Alaska’s ranked-choice voting special election on Aug. 16th to fill its vacant, single seat in the US House is generating two issues I want to flag.
First, in recently campaigning in a tele-rally for Sarah Palin, Donald Trump reportedly told Palin voters “to pick only one” candidate. Palin is competing against a Democrat, Mary Petola, and another Republican backed by the state GOP, Nicholas Begich III. For Palin voters who care whether a Democrat or Republican represents Alaska in the House if Palin does not win, Trump’s advice is obviously bad. Palin supporters would be giving up their power to influence the outcome if Palin does not prevail – and Palin is no better off, of course, if voters vote only for her and throw away their power to rank their second choice.
That Trump is giving Palin voters this advice is important to recognize when we evaluate RCV in this election: voters take their cues from political leaders, especially with a new voting system. If a lot of Palin voters do not provide a second preference, it could well be because this is the message they got from Trump (I do not know if Palin is campaigning on a similar message). It would not necessarily mean voters lack the ability to rank more than one candidate. Particularly with a new voting system, if voters get bad advice from political leaders, they will not use the system as effectively as they can.
Indeed, the failure of Palin voters to rank a second choice, if they prefer a Republican to a Democrat, could mean that the Democrat will win this statewide race. In essence, by ranking only one candidate, the RCV system gets converted back into a plurality-system, in which a candidate lacking majority support can still win.
Second, and of more significance: this race might turn out to demonstrate the large stakes in the choice between two different versions of RCV. One argument for RCV is that it can help identify the true majority winner in a three or more candidate field. But whether RCV does that can depend on the rule used to eliminate the first candidate. In this race, polls suggest Begich would win if pitted head to head with the Democrat. Let’s assume Begich would also beat Palin one on one. That would make him the true majority winner. But if Begich comes in third on the first-round votes, he would be eliminated and never go head to head with either candidate under the current system of RCV used in Alaska and the rest of the US.
The alternative is the Bottom-Two system of RCV. Ned Foley has done tremendous work explaining this and other alternative forms of RCV. Instead of automatically eliminating the candidate who comes in last in the first round, this version compares the votes on all the ballots among the bottom two candidates. As between these two, the candidate who is ranked higher on all the ballots moves on to the next round. The point of this system is it never eliminates a candidate who would win a majority when pitted head to head against any other candidate.
To be concrete, let’s assume in the first round Pelota wins 40% of the vote, Palin 32%, and Begich 28%. Let’s also assume that all of Pelota’s voters rank Begich second. Because Begich, paired head to head with Palin, would then defeat her, he would move on to the final round.
Thus, in Tuesday’s election, a great deal will turn on whether Begich or Palin comes in last in first-place votes (the polling assumption is that Pelota will win a plurality of the first-place votes). If Begich comes in last, he would be eliminated – even though it is possible he would defeat both Palin and Pelota head to head. The true majority winner will then have been eliminated in the first round. Under Bottom-Two, with the assumptions above, if Palin comes in last, Begich would win the election (assuming enough Palin voters rank him second).
There are practical issues to consider with any voting system. But the theoretical case for Bottom-Two is that it does a better job than current versions of RCV in identifying the true majority winner, when one exists in a field with three or more candidates.