“Survivor”-style voting

The TV show uses a worthwhile procedure that could combat polarization.

Two days ago, the Election Law Blog linked to Steven Hill’s Substack piece lamenting that the procedure that Maine Democrats have adopted for choosing a Senate nominee after Graham Platner’s withdrawal is similar to the method of eliminating contestants in the TV show Survivor. Hill is incorrect in this respect, and to understand why, it is important to distinguish two aspects of a voting procedure. One is the precise computational rule for eliminating candidates. The other is the timing and associated mechanical process by which the computational rule for eliminating candidates is implemented.

The Survivor TV show’s computational rule for eliminating contestants is to have voters identify which contestant they think is most deserving of elimination and then to eliminate the contestant with the highest number of these negative votes. By contrast, the procedure that Maine Democrats have adopted to choose among Senate candidates is to have voters at a party convention identify which candidate they think should be the party’s nominee and then to eliminate the candidate with the lowest number of these positive votes.

As a matter of timing and mechanics, both Survivor and Maine Democrats hold new separate rounds of voting after contestants or candidates are eliminated one at a time. It is this aspect of the process in Maine that Steven Hill objects to. He says that holding separate rounds of voting permits jockeying and bargaining among candidates: “Round after round, backroom deals and horse-trading will be taking place between rounds.” Hill favors instead using ranked-choice ballots that will enable the elimination of candidates to occur through an “instant runoff” process, without the need for separate rounds of voting–and thus without the possibility of backroom deals and horse-trading between rounds.

The way an instant runoff works is to eliminate the candidate with the fewest first-choice votes on the ranked-choice ballots, and then redistribute the ballots that ranked the eliminated candidate first to whomever is ranked next on those ballots. This elimination-and-redistribution process is repeated until there is a single winner, and it all happens instantly with the use of modern computers. Thus, instant runoff runoff combines Maine’s computational rule with an automated implementation based on a ranked-choice ballot.

But it is also possible to use ranked-choice ballots to automatically implement Survivor‘s computational rule. In fact, it is known in the literature of electoral systems as the Coombs method. Rather than eliminating the candidate with the fewest first-choice votes (as an instant runoff does), the Coombs method eliminates the candidate with the most last-place votes. The Coombs method then recalculates which of the remaining candidates has the most last-place votes, eliminating that candidate next, and repeats this elimination-and-recalculation procedure until there is a single winner of the election.

You can think of the Coombs method as “instant Survivor voting”–which is why Steven Hill shouldn’t badmouth Survivor‘s voting process when what concerns him is the lack of ranked-choice ballots to avoid negotiations between elimination rounds.

Indeed, Survivor‘s computational rule–which is what the Coombs method uses–has advantages over the instant runoff’s computational rule, as I explore in Chapter Two of this book and discuss more briefly in a previous Common Ground Democracy essay. Eliminating the most disfavored candidate first, as Survivor‘s computational rule does, tends to combat polarization most effectively and elect candidates with the broadest range of appeal or acceptability. In this respect, the Coombs method is similar to Condorcet-based voting methods, which elect a candidate who is preferred by a majority of voters compared to each alternative one-on-one. Indeed, to make a technical point, as long as electoral preferences are “single-peaked” (meaning that on a left-to-right ideological, or blue-to-red partisan, spectrum, voters prefer candidates in order of proximity to their most-preferred candidate), then the Coombs method is Condorcet-consistent (meaning it will elect a Condorcet winner).

Thus, as we work to repair American democracy from the ravages of polarization, it is important not to rule out Survivor-style voting with the use of ranked-choice ballots. Familiarity with the Survivor TV show may make it a more comprehensible and acceptable electoral reform than other potential anti-polarization alternatives. The idea of eliminating weaker candidates until one winner remains is not a bad one and is quite intuitive.

So, don’t confuse the issue as Steven Hill has: if you want an electoral procedure that uses ranked-choice ballots, don’t disparage Survivor voting; instead, consider tabulating your ranked-choice ballots according to the Survivor computational rule.

This post initially appeared on the Election Law Blog.