The number of candidates if Final Five Voting were applied to STV

Voters’ ability/willingness to rank candidates is an issue in ranked-ballot systems. Final Five (Four) Voting (FFV) aims to solve this problem with a plurality winnowing round. Voters are asked to consider only five (four) candidates at the ranking stage. What does this look like in STV?

The minimum number of decisive-round candidates is district magnitude. It cannot be less than 5 in a 5-seat district. So, for two slates of 3 that run in a competitive 5-seat district, voters are asked to consider six options. (I chose that scenario because it feels like an equilibrium.)

This number grows or shrinks in response to the number of slates, the sizes of slates, the number of independents, and the number coming out of an FFV winnowing round.

Someone should make a graph.

Tips for getting started on papers

1. Start a blank Note (iPhone), Google Doc, or whatever. Call it whatever you plan to title your paper — for now. Put ideas here as they come to you.

2. Put the readings/lectures aside, think for a few minutes, and say your honest answer to the prompt. Write it in the blank document.

3. The blank document should sync across your devices. That means you can add to it whenever. You also can delete, re-word, and move text around.

4. When you are ready to write your paper, you can sit down at a computer and work from the document. You might even write the paper on the phone, but that makes citing sources more tedious.

Relative performance of STV and OLPR in providing descriptive representation

I was honored to be asked to share thoughts last week on institutional change and its relationship to the state of politics. I tried to be as brief as possible.

One of my points was that electoral reform largely involves making proposals that politicians may or may not run with later. Diffusion matters, and we may not be able to control that process.

Another was to ask the question in the title of this post. I therefore have been replicating and extending a 1939 study of high-profile U.S. STV elections. Part of the extension is to analyze a larger set of elections. The other is to focus on various dimensions of descriptive representation.

Some early results can be found here. It unfortunately is not possible to combine them with a modern-day analysis of racially polarized voting. (Such an analysis would be needed to establish which candidates were candidates of choice.) However, the secondary literature on the candidates is unusually good.

The simulations are done with a D’Hondt allocation. I also have done them with largest-remainder Droop, which lets us better understand why STV produced the results it did. All of this is still in progress, and I am happy to discuss what I am learning.

I have been thinking a lot about two papers as I work on the project.