[EM] Better Choices for Democracy

[Kristofer Munsterhjelm]

On 2025-06-19 23:12, Ralph Suter via Election-Methods wrote:
> You've oversimplified what they advocate. Their website says:
> 
> 
> "In almost all large-scale elections, the process of comparing pairs of 
> candidates will identify the Consensus Choice, a single candidate who 
> wins all their head-to-head matchups. In the unlikely event that no 
> Consensus Choice exists, the ultimate winner can be determined by one of 
> the following resolution methods:
> 
>      "Margin of Loss Resolution: If there is no Consensus Choice, the 
> candidate whose largest head-to-head loss is smallest is declared the 
> winner.
> 
>      "Number of Wins & Margin of Loss Resolution: The candidate with the 
> most head-to-head wins is declared the winner. In the event that 
> multiple candidates tie for most head-to-head wins, the tie is broken in 
> favor of the one whose largest head-to-head loss is smallest.
> 
>      "Instant Runoff Resolution: If there is no Consensus Choice, 
> Instant Runoff Voting is used to determine the winner."
> 
> My biggest question is why they included instant runoff as one of the 
> resolution methods, especially because on their FAQ page, they explain 
> why it isn't a good method:
> [snip]

The best explanation I can figure out is that they're going for absolute 
simplicity. Minmax and Copeland have a reasonable claim to being as 
simple as possible for (a pairwise matrix-based) Condorcet method; and 
then Condorcet//IRV would be there as the simplest augmentation to IRV 
that makes it Condorcet, for those jurisdictions that already use or are 
familiar with IRV.

I agree that it's not ideal, though. As far as IRV hybrids go, I would 
imagine Benham to have a stronger claim on retaining the desirable 
aspects of Condorcet even in troublesome (cyclical) elections, since it 
passes Smith; while not being all that much harder.

Something like:

"Instant Runoff Resolution: If there is no Consensus Choice among 
candidates not eliminated, perform one step of Instant Runoff: eliminate 
the first preference loser, and try again."

-km