And away we go…!
Baby steps but at least the experiments are continuing and NYC is about as big a laboratory as you can hope for. I think a St. Louis - style approval voting would have been better here but what’s most important is that voters are getting to see and understand how there are better systems than first past the post.
Virginia Republicans recently used RCV to select their candidates for the three top state offices and it was a pretty interesting process.
Innnnnteresting. Never heard of that before: In St. Louis, Voters Will Get To Vote For As Many Candidates As They Want | FiveThirtyEight
There are certainly methods where the voter is allowed to leave candidates unranked, as well as assign the same rank to more than one candidate.
There are many potential systems, but how are politicians going to compromise on the “best” one?
IIRC, the St. Louis election was non-partisan and had been plagued (as many cities are) by winning candidates getting only 30% of the votes in the past using FPTP. St. Louis used approval voting to get down to two finalists, at which point they held a head-to-head runoff. Ballots in the initial round averaged about 1.5 votes, or roughly every other voter voted for two candidates. It would be interesting to see if that average changes in one direction or the other as voters become more accustomed to the process.
If we had had approval voting in your (general your) state’s presidential primary last year, how many candidates would (general) you have voted for?
Let’s not freeze progress with indecision. First, let’s lose FPTP where applicable because it’s almost always the worst.
Hmm. I’m philosophically inclined to “bullet vote” for a single candidate where I strongly support him or her, but in early 2020 I might’ve cast a Dem primary ballot for Cory Booker, Elizabeth Warren and Mayor Pete.
My ballot (Maryland) would have looked pretty similar to this.
I’m in Ohio. Proud to be a Buckeye Democrat.
The problem with ranked choice voting is that it is even more able to be manipulated than FPTP. For example, suppose this is the next ballot for my state’s governor:
- Joe the Regular Repubican
- Bill the Regular Democrat
- Sue the Hard Right Republican
- Sally the Hard Left Democrat
- Dave the Hitler Nazi
- Steve the Stalin Communist
I can read that ballot and know that numbers 3 through 6 are going to be fighting for scraps. The real fight is between the top two. I’m a conservative Republican, so my real preference is this:
- Sue
- Joe
- Bill
- Sally
- Dave and Steve tied for last.
But here is how I maximize my chances for the only realistic candidate that I prefer to win. I vote:
- Joe
- Steve (or Dave)
- Dave (or Steve)
- Sally
- Sue
- Bill
I give my realistic candidate the most possible points and relegate the other realistic candidate to the least. Sure, I gave the communist and the Nazi the 2 and 3 spot, but they won’t win anyways. Better to push the realistic candidates down to the bottom. I put my real favorite down to 5 because I don’t want her gumming up the works for the realistic one that I want to win and I want her to get even fewer votes than the Nazi or communist because others might vote for her and make her a spoiler. I also vote for Sally higher than Sue to hopefully screw up Bill’s votes and help her overcome him.
I’ve never seen a good response to my objection about ranked choice voting.
What if a lot of other people are thinking in a similar fashion?
Indeed. Could cause an issue with ranked choice voting.
It’s also a really good reason for you to not actually vote that way.
Is everyone going to get the message that your or I got not to vote that way?
I’ve no idea.
But you said you would vote that way because you were certain neither the Nazi nor the Stalinist would win. I pointed out one reason you couldn’t be sure of that; and you agreed. So the logic under which you were going to vote the way you described fails; because one of the premises is wrong.
What kind of ranked choice system are you talking about? Because what you are describing doesn’t fit any description I’ve seen of ranked choice voting. In every description I’ve seen ( both in NYC* and elsewhere) , you rank the candidates. But the only time your 2,3,4,5 choice counts is when your higher choices have been eliminated. So if your first choice is Joe, and he and Bill ( the two realistic candidates) are one and two after the first round. If one gets a majority of votes on the first ballot, he wins and that’s the end of it. If neither has a majority, the candidate with the fewest votes is eliminated. Say Dave has fewest votes - he gets eliminated and the #2 choice of the people who ranked him #1 get counted. In this round, Steve has the fewest votes. The people who ranked Steve #1 get their 2nd choice counted and those who had Dave #1 and Steve #2 have their 3rd choice counted. The candidate with the fewest votes that round gets eliminated and so on until there are only 2 left, at which point either one has a majority or they are exactly tied. ( some systems stop as soon as one candidate has a majority) Only one vote counts for each voter , so if your first choice is one of the last two still in the race, your other choices don’t matter.
