to Arrow’s theorem and the Gibbard-Satterthwaite theorem, no election method is perfect, not even approval voting. But with this knowledge, the largest and foremost society of social choice scholars (the Society for Social Choice and Welfare) uses approval voting for its elections. Several other science societies have followed the experts and adopted approval voting.
California should follow the science and amend its recall process to adopt approval voting.
Mahendra Prasad is a Ph.D. candidate in political science at UC Berkeley who researches social choice.”