Approval voting is a poor choice for most political elections (its good for non-secret-ballot group decision-making where the voters are people who can opt-out of participation after the election and an approving vote is a binding commitment by the voter not to do so if any of their "approved" alternatives win.)
Now, if you said the general elections held in the spring (they aren't "primaries" in the usual sense and shouldn't be called that) should use ranked ballots voting and select the Condorcet winner with tie-breaker elections held in the fall, where necessary, as FPTP elections between the members of the Smith set of the spring elections, I'd agree that might be a worthwhile improvement.
I don't agree that approval voting is a poor choice for political elections. I particularly reject any implication (if present) that plurality should be preferred to it.
I recognize that ranked methods have some substantial benefits over approval voting, but there are drawbacks. Considering just the theoretical benefits and drawbacks in this context, I would agree with a preference for ranked methods. However, when we consider the complexity (and resulting decrease in transparency) I've come to prefer approval. Even so, I wouldn't strongly object to a ranked method.
> I don't agree that approval voting is a poor choice for political elections.
It is because "approval", unlike relative preference, doesn't have a consistent meaning across different ballots (which is why it isn't a poor choice in open-ballots where it has a defined meaning, like a commitment to opt-in to the result where otherwise there is an option to opt-out.)
> I particularly reject any implication (if present) that plurality should be preferred to it.
Plurality is also a bad choice for most political elections. Or, for that matter, most elections of any kind. But if you are changing from plurality, approval is about the worst change you could make other than ones that could only be deliberately malicious (e.g., modifying plurality to elect the candidate with the lowest number of first place preference votes.)
> Considering just the theoretical benefits and drawbacks in this context, I would agree with a preference for ranked methods. However, when we consider the complexity (and resulting decrease in transparency) I've come to prefer approval.
I personally have no problem with democracy requiring expecting citizens to deal with concepts more complex than approval voting, since, actually, most of the things they are voting on are things more complex than approval voting.
'It is because "approval", unlike relative preference, doesn't have a consistent meaning across different ballots (which is why it isn't a poor choice in open-ballots where it has a defined meaning, like a commitment to opt-in to the result where otherwise there is an option to opt-out.)'
There is a technical sense in which that's correct, but you're reading more into it than is there. Both approval and ranked choice methods have consistent meaning across ballots when the meaning is taken to be the impact on the election. Neither method expresses how much one candidate is preferred to another.
"Plurality is also a bad choice for most political elections. Or, for that matter, most elections of any kind. But if you are changing from plurality, approval is about the worst change you could make[.]"
I'm not sure I strongly disagree with this, when considering only the act of voting and the theoretical results. I do see the gap between plurality and approval, here, as far larger than the gap between approval and ranked choice (plurality is both a scoring method and a ranked method and so has all the theoretical problems of both). And I do think that other practical concerns are tremendously significant and favor approval (as I mentioned in the previous comment and elaborate on below) over ranked methods.
"I personally have no problem with democracy requiring expecting citizens to deal with concepts more complex than approval voting, since, actually, most of the things they are voting on are things more complex than approval voting."
First and most trivially: otherwise accepting your argument here, your comparison shouldn't be to the complexity of approval voting, but to the complexity of the proposed alternative.
Second (most significantly), I don't mean, mostly, "O NOES VOTERS WILL GET CONFUSED". I mean that counting ranked choice ballots is tremendously ugly compared to counting approval ballots. You have 10 candidates? You're tallying instances of 3.6 million possible ballots! The number that will actually be represented will surely be a fraction of that, but even so this is going to involve more (and uglier) mistakes and more opportunities for fraud (and appearance of fraud) because the system is so much more complicated. Further, running statistical analysis of IRV elections is, in practice, significantly harder than doing the same of other methods (say my connections in opinion research) though this could arguably be a feature.
Third, just because we're hoping they're able to deal with more complex issues doesn't mean adding complexity is a good thing. Complexity typically stacks. If there is a maximum complexity people can deal with, adding more in one place is going to reduce the amount they deal effectively with elsewhere.
