I think that an election method with a real runoff is better than one without. Here’s how I would design it. Collect score ballots from the voters for the first election. Use a range of 0 through 100 by 1. If a majority of the voters give a particular candidate a 100, elect that candidate. Otherwise, schedule a runoff. The candidate with the highest total score from the first election goes into the runoff. Analyze the ballots to determine if there is a pairwise (i. e., “Condorcet”) winner. If there is, and the Condorcet winner is not the same candidate as the one who came out highest in the score election, include the Condorcet winner in the runoff, and that is your runoff, with just those two candidates. If there is no Condorcet winner, or if the Condorcet winner is also the score winner, conduct a round of reweighted range tallying to determine who the second winner would be by that method. Put that person in the runoff. If you want majority rule, that is the design I have to offer. It’s the closest thing possible, I contend, to assuring that the final winner has in some sense a “majority” vote. Real runoffs do not necessarily go the same way as “instant runoffs”. People select themselves for degree of interest in the outcome. Also, when they vote, they are informed by the experience of the prior vote, and by any campaigning and coverage that has happened in the interim.
Much of this opinion is not original to me.