What you seem to be talking about is some sort of points system , where all your votes count and are given differing numbers of points. I searched around and found that it seems to be called positional voting. As far as I can tell, this system is not used in the US for political elections
* In NYC it’s only being used for primaries, not the general election so you wouldn’t be voting for Democrats, Republicans and Stalinists on the same ballot. You’d be voting for different flavors of Democrats to see who gets that line in the general election and there are only two running in the Republican primary, so that won’t really be ranked choice.
Your second and third choices are most likely going to be eliminated after the first round of counting, assuming there even will be a second round. I don’t think you’re grasping the concept and the math of RCV very well at all.
That technically does maximize the chances for your preferred candidate out of the two “realistic” ones. But that’s not what you want. The way you’re maximizing the chance for Joe is by taking away some of the (rare) cases where Sue wins, and giving them to Joe. You’re not decreasing the chance of a Democrat winning, at all.
Assuming that this is IRV (which ranked-choice voting usually is), you should be giving Sue your first vote, followed by Joe. This way, one of two things happens: Either Sue gets eliminated, in which case you get the same outcome as if Joe got your first vote, which is good for you. Or Sue doesn’t get eliminated, and actually manages to win, which is even better for you. This is the big advantage of IRV (or in fact, most other voting systems than the one we use), that it allows you to (usually) vote the way you actually feel, without it (usually) hurting the chances of getting the outcome you want (unfortunately, it’s not perfect, but then it’s been mathematically proven that no system is. But some are a lot better than others).
One oft-neglected requirement of a voting system is that it has to be trusted by the people using it, and that means that people need to understand it. There are some ranked-choice systems that are, mathematically, strictly better than IRV, but in the real world, they wouldn’t work, because they’re complicated enough that most people wouldn’t understand them.
This real-world requirement, I think, narrows down the list of possibilities to first-past-the-post, IRV, or approval voting. Of those three, approval is probably a bit better than IRV in most cases. But approval and IRV are both far superior to first-past-the-post.
Thanks to this and others. It seems I misunderstood how RCV works. However, wouldn’t another issue become a problem?
A Republican candidate would encourage a large number of right learning parties to run, tapping into each individual constituency knowing that they would win the second (or third or the last meaningful vote)? IOW, recruit an anti-abortion party candidate, a pro-gun candidate, a tax cut candidate, etc.
Then the Dems recruit a Green Party candidate, a BLM candidate, etc.
The winner would be whoever could recruit enough candidates to run as sacrificial lambs. It would just be a numbers game, not in votes, but in candidates. Instead of attempting to cast a broad range of support, the major party candidates could use their proxies to say the most outrageous things, distance themselves from those statements, even condemn them, but still get the #2 slot on that ballot (or #11 or whatever matters).
It could get absurd. Say the Republicans have an anti-immigration candidate who advocates shoot to kill orders at the border like the Soviets did at the Berlin Wall. That’s going to fire up some radicals who wouldn’t otherwise vote but show up for that one issue, but those radicals will pick the Republican above the Democrat wherever that choice is made down ballot. It would really revolutionize politics.
It would seem to radicalize politics far more than it is already.
Stated more succinctly. Say in 2024, we have the following ballot:
- Ted Cruz
- Kamala Harris
- Patrick Buchanan clone
- Ralph Nader clone
If we assume like most recent elections that the preference between Cruz and Harris will be razor tight, then the election will come down to the performance of Buchanan clone v. Nader clone. The reasons why one might prefer Buchanan v. Nader will be different than the “real” election going on at the top and should not have such an outsized influence.
Minor hijack, but I’m interested in learning what systems you consider to be superior in a mathematical sense. You’re absolutely right that they wouldn’t be viable in the real world because the majority of plebs are just too dumb, but I might find it useful to apply these systems to more technical niches, such as getting coworkers in my software company to agree on a particular architecture before starting a project, or electing leaders for my social clubs, etc.
Recently I skimmed over the Wikipedia article on quadratic voting, and it looked promising and doesn’t appear to be that difficult to grasp even for mathematically-challenged folks, is that one of the systems you had in mind?