Computers, PR, and Consensus Bias

Unitary Cardinal Voting is my favorite PR method at the moment, but it requires computers to calculate. At a local level, I don’t see this as particularly feasible or likely to be done, and statewide efforts will require a local template somewhere. In this example (!topic/electionscience/Rk4ZGf-s-s8) Unitary Cardinal Voting would pick two centrists and one extremist, which I consider a near-ideal result. The only PR method I can think of that (possibly) wouldn’t require computers is Apportioned Cardinal Voting, but that picks 3 extremists in the example, so I consider it to not be worthy. Any suggestions?

(When I say not worthy, I mean that it seems simpler to just go with a consensus single-winner method than to push for a method which may or may not split the voters and is harder to explain.)


I have some issues with the assumptions which go into the diagram as well at the assigned scoring behaviour. Even if true, it is likely that there is also candidates more extreme than the center of any group.

I am not sure what my system would do. If there are people in the center then hopefully it would choose the second ring.

As for manual vs computer voting. The gold standard is like to compare to STV. What does Australia do?

I think this argument applies to every PR method, except Asset. The math behind any reweighting step necessary for maintaining proportionality will be much more complicated than “add the points.” These Score-PR methods will most likely be considered where people have already established PR as a goal, in which case they knew what they were getting into.

It may be desirable to at least pass one instance of Score PR, so that we can go to FairVote and disprove their claims that RCV is strategically necessary to get to PR. In that case, I favor the Score PR method that is consensus-biased at least within the majority.

1 Like

The argument that RCV is strategically necessary to get to PR will only appeal to people who have established PR as a goal. It’s also not true: choose one party list PR methods such as D’Hondt and Sainte-Laguë are at least as common as multiwinner STV.

Although, choose-one party list would encourage voters to avoid choosing smaller, potentially spoiler-style parties. I don’t know how big a factor that is in pushing people towards STV and thus RCV.

1 Like